article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
unification @xcite is an operation to make two logical expressions equal by finding substitutions into variables .
there are numerous applications in computer science , in particular of ( efficient ) first - order unification , for example in automated reasoning , type checking and verification .
unification algorithms are also extended to higher - order calculi with various equivalence relations .
if equality includes @xmath0-conversion and @xmath1-reduction and perhaps also @xmath2-conversion of a ( typed or untyped ) lambda - calculus , then unification procedures are known ( see e.g. @xcite ) , however , the problem is undecidable @xcite .
our motivation comes from syntactical reasoning on higher - order expressions , with equality being alpha - equivalence of expressions , and where a unification algorithm is demanded as a basic service .
nominal unification is the extension of first - order unification with abstractions .
it unifies expressions w.r.t .
alpha - equivalence , and employs permutations as a clean treatment of renamings .
it is known that nominal unification is decidable in exponential time @xcite , where the complexity of the decision problem is polynomial time @xcite .
it can be seen also from a higher - order perspective @xcite , as equivalent to miller s higher - order pattern unification @xcite .
there are efficient algorithms @xcite , formalizations of nominal unification @xcite , formalizations with extensions to commutation properties within expressions @xcite , and generalizations of nominal unification to narrowing @xcite , and to equivariant ( nominal ) unification @xcite .
we are interested in unification w.r.t .
an additional extension with cyclic let . to the best of our knowledge
, there is no nominal unification algorithm for higher - order expressions permitting also general binding structures like a cyclic let .
the motivation and intended application scenario is as follows : constructing syntactic reasoning algorithms for showing properties of program transformations on higher - order expressions in call - by - need functional languages ( see for example @xcite ) that have a letrec - construct ( also called cyclic let ) @xcite as in haskell @xcite , ( see e.g. @xcite for a discussion on reasoning with more general name binders , and @xcite for a formalization of general binders in isabelle ) .
there may be applications also to coinductive extensions of logic programming @xcite and strict functional languages @xcite .
basically , overlaps of expressions have to be computed ( a variant of critical pairs ) and reduction steps ( under some strategy ) have to be performed . to this end ,
first an expressive higher - order language is required to represent the meta - notation of expressions .
for example , the meta - notation @xmath3 for a beta - reduction is made operational by using unification variables @xmath4 for @xmath5 .
the scoping of @xmath6 and @xmath7 is different , which can be dealt with by nominal techniques .
in fact , a more powerful unification algorithm is required for meta - terms employing recursive letrec - environments .
our main algorithm is derived from first - order unification and nominal unification : from first - order unification we borrowed the decomposition rules , and the sharing method from martelli - montanari - style unification algorithms @xcite .
the adaptations of decomposition for abstractions and the advantageous use of permutations of atoms is derived from nominal unification algorithms .
decomposing letrec - expression requires an extension by a permutation of the bindings in the environment , where , however , one has to take care of scoping . since in contrast to the basic nominal unification , there are nontrivial fixpoints of permutations ( see example [ example : fixpoints - possible ] ) , novel techniques are required and lead to a surprisingly moderate complexity : a fixed - point shifting rule ( fps ) and a redundancy removing rule ( elimfp ) together bound the number of fixpoint equations @xmath8 ( where @xmath9 is a permutation ) using techniques and results from computations in permutation groups .
the application of these techniques is indispensable ( see example [ example : exponential - fps ] ) for obtaining efficiency .
_ results _ : a nominal letrec unification algorithm which is complete and runs in nondeterministic polynomial time ( theorem [ thm : unification - terminates ] ) .
the nominal letrec unification problem is np - complete ( theorem [ thm : matching - np - hard ] ) .
nominal letrec matching is np - complete ( theorem [ thm : matching - in - np],[thm : matching - np - hard ] ) .
nominal letrec matching for dags is in np and outputs substitutions only ( theorem [ thm : nom - dag - matching ] ) , and a very restricted nominal letrec matching problem is graph - isomorphism hard ( theorem [ thm : matching - gi - hard ] ) .
we define the language @xmath10 ( * l*et*r*ec * l*anguage ) of expressions , which is a lambda calculus extended with a recursive let construct . the notation is consistent with @xcite .
the ( infinite ) set of atoms @xmath11 is a set of ( constant ) symbols @xmath12 denoted also with indices ( the variables in lambda - calculus ) .
there is a set @xmath13 of function symbols with arity @xmath14 .
the syntax of the expressions @xmath15 of @xmath10 is : + @xmath16 we also use tuples , which are written as @xmath17 , and which are treated as functional expressions in the language .
we assume that binding atoms @xmath18 in a letrec - expression @xmath19 are pairwise distinct .
sequences of bindings @xmath20 are abbreviated as @xmath21 .
the _ scope _ of atom @xmath22 in @xmath23 is standard : @xmath22 has scope @xmath15 .
the @xmath24-construct has a special scoping rule : in @xmath25 , every free atom @xmath26 in some @xmath27 or @xmath28 is bound by the environment @xmath29 .
this defines the notion of free atoms @xmath30 , bound atoms @xmath31 in expression @xmath15 , and all atoms @xmath32 in @xmath15 .
for an environment @xmath33 , we define the set of letrec - atoms as @xmath34 .
@xmath22 is fresh for @xmath15 _ iff @xmath35 ( also denoted as @xmath36 ) . as an example
, the expression @xmath37 represents an infinite list @xmath38 , where @xmath39 are expressions .
however , since our language @xmath10 is only a fragment of core calculi @xcite , the reader may find more programming examples there .
we will use mappings on atoms from @xmath11 .
a _ swapping _
@xmath40 is a function that maps an atom @xmath22 to atom @xmath41 , atom @xmath41 to @xmath22 , and is the identity on other atoms .
we will also use finite permutations on atoms from @xmath11 , which are represented as a composition of swappings in the algorithms below .
let @xmath42 .
then every finite permutation can be represented by a composition of at most @xmath43 swappings .
composition @xmath44 and inverses @xmath45 can be immediately computed .
permutations @xmath9 operate on expressions simply by recursing on the structure . for a letrec - expression
this is @xmath46 @xmath47 @xmath48 .
note that permutations also change names of bound atoms .
we will use the following definition of @xmath0-equivalence : [ def : alpha - equivalence - sim ] the equivalence @xmath49 on expressions @xmath50 is defined as follows : * @xmath51 . * if @xmath52 for all @xmath53 , then @xmath54 for an @xmath55-ary @xmath56 . * if @xmath57 , then @xmath58 . * if for @xmath59 , @xmath60 , @xmath61 , then @xmath62 . *
@xmath63 iff there is some permutation @xmath64 on @xmath65 , such that @xmath66 .
note that @xmath49 is identical to the equivalence relation generated by @xmath0-equivalence of binding constructs and permutation of bindings in a letrec .
we need fixpoint sets of permutations @xmath9 : we define @xmath67 . in usual nominal unification
, these sets can be characterized by using freshness constraints @xcite . clearly , all these sets and also all finite intersections are nonempty , since at least fresh atoms are elements and since @xmath11 is infinite
. however , in our setting , these sets are nontrivial : [ example : fixpoints - possible ] the @xmath0-equivalence @xmath68 @xmath49 @xmath69 holds , which means that there are expressions @xmath70 in @xmath10 with @xmath71 and @xmath72 . in the following
we will use the results on complexity of operations in permutation groups , see @xcite , and @xcite .
we consider a set @xmath73 of distinct objects ( in our case the atoms ) , the symmetric group @xmath74 ( of size @xmath75 ) of permutations of the objects , and consider its elements , subsets and subgroups .
subgroups are always represented by a set of generators . if @xmath76 is a set of elements ( or generators ) , then @xmath77 denotes the generated subgroup .
some facts are : * permutations can be represented in space linear in @xmath55 . * every subgroup of @xmath74 can be represented by @xmath78 generators .
however , elements in a subgroup may not be representable as a product of polynomially many generators .
the following questions can be answered in polynomial time : * the element - question : @xmath79 ?
, * the subgroup question : @xmath80 . however , intersection of groups and set - stabilizer ( i.e. @xmath81 ) are not known to be computable in polynomial time , since those problems are as hard as graph - isomorphism ( see @xcite ) .
as an extension of @xmath10 , there is also a countably infinite set of ( unification ) variables @xmath82 also denoted perhaps using indices .
the syntax of the language @xmath83 ( * l*et*r*ec * l*anguage e*x*tended ) is @xmath84 @xmath85 is the set of variables and @xmath86 is the set of variables @xmath87 occurring in @xmath15 .
the expression @xmath88 for a non - variable @xmath15 means an operation , which is performed by shifting @xmath9 down , using the simplification @xmath89 @xmath90 @xmath91 , apply it to atoms , where only expressions @xmath92 remain , which are called _
suspensions_. a _ freshness constraint _ in our unification algorithm is of the form @xmath93 , where @xmath15 is an @xmath83-expression , and an _ atomic _ freshness constraint is of the form @xmath94 .
[ def : simplification - freshness ] @xmath95 @xmath96 @xmath97 @xmath98 + @xmath99 { \{a{{{\#}}}({{\tt letrec}}~a_1.s_1;\ldots , a_n.s_n~{{\tt in}}~r)\ } { \ensuremath{\mathaccent\cdot\cup}}\nabla}{\nabla}$ ] + @xmath100 { \{a{{{\#}}}({{\tt letrec}}~a_1.s_1;\ldots , a_n.s_n~{{\tt in}}~r)\ } { \ensuremath{\mathaccent\cdot\cup}}\nabla } { \{a{{{\#}}}s_1,\ldots a{{{\#}}}s_n , a{{{\#}}}r\ } { \ensuremath{\mathaccent\cdot\cup}}\nabla}$ ] @xmath101 an @xmath83-unification problem is a pair @xmath102 , where @xmath103 is a set of equations @xmath104 , and @xmath105 is a set of freshness constraints .
a _ ( ground ) solution _ of @xmath102 is a substitution @xmath64 ( mapping variables in @xmath106 to ground expressions ) , such that @xmath107 for @xmath108 and for all @xmath109 : @xmath110 holds .
+ the decision problem is whether there is a solution for given @xmath102 .
let @xmath102 be an @xmath83-unification problem .
we consider triples @xmath111 , where @xmath112 is a substitution ( compressed as a dag ) mapping variables to @xmath83-expressions , @xmath113 is a set of freshness constraints , and @xmath114 is a set of fixpoint constraints of the form @xmath115 , where @xmath116 .
a triple @xmath111 is a _ unifier _ of @xmath102 , if ( i ) there exists a ground substitution @xmath64 that solves @xmath117 , i.e. , for every @xmath118 in @xmath113 , @xmath119 is valid , and for every @xmath115 in @xmath114 , @xmath120 ; and ( ii ) for every ground substitution @xmath64 that instantiates all variables in @xmath121 which solves @xmath117 , the ground substitution @xmath122 is a solution of @xmath102 .
a set @xmath123 of unifiers is _ complete _ , if every solution @xmath124 is covered by at least one unifier , i.e. there is some unifier @xmath111 in @xmath123 , and a ground substitution @xmath64 , such that @xmath125 for all @xmath126
. we will employ nondeterministic rule - based algorithms computing unifiers : there is a clearly indicated subset of disjunctive ( do nt know non - deterministic ) rules .
the _ collecting variant _ of the algorithm runs and collects all solutions from all alternatives of the disjunctive rules .
the _ decision variant _ guesses one possibility and tries to compute a single unifier . since we want to avoid the exponential size explosion of the robinson - style unification algorithms , keeping the good properties of martelli montanari - style unification algorithms @xcite , but not their notational overhead , we stick to a set of equations as data structure . as a preparation for the algorithm ,
all expressions in equations are exhaustively flattened as follows : @xmath127 plus the equations @xmath128 .
also @xmath129 is replaced by @xmath130 with equation @xmath131 , and @xmath132 is replaced by @xmath133 with the additional equations @xmath134 .
the introduced variables are always fresh ones .
we may denote the resulting set of equations of flattening an equation @xmath135 as @xmath136 .
thus , all expressions in equations are of depth at most 1 , where we do not count the permutation applications in the suspensions . a dependency ordering on @xmath137 is required : if @xmath138 is in @xmath103 , and @xmath15 is not a variable nor a suspension and @xmath139 , then @xmath140 , let @xmath141 be the transitive closure of @xmath142 .
this ordering is only used , if no standard rules and no failure rules ( see definition [ def : failure - rules ] ) apply , hence there are no cycles
. operates on a tuple @xmath143 , where @xmath103 is a set of flattened equations @xmath144 , where we assume that @xmath145 is symmetric , @xmath105 contains freshness constraints , @xmath146 represents the already computed substitution as a list of replacements of the form @xmath147 .
initially @xmath146 is empty .
the final state will be reached , i.e. the output , when @xmath103 only contains fixpoint equations of the form @xmath8 that are non - redundant , and the rule ( output ) fires . in the notation of the rules
, we use @xmath148 $ ] as substitution that replaces @xmath87 by @xmath15 . in the rules
, we may omit @xmath105 or @xmath146 if they are not changed .
we will use a notation `` @xmath149 '' in the consequence part of one rule , perhaps with a set of possibilities , to denote disjunctive ( i.e. do nt know ) nondeterminism . the only nondeterministic rule that requires exploring all alternatives is rule ( 7 ) below .
the other rules can be applied in any order , where it is not necessary to explore alternatives . [
[ standard-1233-and - decomposition - rules-4567 ] ] standard ( 1,2,3,3 ) and decomposition rules ( 4,5,6,7 ) : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + @xmath150 @xmath151 + ( 3)@xmath152 , \nabla[\pi{\cdot}y / x ] , \theta \cup \{x \mapsto \pi{\cdot}y\}}}}$ ] ( 3)@xmath153 , \nabla[y / x ] , \theta \cup \{x \mapsto y\}}}}$ ] + @xmath154 + @xmath155 @xmath156 + @xmath157 + where @xmath64 is a permutation on @xmath65
. + [ [ main - rules ] ] main rules : + + + + + + + + + + + the following rules ( mms ) ( martelli - montanari - simulation ) and ( fps ) ( fixpoint - shift ) will always be immediately followed by a decomposition of the resulting set of equations .
@xmath158 , if @xmath5 are neither variables + nor suspensions .
+ @xmath159 , if @xmath87 is maximal w.r.t .
@xmath160 , @xmath161 , and @xmath15 is neither a variable nor a suspension , and no failure rule ( see below ) is applicable .
+ @xmath162 if @xmath163 + @xmath164 if @xmath103 only consists + of fixpoint - equations .
we assume that the rule will be applied whenever possible .
note that the two rules ( mms ) and ( fps ) , without further precaution , may cause an exponential blow - up in the number of fixpoint - equations .
the rule ( elimfp ) will limit the number of fixpoint equations by exploiting knowledge on operations on permutation groups .
the rule ( output ) terminates an execution on @xmath165 by outputting a unifier @xmath166 .
note that in any case at least one solution is represented : the top symbol of an expression is defined as @xmath167 , @xmath168 , @xmath169 , @xmath170 , @xmath171 , @xmath172 .
let @xmath173 .
_ failure rules of _ [ def : failure - rules ] clash failure : : : if @xmath174 , @xmath175 , @xmath176 , but @xmath177 .
cycle detection : : : if there are equations @xmath178 where @xmath179 , and @xmath180 occurs in @xmath181 for @xmath182 and @xmath6 occurs in @xmath183 .
freshness fail : : : if there is a freshness constraint @xmath184 .
freshness solution fail : : : if there is a freshness constraint @xmath185 , and @xmath186 .
the computation of @xmath187 can be done in polynomial time by iterating over the solution components .
we illustrate the letrec - rule by a ground example without flattening .
let the equation be : @xmath188 select the identity permutation @xmath64 , which results in : @xmath189 ( the freshness constraint @xmath190 holds ) .
then the application of the @xmath191-rule gives @xmath192 ( the freshness constraint @xmath193 holds ) .
the resulting equation is @xmath194 which obviously holds .
[ example : exponential - fps ] this example shows that fps ( together with the standard and decomposition rules ) may give rise to an exponential number of equations on the size of the original problem .
let there be variables @xmath195 and the equations @xmath196 @xmath197 where @xmath198 are permutations .
we prove that this unification problem may give rise to @xmath199 many equations , if the redundancy rule ( elimfp ) is not there .
+ ll the first step is by ( fps ) : & @xmath200 + ll using decomposition and inversion : & @xmath201 [ cols= " < , < " , ] now it is easy to see that all equations @xmath202 are generated , with @xmath203 where @xmath64 is a composition of a subsequence of @xmath204 , which makes @xmath199 equations .
the permutations are pairwise different using an appropriate choice of @xmath205 and @xmath9 .
the starting equations can be constructed using the decomposition rule of abstractions .
[ thm : unification - terminates ] the decision variant of the algorithm runs in nondeterministic polynomial time .
its collecting version returns a complete set of at most exponentially many unifiers , every one represented in polynomial space .
note that we assume that the input equations are flattened before applying the rules , which can be performed in polynomial time .
let @xmath206 be the input , and let @xmath207 .
the execution of a single rule can be done in polynomial time depending on the size of the intermediate state , thus we have to show that the size of the intermediate states remains polynomial and that the number of rule applications is at most polynomial .
the termination measure @xmath208 , which is ordered lexicographically , is as follows : @xmath209 is the number of letrec expressions in @xmath103 , @xmath210 is the number of letrec- , @xmath191-symbols , function - symbols and atoms in @xmath103 , @xmath211 is the number of different variables in @xmath103 , @xmath212 is the number of occurrences of variables in @xmath103 , @xmath213 is the number of equations not of the form @xmath214 , and @xmath215 is the number of equations .
since shifting permutations down and simplification of freshness constraints both terminate and do not increase the measures , we only compare states which are normal forms for shifting down permutations and simplifying freshness constraints .
we assume that the algorithm stops if a failure rule is applicable , and that the rules ( mms ) and ( fps ) are immediately followed by a full decomposition of the results ( or failure ) .
now it is easy to check that the rule applications strictly decrease @xmath124 : the rules ( mms ) and ( fps ) together with the subsequent decomposition strictly decrease @xmath216 . since expressions in equations are flat , ( mms ) does not increase the size : @xmath217 is first replaced by @xmath218 , and the latter is decomposed , which due to flattening results only in equations containing variables and suspensions .
thus @xmath210 is reduced by the size of @xmath219 . in the same way ( fps )
strictly decreases @xmath216 .
in addition @xmath210 is at most @xmath220 , since only the letrec - decomposition rule can add @xmath221-constructs .
the number of fixpoint - equations for every variable @xmath87 is at most @xmath222 for some ( fixed ) @xmath223 , since the number of atoms is never increased , and since we assume that is applied whenever possible .
the size of the permutation group is at most @xmath224 , and so the length of proper subset - chains and hence the maximal number of generators of a subgroupp is at most @xmath225 .
note that the redundancy of generators can be tested in polynomial time depending on the number of atoms .
now we prove a ( global ) upper bound on the number @xmath211 of variables : an application of ( 7 ) may increase @xmath211 at most by @xmath226 .
an application of ( fps ) may increase this number at most by @xmath227 , where the worst case occurs when @xmath15 is a letrec - expression . since ( mms ) and ( fps )
can be applied at most @xmath226 times , the number of variables is smaller than @xmath228 .
the other rules strictly decrease @xmath216 , or they do not increase @xmath216 , and strictly decrease @xmath229 and can be performed in polynomial time . the problematic rule for complexity is ( fps ) , which does not increase @xmath209 and @xmath210 , but may increase @xmath211 , @xmath212 and @xmath215 ( see example [ example : exponential - fps ] ) .
this increase is defeated by the rule ( elimfp ) , which helps to keep the numbers @xmath212 and @xmath215 low .
[ thm : unification - sound - and - complete ] the algorithm is sound and complete .
soundness of the algorithm holds , by easy arguments for every rule , similar as in @xcite , and since the letrec - rule follows the definition of @xmath49 in def .
[ def : alpha - equivalence - sim ] .
a further argument is that the failure rules are sufficient to detect final states without solutions .
completeness requires more arguments .
the decomposition and standard rules ( with the exception of rule ( 7 ) ) , retain the set of solutions . the same for ( mms ) , ( fps ) , and . the nondeterministic rule ( 7 )
provides all possibilities for potential ground solutions .
moreover , the failure rules are not applicable to states that are solvable .
a final output of has at least one ground solution as instance : we can instantiate all variables that remain in @xmath230 by a fresh atom
. then all fixpoint equations are satisfied , since the permutations can not change this atom , and since the ( atomic ) freshness constraints hold
. this ground solution can be represented in polynomial space by using @xmath146 , plus an instance @xmath231 for all remaining variables @xmath87 and a fresh atom @xmath22 , and removing all fixpoint equations and freshness constraints .
[ thm : letrec - unification - in - np ] the nominal letrec - unification problem is in @xmath232 .
this follows from theorems [ thm : unification - terminates ] and [ thm : unification - sound - and - complete ] .
reductions in higher order calculi with letrec , in particular on a meta - notation , require a matching algorithm , matching its left hand side to an expression . consider the ( lbeta)-rule , which is the version of ( beta ) used in call - by - need calculi with sharing @xcite .
@xmath233 an ( lbeta ) step , for example , on @xmath234 is performed by switching to the language @xmath235 and then matching @xmath236 , where @xmath237 is the explicit representation of the binary application operator .
this results in @xmath238 , and the reduction result is the @xmath112-instance of @xmath239 , which is @xmath240 . note that only the sharing power of the recursive environment is used here .
we derive a nominal matching algorithm as a specialization of .
we use nonsymmetric equations written @xmath241 , where @xmath242 is an @xmath83-expression , and @xmath70 does not contain variables .
note that neither freshness constraints nor suspensions are necessary ( and hence no fixpoint equations ) .
we assume that the input is a set of equations of ( plain ) expressions
. the rules of the algorithm are : + @xmath243 @xmath244 @xmath245 + @xmath246 if @xmath247 , otherwise fail .
+ @xmath248 + where @xmath64 is a ( mathematical ) permutation on @xmath65 + @xmath249 if @xmath250 , otherwise fail + the test @xmath250 will be performed by the ( nondeterministic ) matching rules .
clash failure : : : if @xmath174 , @xmath251 , but @xmath177 .
[ thm : matching - in - np ] is sound and complete for nominal letrec matching .
it decides nominal letrec matching in nondeterministic polynomial time .
its collecting version returns a finite complete set of an at most exponential number of matching substitutions , which are of at most polynomial size .
this follows by standard arguments .
nominal letrec matching is np - complete .
the problem is in np , which follows from theorem [ thm : matching - in - np ] .
it is also np - hard , which follows from the ( independent ) theorem [ thm : matching - np - hard ] . a slightly more general situation for matching occurs , when the matching equations @xmath165 are compressed using a dag .
we construct a practically more efficient algorithm from as follows .
first we generate @xmath252 from @xmath165 , which only contains ( plain ) flattened expressions by encoding the dag - nodes as variables together with an equation .
an expression is said @xmath165-ground , if it does not reference variables from @xmath165 ( also via equations ) . in order to avoid suspension ( i.e. to have nicer results ) , the decomposition rule for @xmath191-expressions with different binder names is modified as follows : @xmath253 @xmath254 is @xmath165-ground + the extra conditions @xmath255 and @xmath165-ground can be tested in polynomial time .
the equations @xmath252 are processed applying ( with the mentioned modification ) with the guidance that the right - hand sides of match - equations are also right - hand sides of equations in the decomposition rules .
the resulting matching substitutions can be interpreted as the instantiations into the variables of @xmath165 .
since @xmath165 is a matching problem , the result will be free of fixpoint equations , and there will be no freshness constraints in the solution .
thus we have : [ thm : nom - dag - matching ] the collecting variant of outputs an at most exponential set of dag - compressed substitutions that is complete and where every unifier is represented in polynomial space .
[ thm : matching - np - hard ] nominal letrec matching ( hence also unification ) is np - hard , for two letrec expressions , where subexpressions are free of letrec .
we encode the np - hard problem of finding a hamiltonian cycle in regular graph @xcite : let @xmath256 be the vertexes of the graph , and @xmath257 be the set of edges .
the first environment part is @xmath258 , and a second environment part @xmath259 consists of bindings @xmath260 for every edge @xmath261 for fresh names @xmath41 .
then let @xmath262 @xmath263 representing the graph .
let the second expression encode the question whether there is a hamiltonian cycle in a regular graph as follows .
the first part of the environment is @xmath264 .
the second part is @xmath265 consisting of @xmath266 , and the third part consisting of a number of ( dummy ) entries of the form @xmath267 , where @xmath41 is always a fresh atom for every binding , and @xmath268 are fresh variables . the number of these dummy entries can be easily computed from the number of nodes and the degree of the graph , and it is less than the size of the graph .
then the matching problem is solvable iff the graph has a hamiltonian cycle .
[ thm : letrec - iunification - np - complete ] the nominal letrec - unification problem is np - complete .
this follows from theorems [ thm : letrec - unification - in - np ] and [ thm : matching - np - hard ] .
we say that an expression @xmath70 _ contains garbage _ , iff there is a subexpression @xmath269 , and the environment @xmath21 can be split into two environments @xmath270 , such that @xmath271 is not trivial , and the atoms from @xmath272 do not occur in @xmath259 nor in @xmath28 .
otherwise , the expression is _ free of garbage_. since @xmath0-equivalence of @xmath10-expressions is graph - isomorphism - complete @xcite , but @xmath0-equivalence of garbage - free @xmath10-expressions is polynomial , it is useful to look for improvements of unification and matching for garbage - free expressions . as a remark : graph - isomorphism is known to have complexity between @xmath273 and @xmath232 ; there are arguments that it is weaker than the class of np - complete problems @xcite .
there is also a claim that it is quasi - polynomial @xcite , which means that it requires less than exponential time .
[ thm : matching - gi - hard ] nominal letrec matching with one occurrence of a single variable and a garbage - free target expression is graph - isomorphism - hard . let @xmath274 be two graphs .
let @xmath242 be @xmath275 the encoding of a graph @xmath276 where @xmath271 is the encoding as in the proof of theorem [ thm : matching - np - hard ] and the nodes are encoded as @xmath277 .
then the expression @xmath242 is free of garbage .
let the environment @xmath259 be the encoding of @xmath278 in the expression @xmath279 .
then @xmath70 matches @xmath242 iff the graphs are isomorphic .
hence we have @xmath280-hardness .
if there is an isomorphism of @xmath276 and @xmath278 , then it is easy to see that this bijection leads to an equivalence of the environments , and we can instantiate @xmath87 with @xmath281 .
extending the language by variables @xmath282 that may encode partial letrec - environments would lead to a larger coverage of unification problems in reasoning about the semantics of programming languages .
consider as an example a rule ( llet - e ) that merges letrec environments ( see @xcite ) : + @xmath283 .
+ it can be applied to an expression @xmath284 as follows : the left - hand side @xmath285 of the reduction rule matches @xmath286 with the match : @xmath287 , producing the next expression as an instance of the right hand side @xmath288 , which is : @xmath289 .
note that in the application to extended lambda calculi , a bit more care ( i.e. a further condition ) is needed w.r.t .
scoping in order to get valid reduction results in all cases .
we will now also have partial environments as syntactic objects .
the grammar for the extended language @xmath290 ( * l*et*r*ec * l*anguage e*x*tended with * e*nvironments ) is : @xmath291 we define a matching algorithm , where environment variables may occur in left hand sides .
this algorithm needs a more expressive data structure in equations : a letrec with two environment - components , ( i ) a list of bindings that are already fixed in the correspondence to another environment , and ( ii ) an environment that is not yet fixed .
we denote the fixed bindings as a list , which is the first component . in the notation
we assume that the ( non - fixed ) letrec - environment part on the right hand side may be arbitrarily permuted before the rules are applied .
the justification for this special data structure is the scoping in letrec expressions .
note that suspensions do not occur in this algorithm .
[ def : matchalg - env ] the matching algorithm for expressions where environment variables @xmath282 and expression variables @xmath87 may occur only in the left hand sides of match equations is described below .
the rules are : @xmath243 @xmath292 @xmath245 + @xmath246 if , @xmath255
@xmath293 + if @xmath294 , otherwise fail .
+ @xmath295 + @xmath296 @xmath297 { { \ \unlhd\ } } [ e_1';\ldots;e_n']\ } } { \gamma { \ensuremath{\mathaccent\cdot\cup}}\{e_1 { { \ \unlhd\ } } e_1 ' ; \ldots ; e_n { { \ \unlhd\ } } e_n'\}}}}$ ] + @xmath298 @xmath299 testing @xmath144 and @xmath300 is done with high priority using the ( nondeterministic ) matching rules .
clash failure : : : if @xmath174 , @xmath301 , but @xmath177 .
after successful execution , the result will be a set of match equations with components @xmath302 , and @xmath303 , which represents a matching substitution .
the algorithm [ def : matchalg - env ] ( ) is sound and complete .
it runs in non - deterministic polynomial time .
the corresponding decision problem is np - complete .
the collecting version of returns an at most exponentially large , complete set of representations of matching substitutions , where the representations are of at most polynomial size .
the reasoning for soundness , completeness and termination in polynomial time is a variation of previous arguments .
the nonstandard part is fixing the correspondence of environment parts step - by - step and keeping the scoping .
we constructed a nominal letrec unification algorithm , several nominal letrec matching algorithms for variants , which all run in nondeterministic polynomial time .
future research is to investigate extensions with environment variables @xmath282 , and to investigate nominal matching together with equational theories .
jeannin , j.b . ,
kozen , d. , silva , a. : cocaml : programming with coinductive types . tech . rep .
http://hdl.handle.net/1813/30798 , computing and information science , cornell university ( december 2012 ) , fundamenta informaticae , to appear luks , e.m .
: permutation groups and polynomial - time computation . in : finkelstein , l. ,
kantor , w.m .
groups and computation , proceedings of a dimacs workshop .
dimacs , vol .
dimacs / ams ( 1991 ) | a sound and complete algorithm for nominal unification of higher - order expressions with a recursive let is described , and shown to run in non - deterministic polynomial time .
we also explore specializations like nominal letrec - matching for plain expressions and for dags and determine the complexity of corresponding unification problems .
keywords : nominal unification , lambda calculus , higher - order expressions , recursive let , operational semantics |
xmmu122939.7 + 075333 , which is coincident with a spectrally confirmed globular cluster ( rz2109 ) in the massive virgo elliptical galaxy ngc4472 @xcite , was the first x - ray emitting globular cluster identified to contain a black hole ( bh ) system , based on data from x - ray multiple mirror - newton ( xmm ) satellite @xcite .
the peak x - ray luminosity of the source at discovery , l@xmath5ergs@xmath1 , is more than 10 times the eddington luminosity ( l@xmath6 ) of a neutron star .
this high luminosity makes the source an ultraluminous x - ray source ( ulx ; a non - nuclear x - ray source with l@xmath7ergs@xmath1 , assuming isotropic emission ) .
bf moreover , its luminosity varied by a factor of seven in just a few hours , thereby ruling out the possibility that the system is a superposition of neutron stars and confirming it as a likely bh system .
@xcite carried out spectroscopic analysis of xmmu122939.7 + 075333 using _ xmm - newton _
data @xcite and _ chandra _ data .
they found that the spectra were well fit by a multicolour disk blackbody model and had an inner disk temperature of about _ _
k__t@xmath8= 0.15kev with the same underlying continuum whether the source was in a bright or faint phase , but with varying absorption column density .
the results of this spectral analysis support the idea that the x - ray variability was due to a change in absorption column density .
further investigation by @xcite using four short _ chandra _ and _ swift _ observations of xmmu122939.7 + 075333 found that its luminosity appeared to be declining over a period of several years .
xmmu122939.7 + 075333 has also been studied extensively at optical wavelengths .
@xcite found [ oiii]@xmath95007 emission associated with the system .
optical spectroscopic follow up with higher signal - to - noise by @xcite revealed both [ oiii]@xmath95007 and [ oiii]@xmath94949 emission from xmmu122939.7 + 075333 .
in addition to the narrow ( few hundred kms@xmath1 ) core already found by @xcite , these emission lines were found to have very broad wings with a velocity width of about 2000kms@xmath1 and were therefore blended .
the total emission line luminosity in the @xcite spectrum was found to be 1.4@xmath10ergs@xmath1 , with a flux ratio of [ oiii ] to h@xmath11 of at least 30 .
a long term spectroscopic campaign was carried out by @xcite to monitor the variability and velocity structure of the [ oiii ] emission from xmmu122939.7 + 075333 . while the [ oiii ] velocity structure was relatively constant over a time span of more than 400 days , the luminosity of the emission lines declined by approximately 10% .
most recently , @xcite used spatially resolved spectroscopic data from _ hst_/stis to determine the spatial structure of the [ oiii ] emission .
they found that the half - light radius of the gas is between 3 and 7pc and that the light profile of the oxygen gas emission is consistent with the light profile of rz2109 s continuum .
the high ratio of [ oiii ] to h@xmath11 determined by @xcite suggests that the donor star is likely to be a hydrogen deficient star such as a white dwarf ( wd ) , as first suggested by @xcite .
most wds are composed almost entirely ( @xmath12 by mass ) of carbon and oxygen ( e.g. , * ? ? ?
thus we would expect that the matter accreted from such a donor would be rich in carbon and oxygen and poor in hydrogen .
this is in keeping with the findings of @xcite , who showed that the optical spectrum associated with xmmu122939.7 + 075333 is poor in hydrogen and rich in oxygen .
on the basis of evolutionary arguments @xcite suggested that the system is most likely a hierarchical triple star system , with an inner stellar mass bh - wd binary . in this scenario , it is thought that the outer star in the triple plays an important role in the evolution of the inner binary through the kozai mechanism @xcite .
@xcite argued that we would expect @xmath13 ( and therefore the x - ray flux of the binary ) to vary significantly on the time - scale for which @xmath14 varies due to the kozai mechanism .
they showed that for representative values of the system parameters , the period of @xmath14 variations is about 40 years ; we then expect the x - ray flux of the inner binary to vary on a similar timescale .
@xcite studied x - ray data for xmmu122939.7 + 075333 that spanned 30 years and they show that long term variation in flux is consistent with it declining on the timescale of the period of eccentricity .
thus the scenario of xmmu122939.7 + 075333 as a triple system with an inner wd - bh binary was favoured .
the x - ray variability is likely to be due to a variable absorption column density caused by an accretion disk which has the same composition as the wd donor star .
the [ oiii ] emission is likely to be from the disk wind , driven by moderately super - eddington accretion onto the central compact object .
the disk wind is then photoionised by the x - ray emission from the central source .
such a system would have several observational features and parameters that could be observed to test the veracity of this model .
a wd - bh binary is expected to be compact and have a short orbital period @xcite .
this gives rise to a high accretion rate , a fully ionised accretion disk and thus a persistently luminous x - ray source .
the wd would have supplied the accretion disk with material rich in oxygen and carbon .
it may therefore be possible to detect x - ray and/ or ultraviolet emission lines from carbon or oxygen in the source spectra .
if the photoionised accretion disk wind is variable , we expect to see aperiodic modulations of the x - ray luminosity due to obscuration by a changing absorption column density @xcite .
since soft x - rays ( @xmath41kev ) are absorbed preferentially , this means that the source spectrum should be softer when the source is more luminous .
recent deep observations of xmmu122939 + 075333 have made it possible to probe some of these questions . in this paper
we explore both the temporal and spectral properties of this source to improve our understanding of its nature .
the two most recent deep observations of xmmu122939 + 075333 were obtained from the _ chandra _ telescope ( obs i d 12889 and 12888 ) .
they were taken one week apart on 14 and 21 february 2011 respectively , with exposure times of 140ks and 160ks respectively .
the data were first reprocessed using _
calibration data from march 2011 to create new level two events files with the ciao _
acis_reprocess_events _ script .
the images were then checked for background flares .
specextract _ to produce source and background spectra and _ dmextract _ to make light curves . to establish whether the source is persistently bright , we look to a long term x - ray light curve spanning several years .
the long term light curve of the source was produced using data from the _ chandra _ , _ xmm _ and _ rosat _ telescopes .
the light curves for obs i d 12889 and 12888 are used to explore the short term variation of the x - ray flux . to test whether varying intrinsic absorption plays a significant role in flux variation , the most recent _ _
chandra__observations were split into bright and faint phases ( with the dividing line being the mean count rate of @xmath151.3@xmath16 to more accurately analyse the data .
we analyse the light curves to determine whether the flux variations are between faint and bright phases is significant and , if so , whether it is related to a change in absorption column by calculating hardness ratios . to do this , the counts in the bright and faint phases were divided into hard ( 1@xmath175kev ) and soft ( 0.5@xmath171kev ) energy bands , such that there were a roughly even number of counts in each band for the bright phase .
xmmu122939 + 075333 is situated near the edge of ccd chip acis-5 in obs i d 12889 and 12888 . to calculate the source count rate accurately , we made exposure maps of the chip using _ mkexpmap _ and used them to correct for the decreased effective area near the chip gaps .
the hardness ratio , @xmath18 where @xmath19 and @xmath20 are the number of counts in the hard and soft bands respectively , was then determined for the bright and soft phase of each observation .
we also compute the probability that the changes in hardness ratio between faint and bright phases happened by chance , using the binomial probability distribution . in order to compare the spectral properties of xmmu122939 + 075333 to those of other black hole binaries
, we model the spectra with the standard spectral model often used for galactic bh binaries as well as ulxs ; i.e. a model consisting of a disk blackbody ( diskbb ) continuum plus a powerlaw ( pl ) component .
the obs i d 12888 and 12889 spectra were both fit with this model .
we also include 100ks _ xmm _ observation of the source taken in 2004 ( hereafter referred to as xmm04 ; * ? ? ?
* ) in out spectral analysis .
the optical spectra of xmmu122939 + 075333 shows strong [ oiii ] emission . to establish whether oxygen emission is also present in the x - ray spectra of the source , we attempt to model the excess of soft counts as a an emission line , using a gaussian spectral model component . the obs i d 12889 and 12888 spectra and xmm04 were fit jointly with the same models so that the gaussian component could be tied to the same value across all three data sets .
the data sets were fit simultaneously with the same continuum model ( either diskbb only , pl only , or diskbb plus pl ) , with the associated continuum spectral parameters ( e.g. , photon index , @xmath21 ) free to vary between each data set .
the _ f _ test was carried out on the gaussian line component to determine its significance .
the spectral analysis was carried out using the interactive spectral interpretation system ( isis ) version 1.6.1 .
only data in the well calibrated , low background range of 0.5@xmath175kev were used .
this energy range is not standard ; many other groups use an energy range up to 8 or 10kev .
however , after inspecting our data , we found that our chosen energy range resulted in the best quality spectra .
the gehrels statistic was used to calculate errors as it gives a more reliable fit to data with low count rates @xcite .
the phabs absorption model was used to fit foreground absorption with the galactic absorption column density , @xmath22@xmath23 and default isis ( xspec ) abundances and cross sections .
we do not have a sufficient number of counts in the spectra to allow for the fitting of models where both the spectral parameters and the absorption column density are free to vary .
we thus keep the absorption column density frozen to the galactic foreground value .
the interstellar medium of elliptical galaxies contains very little cool molecular or atomic gas ( see e.g. , * ? ? ?
* ) , so any absorption from ngc4472 itself would be negligible . for the cases where the flux variation between bright and faint spectra was found to be significant , the spectra were analysed separately .
the bright and faint phase spectra were grouped to a minimum signal - to - noise ratio of three and two per bin respectively .
these spectra were then separately fit with pl or diskbb continuum models @xcite and a gaussian component to fit an excess of counts at @xmath150.6kev . to further test whether the flux variation between bright and faint spectra could be due to a change in absorption column density
, we also fit the faint spectra with the best fit parameters of their respective bright spectrum and allow @xmath24 to vary .
the long term light curve was made up of data from the most recent deep _ chandra _ observations of xmmu122939.7 + 075333 as well as archival data from _ xmm _ , _ chandra _ , and _
rosat_. the flux from a _
data set taken in 2000 was determined using isis and the spectral parameters from @xcite .
the flux for the 2004 _ xmm _ observation was taken from the bright phase spectrum ( count rate @xmath25 0.04 counts / s ) of xmm04 only and calculated in isis using the spectral parameters found in @xcite .
this was done to be consistent with the work of @xcite and @xcite .
the 2002 _ xmm _ and 2010 _ chandra _ @xcite fluxes and errors were calculated from their respective count rate or flux errors using the w3pimms tool , assuming a @xmath21= 1.7 power law spectral model .
the _ rosat _ data @xcite had poor spectral resolution and w3pimms was used to calculate the flux from the count rate .
@xcite found that the count rate for the _ rosat _ data was consistent with that of the bright xmm04 data ; they therefore used the xmm04 inner disk temperature value ( 0.2kev ) to fit the _ rosat _ data .
the _ rosat _
luminosity is a factor of 6 higher when a 1kev blackbody temperature is assumed compared to a temperature of 0.2kev .
the errors for this source are thus dominated by the uncertain energy to counts conversion for this data .
@xcite estimate that the luminosity of the source at the time of the _ rosat _ observation is between 10@xmath26 and 10@xmath27 ergs@xmath1 . in this work
we used a 0.2kev blackbody to estimate the flux of the _ rosat _ data .
all the data plotted in the light curve , shown in fig.[long_lc ] , are the unabsorbed x - ray luminosities for the 0.210kev range .
the _ rosat _ , 2002 _ xmm _ and 2010 _ chandra _ errors were calculated from their respective count rate or flux errors using the w3pimms tool .
the dominant source of error for the _ rosat _ luminosity is due to uncertainty in converting counts to energy ( see * ? ? ?
* ; * ? ? ?
the errors in these luminosity measurements do not account for the uncertainty in the power law slope .
the source displays periods of brightening ( by a factor of roughly three from the mean count rate ) on the timescale of hours in the light curves from obs i d 12289 and 12888 .
these source light curves were fit with a constant count rate , which was different for each light curve .
this yielded a @xmath28 of 35.95/34 and 68.86/40 for obs i d 12889 and 12888 respectively .
a constant count rate fit to the light curve for obs i d 12889 can not be ruled out at greater than the 37% level ; we did not detect significant variability over this observation .
we will therefore not discuss the intra - observational variability of obs i d 12889 any further in this paper .
however , for obs i d 12888 , the straight line can be ruled out at greater than the 99% level ; the source is displaying significant flux variation over this time period .
the count rate threshold between the bright and faint phases is 1.3@xmath16counts / s .
we use the binomial probability distribution to determine whether or not the change in hr between bright and faint phases is significant , with each photon simply being assigned a probability of being in the soft or hard band . in our case
, the null hypothesis is that the hr of the faint phases is only different from that of the bright ones by chance ( given that in the bright phases the chances of a photon falling into the hard or soft band is 50% ) .
hardness ratio data as well as the probabilities are displayed in table[hr_table ] .
we find that the hardness ratios for the bright and faint phases of obs i d 12888 are [email protected] and [email protected] .
the null hypothesis probability for obs i d 12888 is 0.006 ; thus the difference in hardness between the faint and bright phases is statistically significant .
[ cols="<,<",options="header " , ] [ [ joint - fits - of - obs - id-12889 - 12888-and - xmm04-full - spectra ] ] joint fits of obs i d 12889 , 12888 , and xmm04 full spectra
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for the joint fit of obs i d 12889 , 12888 and xmm04 we compared the diskbb only , pl only , and diskbb plus pl continuum models with and without the gaussian component .
we then tested the significance of the line component using an f test .
we note that these data fulfil the criteria for the appropriate use of the f test : the line energy is well - known in advance as the emission is assumed to be from oviii and and the line width should be much smaller than the energy resolution of the ccds on chandra and xmm .
so , while warnings about the use of the f test have been made in recent years @xcite , we are comfortable that it should provide , in this case , a good estimator of the significance of the feature . for the model consisting of a diskbb continuum only , the inner disk temperatures were 0.3@xmath30kev , 0.5@xmath31 and 0.3@xmath32kev for obs i d 12889 , 12888 , and xmm04 respectively . when the gaussian component was added to the diskbb continuum model
, the inner disk temperature , t@xmath8 for obs i d 12889 , 12888 , and xmm04 were found to be 0.3@xmath33kev , 0.7@xmath34kev and 0.4@xmath30kev respectively .
an emission line component with a line energy of 0.66@xmath35kev was found . in this fit ,
the line width was restricted to [email protected] and a pw of 6.2@xmath36kev was found .
the ews for obs i d 12889 , 12888 , and xmm04 are 0.85kev , 0.17kev and 0.10kev respectively .
the reduced @xmath37 is 0.88 for the diskbb only model and 0.76 for the gaussian plus diskbb model .
the f test revealed that the emission line is significant , with an f statistic of 8.24 and null hypothesis probability of 4.6@xmath38 .
the pl - only continuum model yielded the following spectral parameters for obs i d 12889 , 12888 , and xmm04 respectively : 3.5@xmath39 , 2.8@xmath40 and 2.7@xmath41 .
when the gaussian component was included in the model , an emission line with an energy of 0.66@xmath42kev and a pw of 1.92@xmath43 @xmath44kev was found .
the spectral parameters for the continuum for this model are 3.10@xmath45 , @xmath46 and 2.5@xmath41 for obs i d 12889 , 12888 and xmm04 respectively .
the ews are 0.32kev , 0.10kev and 0.07kev for obs i d 12889 , 12888 , and xmm04 respectively .
the pl - only model has a reduced @xmath37 of 0.73 whereas the gaussian plus pl model had a reduced @xmath37 of 0.66 .
the f test results show that the line is significant ; the f statistic is 5.91 and the null hypothesis probability is 8.22@xmath47 . in the gaussian plus diskbb and pl continuum model , several of the continuum components
were not well constrained , but the line parameters are robust and very similar to what is found for the other spectral fits .
t@xmath8 for obs i d 12889 , 12888 , and xmm04 were found to be 997.0@xmath48kev , 0.08@xmath49kev , and 0.3@xmath50kev respectively .
the values for @xmath21 are 4.8@xmath51 , 1.6@xmath52 , and 0.2@xmath53 .
the gaussian component had a line energy of 0.66@xmath42kev and pw of 0.01@xmath54kev .
the ew values were 0.05@xmath55kev , 0.11@xmath56kev , and 0.06@xmath32kev for obs i d 12889 , obs i d 12888 , and xmm04 respectively .
the reduced @xmath37 for the gaussian plus continuum model is 0.57 .
we obtain the reduced @xmath37 for the joint fit of the diskbb plus pl continuum model by summing the respective reduced @xmath37 values from the individual fits ; this gives a reduced @xmath37 of 0.62 .
the f statistic is then 4.79 and the null hypothesis probability is 0.003 .
the line is therefore fairly significant .
the spectral parameters for these models are are summarised in table[joint_fit ] .
the spectra for the diskbb only and pl only continuum model joint fits are shown in fig.[joint_fits_pl ] and fig.[joint_fits_disk ] .
we assume that the difference between the bright and faint phases in obs i d 12888 are due to a variable absorption column density ,
i.e. , the underlying continuum is essentially the same and the difference in flux is due to the soft photons being preferentially absorbed in the faint phase .
we can test this assumption further by fitting the faint phase spectrum using the best fit continuum spectral parameters from the respective bright phase spectrum and leaving the absorption free to vary .
if our assumption is correct , then the fits should be statistically acceptable and the absorption column density should be higher than the galactic foreground value .
we found that this model provided statistically acceptable fits for the faint spectrum , with @xmath37/@xmath57 for obs i d 12888 .
the faint phase spectrum of obs i d 12888 was found to have a higher absorption value of 2.5@xmath58@xmath23 and a 90% confidence interval of @xmath59@xmath23 .
we find that during the period from 1994 to 2011 , the source has @xmath60ergs@xmath1 . in addition , the source luminosity has increased significantly between the observation of @xcite in february 2010 and the most recent observations in february 2011 . this large increase in luminosity over a period of a year argues against the idea that the long term flux variations are caused by ( just ) the kozai mechanism . this intra - observational variability seen in obs
i d 12888 is too short to be due to spectral state changes , which have time scales on the order of days to weeks for stellar mass black hole binary systems ( see e.g. , * ? ? ? * ) .
these variations could , however , be easily explained by a variable absorption column density .
this result is consistent with the explanation discussed in 1 : that a variable absorption column density is responsible for the decrease in x - ray luminosity due to the preferential absorption of soft x - ray photons .
@xcite also found that the spectral changes between the faint and bright phases in xmm04 were due to a lack of soft photons in the faint phase , which is consistent with our result for obs i d 12888 .
[ [ bright - phase - spectrum-1 ] ] bright phase spectrum + + + + + + + + + + + + + + + + + + + + + + the spectral parameters for the diskbb plus pl model are consistent with the system having a cool disk ( t@xmath8=@xmath61kev ) and hard power law component ( @xmath21=2.06@xmath62 ) as seen in other ulxs ( see e.g. , * ? ? ?
this value of t@xmath8 is not consistent with the spectral parameter values found for galactic black hole binaries , which are typically in the range 0.7 1.5kev ( see e.g. * ? ? ?
[ [ obs - id-12889-and-12888-and - xmm04-full - spectra ] ] obs i d 12889 and 12888 and xmm04 full spectra
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + these parameter values are in keeping with the results found for other ulxs ; i.e. a cool disk ( t@xmath63kev ) and a hard pl component ( @xmath64 ) ( see e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the diskbb plus pl model has fit the apparent hard excess much better than the previous models that only include one continuum component ( see fig.[disk_pl_spectra ] ) .
the disk blackbody plus power law model therefore provides a good fit to the spectra of xmmu122939.7 + 075333 , implying that the source is behaving like a typical ulx .
we note that the apparent excess of photons at around 1kev still remains , even with this two continuum component model .
while this two component continuum model can fit ulx spectra with low signal - to - noise , @xcite showed that high quality ulx spectra with at least 10000 counts are not well fit by this model .
instead , these high signal - to - noise spectra are well fit a model consisting of a disc plus a comptonized corona component . the characteristic disk temperature , t@xmath65 , for this model is found to be @xmath660.5kev , which like the disk temperatures found for the diskbb plus pl models , is also significantly cooler than that that of the galactic bhbs .
since very high quality spectra are required to properly fit the @xcite models , we do not attempt to carry out this analysis on our data sets . [
[ bright - and - faint - phase - spectra ] ] bright and faint phase spectra + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the energy for gaussian component found in the models are consistent with emission from highly ionised oxygen , oviii , with its line energy of 0.65kev
. the oviii edge absorption feature is located at 0.87kev .
we have not attempted to model this feature due to the low signal - to - noise of the data , but can not rule out its presence . however , because the fitted emission lines are strong and narrow , the absence or presence of the absorption edge should not affect whether the emission line is detected .
[ [ obs - id-12889 - 12888-and - xmm04-full - spectra-1 ] ] obs i d 12889 , 12888 , and xmm04 full spectra + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + based on the results of the f test for these spectral fits , the gaussian component is found to be fairly significant .
in all the joint fits , the gaussian component is narrow and unresolved . in the model with two continuum components , the gaussian component is well constrained , with reasonable parameter values .
however the continuum parameters in this model are degenerate with one another .
given that the disk blackbody component peaks outside of the _ chandra _ energy range , it is very difficult to distinguish between a cool diskbb and a steep pl continuum model .
the emission line seen here is unlikely to be due to reflection from the accretion disk because it is so luminous ( see * ? ? ? * who find an emission line with an equivalent width of 40ev from a galactic ultracompact x - ray binary . ) .
the emission could , however be caused by either collisional excitation or photoionisation . due to the low signal - to - noise ratio of our data , we are not able to draw further conclusions about the origin of this spectral feature , except to say that this line emission seen in xmmu122939.7 + 075333 and not in other nearby ulxs may be due to the much higher oxygen abundance in this system s donor star compared to systems fed by massive stars .
these results suggest that the soft excess could be due to spectral line emission .
we argue that this line emission could arise from highly ionised oxygen ( oviii ) , which is in line with our theory that the accretion disk of xmmu122939.7 + 075333 is rich in oxygen due to the companion star being a wd .
other ulxs also show soft , line - like x - ray emission .
for instance , @xcite found that the spectra of ngc5408 x-1 contained line - like emission at @xmath150.6kev .
they suggested that this emission could be due to the presence of highly ionised oxygen and iron around the source .
subsequently , @xcite showed that this feature is well modelled by an optically thin plasma model component with temperature @xmath67kev .
ngc5204 x-1 was found to have an excess of counts at around 1kev @xcite .
it was found that this excess could be modelled either as a thermal plasma component with @xmath68kev or as a broad gaussian line with @xmath69kev .
@xcite found that the soft excess in ngc4395 x-1 could be modelled as a thermal plasma with @xmath70kev .
@xcite modelled the soft excess in ngc5408 x-1 and ngc6946 x-1 as blue - shifted , absorption lines due to the radiatively driven wind which is expected to accompany the accretion scenario in ulxs @xcite .
they find that modelling the spectra in this way gives statistically poorer fits than modelling with a thermal plasma .
however , they still favoured broad absorption features from a disk wind over thermal plasmas , and argued that more sophisticated modelling is needed . in our system , we argue that the features are more likely to be emission lines from the same plasma that produces the o[viii ] emission lines in the optical , but we agree that several plausible scenarios could produce the observed results , and some combination of better data and more sophisticated modelling would be required to distinguish between the various possibilities .
our spectra also show hints of other features besides the excess at 0.65kev .
both the obs i d 12889 and 12888 spectra show suggestive evidence for a hard ( @xmath712kev ) excess .
@xcite also report a hard excess in their spectral analysis of the source . this feature may require an additional continuum component in order to fit properly .
however , our data do not have enough counts to do this without increasing the degeneracy of the model significantly .
there also appears to be an excess of photons at @xmath151kev in the _ chandra _ and _ xmm _ spectra .
narrow spectral features have been seen at these energies for lmc x-4 @xcite .
the narrow feature arises from emission from ne x ly@xmath72 ( 1.02 kev ) .
however , since the standard ulx spectral model gives statistically acceptable fits to the spectra of xmmu122939.7 + 075333 , we can not say that the line plus continuum spectral modelled is preferred over the standard ulx model .
we can therefore not distinguish between the scenario where the soft excess is due to highly ionised oxygen or a cool disk component , using these data sets .
the result of the varying absorption model indicate that the absorption column density is significantly higher during the faint phase of obs i d 12888 than during the bright phase .
this result is in accord with the findings of the flux variation analysis as well as the hr calculations .
it therefore appears that over the course of obs i d 12888 , the source did undergo significant variations in flux , hr and absorption column density between the faint and bright phase .
@xcite , using bright and faint phase spectra from xmm04 , also found that the faint phase could be well modelled by using the best fit spectral parameters for the bright phase and letting the absorption column density be free to vary .
they found a significantly increased value of @xmath24 with this model .
in this paper we have shown that xmmu122939.7 + 075333 appears to be a persistently luminous x - ray source .
the short term , intra - observational flux variability is too rapid to be caused by spectral state changes and can more easily be explained by a changing absorption column density .
the significant increase in luminosity over a the period of a year ( 2010 to 2011 ) is not consistent with the idea that the long term , large amplitude flux modulations displayed by xmmu122939.7 + 075333 are caused by the kozai mechanism alone .
however , this does not rule out the possibility that there may be other variations in addition to the decay expected from the kozai mechanism .
for instance , @xcite have shown that periodic modulation of the eccentricity of a close binary can be forced by a third body , even in the absence of the large inclination angle required for kozai resonances to develop .
we also show that source has an excess of soft photons when the source is in a bright phase , which is consistent with previous analysis carried out on this system .
the excess of soft x - ray photons can be modelled as emission from oviii ( 0.65kev ) from the source
. however , the standard spectral model ( disk blackbody plus power law component ) also gives acceptable fits to the data .
both sets of spectral models suggest that xmmu122939.7 + 075333 is indeed bh binary . at this point we do not have sufficient evidence to prefer one over the other .
tdj acknowledges support from a stobie - salt studentship , funded jointly by the nrf of south africa , the british council and the university of southampton .
this research has made use of data obtained from the chandra data archive and software provided by the chandra x - ray center ( cxc ) in the application packages ciao and chips , etc .
this material is based upon work supported by the national aeronautics and space administration under grant no .
go1 - 12160x .
grs acknowledges the support of an nserc discovery grant .
tdj would also like to thank texas tech university for its hospitality while this work was being finished .
lastly , the authors would like to thank daniel plant , pablo cassatella , joey neilsen , matthew middleton , and dominic walton for their assistance and useful comments . 99 antognini , j. m. , shappee , b. j. , thompson , t. a. , & amaro - seoane , p. 2014
, , 439 , 1079 ballantyne , d. r. , purvis , j. d. , strausbaugh , r. g. , & hickox , r. c. 2012 , , 747 , l35 begelman , m. c. , king , a. r. , & pringle , j. e. 2006 , , 370 , 399 bildsten , l. , & deloye , c. j. 2004 , , 607 , l119 caballero - garca , m. d. , & fabian , a. c. 2010 , , 402 , 2559 colbert , e. j. m. , & ptak , a. f. 2002 , , 143 , 25 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 dufour , p. 2011
, white dwarf atmospheres and circumstellar environments , 53 t. m. , & gallo , e. 2004 , , 355 , 1105 feng , h. , & soria , r. 2011 , , 55 , 166 gehrels , n. 1986 , , 303 , 336 gladstone , j. c. , roberts , t. p. , & done , c. 2009 , , 397 , 1836 gnedin , o. y. , maccarone , t. j. , psaltis , d. , & zepf , s. e. 2009 , , 705 , l168 hung , l .- w . , hickox , r. c. , boroson , b. s. , & vrtilek , s. d. 2010 , , 720 , 1202 hut , p. , & paczynski , b. 1984 , , 284 , 675 irwin , j. a. , brink , t. g. , bregman , j. n. , & roberts , t. p. 2010
, , 712 , l1 ivanova , n. , chaichenets , s. , fregeau , j. , et al . 2010 , , 717 , 948 jenkins , l. p. , roberts , t. p. , warwick , r. s. , kilgard , r. e. , & ward , m. j. 2005 , , 357 , 401 kajava , j. j. e. , & poutanen , j. 2009 , , 398 , 1450 king , a. r. , davies , m. b. , ward , m. j. , fabbiano , g. , & elvis , m. 2001 , , 552 , l109 kozai , y. 1962 , , 67 , 591 maccarone , t. j. 2003 , , 409 , 697 maccarone , t. j. , kundu , a. , zepf , s. e. , & rhode , k. l. 2007 , , 445 , 183 maccarone , t. j. , kundu , a. , zepf , s. e. , & rhode , k. l. 2010 , , 409 , l84 madej , o. k. , jonker , p. g. , fabian , a. c. , et al .
2010 , , 407 , l11 mcclintock , j. e. , & remillard , r. a. 2006 , compact stellar x - ray sources , 157 middleton , m. j. , roberts , t. p. , done , c. , & jackson , f. e. 2011 , , 411 , 644 middleton , m. j. , walton , d. j. , roberts , t. p. , & heil , l. 2014 , , 438 , l51 miller , j. m. , fabbiano , g. , miller , m. c. , & fabian , a. c. 2003 , , 585 , l37 mitsuda , k. , inoue , h. , koyama , k. , et al .
1984 , , 36 , 741 mitsuda , k. , inoue , h. , nakamura , n. , & tanaka , y. 1989 , , 41 , 97 mucciarelli , p. , zampieri , l. , turolla , r. , & treves , a. 2004 , memorie della societa astronomica italiana supplementi , 5 , 276 neilsen , j. , lee , j. c. , nowak , m. a. , dennerl , k. , & vrtilek , s. d. 2009 , , 696 , 182 peacock , m. b. , zepf , s. e. , kundu , a. , et al .
2012 , , 759 , 126 portegies zwart , s. f. , dewi , j. , & maccarone , t. 2004 , , 355 , 413 poutanen , j. , lipunova , g. , fabrika , s. , butkevich , a. g. , & abolmasov , p. 2007
, , 377 , 1187 proga , d. , stone , j. m. , & drew , j. e. 1998 , , 295 , 595 protassov , r. , van dyk , d. a. , connors , a. , kashyap , v. l. , & siemiginowska , a. 2002 , , 571 , 545 rhode , k. l. , & zepf , s. e. 2001 , , 121 , 210 ripamonti , e. , & mapelli , m. 2012 , , 423 , 1144 roberts , t. p. , warwick , r. s. , ward , m. j. , goad , m. r. , & jenkins , l. p. 2005
, , 357 , 1363 roberts , t. p. , kilgard , r. e. , warwick , r. s. , goad , m. r. , & ward , m. j. 2006 , , 371 , 1877 roberts , t. p. 2007
, , 311 , 203 rosswog , s. , ramirez - ruiz , e. , & hix , w. r. 2009 , , 695 , 404 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 shih , i. c. , maccarone , t. j. , kundu , a. , & zepf , s. e. 2008 , , 386 , 2075 steele , m. m. , zepf , s. e. , kundu , a. , et al .
2011 , , 739 , 95 stobbart , a .- m . , roberts , t. p. , & wilms , j. 2006 , , 368 , 397 watson , m. g. , schrder , a. c. , fyfe , d. , et al .
2009 , , 493 , 339 welch , g. a. , sage , l. j. , & young , l. m. 2010 , , 725 , 100 yu , w. , & yan , z. 2009 , , 701 , 1940 zepf , s. e. , maccarone , t. j. , bergond , g. , et al .
2007 , , 669 , l69 zepf , s. e. , stern , d. , maccarone , t. j. , et al .
2008 , , 683 , l139 | in this paper we discuss the luminosity modulations and spectral analysis results of the recent deep observations of xmmu122939.7 + 075333 , the first black hole discovered in a globular cluster .
the source has been detected many times , typically with l@xmath0ergs@xmath1 , but in a 2010 observation had faded to @xmath2ergs@xmath1 . in our 2011 observations , it has rebrightened to l@xmath3ergs@xmath1 .
this significant increase in luminosity over a a relatively short time period is not consistent with the idea that the long term flux modulations displayed by xmmu122939.7 + 075333 are caused by the kozai mechanism alone as had previously been suggested . instead ,
given that the source shows `` spiky '' behaviour in its light curve , it seems likely that the faintness in 2010 was a result of a short observation that did not catch any bright epochs .
we also find that when the source is brighter than average , it has an excess of soft ( @xmath40.7kev ) photons .
the spectral analysis reveals strong , albeit model - dependent , evidence of emission from highly ionised oxygen ( oviii ) when the source is brighter than average .
[ firstpage ] stars : x - rays : binaries galaxies : individual : ngc4472 x - rays : galaxies stars : optical |
the ferrimagnetic metal cucr@xmath0se@xmath1 shows the highest critical temperature t@xmath7=430k among chromium spinel chalcogenides .
@xcite although the lattice constants of the ferromagnetic semiconductors cdcr@xmath0s@xmath1 and cdcr@xmath0se@xmath1 differ only by 1 - 4@xmath8 , their ferromagnetism is considerably weakened by the complete filling of the valance band as reflected by t@xmath7=85k and 129k , respectively @xcite .
cucr@xmath0se@xmath1 exhibits large magneto - optical kerr effect ( moke ) in the near infrared photon energy region at room temperature which makes this compound a promising candidate for magneto - optical devices .
@xcite materials from the same family show interesting magneto - transport phenomena like the colossal magnetoresistance@xcite in fe@xmath9cu@xmath10cr@xmath0s@xmath1 , and the colossal magnetocapacitance@xcite in cdcr@xmath0s@xmath1 .
moreover , recent band structure calculations indicate that cucr@xmath0se@xmath1 is almost half - metallic,@xcite and the density of states for spin down electrons can be fully suppressed with cadmium doping , i.e. a perfect half - metallic situation can be realized .
@xcite the strong ferrimagnetism in cucr@xmath0se@xmath1 was first explained by lotgering and stapele assuming the mixed - valence state of cr@xmath11 and cr@xmath12 with the monovalent cu@xmath13 ; thus , this compound was classified as a d - metal with closed se 4p shell .
@xcite in this picture only the chromium sites are magnetic and the double exchange mechanism between the cr@xmath11 and the cr@xmath12 ions align their magnetic moment parallel . however , early neutron diffration studies indicated that each chromium is in the cr@xmath11 state .
@xcite later , goodenough proposed the copper ions to be divalent cu@xmath14 and as a source of the magnetism the 90 degree superexchange to be responsible for the coupling between the cr@xmath11 ions through the completely filled se 4p states .
@xcite the recent xmcd measurements by kimura et al.@xcite settled the long standing issue of the valance state of cucr@xmath0se@xmath1 .
they have confirmed the cr@xmath11 state , however , they have found almost monovalent copper and a delocalized hole in the se 4p band with a magnetic moment anti - parallel to the moment of the cr@xmath11 ions . based on these experimental results sarma et al.@xcite
have interpreted the ferrimagnetism in terms of a kinetic - energy driven mechanism in which the hybridization between the localized cr@xmath11 ions and the delocalized se 4p band results a hole mediated exchange .
their density functional calculation indicates the appearance of a hybridization induced hump - like structure at the fermi energy only for the up spin states in accordance with other band structure calculations .
@xcite in order to have a deeper understanding of the electronic structure of cucr@xmath0se@xmath1 we have investigated the low - energy ( e=0.1 - 4ev ) charge excitations over the temperature range of t=10 - 300k by determining both diagonal and off - diagonal element of the optical conductivity tensor . beside the low - energy response of the metallic carriers and the charge transfer excitations above @xmath15ev , we have found a sharp and distinct optical transition in the mid - infrared region with large magneto - optical activity .
single crystals of cucr@xmath0se@xmath1 were grown by the chemical vapor transport method .
details of the preparation and the structure characterization were given elsewhere .
@xcite all the optical measurements were carried out with nearly normal incidence on the as - grown ( 111 ) surface . in order to determine the diagonal optical conductivity , reflectivity spectra
was measured over a broad energy range ( e=0.08 - 26ev and e=0.08 - 5ev at room and low temperature , respectively ) to facilitate the proper kramers - kronig transformation .
we have measured the complex magneto - optical kerr angle @xmath16 , which allows the direct determination of the off - diagonal conductivity , between e=0.12 - 4ev with a polarization modulation technique .
@xcite in the mid - infrared region a fourier transform infrared spectrometer was combined with a znse photoelastic modulator ( hinds , ii / zs50 ) @xcite to perform measurements down to as low energy as e=0.12ev , while above @xmath17ev a caf@xmath0 photoelastic modulator ( hinds , i / cf50 ) and a grating spectrometer was used .
the external magnetic field [email protected] t was applied by a permanent magnet along the [ 111 ] easy axis of the magnetization and also parallel to the propagation direction of the light . in the above arrangement
, the conductivity tensor has the following form : @xmath19,\ ] ] in our notation the x , y , z directions do not correspond to the main cubic axes , as z is chosen parallel to the [ 111 ] easy axis .
se@xmath1 at various temperatures .
three main features are indicated by dashed lines .
the first charge transfer excitation ( se 4p@xmath20cr 3d ) appears at @xmath2@xmath5@xmath21ev with a low - energy shoulder at @xmath2@xmath5@xmath22ev due to on - site d - d transition of the chromium ions .
in addition to the low - energy metallic peak , a strong transition is present in the mid - infrared region ( at @xmath2@xmath5@xmath6ev ) with a characteristic width of @xmath23@xmath3@xmath6ev . in the inset
the maximum of the loss - function indicates the plasma frequency at @xmath24@xmath3@xmath25ev.,width=288 ]
the temperature dependence of the reflectivity and the diagonal optical conductivity spectra is shown in fig .
we identified the different contributions to the optical conductivity as follows .
the first charge transfer peak is centered around e=2.75ev , that we assign to se 4p @xmath20 cr 3d transition in agreement with former optical data on a broad variety of chromium spinel oxides and chalcogenides .
@xcite it has a low - energy shoulder located at e=1.9ev which likely originates from the on - site chromium d - d transition since this structure is common for chromium spinels insensitive to the change of the other cation .
@xcite this originally dipole - forbidden transition becomes allowed by the hybridization with the ligand , which results in the fairly small oscillatory strength .
the spectral structures become distinct as the temperature decreases . in the low - energy region ( @xmath2@xmath26@xmath27ev )
metallic conductivity appears .
although it does not closely follow a drude - like behaviour the scattering rate is estimated to be @xmath28ev at t=10k the small residual resistivity@xcite @xmath29@xmath30 cm is not typical of bad metals with strongly correlated d band . between the low - energy metallic term and the first charge transfer excitation , a distinct peak appears at @xmath2@xmath5@xmath6ev . at low temperatures
it becomes clearly distinguishable from the excitation of the free carriers . in spite of its closeness to the metallic continuum it becomes sharp at low temperatures characterized by a width of @xmath23@xmath3@xmath6ev width . in the inset of fig .
[ fig1 ] the maximum of the loss - function signals a plasma frequency of @xmath311ev , which only increase by 4@xmath8 as the temperature decrease to t=10k .
this implies merely tiny changes in the carrier concentration as a function of temperature .
@xmath5@xmath25ev due to the plasma resonance and its maximal value increase to @xmath32@xmath5@xmath33 as the temperature decrease to t=10k corresponding to the sharpening of the plasma edge .
the kerr rotation reaches the value @xmath34@xmath5@xmath35 in the same region.,width=268 ] the magneto - optical kerr spectra are presented in fig .
[ fig2 ] at room temperature and at the lowest temperature @xmath36@xmath5@xmath37k .
these results are in good agreements with kerr spectra previously reported at room - temperature for @xmath2@xmath38@xmath39ev .
@xcite the moke signal reaches its maximum around @xmath2@xmath5@xmath25ev , where kerr ellipticity exhibits a peak while kerr rotation has a dispersive line shape with the maximal values of @xmath32=1.9@xmath40 and @xmath34=-1@xmath40 , respectively .
although the magnetization is almost constant below room temperature , the moke is enhanced by 45@xmath8 down to @xmath36@xmath5@xmath37k . from the complex kerr angle
, we have calculated the off - diagonal conductivity according to the following relation : @xmath41 where @xmath42 and @xmath43 are the elements of the complex optical conductivity tensor .
the kerr rotation and ellipticity describe the phase - shift and the intensity difference , respectively , between left and right circularly polarized light upon normal - incidence reflection from a magnetic surface .
the corresponding results are presented in fig .
[ fig3 ] . in spite of the large moke around @xmath2@xmath5@xmath25ev ,
neither the off - diagonal nor the diagonal conductivity show any specific optical excitation in this energy region .
the large enhancement of the moke signal corresponds to the plasma resonance at @xmath31@xmath25ev ; it is caused by the strong minimum of the denominator of eq .
[ sxy_formula ] rather than by an increase of the off - diagonal conductivity.@xcite the almost perfect cancelation of this resonance in the off - diagonal conductivity indicates the properness of the kramers - kronig transformation for the reflectivity .
when the optical excitations are broad compared to the magnetically induced splitting of these transitions for the two circular polarizations , the kerr parameters @xmath34 and @xmath32 are proportional to the derivative of the reflectivity : @xmath44 where @xmath45=r@xmath46e@xmath47 are the fresnel coefficients for the right and left circularly polarized photons and @xmath48@xmath5@xmath49 and @xmath50 are the reflectivity and the corresponding phase .
the sudden decrease of the reflectivity near the plasma edge , which generate the large moke signal , is sensitive to the slope of the reflectivity which becomes steeper as the life - time increases toward low temperatures , causing considerable temperature dependence in the region of the plasma resonance . .
the transition at @xmath2@xmath5@xmath6ev shows large while the se 4p@xmath20cr 3d charge transfer excitation has also considerable magneto - optical activity.,width=288 ] the real part of the off - diagonal conductivity ( shown in fig .
[ fig3 ] ) is dominated by two main structures , namely a broader hump around @xmath2@xmath5@xmath21ev and a resonance - like peak centered at @xmath2@xmath5@xmath6ev . the dispersive line - shape in the imaginary part of the off - diagonal conductivity and also the corresponding large values im@xmath51=53@xmath52 and im@xmath51=110@xmath52 at @xmath36@xmath5@xmath37k ,
respectively are suggestive of parity allowed transitions , which is reasonable for the se 4p@xmath20cr 3d charge transfer transition at @xmath2@xmath5@xmath21ev . the magnitude of the low - energy part of the off - diagonal conductivity enlarged in fig . [ fig4 ] .
is very close to that of the dc hall effect obtained in the same magnetic field,@xcite except for @xmath36@xmath5@xmath37k where @xmath53@xmath5@xmath54@xmath52 .
as the temperature decreases the low - energy tail of the real part is considerably reduced in contrast to the temperature independent behavior of the magnetization .
this may indicate that a non - perturbative treatment of the spin - orbit coupling is also necessary to describe the low - energy off - diagonal conductivity as it was formerly proposed for the dc anomalous hall effect.@xcite the relations between the elements of the optical conductivity tensor and the underlying microscopic optical processes are described by the kubo formula : @xmath55f(\varepsilon_i ) \{|\langle f|\pi_+|i\rangle|^2 + \nonumber\\ + |\langle f|\pi_-|i\rangle|^2\ } [ \delta(\omega_{fi}-\omega)+\delta(\omega_{fi}+\omega ) ] , \nonumber\\ im\{\sigma_{xy}\}=\frac{\pi e^2}{4m^2v\hbar\omega}\sum_{i , f } [ 1-f(\varepsilon_f)]f(\varepsilon_i ) \{|\langle f|\pi_+|i\rangle|^2- \nonumber\\ - |\langle f|\pi_-|i\rangle|^2\ } [ \delta(\omega_{fi}-\omega)+\delta(\omega_{fi}+\omega)],\label{kubo_formula}\end{aligned}\ ] ] where @xmath56 are the momentum operators in circular basis .
the real part of the diagonal optical conductivity is proportional to the joint density of states for the occupied and unoccupied states multiplied by the electric dipole matrix elements , therefore , it describes the absorption of light for left and right circularly polarized photons in average . on the other hand , the off - diagonal conductivity is the difference between the absorption spectra corresponding the two circular polarizations .
contributions from electric dipole processes to off - diagonal optical conductivity are remarkable in ferromagnetic materials due to the orbital magnetization induced by the spontaneous spin polarization via the spin - orbit interaction similarly to the anomalous hall effect in the dc limit . and an e@xmath57 band .
the hybridization between the selenium and chromium induce a gap just below the fermi energy and a holes appears in the majority spin channel .
the arrows indicate the excitations observed both in the diagonal and off - diagonal conductivity spectra with the same labels as used in fig . 1 and 3.,width=316 ] to understand the origin of the optical excitations ,
the results of the recent band structure calculations@xcite are summarized schematically in fig .
the chromium d - band is split by the cubic crystal field to a t@xmath58 and an e@xmath57 band .
the selenium 4p and the copper 3d states are fully mixed with each other . furthermore , the hybridization between the selenium 4p and the chromium t@xmath58 induces a gap just below the fermi energy in the majority spin channel . as a consequence ,
states from the se 4p band are shifted above the fermi level ( referred to as `` hump in the density of states '' in the introduction ) , thus holes appears in the majority spin channel .
the main optical transitions observed in the experiments are also indicated in the figure .
our optical and magneto - optical study confirms the results of the band structure calculations both in the close vicinity of the fermi energy and on a few ev large scale . in agreement with the theoretical results
, we explain the transition at @xmath2@xmath5@xmath6ev as excitations trough the hybridization induced gap .
the high oscillator strength is due to the parity difference between the initial and the final states .
the large off - diagonal conductivity is possibly owing to the strong spin - orbit coupling for the delocalized electrons with strong selenium character ( e@xmath59@xmath3@xmath6ev ) and the highly spin polarized states in the @xmath601ev vicinity of the fermi level.@xcite the first charge transfer excitations around @xmath2@xmath5@xmath21ev , with a remarkable oscillator strength and magneto - optical activity are attributed to se 4p@xmath20cr 3d transition , while the excitation at @xmath2@xmath5@xmath22ev is assigned to on - site d - d transitions of chromium ions .
these are in overall agreement with the numerical calculations , although the transition energies are somewhat higher in the experiment .
@xcite the contribution of the d - d transition to the off - diagonal conductivity is small compared to that of the charge transfer transition . besides the reduced oscillator strength of the d - d transition due to its dipole forbidden nature ,
it is likely caused by the fairly small spin - orbit coupling for chromium ( e@xmath59@xmath5@xmath61ev ) .
@xcite the measurement of the off - diagonal conductivity greatly helps to extract the optical transition found around @xmath2@xmath5@xmath6ev since its large magneto - optical activity dominates over the contribution from the metallic charge carriers ( damped cyclotron resonance ) contrary to the case of the diagonal conductivity .
we have measured the reflectivity and magneto - optical kerr effect over a broad energy range ( e=0.08 - 26ev and e=0.1 - 4ev , respectively ) at various temperatures down to t=10k and evaluated the elements of the optical conductivity tensor .
the diagonal and off - diagonal optical conductivity spectra determined from the experiments are consistent with the results of former band structure calculations@xcite over the whole energy region . at low energies
a metallic peak is present , while the e@xmath622ev region is dominated by the first charge transfer transition se 4p@xmath20cr 3d .
moreover , we have observed a distinct optical transition around @xmath2@xmath5@xmath6ev that we attribute to excitations through the se - cr hybridization induced gap .
this transition has huge magneto - optical activity due to the high spin polarization in the @xmath601ev vicinity of the fermi level and its parity - allowed nature .
the corresponding sharp feature in the off - diagonal conductivity dominates over the contribution from the metallic electrons .
large enhancement of the kerr effect was also observed around the plasma edge .
landolt - brnstein , _ magnetic and other properties of oxides and related compounds _ , vol iii .
part 12b , springer , berlin ( 1988 ) and the references there .
h. brndle , j. schoenes , p. wachter , f. hulliger and w. reim , j. magn . magn . mat . *
93 * , 207 ( 1991 ) .
a. p. ramirez , r. j. cava and j. krajewski , nature * 386 * , 156 ( 1997 ) .
j. hemberger , p. lunkenheimer , r. fichtl , h .- a .
krug von nidda , v. tsurkan and a. loidl , nature * 434 * , 364 ( 2005 ) .
v. n. antonov , v. p. antropov , b. n. harmon , a. n. yaresko and a. ya .
perlov , phys .
b * 59 * , 14552 ( 1999 ) .
t. saha - dasgupta , molly de raychaudhury and d. d. sarma , phys .
b * 76 * , 054441 ( 2007 ) . y
a. wang , a. gupta , m. chshiev and w. h. butler , appl .
92 * , 062507 ( 2008 ) .
f. k. lotgering and r.p .
stapele , solid state commun . * 5 * , 143 ( 1967 ) . c. colominas , phys
. rev . * 153 * , 558 ( 1966 ) .
j. b. goodenough , j. phys .
solids * 30 * , 261 ( 1969 ) .
a. kimura , j. matsuno , j. okabayashi , a. fujimori , t. shishidou , e. kulatov and t. kanomata , phys .
b * 63 * , 224420 ( 2001 ) .
k. ohgushi , y. okimoto , t. ogasawara , s. miyasaka and y. tokura , j. phys .
* 77 * , 034713 ( 2008 ) . k. sato , jpn .
. phys . * 20 * , 2403 ( 1981 ) .
h. feil and c. haas , phys .
lett . * 58 * , 65 ( 1987 ) .
wei - li lee , s. watauchi , v. l. miller , r. j. cava and n. p. ong , science * 303 * , 1647 ( 2004 ) . y. yao , y. liang , d. xiao , q. niu , shun - qing shen , x. dai , and z. fang , * 75 * , 020401(r ) ( 2007 ) . | diagonal and off - diagonal optical conductivity spectra have been determined form the measured reflectivity and magneto - optical kerr effect ( moke ) over a broad range of photon energy in the itinerant ferromagnetic phase of cucr@xmath0se@xmath1 at various temperatures down to t=10k . besides the low - energy metallic contribution and the lower - lying charge transfer transition at @xmath2@xmath3@xmath4ev ,
a sharp and distinct optical transition was observed in the mid - infrared region around @xmath2@xmath5@xmath6ev with huge magneto - optical activity .
this excitation is attributed to a parity allowed transition through the se - cr hybridization - induced gap in the majority spin channel .
the large off - diagonal conductivity is explained by the high spin polarization in the vicinity of the fermi level and the strong spin - orbit interaction for the related charge carriers .
the results are discussed in connection with band structure calculations . |
the modification of the qcd vacuum , and of its thermal excitations as well , under the influence of external fields , is an attractive topic .
firstly , it is extremely interesting to understand how an external field can modify the main characteristics of confinement and spontaneous chiral symmetry breaking .
secondly , strong magnetic fields , with order of magnitude between @xmath0 and @xmath1 , might be produced in the very first moments of the non - central heavy ion collisions @xcite . in this case
, it has been argued that the non - trivial topological structure of thermal qcd gives rise to chiral magnetic effect ( cme ) @xcite .
an useful approach to the physics of strong interactions in external magnetic fields is the use of some model . among them ,
the nambu - jona lasinio ( njl ) model @xcite is quite popular , see refs .
@xcite for reviews . in this model ,
the qcd gluon - mediated interactions are replaced by effective interactions among quarks , which are built in order to respect the global symmetries of qcd . on the other hand
, the njl model lacks confinement of color .
it is well known that color confinement can be described in terms of the center symmetry of the color gauge group and of the polyakov loop @xcite , which is an order parameter for the center symmetry .
motivated by this property , the polyakov loop extended nambu - jona lasinio model ( p - njl model ) has been introduced @xcite , in which the concept of statistical confinement replaces that of the true confinement of qcd , and an effective interaction among the chiral condensate and the polyakov loop is achieved by a covariant coupling of quarks with a background temporal gluon field . in the literature , there are several studies about various aspects of the p - njl model @xcite .
lattice studies on the response of the qcd ground state to external magnetic and chromomagnetic fields can be found in @xcite .
previous studies of qcd in magnetic fields , and of qcd - like theories as well , can be found in refs .
@xcite . beside the polyakov loop
, it has been suggested @xcite that another observable which is an order parameter for the center symmetry , hence for confinement , is the dressed polyakov loop .
the dressed polyakov loop has been computed in refs .
@xcite within the scheme of truncated schwinger - dyson equations ; within the nambu - jona lasinio model , @xmath2 has been computed at finite temperature and chemical potential in @xcite .
finally , the dressed polyakov loop has been computed within the pnjl model in @xcite at finite temperature , and in @xcite at finite temperature with strong magnetic field . in this talk , i present results obtained in ref .
@xcite about the phase structure and the dressed polyakov loop of hot two massive flavor quark matter at zero chemical potential , in an external magnetic field . to compute the effective potential , i rely on the pnjl model of strongly interacting quarks .
i consider here two flavor quark matter whose lagrangian density is specified as @xcite @xmath3 \nonumber \\ & & + g_8\left[(\bar q q)^2 + ( \bar q i \gamma_5 \bm\tau q)^2\right]^2~. \label{eq : lagr}\end{aligned}\ ] ] the covariant derivative embeds the quark coupling to the external magnetic field and to the background gluon field as well , see below . in eq .
, @xmath4 represents a quark field in the fundamental representation of color and flavor ( indices are suppressed for notational simplicity ) ; @xmath5 is a vector of pauli matrices in flavor space ; @xmath6 is the bare quark mass , which is fixed to reproduce the pion mass in the vacuum , @xmath7 mev .
the model at hand is called polyakov loop extended nambu - jona lasinio model ( pnjl in the following ) , since a coupling of the chiral condensate and the polyakov loop is introduced via the covariant derivative in eq .
, see below . in this study ,
i limit myself to the one - loop approximation for the partition function . in order to couple the polyakov loop to the quark fields , it is customary , in the pnjl model , to introduce a background temporal , static and homogeneous euclidean gluon field , @xmath8 , in terms of which the polyakov loop is given by @xmath9/3 $ ] .
@xmath8 is coupled to the quarks via the covariant derivative , see eq .
( 1 ) ; as a consequence , a coupling among the quark fields and the polyakov loop arises naturally when the integration over fermion fields in the partition function is performed .
i work in the landau gauge , and take the magnetic field homogeneous , static and aligned with the positive @xmath10axe .
moreover , i take twisted fermion boundary conditions along the compact temporal direction , @xmath11~ , \label{eq : phi}\ ] ] while for spatial directions the usual periodic boundary condition is taken . the one - loop thermodynamic potential in the general case of twisted boundary conditions
is given by @xcite @xmath12 in the previous equation , @xmath13 ; @xmath14 is a non - negative integer which labels the landau level ; @xmath15 counts the degeneracy of the @xmath16th landau level .
moreover , @xmath17 , with @xmath18 . the arguments of the thermal exponentials are defined as @xmath19 , with @xmath20 defined in eq . .
the vacuum part of the thermodynamic potential , @xmath21 , is ultraviolet divergent . in this study ,
i use a smooth regularization procedure by introducing a form factor @xmath22 in the diverging zero - point energy . the potential term @xmath23 $ ] in eq .
is built by hand in order to reproduce the pure gluonic lattice data @xcite . among several different potential choices
i adopt the logarithmic form of @xcite .
following @xcite , i introduce the dual quark condensate , @xmath24 where @xmath25 is an integer .
the expectation value @xmath26 denotes the path integral over gauge field configurations .
the case @xmath27 is called the _ dressed _ polyakov loop . for my later convenience ,
i scale the definition of the dressed polyakov loop in eq . , and introduce @xmath28 where @xmath29 corresponds to the expectation value of the @xmath30 field computed keeping twisted boundary conditions for fermions .
dressed polyakov loop ( left panel ) and its effective susceptibility ( right panel ) as a function of temperature and magnetic field strength.,title="fig:",width=302 ] dressed polyakov loop ( left panel ) and its effective susceptibility ( right panel ) as a function of temperature and magnetic field strength.,title="fig:",width=302 ] in the left panel of fig . [ fig : dpm ] , i collect the results for the dressed polyakov loop as a function of temperature and magnetic field strength . in the right panel of the same figure ,
i plot the data of the effective susceptibility , @xmath31 .
the bifurcation of the effective susceptibility at large value of the magnetic field strength arises because the dressed polyakov loop is capable to feel both the deconfinement and the chiral crossovers , which are split of almost @xmath32 if the magnetic field strength is large enough .
phase diagram of the pnjl model in magnetic field . dashed line denotes the polyakov loop crossover ; dot - dashed line corresponds to the chiral crossover .
the shaded area is the region , in the @xmath33 plane , in which quark matter is not statistically confined , but chiral symmetry is still broken by the chiral condensate .
temperatures on the vertical axes are measured in units of the pseudo - critical temperature at zero field , which is @xmath34 mev .
, width=302 ] in figure [ fig : pd7 ] , i collect the results on the pseudo - critical temperatures for chiral and polyakov loop crossovers , in the form of phase diagrams in the @xmath33 plane .
the dashed line denotes the polyakov loop crossover , and the dot - dashed line corresponds to the chiral crossover .
the shaded area is the region , in the @xmath33 plane , in which quark matter is not statistically confined , but chiral symmetry is still broken by the chiral condensate .
temperature on the vertical axes are measured in units of the pseudo - critical temperature at zero field , which is @xmath34 mev .
we fit our data on the pseudo - critical temperatures by the law @xmath35 where @xmath36 , and @xmath34 mev .
the numerical values for @xmath37 are @xmath38 , @xmath39 ; for @xmath40 the numerical coefficients are @xmath41 , @xmath42 .
it is instructive to compare this result with those obtained in a different model .
the shape of the phase diagram drawn in fig .
[ fig : pd7 ] is similar to that drawn by the polyakov extended quark - meson model , see e.g. fig .
. @xcite . in that reference ,
an interpretation of the split in terms of the interplay among vacuum and thermal contribution , is given .
i totally agree with those arguments , which are reproduced within the pnjl model as well , as the results on critical temperatures show .
in this talk , i have reported about the computation of the dressed polyakov loop and of the phase diagram of hot quark matter under the influence of a strong external magnetic field .
the results on the dressed polyakov loop , @xmath2 , in magnetic field show that this quantity is capable to describe both polyakov loop and chiral crossovers .
this is resumed in the double peak structure of its effective susceptibility .
we measure an increase of both deconfinement and chiral crossovers ; the tiny split of the two critical temperatures is of the order of @xmath43 for the largest value of the magnetic field strength considered here .
i acknowledge the organizers of qcd@work10 for the opportunity to give a talk .
i acknowledge r. gatto for his fruitful collaboration in ref .
@xcite , on which this talk is mainly based . moreover , i acknowledge correspondence with m. huang and s. nicotri , and in particular with m. delia .
furthermore , i acknowledge stimulating discussions with l. campanelli and k. fukushima . this work is supported by jsps under the contract number p09028 .
the numerical calculations were carried out on altix3700 bx2 at yitp in kyoto university .
d. e. kharzeev , l. d. mclerran and h. j. warringa , nucl . phys .
a * 803 * , 227 ( 2008 ) [ arxiv:0711.0950 [ hep - ph ] ] .
v. skokov , a. y. illarionov and v. toneev , int .
j. mod .
a * 24 * , 5925 ( 2009 ) [ arxiv:0907.1396 [ nucl - th ] ] .
p. v. buividovich , m. n. chernodub , e. v. luschevskaya and m. i. polikarpov , phys .
d * 80 * , 054503 ( 2009 ) [ arxiv:0907.0494 [ hep - lat ] ] ; m. abramczyk , t. blum , g. petropoulos and r. zhou , arxiv:0911.1348 [ hep - lat ] .
u. vogl and w. weise , prog
. part .
* 27 * , 195 ( 1991 ) ; s. p. klevansky , rev .
* 64 * , 649 ( 1992 ) ; t. hatsuda and t. kunihiro , phys . rept . *
247 * , 221 ( 1994 ) ; m. buballa , phys . rept . *
407 * , 205 ( 2005 ) .
a. m. polyakov , phys .
b * 72 * , 477 ( 1978 ) ; l. susskind , phys .
d * 20 * , 2610 ( 1979 ) ; b. svetitsky and l. g. yaffe , nucl .
b * 210 * , 423 ( 1982 ) ; b. svetitsky , phys .
rept . * 132 * , 1 ( 1986 ) .
k. fukushima , phys .
d * 77 * , 114028 ( 2008 ) [ erratum - ibid .
d * 78 * , 039902 ( 2008 ) ] ; m. ciminale , r. gatto , n. d. ippolito , g. nardulli and m. ruggieri , phys .
d * 77 * , 054023 ( 2008 ) ; w. j. fu , z. zhang and y. x. liu , phys . rev .
d * 77 * , 014006 ( 2008 ) ; t. hell , s. rossner , m. cristoforetti and w. weise , phys .
d * 81 * , 074034 ( 2010 ) .
l. mclerran and r. d. pisarski , nucl .
a * 796 * , 83 ( 2007 ) [ arxiv:0706.2191 [ hep - ph ] ] .
y. sakai , k. kashiwa , h. kouno and m. yahiro , phys .
d * 77 * , 051901 ( 2008 ) ; phys .
d * 78 * , 036001 ( 2008 ) .
y. sakai , k. kashiwa , h. kouno , m. matsuzaki and m. yahiro , arxiv:0902.0487 [ hep - ph ] .
k. kashiwa , h. kouno and m. yahiro , phys .
d * 80 * , 117901 ( 2009 ) .
h. abuki , m. ciminale , r. gatto , n. d. ippolito , g. nardulli and m. ruggieri , phys .
d * 78 * , 014002 ( 2008 ) ; h. abuki , m. ciminale , r. gatto and m. ruggieri , phys . rev .
d * 79 * , 034021 ( 2009 ) [ arxiv:0811.1512 [ hep - ph ] ] .
k. kashiwa , h. kouno , m. matsuzaki and m. yahiro , phys .
b * 662 * , 26 ( 2008 ) [ arxiv:0710.2180 [ hep - ph ] ] .
r. gatto and m. ruggieri , arxiv:1007.0790 [ hep - ph ] .
m. delia , s. mukherjee and f. sanfilippo , arxiv:1005.5365 [ hep - lat ] .
p. v. buividovich , m. n. chernodub , e. v. luschevskaya and m. i. polikarpov , phys .
d * 81 * , 036007 ( 2010 ) [ arxiv:0909.2350 [ hep - ph ] ] .
p. v. buividovich , m. n. chernodub , e. v. luschevskaya and m. i. polikarpov , phys .
b * 682 * , 484 ( 2010 ) [ arxiv:0812.1740 [ hep - lat ] ] .
p. cea and l. cosmai , jhep * 0302 * , 031 ( 2003 ) [ arxiv : hep - lat/0204023 ] ; p. cea and l. cosmai , jhep * 0508 * , 079 ( 2005 ) [ arxiv : hep - lat/0505007 ] .
p. cea , l. cosmai and m. delia , jhep * 0712 * , 097 ( 2007 ) [ arxiv:0707.1149 [ hep - lat ] ] .
s. p. klevansky and r. h. lemmer , phys .
d * 39 * , 3478 ( 1989 ) ; h. suganuma and t. tatsumi , annals phys .
* 208 * , 470 ( 1991 ) ; i. a. shushpanov and a. v. smilga , phys .
b * 402 * , 351 ( 1997 ) [ arxiv : hep - ph/9703201 ] ; d. n. kabat , k. m. lee and e. j. weinberg , phys .
d * 66 * , 014004 ( 2002 ) [ arxiv : hep - ph/0204120 ] ; t. inagaki , d. kimura and t. murata , prog .
theor . phys .
* 111 * , 371 ( 2004 ) [ arxiv : hep - ph/0312005 ] ; t. d. cohen , d. a. mcgady and e. s. werbos , phys . rev .
c * 76 * , 055201 ( 2007 ) [ arxiv:0706.3208 [ hep - ph ] ] .
v. p. gusynin , v. a. miransky and i. a. shovkovy , nucl .
b * 462 * , 249 ( 1996 ) [ arxiv : hep - ph/9509320 ] ; nucl .
b * 563 * , 361 ( 1999 ) [ arxiv : hep - ph/9908320 ] ; g. w. semenoff , i. a. shovkovy and l. c. r. wijewardhana , phys .
d * 60 * , 105024 ( 1999 ) [ arxiv : hep - th/9905116 ] ; v. a. miransky and i. a. shovkovy , phys . rev .
d * 66 * , 045006 ( 2002 ) [ arxiv : hep - ph/0205348 ] .
k. g. klimenko , theor .
phys . * 89 * , 1161 ( 1992 ) [ teor . mat .
89 * , 211 ( 1991 ) ] ; k. g. klimenko , z. phys .
c * 54 * , 323 ( 1992 ) ; k. g. klimenko , theor .
* 90 * , 1 ( 1992 ) [ teor .
* 90 * , 3 ( 1992 ) ] .
n. o. agasian and s. m. fedorov , phys .
b * 663 * , 445 ( 2008 ) [ arxiv:0803.3156 [ hep - ph ] ] ; e. s. fraga and a. j. mizher , phys .
d * 78 * , 025016 ( 2008 ) [ arxiv:0804.1452 [ hep - ph ] ] .
k. fukushima , m. ruggieri and r. gatto , arxiv:1003.0047 [ hep - ph ] .
a. j. mizher , m. n. chernodub and e. s. fraga , arxiv:1004.2712 [ hep - ph ] . l. campanelli and m. ruggieri , phys .
d * 80 * , 034014 ( 2009 ) [ arxiv:0905.0853 [ hep - ph ] ] . | in this talk , i review the computation of the phase diagram of hot quark matter in strong magnetic field , at zero baryon density , within an effective model of quantum chromodynamics . |
a number of observations suggested that phenomena in most active regions in the universe are related to black holes .
some of most active objects in the universe , for example , active galactic nuclei ( agns ) , micro - quasars ( black hole binaries ) , and gamma - ray bursts ( grbs ) , emit relativistic jets @xcite .
it is believed that these relativistic jets are caused by the drastic phenomena around the black holes at the centers of these objects .
the possible energy sources of the drastic phenomena are gravitational energy of the matter falling toward the black hole and rotational energy of the black hole itself .
recently , numerical simulations of general relativistic mhd ( grmhd ) have suggested that the relativistic jet is launched from the vicinity of the black hole , i.e. inside of the ergosphere @xcite , and some long - term simulations showed that the energy is seemed to be supplied from the rotational energy of the black hole @xcite . in these grmhd simulations
, the black hole rotational energy seems to be extracted through the magnetic field flux tubes due to the so called blandford - znajek mechanism @xcite .
it was proposed as the mechanism in the force - free condition , by which the rotational energy of the black hole is extracted directly through the horizon along the magnetic flux tubes . however , in principle , causality prohibits the outward transportation of any material , energy , and information across the horizon .
thus , as pointed out by @xcite , the blandford - znajek mechanism seems to be contradictory to the causality . on the other hand , in the penrose process
, the black hole spin energy is extracted causally due to the negative energy - at - infinity ( or just called energy " ) of a particle caused by fission @xcite . @xcite and
@xcite found the axisymmetric steady - state solution of ideal mhd plasma inflow with the negative energy toward the rotating black hole .
when the negative energy of the inflowing plasma in the ergosphere is swallowed by the black hole , the black hole rotation energy decreases , that is , the black hole energy is extracted , just like the penrose process .
the difference between the ideal mhd mechanism and the penrose process is that the negative energy is produced by the magnetic tension force in the ideal mhd inflow , while in the penrose process it is caused by the fission of a particle .
this mhd energy extraction mechanism is called mhd penrose process " ( see table [ clasmech ] ) .
the mhd penrose process was mimicked and confirmed by the grmhd simulations of initially uniform , very strongly magnetized plasma around a rapidly rotating black hole , which showed that the negative energy of plasma is produced quickly in the ergosphere @xcite . however , because of the short time duration of the simulation , the numerical solution is far from a stationary state .
@xcite performed a long term grmhd simulations with the similar initial situation of koide ( 2003 ) , and confirmed the mhd penrose process in the early stage .
furthermore , he found that the mhd penrose process is a transient phenomenon and alternately the outward electromagnetic energy flux through the horizon stationarily appears almost everywhere with the exception of a very thin equatorial belt .
he remarked that the pure electromagnetic mechanism with ideal mhd condition continues to operate to extract the rotational energy of the black hole . strictly speaking
, this electromagnetic mechanism should be distinguished from the original blandford - znajek mechanism because the original mechanism is derived with the force - free condition , while the electromagnetic energy extraction mechanism was shown with the ideal mhd simulations . in this paper
, we call the mechanism shown by the simulations mhd blandford - znajek mechanism " while the original mechanism is called force - free blandford - znajek mechanism " . considering the numerical results
, @xcite discussed the electromagnetic extraction mechanism of the black hole energy , including the force - free blandford - znajek mechanism , mhd penrose mechanism , and superradiance in the wide view .
however , unfortunately , the convincing explanation with respect to the causality of these mechanisms , which should also yield the conditions of the mechanisms , is not given except for the mhd penrose process @xcite .
@xcite pointed out that the force - free blandford - znajek mechanism uses the _ negative electromagnetic energy - at - infinity _ to extract the spin energy of the black hole .
this point of view was discussed extensively by @xcite and @xcite for ideal mhd and force - free blandford - znajek mechanisms , respectively . however , it is often difficult to build physical intuition on the mhd / force - free blandford - znajek mechanisms with causality . here , we present an intuitive formula for the electromagnetic mechanism of the energy extraction from the rotating black hole to aid in building the physical intuition on the mechanisms .
the formula is also applicable to other electromagnetic mechanisms like the mhd penrose process @xcite and super - radiance @xcite . in section [ sec2 ] ,
we review the energy and angular momentum transport of electromagnetic field around the black holes briefly but sufficiently . in section [ sec3 ]
, we explain the electromagnetic mechanisms of the black hole energy extraction , that is , the force - free blandford - znajek mechanism , mhd blandford - znajek mechanism , and superradiance within causality .
we summarize our explanation about the energy extraction mechanisms from the black hole including the both blandford - znajek mechanisms in section [ sec4 ] .
we review the electromagnetic energy and angular momentum transport in the space - time @xmath0 around a spinning black hole based on so called 3 + 1 formalism " .
the scale of a small element in the space - time around the rotating black hole is given by @xmath1 .\ ] ] here , we have @xmath2 , @xmath3 , @xmath4 , @xmath5 , where greek indices @xmath6 run from 0 to 3 and roman indices @xmath7 run from 1 to 3 . through this paper
, we use the natural unit system , where the light speed , electric permittivity , and magnetic permeability in vacuum are unity : @xmath8 , @xmath9 , and @xmath10 .
when we define the lapse function @xmath11 and shift vector @xmath12 by @xmath13 the line element @xmath14 is written as @xmath15 the determinant of the matrix with elements @xmath16 is given by @xmath17 , and the contravariant metric is written explicitly as @xmath18 and @xmath19 where @xmath20 is the kronecker @xmath21 symbol .
the relativistic maxwell equations are @xmath22 where @xmath23 is the covariant derivative , @xmath24 is the electromagnetic field - strength tensor , and @xmath25 is the dual tensor of @xmath24 , @xmath26 ( @xmath27 is the levi - civita anti - symmetric tensor , which is a tensor density of weight -1 ) , and @xmath28 is the electric 4-current density ( @xmath29 is the electric charge density)@xcite .
the electric field @xmath30 and the magnetic field @xmath31 are given by @xmath32 @xmath33 and @xmath34 , @xmath35 , @xmath36 or @xmath37 , respectively . using the 4-vector potential @xmath38 ,
we have @xmath39 because of the symmetry of christoffel symbols , @xmath40 . the electromagnetic energy - momentum tensor @xmath41 is given by @xmath42 the total energy - momentum tensor @xmath43 is @xmath44 where @xmath45 , @xmath46 , and @xmath47 are the proper pressure , the proper enthalpy density , and the 4-velocity of the plasma , respectively .
the energy momentum conservation law is given by @xmath48 the force - free condition is @xmath49 and the general relativistic ohm s law is @xmath50 , \label{ohmlaw}\ ] ] where @xmath51 is the resistivity of the plasma .
the ideal mhd condition is given by setting @xmath52 , @xmath53 . here
, we introduce a local coordinate frame , so called `` fiducial observer ( fido ) frame '' , @xmath54 . using the local coordinates of the frame @xmath55
, the line element becomes @xmath56 where @xmath57 is the metric of minkowski space - time . comparing this metric with eq .
( [ eqlinelement ] ) , we get @xmath58 and we have partial derivative relations , @xmath59 then , a contravariant vector @xmath60 in the fido frame of an arbitrary contravariant vector @xmath61 in the global coordinates @xmath62 is written as , @xmath63 and the covariant vector @xmath64 is @xmath65 we use the quantities observed by the fido frame because they can be treated intuitively and yield formulae more easily .
this is because the relations between the variables in the fido frame are the same as these in the special theory of relativity and similar to the newtonian relation . using the quantities of electromagnetic field in the fido frame ,
maxwell equations are written by the following 3 + 1 formalism , @xmath66 , \label{cmfa}\ ] ] @xmath67 , \label{cmam}\ ] ] @xmath68 @xmath69 where @xmath70 .
for the convenience , we introduce the derivatives of arbitrary a three - vector field @xmath71 and an arbitrary scalar field @xmath72 measured by the fido frame as @xmath73 @xmath74 @xmath75 we express maxwell equations in vector forms as , @xmath76 , \label{vecfa}\ ] ] @xmath77 , \label{vecam}\ ] ] @xmath78 @xmath79 where @xmath80 , @xmath81 , @xmath82 , and @xmath83 . + the 3 + 1 form of the force - free condition is @xmath84 and ohm s law is written by @xmath85 , \label{vecimhd}\ ] ] where @xmath86 is the lorentz factor , @xmath87 is the 3-velocity , and @xmath88 is the electric charge density observed by the plasma - rest frame ( the proper electric charge density ) .
the conservation equation of the electric charge is derived by eqs .
( [ vecam ] ) and ( [ vecec ] ) as @xmath89 = 0 . \label{vecformchrgcon}\ ] ] we present the equations of energy and angular momentum conservation around a spinning black hole . when @xmath90 is a killing vector , we have an conservation law associated with eq .
( [ eqenmo ] ) @xmath91 because of @xmath92 , this equation yields @xmath93 using the killing vector @xmath94 , we have the conservation law of energy @xmath95 where @xmath96 is called _ energy - at - infinity _ ( or just `` energy '' ) density and @xmath97 is the @xmath98-th component of energy flux density . here
, we also express these quantities in the fido frame as @xmath99 @xmath100 , \ ] ] where @xmath101 , @xmath86 , and @xmath102 .
we separate these quantities into the hydrodynamic and electromagnetic components : @xmath103 @xmath104 where @xmath105 @xmath106 @xmath107 @xmath108^i , \ ] ] where the subscripts ` hyd ' and ` em ' indicate hydrodynamic and electromagnetic components , respectively , and @xmath109 , @xmath110 . here
, @xmath111 can be regarded as the poynting vector . the general relativistic maxwell equations ( [ vecfa])-([vecec ] ) yield @xmath112 where @xmath113 is the lorentz force density .
if @xmath114 is the killing vector for the azimuthal direction , we have the equation of angular momentum conservation @xmath115 where @xmath116 and @xmath117 are the total angular momentum density and the angular momentum flux density , respectively . using the quantities measured in the fido frame
, we have @xmath118 @xmath119 these variables also can be divided into the hydrodynamic and electromagnetic components , denoted by the subscripts ` hyd ' and ` em ' , as follows : @xmath120 @xmath121 where @xmath122 @xmath123 @xmath124 , \ ] ] @xmath125 .\ ] ] in this case , from eqs .
( [ apeai ] ) and ( [ apam ] ) , we have a relation of the energy and the angular momentum , @xmath126 , \label{releaianm}\ ] ] when @xmath127 .
furthermore , eqs ( [ eaikin ] ) , ( [ eaiem ] ) , ( [ amkin ] ) , ( [ amem ] ) yield @xmath128 @xmath129 the general relativistic maxwell equations ( [ vecfa])-([vecec ] ) read @xmath130 from now on , we consider the electromagnetic energy transport when we have the relation between the electric field and magnetic field as @xmath131 here , @xmath132 is a certain vector field and does not always mean the real velocity , while in the ideal mhd case , it is identified by the plasma velocity @xmath133 .
it is noted that the drift velocity due to the electric field @xmath134 , @xmath135 can be used as one of vector of @xmath132 .
intuitively , @xmath132 is regarded as the velocity of the magnetic field lines , while this intuition is not rigorous because we can not identify the magnetic field lines at the different times .
however , we have not so serious contradiction with the interpretation and we often recognize @xmath132 as the velocity of the field line implicitly . using eq .
( [ eandb ] ) , we have @xmath136 ( \hat{b})^2 , \label{eeminf}\end{aligned}\ ] ] where @xmath137 is the component of @xmath132 perpendicular to the magnetic field @xmath138 , @xmath139 , @xmath140 , @xmath141 . here
, we used the relations , @xmath142 , @xmath143 .
with respect to the energy transport flux density , we have @xmath144 . \label{semeq}\ ] ] using eqs .
( [ eeminf ] ) and ( [ semeq ] ) , we obtain @xmath145 with respect to the angular momentum of the electromagnetic field , assuming eq .
( [ eandb ] ) , we have @xmath146 .\end{aligned}\ ] ] in this case , we also have @xmath147
in this subsection , we consider the energy transport near the horizon in the force - free limit case , which is assumed in the original work of blandford - znajek mechanism @xcite . here
, we use the kerr metric for space - time @xmath148 with @xmath149 in this section .
the condition of force - free , @xmath150 , reads @xmath151 this means no energy and momentum transforms between the electromagnetic field and plasma . in such a case , we can write the electromagnetic field by eq .
( [ eandb ] ) . this is because when @xmath152 , we have @xmath153 , and confirm eq .
( [ eandb ] ) with @xmath154 .
when @xmath155 , we have @xmath156 , that is @xmath157 . furthermore , because of eq .
( [ ffcone ] ) , we have @xmath158 , and then @xmath159 . we confirm eq .
( [ eandb ] ) with @xmath160 . in the steady - state and axisymmetry case ,
( [ cmfa ] ) , ( [ divb0 ] ) , ( [ ffcone ] ) , and ( [ ffconj ] ) yield @xmath161 where @xmath162 is a constant along the magnetic flux surface , @xmath163 corresponds to the distance from the @xmath164 axis , @xmath165 is the unit vector for azimuthal direction @xcite . because the triangle of @xmath137 and @xmath132 and the triangle of @xmath166 and @xmath138 are similar ( fig .
[ bzvfperp])we found the following relation , @xmath167 here , we define @xmath166 and @xmath168 as the poloidal and azimuthal components of magnetic field @xmath138 , respectively .
then , we have @xmath169 the znajek boundary condition at the horizon @xcite is expressed as @xmath170 then , very near the horizon , we also have @xmath171 where
@xmath172 " means asymptotic equivalence . in the limit toward the horizon ( @xmath173
, @xmath174 is the radius of the black hole ) , we have @xmath175 when @xmath176 , and then we found @xmath177 . here , we write the value of @xmath178 at the horizon by @xmath179 . eventually , using eq .
( [ frm4sem ] ) we obtain very near the horizon , @xmath180 the directions of @xmath132 of the cases of @xmath181 and @xmath182 are opposite because of eq .
( [ vhf2hz ] ) , and the slope of the magnetic field lines in the two cases are also opposite ( fig .
[ bh_horizon ] ) .
then , the direction of @xmath137 is always directed toward the black hole inner region when @xmath176 .
then , when @xmath183 , the electromagnetic energy flux is directed outward and the energy of the black hole is extracted through the horizon .
next , we determine the condition of the negative energy @xmath183 at the horizon .
when @xmath176 , @xmath184 is directed toward the black hole horizon in both cases of @xmath181 and @xmath182 . because the triangle of @xmath137 , @xmath132 and the triangle of @xmath166 , @xmath138 are similar ( fig .
[ bzvfperp ] ) , we found @xmath185 , and then we obtain @xmath186 finally , using the second equation of eq .
( [ diff_alpbet ] ) and eq .
( [ vhf2hz ] ) we get @xmath187 \alpha \hat{b}^2 \\ = \frac{1}{2 } \frac{\alpha^2 + 2 r^2 \omega_{\rm f } ( \omega_{\rm f } - \hat{\omega}_\phi)}{\alpha^2 + r^2 ( \omega_{\rm f } - \hat{\omega}_\phi)^2 } \alpha \hat{b}^2.\ ] ] at the horizon , ( @xmath188 @xmath189 ) , eqs . ( [ vhf2hz ] ) and ( [ znajekcond ] ) yield @xmath190 because @xmath191 , where @xmath192 is the value of @xmath193 at the horizon .
eventually , at the horizon , we found @xmath194 where @xmath195 is the value of @xmath196 at the horizon .
it is noted that the radial component of the electromagnetic energy flux is identical to the simple equation given by @xcite ( eq . ( 34 ) in the paper ) , if we set the force - free condition at the horizon , @xmath197 .
then , when @xmath198 , the negative energy of the electromagnetic field is realized ( @xmath183 ) and the rotation energy of the black hole is extracted .
this is exactly the same condition of the blandford - znajek mechanism .
this suggests even in the blandford - znajek mechanism , to extract the black hole rotational energy , the negative energy of the electromagnetic field is utilized as a mediator . in conclusion , putting the negative electromagnetic energy into the black hole , the black hole rotational energy is extracted causally in the blandford - znajek mechanism .
sometimes the energy extraction of the rotating black hole is intuitively explained by the torque of magnetic field at the horizon .
this intuitive explanation is not appropriate with respect causality . because at the horizon no torque affects the matter and field outside of the horizon from these inside of the horizon .
( [ form2eemi ] ) and ( [ form2sem ] ) suggest that the falling - down of the negative ( electromagnetic ) energy into the black hole could decrease the black hole energy to extract the black hole energy .
we consider the ideal mhd case in the space - time around the spinning black hole .
we assume the situation is stationary and axisymmetric as the same as the force - free case in the previous section .
in such a case , the magnetic flux surfaces are stationary and axisymmetric and are expressed as constant azimuthal component of vector potential , @xmath199 .
we introduce the new coordinate system @xmath200 , where @xmath201 is the time of kerr space - time , @xmath202 is the azimuthal coordinate , @xmath203 , and the coordinate @xmath204 is set outwardly along the intersection line of a magnetic surface and the meridian plane ( @xmath202= const.)(fig .
[ idealmhdcase ] ) . here , we set the coordinates @xmath204 so that it is perpendicular to the coordinate @xmath205 .
the @xmath204 coordinate at the horizon is @xmath206 .
essentially , this coordinate system corresponds to the boyer - linquist coordinate @xmath207 where @xmath208 , @xmath209 , @xmath210 , and @xmath211 .
then , the length of a line element in the space - time of the rotating black hole is given by @xmath212 we assume the ideal mhd condition , @xmath213 , which yields @xmath214 using the coordinates @xmath215 , eqs .
( [ cmfa])([dive ] ) , ( [ coen ] ) , ( [ coam ] ) , and ( [ idealmhdcond ] ) yield the following conservation variables along the magnetic surface : @xmath216 , \label{constwf}\\ l(\psi ) & = & h_\phi \left [ \frac{\mathfrak h}{\rho } \hat{u}^\phi - \frac{\alpha}{\dot{m } } \hat{b}^\phi \right ] , \label{constl } \\
h(\psi ) & = & \frac{\mathfrak h}{\rho } [ \alpha \hat{\gamma } - h_\phi ( \omega_{\rm f } - \omega_{\phi } ) ] = \frac{\mathfrak h}{\rho } \alpha ( \hat{\gamma } - \hat{v}^\phi_{\rm f } \hat{u}^\phi ) .
\label{consth}\end{aligned}\ ] ] it is noted that quantities with hats are variables observed by the fido frame .
it is also noted that the distribution of @xmath205 is determined by the transverse equation called the grad - shafranov equation"@xcite .
recently , numerical simulations of grmhd provide the more complete feature of the mechanism like the distribution of poynting flux over the event horizon , the relative importance of negative energy - at - infinity fluid and electromagnetic field , the energy flux from the black hole to the disk through the magnetic field lines , etc .
it is noted that the numerical , time - dependent simulations showed that magneto - rotational instability ( mri ) always causes fluctuations and no steady state of plasma and magnetic field is found . at the black hole horizon ,
the lapse function @xmath11 becomes 0 , @xmath217 becomes infinite , while @xmath218 , @xmath219 , @xmath220 are finite except on the @xmath164 axis .
hereafter , we discuss the quantities along a certain fixed magnetic flux surface @xmath221 .
because the horizon is not a real singular surface , and the density @xmath222 and pressure @xmath45 are measured by the plasma rest frame , @xmath222 and @xmath45 should be finite at the horizon .
then , from eqs .
( [ constmd ] ) and ( [ constbs ] ) , @xmath223 and @xmath224 must be finite at the horizon , where we write @xmath224 at the horizon by @xmath225 . at the horizon
, the plasma falls vertically to the horizon at the light velocity , is finite , because @xmath226 is finite .
then , from eq .
( [ constl ] ) , @xmath227 is finite . at the horizon , because @xmath223 is finite and @xmath188 , @xmath228 and @xmath229 are infinite .
then , @xmath230 becomes zero at the horizon and @xmath231 .
finally , at the horizon , @xmath232 .
] @xmath233 , @xmath234 , and second equation of eq .
( [ diff_alpbet ] ) and eq .
( [ constwf ] ) yield @xmath235 using eqs .
( [ idealmhdcond ] ) and ( [ constwf ] ) , we have @xmath236 where we put @xmath237 and @xmath165 , @xmath238 are the unit base vectors along the @xmath202 and
@xmath204 coordinates , respectively . very near the horizon , we have @xmath239 eqs .
( [ bnd2hz ] ) and ( [ eqvf ] ) present the geometrical disposition of vectors @xmath138 and @xmath132 , as shown in fig .
[ bh_horizon ] .
when @xmath176 , we found that the vector of @xmath240 is always directed toward the black hole inner region .
intuitively , at the horizon of the rotating black hole , the plasma falls into the black hole radially with the speed of light ( @xmath241 , @xmath242 at @xmath243 ) .
when the azimuthal component of magnetic field is finite outside of the horizon and stationary , the magnetic field lines are twisted extremely strongly near the horizon in appearance because of eq .
( [ constwf ] ) where @xmath244 is uniform along the magnetic surface and @xmath11 vanishes at the horizon .
this is due to difference in the lapse of time near the black hole and is apparent feature in the kerr metric .
in such case , the perpendicular component of the velocity to the magnetic field is identical to the plasma velocity and then we have @xmath245 at the horizon . and @xmath132 and the triangle of @xmath246 and @xmath138 ( fig .
[ bzvfperp ] ) , we found @xmath247 , where @xmath248 and @xmath249 .
very near the horizon , because @xmath250 is much larger than @xmath224 , we have @xmath251 and then @xmath252 . using eqs .
( [ bnd2hz ] ) and ( [ eqvf ] ) , at the horizon we confirm @xmath253 . ]
then , the electromagnetic energy flux density at the horizon is given by @xmath254 from eq .
( [ frm4sem ] ) .
when @xmath255 becomes negative at the horizon , the electromagnetic energy is transported outwardly through the horizon when @xmath176 , because @xmath137 is always directed inwardly toward the black hole inner region ( see fig .
[ bh_horizon ] ) . here ,
because @xmath137 vanishes if @xmath256 , no electromagnetic output is expected , then we consider only the case of @xmath176 case .
as shown in eq .
( [ eeminf ] ) , the electromagnetic energy - at - infinity density is given by @xmath257 ( \hat{b})^2 .\ ] ] with fig .
[ bzvfperp ] , we found @xmath258 , and then we have @xmath259 . at the horizon , using eqs .
( [ eqvf ] ) and ( [ eqvfperp ] ) , we have @xmath260 using eq .
( [ bnd2hz ] ) , we also have @xmath261 at the horizon because of @xmath262 . eventually , we obtain @xmath263 this clearly shows that when @xmath198 , @xmath255 becomes negative and the electromagnetic energy flux directs outward through the horizon .
it is surprising that not only the condition of the electromagnetic energy extraction from the black hole but also the expression of energy density and the energy flux density at the horizon are the same as those of the blandford - znajek mechanism ( force - free case ) . in the above two cases of electromagnetic extraction of the black hole rotational energy , the negative electromagnetic energy - at - infinity
is required as a mediator to extract the black hole rotational energy through the horizon causally . as shown in eq .
( [ eeminfang ] ) , we have @xmath264 , where @xmath265 is the electromagnetic energy density in the fido frame . to realize the negative electromagnetic energy , the angular momentum of the electromagnetic field @xmath266 should become less than @xmath267 .
locally the angular momentum should be conserved because of eq .
( [ coam ] ) and then redistribution of the angular momentum is required . in the penrose process ,
fission of a particle is utilized for redistribution of the angular momentum and production of a particle with negative energy - at - infinity .
equation ( [ tram ] ) indicates that dynamically only the magnetic force ( the magnetic tension in the axisymmetric case ) and the lorentz force can redistribute the electromagnetic angular momentum . in the ideal mhd case ,
magnetic tension plays an important role to redistribute the electromagnetic angular momentum and realize the negative electromagnetic energy .
this mechanism of energy extraction with negative electromagnetic energy is often confused with the ( original ) blandford - znajek mechanism , where the force - free condition is used , as we did in section [ sec1 ] . however , strictly speaking , they should be distinguished . from now on , in this paper , we call the ideal mhd process with the negative electromagnetic energy mhd blandford - znajek mechanism " . in the mhd blandford - znajek mechanism , we have to take the hydrodynamic energy flux of the plasma flow into account to discuss the net energy flux from / into the black hole .
in fact , in the ideal mhd case , the black hole rotational energy can be also extracted with the negative _ hydrodynamic _ energy of the plasma .
the hydrodynamic energy flux density is @xmath268 then , near the horizon if the plasma with @xmath269 falls into the black hole , the energy is transported outwardly through the horizon because @xmath188 at the horizon . if @xmath270 is directed outward , @xmath271 must be smaller than zero to extract the black hole rotational energy .
this extraction mechanism of black hole rotational energy is called `` mhd penrose process '' @xcite .
the hydrodynamic energy is given by @xmath272 where @xmath273 and @xmath274 is the hydrodynamic angular momentum density .
to realized the negative hydrodynamic energy , @xmath275 .
the angular momentum is conserved and the redistribution of hydrodynamic angular momentum is also required .
the redistribution of hydrodynamic angular momentum is caused by the lorentz force shown in eq .
( [ tram ] ) . to distinguish the mhd blandford - znajek mechanism and mhd penrose process
, we should observe the electromagnetic and hydrodynamic energy - at - infinity density ( @xmath255 and @xmath276 ) .
if the electromagnetic energy plays a main role to extract the black hole energy , we recognize the process as the mhd blandford - znajek mechanism . on the other hand
, the hydrodynamic or plasma energy has an important role for the extraction , it is recognized as the mhd penrose process . in the real cases , both of them are possible , while some long - term simulations indicate that mhd penrose process is transient and the mhd blandford - znajek mechanism is dominant in the later phase of the simulations @xcite
. the electromagnetic extraction mechanisms of black hole rotational energy picked up in this paper are restricted to those in the steady - state , axisymmetric cases .
recently , the long term grmhd simulations showed 3-d dynamics of plasma interacting with the magnetic field around the rotating black hole @xcite .
strictly speaking , the results of this paper are not applicable to the time - dependent , axiasymmetric numerical results .
the generalization of the results of this paper for such time - dependent , axiasymmetric numerical results is required .
we mention the electromagnetic wave energy transport through the horizon briefly .
we use the kerr metric for the space - time @xmath148 around the spinning black hole , where we set @xmath149 .
we consider the stationary solution of the electromagnetic wave in the vacuum , where each component of the electromagnetic field is proportional to @xmath277 ( @xmath278 is a function of @xmath279 and @xmath280 ) .
we use the short wavelength limit of the electromagnetic wave , @xmath281 ( @xmath282 , @xmath283 ) , where @xmath284 is the wavenumber of the electromagnetic wave in a local region , which is fixed at the global coordinates . in a vacuum ( @xmath285@xmath286 ) , eqs .
( [ vecfa])([vecec ] ) in the fido frame yield @xmath287 where @xmath288 and @xmath289 are the wave number and angular frequency of the electromagnetic wave in the fido frame .
these equations read the dispersion relation , @xmath290 and the relation @xmath291 . in this case
, we identify @xmath292 because of @xmath293 , using eq .
( [ frm4sem ] ) , we have @xmath294 when electromagnetic wave passes through the horizon and enters into the black hole , if @xmath295 is negative , the rotational energy of the black hole decreases . in this case , eqs .
( [ angmome2andb ] ) , ( [ eeminfang ] ) , and ( [ vfperpideal ] ) read @xmath296 very near the horizon , we have @xmath297 . because the 4-wavenumber @xmath298 is the covariant vector , using eq .
( [ covarianttransform ] ) , we have @xmath299 then , the energy density of the electromagnetic wave very near the horizon is approximately given by @xmath300 when @xmath301 , the negative energy at the horizon appears and the rotational energy of the black hole is extracted .
this extraction mechanism corresponds to the superradiance " . to produce the negative energy of the electromagnetic wave ,
the redistribution of the angular momentum is required . to understand the redistribution process , we have to consider the structure of the solution of the electromagnetic wave in the ergosphere .
in this paper , we showed simple formulae ( eqs .
( [ frm4sem ] ) and ( [ eeminfang ] ) ) to aid in building physical intuition on the causal extraction mechanism of the black hole energy by the electromagnetic fields with the negative electromagnetic energy produced in the ergosphere . in three cases of
force - free , ideal mhd conditions and electromagnetic wave in vacuum , at the horizon we found that @xmath302 and then we have @xmath303 .
to extract the black hole rotational energy causally , we have to put the negative electromagnetic energy down into the black hole through the horizon . to produce the negative electromagnetic energy , because of the angular momentum conservation ( [ eeminfang ] )
, we should redistribute the angular momentum of the electromagnetic field , where we require the negative electromagnetic angular momentum density , @xmath304 at the horizon ( see eq .
( [ eeminfang ] ) ) . to realize the negative angular momentum azimuthal component ,
the angular momentum should be redistributed because the total angular momentum is conserved .
the redistribution of the angular momentum of the electromagnetic field is caused by the electromagnetic torque @xcite .
this point of view is originated on the penrose process @xcite , which uses negative mechanical energy of a particle .
in fact , equations of the energies of matter and electromagnetic field have similar forms as shown in eqs .
( [ eaikinang ] ) and ( [ eaiemang ] ) . with the viewpoint , in general , we classify the known mechanisms of energy extraction from the black hole as shown in table [ clasmech ] .
the penrose process is well known and is shortly mentioned in section [ sec1 ] .
the blandford - znajek mechanism , mhd blandford - znajek mechanism , and the mhd penrose process were explained in the previous sections .
we showed that in all electromagnetic mechanisms of the energy extraction from the spinning black hole , the negative electromagnetic energy is utilized as a mediator for the causal energy extraction through the horizon .
we confirmed that the condition of the energy extraction is given by the realization condition of the negative energy at the horizon .
the magnetic penrose process was not discussed in this paper . in the magnetic penrose process
, a particle interacts with the electromagnetic field and falls to the negative energy orbit .
the negative energy of the particle is used to extract the black hole rotational energy .
this is just the penrose process with the electromagnetic interaction instead of fission .
the superradiance was mentioned in subsection [ superradiance ] .
we found the electromagnetic wave with negative energy is used to extract the black hole rotational energy .
we also add the energy extraction mechanism with the magnetic reconnection in the ergosphere in table [ clasmech ] @xcite .
we discuss the coincidence of the formulae of the energy density and the energy flux density of the electromagnetic field at the horizon for the force - free and mhd blandford - znajek mechanisms as shown by eqs .
( [ form2eemi ] ) , ( [ form2sem ] ) , and ( [ eemhzmhd ] ) , ( [ semhzmhd ] ) in sections [ ffefbz ] and [ mhdbz ] , although the conditions of the two mechanisms are different . on a posteriori reasoning
, we have the coincident expressions of the electric field @xmath305 , @xmath306 in the assumption of stationary , axisymmetric conditions for the both cases .
furthermore , we have the coincident boundary condition at the horizon @xmath307 and @xmath308 for the both cases . these leading equations for the both cases
are the same and then we have the coincident formulae for the both mechanisms . here
, we remark on the overlap of the ideal mhd and force - free conditions .
the conditions of ideal mhd ( eq .
( [ idealmhdcond ] ) ) and force - free ( eq .
( [ ffconj ] ) ) can both be satisfied if @xmath309 and @xmath310 , where @xmath311 is a vector parallel to the magnetic field @xmath138 .
the vector @xmath311 corresponds to the net current density along the magnetic field lines at the plasma rest - frame .
alternatively , in ideal mhd simulations , the force - free " condition is often defined by @xmath312 even if @xmath313 is not parallel to @xmath138 . in the astrophysical situation like agns ,
which mechanism is mostly expected to extract the black hole rotational energy and activate the region near the black hole ?
we think the mhd blandford - znajek mechanism is most promising process rather than the original blandford - znajek mechanism .
because the plasma near the black hole is expected to be relativistically hot , the plasma beta @xmath314 never vanishes . of course , the original blandford - znajek mechanism is applicable as an approximation with respect to the very strong magnetic field case .
such very low plasma beta is expected at the higher - latitude of the black hole magnetosphere and the fast component of a relativsitc jet .
lllll penrose process & mechanical energy of particle & force of particle fission & mechanical energy of particle & @xcite + & & & & @xcite + & electromagnetic energy & & electromagnetic energy & @xcite + & electromagnetic energy & & & + mhd penrose process & mechanical energy of plasma & & & + & mechanical energy of plasmoid & & mechanical energy of plasmoid & @xcite + superradiance & & & & + beskin , v. s. , & kuznetsova , i. v. 2000 , il nuovo cimento b , * 115 * , 795 biretta , j. a. , sparks , w. b. , & macchetto , f. 1999 , , * 520 * , 621 blandford , r. d. & znajek , r. 1977 , , * 179 * , 433 gammie , c. f. , shapiro , s. l. , & mckinney , j. c. 2004 , , * 602 * , 312 hawley , j. f. & krolik , j. h. 2006 , , * 641 * , 103 hirotani , k. , takahashi , m. , nitta , s .- y . , & tomimatsu , a. 1992 , , * 386 * , 455 jackson , j. d. 1979 , _ classical electrodynamics _ ( john wiley & sons , new york ) .
koide , s. , shibata , k. , kudoh , t. , & meier , d. l. 2002 , science , * 295 * , 1688 koide , s. 2003 , phys . rev .
d. * 67 * , 104010 koide , s. 2004 , , l45 koide , s. , kudoh , t. , & shibata , k. 2006 , , * 74 * , 044005 koide , s. 2009 , , * 696 * , 2220 komissarov , s. s. 2005 , , * 359 * , 801 komissarov , s. s. 2009 , j. korean physical society , * 54 * , 2503 krolik , j. h. , hawley , j. f. , & hirose , s. 2005 , , * 622 * , 1008 kulkarni , s. r. 1999 , nature * 398 * , 389 lasota , j .- p . , gourgoulhon , e. , abramovicz , m. tchekhovskoy , a. & narayan , r. 2014 , , * 89 * , 024041 .
lightman , a. p. , press , w. h. , price , r. h. , & teukolsky , s. a. 1975 , _ problem book in relativity and gravitation _ , ( princeton univ . press , princeton ) .
mckinney , j. c. 2006 , , * 368 * , 1561 mckinney , j. c. & gammie , c. f. 2004 , , * 611 * , 977 mckinney , j. c. , & tchekhovskoy , a. & blandford , r. d. 2012 , , * 423 * , 3083 miniutti , g , et al .
2007 , , * 59 * , 351 mirabel , i. f. , & rodriguez , l. f. 1994 , nature * 374 * , 141 .
pearson , t. j. , & zensus , j. a. 1987 , in _ superluminal radio sources _ , edited by j. a. zensus and t. j. pearson ( cambridge univ . ,
london ) , p. 1 .
r. 1969 , nuovo cimento * 1 * , 252 press , w. h. , teukolsky , s. a. 1972 , nature , * 238 * , 211 punsly , b. & coroniti , f. v. 1989 , phys . rev .
d , * 40 * , 3834 punsly , b. , coroniti , f. v. 1990a , , * 350 * , 518 .
punsly , b. , coroniti , f. v. 1990b , , * 354 * , 583 .
takahashi , m. , nitta , s. , tatematsu , y. , & tomimatsu , a. 1990 , , * 363 * , 206 teukolsky , s. a. & press , w. h. 1974 , , * 193 * , 443 tingay , s. j. _ et al . _ , 1995 , nature * 374 * , 141 wagh , s. m. & dadhich , n. 1989 , physics reports , * 183 * , 137 znajek , r. 1977 , , * 179 * , 457 | recent general relativistic mhd simulations have suggested that relativistic jets from active galactic nuclei ( agns ) have been powered by rotational energy of central black holes . some mechanisms of extraction of black hole rotational energy have been proposed , like the penrose process , blandford - znajek mechanism , mhd penrose process , and superradiance .
the blandford - znajek mechanism is the most promising mechanism for the engines of the relativistic jets from agns .
however , an intuitive interpretation of this mechanism with causality is not clarified yet , while the penrose process has a clear interpretation for the causal energy extraction from the black hole with negative energy . in this paper
, we present a formula to build physical intuition so that in the blandford - znajek mechanism as well as other electromagnetic processes , negative electromagnetic energy plays an important role to extract the rotational energy of the black holes causally . |
v821cas ( bd + 52@xmath11 3571 , tycho 4001 - 1445 - 1 , [email protected] , ( b - v)=0.11 ) was discovered to be an eclipsing binary of early a spectral type , with a variability period of @xmath131.8 days by the hipparcos satellite ( esa , 1997 ) .
deirmenci et al .
( 2003 , 2007 ) determined an improved light curve and listed 25 photographic times of minima from which they derived an improved eclipse ephemeris .
they obtained _ bvr _ light curves , and established that the orbit is eccentric from the displacement of the secondary minimum .
its components do not show intrinsic variability , they are well - separated , and well within their roche lobes , which values them proper tests for stellar models .
several times of minima for v821cas have appeared in the literature since , but no radial velocity measurements and multi - band light curves have been published . the aim of this work is to obtain the absolute parameters of the system . along the next section
we first present our spectroscopic data and the set of _ ubvri _ light curves . making use of the spectra
we determine the parameters of the radial velocity curve and perform a reliable estimation of the spectral types for both components .
then , using the results from the spectroscopy together with the our photometric data we analyze the light curves to obtain reliable solutions that allow to determine the rest of the parameters of the system .
we report here new photometry of v821cas in the bessell _ ubvri _ bands .
the photometric accuracy and the phase coverage ( over 350 observations ) are sufficient to guarantee a reliable determination of the light curve parameters thus permitting a critical evaluation of stellar models .
the observations were carried out with the 0.40-m telescope located on mt .
bakirlitepe in september of 2007 at the tbitak national observatory ( tug , located in south of turkey ) .
the telescope is equipped with a apogee 1k@xmath141k ccd ( binned 2@xmath142 ) and standard bessel _ ubvri
_ filters .
the instrument with attached camera provide a field - of - view of [email protected]@[email protected] . by placing v821cas very close to center of the ccd to get highest accuracy
, we managed to strategically locate the binary on the chip together with two other stars of similar apparent magnitude .
as comparison star we selected bd + 52@xmath113575 . the check star chosen to be bd + 52@xmath113580 .
both stars passed respective tests for intrinsic photometric variability and proved to be stable during time span of our observations .
we collected a total of 1600 points in the ubvri bands .
the resulting v - band magnitude differences ( var - comp ) are listed in table 1 ( available in electronic form at the cds ) .
a typical precision of the differential magnitudes is about 0.008 mag per measurement .
standard iraf tasks were used to remove the electronic bias and to perform the flat - fielding corrections .
the iraf task imalign was used to remove the differences in the pixel locations of the stellar images and to place all the ccd images on the same relative coordinate systems .
the data were analyzed using another iraf task phot with no differential extinction effects taken into account given the relative small separation between the target and the comparison and check stars .
the phase - folded light curves for the whole observations are shown in fig .
5 with all bands .
spectroscopic observations have been performed with the chelle spectrograph ( fresco ) at the 91-cm telescope of catania astrophysical observatory .
the spectrograph is fed by the telescope through an optical fibre ( @xmath16@xmath17 , 100 @xmath18 m core diameter ) and is located , in a stable position , in the room below the dome level .
spectra were recorded on a ccd camera equipped with a thinned back illuminated site ccd of 1k@xmath141k pixels ( size 24@xmath1424 @xmath18 m ) .
the cross - dispersed chelle configuration yields a resolution of about 20000 , as deduced from the full width at half maximum of the lines of the th ar calibration lamp .
the spectra cover the wavelength range from 4300 to 6650 , split into 19 orders . in this spectral region , and in particular in the blue portion of the spectrum , there are several lines useful for the measure of radial velocity , as well as for spectral classification of the stars .
the data reduction was performed by using the chelle task of iraf package following the standard steps : background subtraction , division by a flat field spectrum given by a halogen lamp , wavelength calibration using the emission lines of a th - ar lamp , and normalization to the continuum through a polynomial fit .
sixteen spectra of v821cas were collected during the 20 observing nights between august 2 and 23 , 2006 .
typical exposure times for the v821cas spectroscopic observations were between 2400 and 2600 s. the signal - to - noise ratio ( @xmath19 ) achieved was between 70 and 115 , depending on atmospheric condition .
@xmath20 lyr ( a0v ) , 59 her ( a3iv ) , @xmath21 psc ( f7v ) , hd 27962 ( a2iv ) , and @xmath22 her were observed during each run as radial velocity and/or rotational velocity templates .
the average @xmath19 at continuum in the spectral region of interest was 150200 for the standard stars .
double - lined spectroscopic binaries reveal two peaks , displacing back and forth , in the cross - correlation function ( ccf ) between variable and the radial velocity template spectrum as seen in fig . 1 .
the location of the peaks allows to measure of the radial velocity of each component at the time of observation .
the cross - correlation technique applied to digitized spectra is now one of the standard tools for the measurement of radial velocities in close binary systems .
the radial velocities of v821cas were obtained by cross correlating of chelle orders of v821cas spectra with the spectra of the bright radial velocity standard stars @xmath20 lyr ( a0v ) , 59 her ( a3iv ) and @xmath21 psc ( f7v ) ( nordstrm et al . , 2004 ) .
for this purpose the iraf task was used .
1 shows examples of ccfs of v821cas near the first and second quadrature .
the two non - blended peaks correspond to each component of v821cas .
we applied the cross - correlation technique to five wavelength regions with well - defined absorption lines of the primary and secondary components .
these regions include the following lines : siiii 4568 , mgii 4481 , hei 5016 , hei 4917 , hei 5876 .
the stronger ccf peak corresponds to the more massive component that also has a larger contribution to the observed spectrum . to better evaluate the centroids of the peaks ( i.e. the radial velocity difference between the target and the template ) , we adopted two separate gaussian fits for the case of significant peak separation .
the radial velocity measurements , listed in table 2 together with their standard errors , are weighted means of the individual values deduced from each order ( see , e.g. , frasca et al . 2006 ) .
the observational points and their error bars are displayed in fig .
2 as a function of orbital phase as calculated by means of the ephemeris based on the photometric times of the primary eclipse described in deirmenci et al .
( 2007 ) . the first detailed solution of both radial velocity curves of v821cas components is presented in this study .
we found the semi - amplitude of the more massive , more luminous component to be k@xmath0=120@xmath12 km s@xmath8 and k@xmath3=150@xmath12 km s@xmath8 for the secondary component .
we have used our spectra to classify the primary component of v821cas . for this purpose
we have degraded the spectral resolution from 20000 to 3000 , by convolving them with a gaussian kernel of the appropriate width , and we have measured the equivalent width ( @xmath23 ) of photospheric absorption lines useful for the spectral classification . we have followed the procedures of hernndez et al .
( 2004 ) , choosing hydrogen and helium lines in the blue - wavelength region , where the contribution of the secondary component to the observed spectrum is negligible . from several spectra we measured @xmath24 , @xmath25 , @xmath26 , and @xmath27 . from the calibration relations @xmath23spectral - type of hernndez et al
( 2004 ) , we have derived a spectral type a1v with an uncertainty of about 1.0 spectral subclass . the effective temperature deduced from the calibrations of drilling & landolt ( 2000 ) or de jager & nieuwenhuijzen ( 1987 ) is about 9400k ( a1v ) .
the spectral - type uncertainty leads to a temperature error of @xmath28k .
.radial velocities of the v821cas components .
the columns give the heliocentric julian date , the orbital phase ( according to the ephemeris given by deirmenci et al .
2006 ) , the radial velocities of the two components with the corresponding errors , and the average s / n of the spectrum . [ cols="^,^,>,^,>,^,^ " , ] + * newly reduced hipparcos catalogue ( van leeuwen 2007 ) +
combination of the parameters obtained from light curves and rvs yield the absolute dimensions of the system , which are presented in table 6
. the standard deviations of the parameters have been determined by jktabsdimjkt / codes.html ] code , which calculates distance and other physical parameters using several different sources for bolometric corrections ( southworth et al .
the radii of the components were estimated with uncertainties of 1.2 % and 1.6%
. however , the uncertainties on the masses , being 3.3 % and 3.6 % , are slightly larger than the criteria of andersen ( 1991 ) for the necessary precision of absolute dimensions of stars to be used for comparison with theoretical models .
an inspection of the temperatures , masses and radii of the component stars reveals a binary system composed of two main - sequence stars . the temperature @xmath29 k , mass @xmath30 m@xmath2 and radius @xmath31 r@xmath2 of the primary are consistent with the spectral type of a1v , and the temperature @xmath32 k , mass @xmath33 m@xmath2 and radius @xmath34 r@xmath2 of the secondary are consistent with an a4v spectral type star .
the colour excess @xmath35 for a star may also be determined from the wide - band photometry . for calculation
the reddening - free @xmath36 we used the well - known equation below , @xmath37 \times ( b - v),\ ] ] for stars from o9 to a2
. the average ratio of colour excesses for stars from o8 to a2 was adopted as @xmath38=0.72\pm0.03 $ ] ( hovhannessian 2004 ) . the standard relation between @xmath39 and @xmath40
was also given in hovhannessian s study .
for calculation of the total absorption in the visual magnitude the following relation , the ratio of selective - to - total extinction in the @xmath41 band , @xmath42 = 3.30 + 0.28(b - v)_{0 } + 0.04 e(b - v),\ ] ] given by drilling & landolt ( 2001 ) was adopted .
using the observed visual magnitude and color indexes given by oja ( 1985 ) as v=8.26 , u - b=0.07 and b - v=0.11 mag we compute the reddening - free parameter as @xmath43 . using the tables given by hovhannessian ( 2004 ) we find the intrinsic colour index of @xmath44 mag and then the colour excess of @xmath45 mag for the system .
the colours of the system were the average of three measurements .
we should mention that there is no clue about the date of observations obtained , i.e. orbital phase .
this value is in agreement with the e(b - v)=0.147 determined from the spectra given in 3.2 . using the two @xmath35 values derived from photometric and spectroscopic data we calculated the de reddening distance modulus of the system . to estimate the bolometric magnitudes of the components we adopted m@xmath46=4.74 mag for the sun . using the bolometric corrections given by drilling & landolt ( 2000 ) and girardi et al .
( 2002 ) we estimate the distance to the system as 260 and 274 pc , respectively , with an uncertainty of 12 pc . however , the average distance to the system is estimated to be 206@xmath47 pc from the trigonometric parallax measured by the hipparcos mission .
our detection of the apsidal motion rate of the v821cas in 4 provides the opportunity to test the models of stellar internal structure .
in addition to the classical newtonian contribution , the observed rate of rotation of apsides includes also the contributions arising from general relativity .
the theory of general relativity estimates the relativistic contribution to observed rate as the following einstein formula : @xmath48 where m@xmath49(@xmath50 ) denotes the individual masses of the components in solar mass , @xmath51 is the orbital eccentricity ( from radial velocity analysis ) and @xmath52 is the anomalistic period of the system in days .
use of equation ( 4 ) for v821cas yields the relativistic contribution of [email protected] cycle@xmath8 which is only 6 per cent of the observed rate . removing the relativistic contribution from the observed rate of apsidal motion
, one can calculate the observational average value of internal structure constant ( @xmath53 ) using the following formula : @xmath54 where c@xmath55(@xmath50 ) are the functions of the orbital eccentricity , fractional radii and masses of the components and ratio of the rotational velocities of the component stars to the keplerian velocity .
the observed average value of @xmath56 is found to be [email protected] from eq .
the theoretical internal structure constants for the components ( @xmath57 ) are taken from the theoretical calculations of claret ( 2004 ) . among the tabulated values ,
interpolation for the masses and @xmath58 of the component star v821cas yielded -2.48 and -2.37 for the primary and secondary components , respectively .
the mean theoretical value of internal structure constant was computed as @xmath59=-2.43 which is in agreement with that of observed within 2@xmath60 level .
the internal structure constant obtained from the o - c analysis is smaller by 35 % than predicted from the theory .
the present analysis shows that the component stars are more concentrated in mass than predicted by theoretical calculations . in fig .
6 , we plot the location of v821cas stellar components in @xmath61 diagram . the evolutionary tracks for masses 1.6 , 2.0 m@xmath2 and @xmath62 ( girardi et al .
2002 ) are also shown in this figure .
higher metallicity tracks given by girardi et al .
do not match with the observed properties of the components .
an isochrone corresponding to an age of 5.6@xmath63 yr , computed from girardi et al .
( 2002)lgirardi / cgi - bin / cmd ] is also shown in fig .
the effective temperature of the primary component appears to be slightly higher than the model for 2 m@xmath2 .
however , when the uncertainties of the effective temperatures of the components are taken into account it is seen that the deviations from the models are less than 1@xmath60 level .
the diagram suggests that both components of v821cas are main sequence stars , relatively young with an age of about 560 myr .
to study the kinematical properties of v821cas , we used the system s center - of - mass velocity , distance and proper motion values , which are given in table 6 .
the proper motion data were taken from newly reduced hipparcos catalogue ( van leeuwen 2007 ) , whereas the center - of - mass velocity and distance are obtained in this study .
the system s space velocity was calculated using johnson & soderblom s ( 1987 ) algorithm .
the u , v and w space velocity components and their errors were obtained and given in table 6 . to obtain the space velocity precisely the first - order galactic differential rotation correction
was taken into account ( mihalas & binney 1981 ) , and -1.08 and 0.65 kms@xmath8 differential corrections were applied to u and v space velocity components , respectively .
the w velocity is not affected in this first - order approximation . as for the lsr correction , mihalas & binney s ( 1981 ) values ( 9 , 12 ,
7)@xmath2 kms@xmath8 were used and the final space velocity of v821cas was obtained as @xmath64 kms@xmath8 .
this value is in agreement with other young stars space velocities . to determine the population type of v821cas the galactic orbit of the system
was examined .
using dinescu et al .
( 1999 ) n - body code , the system s apogalactic ( @xmath65 ) and perigalactic ( @xmath66 ) distances were obtained as 9.09 and 8.07 kpc , respectively .
also , the maximum possible vertical separation from the galactic plane is @xmath67=50 pc for the system .
when determining the ellipticity the following formula was used : @xmath68 the ellipticity was calculated as @xmath69 .
this value shows that v821cas is orbiting the galaxy in an almost circular orbit and that the system belongs to the young thin - disc population .
we thank prof .
g. strazzulla , director of the catania astrophysical observatory , and dr .
g. leto , responsible for the m. g. fracastoro observing station for their warm hospitality and allowance of telescope time for the observations .
in addition , is grateful to all the people working at the catania astrophysical observatory for creating a stimulating and enjoyable atmosphere and , in particular , to the technical staff of the oac , namely p. bruno , g. carbonaro , a. distefano , m. miraglia , a. miccich , and g. occhipinti , for the valuable support in carrying out the observations .
we also thank to tbitak for a partial support in using t40 with project number tug - t40.000.111 .
ebiltem ege university science foundation project no:08/bil/0.27 and turkish scientific and technical research council for supporting this work through grant nr .
we also thank to k. b. cokunolu for his contributions .
we are also grateful to the anonymous referee whose comments helped to improve this paper .
this research has been also partially supported by inaf and italian miur .
this research has been made use of the ads and cds databases , operated at the cds , strasbourg , france .
ak h. , filiz n. , 2003 , ibvs , 5462 andersen j. , 1991 , a&arv , 3 , 91 baki v. , baki h. , erdem a. , iek c. , demircan o. , budding e. , 2003 , ibvs , 5464 bessell m. s. , castelli f. , plez b. , 1998 , a&a , 337 , 321 bilir s. , ak t. , soydugan e. , soydugan f. , yaz e. , ak f. , eker z. , demircan o. , helvaci m. , 2008 , an , 329 , 835 bulut i. , demircan o. , 2003 , ibvs , 5476 brat l. , zejda m. , svoboda p. , 2007
, b.r.n.o .
contributions # 34 claret a. , 2004 , a&a , 424 , 919 cutri r. m. , et al . , 2003 , the irsa _
2mass _ all - sky point source catalog , nasa / ipac infrared science archive . http://irsa.ipac.caltech.edu/applications/gator/ de jager c. , nieuwenhuijzen h. , 1987 , a&a , 177 , 217 deirmeci . l. , bozkurt z. , yakut k. et al . , 2003 ,
ibvs , 5386 deirmeci . l. , bozkurt z. , yakut k. et al . , 2007 , newa , 322 , 326 dinescu , d. i. , girardi , t. m. , van altena , w. f. , 1999 , aj , 117 , 1792 drilling j. s. , landolt a. u. , 2000 , allen s astrophysical quantities , 4th ed .
edited by arthur n. cox .
isbn : 0 - 387 - 98746 - 0 .
publisher : new york : aip press ; springer , 2000 , p.381 esa 1997 , esa sp-1200 , the hipparcos and tycho catalogues .
esa , noordwijk frasca a. , guillout p. , marilli e. , freire ferrero r. , biazzo k. , klutsch a. , 2006 , a&a , 454 , 301 gimenez a. , garcia - pelayo j. m. , 1983 , ap&ss , 92 , 203 girardi l. , bertelli g. , bressan a. , chiosi c. , groenewegen m. a. t. , marigo p. , salasnich b. , weiss a. , 2002 , a&a , 391 , 195 hernndez j. , calvet n. , briceo c. , hartmann l. , berlind p. , 2004
, aj , 127 , 1682 hovhannessian r. kh . , 2004 ,
ap , 47 , 499 johnson d. r. h. , soderblom d. r. , 1987 , aj , 93 , 864 mihalas d. , binney j. , 1981 . in galactic astronomy , 2nd edition , freeman , san fransisco , p.181 munari u. , zwitter t. , 1997 , a&a , 318 , 269 nordstrm b. , mayor m. , holmberg j. , pont f. , jorgensen b. r. , olsen e. h. , udry s. , mowlavi n. , 2004 , a&a , 418 , 989 oja t. , 1985 , ap&ss , 59 , 461 otero s. a. , 2005 , ibvs , 5631 pra a. , zwitter t. , 2005 , apj , 628 , 426 queloz d. , allain s. , mermilliod j .- c . , bouvier j. , mayor m. , 1998 , a&a , 335 , 183 royer f. , zorec j. , fremat y. , 2004 , `` the a - star puzzle '' , held in poprad , slovakia , july 8 - 13 , 2004 . edited by j. zverko , j. ziznovsky , s.j .
adelman , and w.w .
weiss , iau symposium , no .
cambridge , uk : cambridge university press ,
p.109 schlegel d. j. , finkbeiner d. p. , davis m. , 1998 , apj , 500 , 525 southworth j. , smalley b. , maxted p. f. l. , claret a. , etzel p. b. , 2005 , mnras , 363 , 529 van leeuwen f. , 2007 , a&a , 474 , 653 | we present ubvri photometric measurements and spectroscopic observations of the double - lined eclipsing binary v821cas .
the radial velocities were obtained by means of the cross - correlation technique .
simultaneous analyses of the multi - band light curves and rvs give the absolute parameters for the stars as : m@[email protected] m@xmath2 , m@[email protected] m@xmath2 , r@[email protected] r@xmath2 , r@[email protected] r@xmath2 , t@xmath4=9400@xmath1400 k , and t@xmath5=8600@xmath1400 k. analysis of the o - c residuals yielded an apsidal motion in the binary at a rate of @[email protected]@[email protected] cycle@xmath8 , corresponding to an apsidal period of u=118@xmath119 yr . subtracting the relativistic contribution we find that @xmath9=-2.590 which is in agreement with the value predicted by theoretical models .
comparison with current stellar evolution models gives an age of @xmath10 yr for the system .
[ firstpage ] binaries : stars : close - binaries : eclipsing - binaries : general - binaries : spectroscopic - stars : individual : v821cas |
the accurate computation of thermal quantum rates is a major challenge in theoretical chemistry , as a purely classical description of the kinetics fails to capture zero - point energy , tunnelling , and phase effects@xcite .
exact solutions using correlation functions , developed by yamamoto , miller , and others@xcite are only tractable for small or model systems , as the difficulty of computation scales exponentially with the size of the system .
consequently , numerous approximate treatments have been developed , which can be broadly classed as those seeking an accurate description of the quantum statistics without direct calculation of the dynamics , and those which also seek to use an approximate quantum dynamics .
methods in the first category include instanton theory@xcite , `` quantum instanton''@xcite , and various transition - state theory ( tst ) approaches@xcite .
of many approximate quantum dynamics methods , particularly successful ones include the linearized semiclassical initial - value representation ( lsc - ivr ) @xcite , centroid molecular dynamics ( cmd)@xcite , and ring polymer molecular dynamics ( rpmd)@xcite .
rpmd has been very successful for the computation of thermal quantum rates in condensed - phase processes , due to the possibility of implementation in complex systems such as ( proton - coupled ) electron transfer reaction dynamics or enzyme catalysis,@xcite and especially in small gas - phase systems@xcite where comparison with exact quantum rates and experimental data has demonstrated that rpmd rate theory is a consistent and reliable approach with a high level of accuracy .
these numerical results have shown that rpmd rate theory is exact in the high - temperature limit ( which can also be shown algebraically@xcite ) , reliable at intermediate temperatures , and more accurate than other approximate methods in the deep tunnelling regime ( see eq .
below ) , where it is within a factor of 23 of the exact quantum result .
rpmd also captures zero - point energy effects,@xcite and provides very accurate estimates for barrierless reactions@xcite .
it has been found to systematically overestimate thermal rates for asymmetric reactions and underestimate them for symmetric ( and quasisymmetric ) reactions in the deep tunnelling regime ( note that zero - point energy effects along the reaction coordinate must be taken into account when assigning the reaction symmetry.)@xcite recently a general code for rpmd calculations ( rpmdrate ) has been developed.@xcite another appealing feature of rpmd rate theory is its rigorous independence to the location of the dividing surface between products and reactants@xcite , a property shared by classical rate theory and the exact quantum rate@xcite , but not by many transition - state theory approaches .
the @xmath2 , tst limit of rpmd ( rpmd - tst ) is identical to true qtst : the instantaneous thermal quantum flux through a position - space dividing surface which is equal to the exact quantum rate in the absence of recrossing@xcite .
a corollary of this is that rpmd will be exact for a parabolic barrier ( where there is no recrossing of the optimal dividing surface by rpmd dynamics or quantum dynamics , and qtst is therefore also exact)@xcite . when the centroid is used as the dividing surface ( see eq .
below ) , rpmd - tst reduces to the earlier theory of centroid - tst@xcite , which is a good approximation for symmetric barriers but significantly overestimates the rate for asymmetric barriers at low temperatures@xcite .
this effect is attributable to the centroid being a poor dividing surface beneath the ` crossover ' temperature into deep tunnelling@xcite . in this ` deep tunnelling ' regime , rpmd - tst has a close relationship to semiclassical `` i m f '' instanton theory@xcite , which has been very successful for calculating rates beneath the crossover temperature , though has no first - principles derivation@xcite and was recently shown to be less accurate than qtst when applied to realistic multidimensional reactions@xcite . very recently , both cmd and rpmd have been obtained from the exact quantum kubo - transformed@xcite time - correlation function ( with explicit error terms ) via a boltzmann - conserving `` matsubara dynamics''@xcite which considers evolution of the low - frequency , smooth `` matsubara '' modes of the path integral@xcite .
matsubara dynamics suffers from the sign problem and is not presently amenable to computation on large systems .
however , by taking a mean - field approximation to the centroid dynamics , such that fluctuations around the centroid are discarded , one obtains cmd.@xcite alternatively , if the momentum contour is moved into the complex plane in order to make the quantum boltzmann distribution real , a complex liouvillian arises , the imaginary part of which only affects the higher , non - centroid , normal modes . discarding the imaginary liouvillian leads to spurious springs in the dynamics and gives rpmd.@xcite consequently
, rpmd will be a reasonable approximation to matsubara dynamics , provided that the timescale over which the resultant dynamics is required ( the timescale of ` falling off ' the barrier in rate theory ) is shorter than the timescale over which the springs ` contaminate ' the dynamics of interest ( in rate theory , this is usually coupling of the springs in the higher normal modes to the motion of the centroid dividing surface via anharmonicity in the potential ) .
both rpmd and cmd are inaccurate for the computation of multidimensional spectra : the neglect of fluctuations in cmd leads to the `` curvature problem '' where the spectrum is red - shifted and broadened , whereas in rpmd the springs couple to the external potential leading to `` spurious resonances''@xcite . recently
, this problem has been solved by attaching a langevin thermostat@xcite to the internal modes of the ring polymer@xcite ( which had previously been used for the computation of statistical properties@xcite ) , and the resulting thermostatted rpmd ( trpmd ) had neither the curvature nor resonance problem .
the success of rpmd for rate calculation , and the attachment of a thermostat for improving its computation of spectra , naturally motivates studying whether trpmd will be superior for the computation of thermal quantum rates to rpmd ( and other approximate theories)@xcite , which this article investigates . given that rpmd is one of the most accurate approximate methods for systems where the quantum rates are available for comparison , further improvements would be of considerable benefit to the field .
we firstly review trpmd dynamics in section [ ssec : trpmd ] , followed by developing trpmd rate theory in section [ ssec : rate ] .
to predict the behaviour of the rpmd rate compared to the trpmd rate , we apply one - dimensional kramers theory@xcite to the ring - polymer potential energy surface in section [ ssec : kram ] . numerical results in section [ sec : num ]
apply trpmd to the symmetric and asymmetric eckart barriers followed by representative bimolecular reactions : h+h@xmath0 ( symmetric ) , d+muh ( quasisymmetrical ) , h+ch@xmath1 ( prototypical polyatomic reaction ) and f+h@xmath0 ( asymmetric and highly anharmonic ) .
conclusions and avenues for further research are presented in section [ sec : con ] .
for simplicity we consider a one - dimensional system @xmath3 with position @xmath4 and associated momentum @xmath5 at inverse temperature @xmath6 , where the @xmath7-bead ring - polymer hamiltonian is@xcite @xmath8 with the ring - polymer potential @xmath9 and the frequency of the ring - polymer springs @xmath10 , where @xmath11 .
generalization to further dimensions follows immediately , and merely requires more indices.@xcite the ring polymer is time - evolved by propagating stochastic trajectories using trpmd dynamics@xcite , @xmath12 where @xmath13 is the vector of bead positions and @xmath14 the vector of bead momenta , with @xmath15 the grad operator in position - space , @xmath16 a vector of @xmath7 uniform gaussian deviates with zero mean and unit variance , and @xmath17 the @xmath18 positive semi - definite friction matrix@xcite .
the fokker - planck operator corresponding to the trpmd dynamics in eqs . and is@xcite @xmath19 ( where the arrows correspond to the direction in which the derivative acts@xcite ) and for any @xmath17
, trpmd dynamics will conserve the quantum boltzmann distribution ( @xmath20 ) , a feature shared by rpmd and cmd but not some other approximate methods such as lsc - ivr@xcite .
we then show in appendix [ ap : db ] that trpmd obeys detailed balance , such that the trpmd correlation function is invariant to swapping the operators at zero time and finite time , and changing the sign of the momenta .
the time - evolution of an observable is given by the adjoint of eq .
, @xcite @xmath21 in the zero - friction limit , @xmath22 and @xmath23 , where @xmath24 is the adjoint of the liouvillian corresponding to deterministic ring - polymer trajectories@xcite .
we assume the standard depiction of rate dynamics , with a thermal distribution of reactants and a dividing surface in position space . in what follows
we assume scattering dynamics , with the potential tending to a constant value at large separation of products and reactants .
the methodology is then immediately applicable to condensed phase systems subject to the usual caveat that there is sufficient separation of timescales between reaction and equilibration.@xcite the exact quantum rate can be formally given as the long - time limit of the flux - side time - correlation function@xcite @xmath25 where @xmath26 is the partition function in the reactant region and limit . ]
@xmath27 { \label{eq : cfsqm}}\end{aligned}\ ] ] with @xmath28 and @xmath29 the quantum flux and side operators respectively , and @xmath30 the hamiltonian for the system . the quantum rate
can equivalently be given as minus the long - time limit of the time - derivative of the side - side correlation function , or the integral over the flux - flux correlation function@xcite .
the trpmd side - side correlation function is @xmath31 h[f ( { { \bf q } } _ t ) ] { \label{eq : css}}\end{aligned}\ ] ] where @xmath32 and likewise for @xmath33 , and @xmath34 is obtained by evolution of @xmath35 for time @xmath36 with trpmd dynamics .
the ring polymer reaction co - ordinate @xmath37 is defined such that the dividing surface is at @xmath38 , and that @xmath39 corresponds to products and @xmath40 to reactants .
direct differentiation of the side - side correlation function using the fokker planck operator in eq .
yields the trpmd flux - side time - correlation function @xmath41}s_n ( { { \bf p } } , { { \bf q } } ) h[f ( { { \bf q } } _ t ) ] { \label{eq : cfs}}\end{aligned}\ ] ] where @xmath42 is the flux perpendicular to @xmath37 at time @xmath43 , @xmath44 we approximate the long - time limit of the quantum flux - side time - correlation function in eq . as the long - time limit of the trpmd flux - side time - correlation function in eq . , leading to the trpmd approximation to the quantum rate as @xmath45 the flux - side time - correlation function eq . will decay from an initial tst ( @xmath2 ) value to a plateau , which ( for a gas - phase scattering reaction with no friction on motion out of the reactant or product channel ) will extend to infinity . for condensed - phase reactions
( and gas - phase reactions with friction in exit channels ) a rate is defined provided that there is sufficient separation of timescales between reaction and equilibration to define a plateau in @xmath46,@xcite which at very long times ( of the order @xmath47 for a unimolecular reaction ) tends to zero . further differentiation of the flux - side time - correlation function ( with the adjoint of the fokker - planck operator in eq . )
yields the trpmd flux - flux correlation function @xmath48}s_n ( { { \bf p } } , { { \bf q } } ) \delta[f ( { { \bf q } } _ t ) ] s_n ( { { \bf p } } _ t , { { \bf q } } _ t ) { \label{eq : cff}}\end{aligned}\ ] ] which , by construction , must be zero in the plateau region , during which no trajectories recross the dividing surface .
like rpmd rate theory , trpmd has the appealing feature that its short - time ( tst ) limit is identical to true quantum transition - state theory ( qtst ) , as can be observed by applying the short - time limit of the fokker - planck propagator @xmath49 to @xmath37 , yielding@xcite @xmath50 where @xmath51 is the qtst rate@xcite . in appendix
[ ap : ind ] we then show that the trpmd rate in eq . is rigorously independent of the location of the dividing surface .
consequently , the trpmd rate will equal the exact quantum rate in the absence of recrossing of the optimal dividing surface ( and those orthogonal to it in path - integral space ) by either the exact quantum or trpmd dynamics.@xcite we also note that eq
. holds regardless of the value of the friction matrix @xmath17 and that recrossing of individual ( stochastic ) trajectories can only reduce the trpmd rate from the qtst value , and hence qtst is an upper bound to the long - time trpmd rate . in the following calculations we use a friction matrix which corresponds to damping of the free ring polymer vibrational frequencies , and which has been used in previous studies of trpmd for spectra.@xcite for an orthogonal transformation matrix @xmath52 such that @xmath53 where @xmath54 is the spring matrix in eq
@xmath55 , the friction matrix is given by @xmath56 here @xmath57 is an adjustable parameter , with @xmath58 giving critical damping of the free ring polymer vibrations , @xmath59 corresponding to optimal sampling of the free ring polymer potential energy , and @xmath60 corresponding to zero friction ( i.e. rpmd).@xcite a crucial consequence of this choice of friction matrix is that the centroid of the ring polymer is unthermostatted , and the short - time error of trpmd from exact quantum dynamics is therefore @xmath61 , the same as rpmd.@xcite to provide a qualitative description of the effect of friction on the trpmd transmission coefficient , we apply classical kramers theory@xcite in the extended @xmath62-dimensional ring polymer space , governed by dynamics on the ( temperature - dependent ) ring - polymer potential energy surface in eq . .
since the short - time limit of trpmd rate theory is equal to qtst , and its long - time limit invariant to the location of the dividing surface , trpmd will give the qtst rate through the optimal dividing surface ( defined as the surface which minimises @xmath63)@xcite , weighted by any recrossings of that dividing surface by the respective dynamics .
we express this using the bennett - chandler factorization@xcite , @xmath64 where @xmath65 is the qtst rate , the asterisk denotes that the optimal dividing surface @xmath66 is used and the trpmd transmission coefficient is @xmath67 s_n^ * ( { { \bf p } } , { { \bf q } } ) h[f^ * ( { { \bf q } } _ t ) ] } { \int d { { \bf p } } \int d { { \bf q } } \ e^{-{\beta_n}h_n ( { { \bf p } } , { { \bf q } } ) } \delta[f^ * ( { { \bf q } } ) ] s_n^ * ( { { \bf p } } , { { \bf q } } ) h[s_n^ * ( { { \bf p } } , { { \bf q } } ) ] } { \label{eq : transcoef}}\end{aligned}\ ] ] with analogous expressions to eqs . and
for rpmd . to examine the explicit effect of friction on the trpmd rate we define the ratio @xmath68 and from eq .
@xmath69 we then assume that the recrossing dynamics is dominated by one - dimensional motion through a parabolic saddle point on the ring - polymer potential energy surface , in which case the trpmd transmission coefficient can be approximated by the kramers expression@xcite @xmath70 where formally @xmath71 , with @xmath72 the friction along the reaction co - ordinate and @xmath73 the barrier frequency in ring - polymer space . for a general @xmath74-dimensional system finding @xmath66 and thereby computing @xmath72 and @xmath73
is largely intractable .
however , we expect @xmath75 , and therefore define @xmath76 where the dimensionless parameter @xmath77 is expected to be independent of @xmath57 for a given system and temperature , and represents the sensitivity of the trpmd rate to friction .
we further approximate that there is minimal recrossing of the optimal dividing surface by the ( unthermostatted ) ring polymer trajectories such that @xmath78 , ) fewer than 20% of trajectories recrossed . ]
leading to @xmath79 equation relates the ratio of the trpmd and rpmd rates as a function of @xmath57 with one parameter @xmath77 , and without requiring knowledge of the precise location of the optimal dividing surface @xmath66 .
however , we can use general observations concerning which ring - polymer normal modes contribute to @xmath66 to determine the likely sensitivity of the trpmd rate to friction . above the crossover temperature into deep tunnelling , defined by@xcite @xmath80 where @xmath81 is the barrier frequency in the external potential @xmath82 , the optimal dividing surface is well approximated by the centroid@xcite @xmath83 where @xmath84 is the maximum in @xmath82 . as the centroid is not thermostatted ( since @xmath85 ) , in this regime @xmath86 and we therefore predict from eq . that the rate will be independent of @xmath57 , i.e. @xmath87 . beneath the crossover temperature , the saddle point on the ring - polymer potential energy surface bends into the space of the first degenerate pair of normal modes.@xcite for symmetric systems , the optimal dividing surface is still the centroid expression in eq . and ( insofar as the reaction dynamics can be considered one - dimensional ) @xmath88 , so @xmath87 . for asymmetric reactions ,
the optimal dividing surface is now a function of both the centroid and first degenerate pair of normal modes ( which are thermostatted)@xcite , and we expect @xmath89 . from eq
. the trpmd rate will decrease linearly with @xmath57 for small @xmath57 , for large friction as @xmath90 , and the ratio of the trpmd to rpmd rates to be a convex function of @xmath57 .
this behaviour would also be expected for symmetric reactions beneath the second crossover temperature where the optimal dividing surface bends into the space of the second degenerate pair of normal modes.@xcite in all cases one would expect that increasing friction would either have no effect on the rate , or at sufficiently low temperatures cause it to decrease .
it should be stressed that eq .
is a considerable simplification of the trpmd dynamics and is not expected to be reliable in systems where the ring polymer potential energy surface is highly anharmonic or skewed ( such as f+h@xmath0 investigated below ) .
in fact , even for a one - dimensional system , the minimum energy path on the @xmath7-dimensional ring polymer potential energy surface shows a significant skew beneath the crossover temperature@xcite .
the utility of eq . lies in its simplicity and qualitative description of friction - induced recrossing .
we initially study the benchmark one - dimensional symmetric and asymmetric eckart barriers before progressing to the multidimensional reactions h+h@xmath0 ( symmetric ) , d+muh ( quasisymmetrical ) , h+ch@xmath1 ( asymmetric , polyatomic ) and f+h@xmath0 ( asymmetric , anharmonic ) .
the methodology for computation of trpmd reaction rates is identical to that of rpmd@xcite , except for the thermostat attached to the internal normal modes of the ring polymer , achieved using the algorithm in ref . .
the bennett - chandler@xcite factorization was employed , and the same dynamics can be used for thermodymamic integration along the reaction co - ordinate ( to calculate the qtst rate ) as to propagate trajectories ( to calculate the transmission coefficient).@xcite [ cols="<,<,^,^,^,<,^,^,^ " , ] we firstly examine the symmetric eckart barrier@xcite , @xmath91 and to facilitate comparison with the literature@xcite , use parameters to model the h+h@xmath0 reaction : @xmath92ev , @xmath93 , and @xmath94 , leading to a crossover temperature of @xmath95 .
the centroid reaction co - ordinate of eq . was used throughout .
results for a variety of temperatures and values of friction parameter @xmath57 are presented in fig .
[ fig : sym ] , and values of @xmath96 obtained by nonlinear least squares in table [ tab : tar ] .
( red crosses ) , fitted kramers curve ( green dashes ) and quantum result ( black line ) .
@xmath97 is quoted in units of @xmath98 and the crossover temperature is @xmath95.,scaledwidth=50.0% ] [ fig : sym ] slightly beneath the crossover temperature ( @xmath99 ) , the trpmd rate is indepedent of the value of friction ( @xmath100 ) , as predicted by kramers theory .
some sensitivity to @xmath57 is seen before twice the crossover temperature , which is likely to be a breakdown of the one - dimensional assumption of kramers theory ; while the centroid is the optimal dividing surface , the minimum energy path bends into the space of the ( thermostatted ) lowest pair of normal modes@xcite .
beneath twice the crossover temperature the friction parameter has a significant effect on the rate , as to be expected from the second degenerate pair of normal modes becoming part of the optimal dividing surface@xcite .
the functional form of @xmath101 is also in accordance with the predictions of kramers theory , monotonically decreasing as @xmath57 rises , and being a convex function of @xmath57 . since rpmd underestimates the rate for this symmetric reaction ( and many others@xcite ) , adding friction to rpmd decreases its accuracy in approximating the quantum rate for this system .
the asymmetric eckart barrier is given by@xcite @xmath102 where @xmath103 , @xmath104 and @xmath105 in atomic units ( @xmath106 ) , giving a crossover temperature of @xmath107 .
to facilitate comparison with previous literature@xcite the results are presented in fig .
[ fig : asymlam ] as the ratio @xmath108 and @xmath96 values in table [ tab : tar ] .
[ eq . ] , showing the trpmd result as a function of @xmath57 ( red crosses ) , fitted kramers curve ( green dashes ) and quantum result ( black line ) .
the crossover temperature is @xmath107a.u.,scaledwidth=50.0% ] [ fig : asymlam ] above the crossover temperature , trpmd is invariant to the value of the friction parameter , and beneath the crossover temperature , increasing @xmath57 results in a decrease in the rate , such that trpmd is closer to the exact quantum result than rpmd for all @xmath109 in this system .
the decrease in the trpmd rate with @xmath57 is qualitatively described by the crude kramers approximation ( see fig . [
fig : asymlam ] ) , and it therefore seems that the improved accuracy of trpmd could be a fortuitous cancellation between the overestimation of the quantum rate by qtst , and the friction - induced recrossing of the optimal dividing surface by trpmd trajectories . there is no particular _ a priori _ reason to suppose that one value of @xmath57 should provide superior results ; from fig .
[ fig : asymlam ] , at @xmath110 a friction parameter of @xmath111 causes trpmd to equal the quantum result to within graphical accuracy , whereas at @xmath112 this value of friction parameter causes overestimation of the rate , and further calculations ( not shown ) show that @xmath113 is needed for trpmd and the quantum rates to agree .
the numerical results also show a slightly higher curvature in @xmath114 as a function of @xmath57 than eq . would predict , suggesting that the trpmd rate reaches an asymptote at a finite value , rather than at zero as the kramers model would suggest .
we suspect this is a breakdown of one - dimensional kramers theory , since in the @xmath115 limit the system can still react via the unthermostatted centroid co - ordinate , but may have to surmount a higher barrier on the ring polymer potential energy surface .
, as a function of the dividing surface @xmath84.,scaledwidth=50.0% ] [ fig : dsinv ] we then investigate the effect of changing the location of the centroid dividing surface on the trpmd rate .
rpmd is already known to be invariant to the location of the dividing surface , and we therefore choose a system for which trpmd and rpmd are likely to differ the most , namely a low - temperature , asymmetric system where there is expected to be significant involvement of the thermostatted lowest degenerate pair of normal modes in crossing the barrier .
the asymmetric eckart barrier at @xmath112 is therefore used as a particularly harsh test , with the result plotted in fig .
[ fig : dsinv ] .
although the centroid - density qtst result varies by almost a factor of six across the range of dividing surfaces considered ( @xmath116a.u . ) , both the trpmd and rpmd rates are invariant to the location of the dividing surface .
we also observe that , even with the optimal dividing surface , centroid - density qtst significantly overestimates the exact rate@xcite .
llllll parameter & & explanation + & h + h@xmath0 & d + muh & f + h@xmath0 & h + ch@xmath1 + + & & 250 ; 300 ; 500 & temperature ( k ) + & 128 & 512 & 384 ( 200 k ) & 192 ( 250 k ) & number of beads in the trpmd calculations + & & & 256 ( 300 k ) & 128 ( 300 k ) + & & & 64 ( 500 k ) & 64 ( 500 k ) + + @xmath117 & 30 & 30 & 30 & 30 & dividing surface @xmath118 parameter ( @xmath119 ) + @xmath120 & 1 & 1 & 1 & 1 & number of forming and breaking bonds + @xmath121 & 2 & 1 & 2 & 4 & number of equivalent product channels + + & & thermostat for the qtst calculations + & & friction coefficient for the recrossing factor calculations + + @xmath122 & 111 & 111 & 111 & 111 & number of windows + @xmath123 & -0.05 & -0.05 & -0.05 & -0.05 & center of the first window + @xmath124 & 0.01 & 0.01 & 0.01 & 0.01 & window spacing step + @xmath125 & 1.05 & 1.05 & 1.05 & 1.05 & center of the last window + @xmath126 & 0.0001 & 0.0001 & 0.0001 & 0.0001 & time step ( ps ) + @xmath127 & 2.72 & 2.72 & 2.72 & 2.72 & umbrella force constant ( ( t / k ) ev ) + @xmath128 & 200 & 200 & 200 & 200 & number of trajectories + @xmath129 & 20 & 20 & 20 & 20 & equilibration period ( ps ) + @xmath130 & 100 & 100 & 100 & 100 & sampling period in each trajectory ( ps ) + @xmath131 & @xmath132 & @xmath132 & @xmath132 & @xmath132 & total number of sampling points + + @xmath133 & -0.02 & -0.02 & -0.02 & -0.02 & start of umbrella integration + @xmath134 & 1.0000@xmath135 & 0.9912 ( 200 k)@xmath135 & 0.9671 ( 200 k)@xmath135 & 1.0093 ( 250 k)@xmath135 & end of umbrella integration + & & 0.9904 ( 300 k)@xmath135 & 0.9885 ( 300 k)@xmath135 & 1.0074 ( 300 k)@xmath135 & + & & 0.9837 ( 500
k)@xmath135 & 0.9947 ( 500 k)@xmath135 & 1.0026 ( 500
k)@xmath135 & + @xmath136 & 5000 & 5000 & 5000 & 5000 & number of bins + + @xmath126 & 0.0001 & 0.00003 & 0.0001 & 0.0001 & time step ( ps ) + @xmath129 & 20 & 20 & 20 & 20 & equilibration period ( ps ) in the constrained ( parent ) + & & & & & trajectory + @xmath137 & 100000 & 100000 & 500000 & 500000 & total number of unconstrained ( child ) trajectories + @xmath138 & 20 & 20 & 20 & 20 & sampling increment along the parent trajectory ( ps ) + @xmath139 & 100 & 100 & 100 & 100 & number of child trajectories per one + & & & & & initially constrained configuration + @xmath140 & 0.05 & 0.2 & 0.2 & 0.1 & length of child trajectories ( ps ) + + @xmath135 detected automatically by rpmdrate .
+ the results are calculated using adapted rpmdrate code@xcite , with details summarized in table [ tab1 ] . in the calculations reported below we used the potential energy surface developed by boothroyd et al .
( bkmp2 pes ) for h+h@xmath0 and d+muh,@xcite the stark werner ( sw ) potential energy surface for f+h@xmath0,@xcite and the pes-2008 potential energy surface developed by corchado et al . for h+ch@xmath1.@xcite the computation of the free energy was achieved using umbrella integration@xcite with trpmd and checked against standard umbrella integration with an andersen thermostat@xcite .
reaction as a function of @xmath57 .
kramers is the fitted kramers curve ( see text ) .
the crossover temperature is 345k.,scaledwidth=50.0% ] [ fig : hhh ] h+h@xmath0 represents the simplest atom - diatom scattering reaction and has been the subject of numerous studies@xcite .
the pes is symmetric and with a relatively large skew angle ( 60 ) , and a crossover temperature of 345k .
the results in fig .
[ fig : hhh ] show that the rate is essentially invariant to the value of @xmath57 above the crossover temperature . at 300k
there is a slight decrease in the rate with increasing friction from 0 to 1.5 ( @xmath14125 @xmath142 ) , and this is far more pronounced at 200k where the @xmath143 result is almost half that of the @xmath60 ( rpmd ) result . , but for the d+muh reaction with a high crossover temperature of 860k.,scaledwidth=50.0% ] [ fig : dmuh ] d+muh is `` quasisymmetrical '' since dmu and muh have very similar zero - point energies , and one would therefore expect the rpmd rate to underestimate the exact quantum rate@xcite . since it is mu - transfer the crossover temperature is very high ( 860 k ) and therefore this reaction can be considered as a stress test for the deep tunneling regime . the results in fig .
[ fig : dmuh ] show that friction in the trpmd dynamics causes further underestimation of the rate , especially at low temperatures ; for @xmath143 at 200k , trpmd underestimates the exact quantum rate by almost an order of magnitude , and even at 500k it decreases by @xmath14440% over the range of @xmath145 explored here .
the crossover temperature is 341k.,scaledwidth=50.0% ] [ fig : hch4 ] as an example of a typical asymmetric reaction , results for h+ch@xmath1 are plotted in fig .
[ fig : hch4 ] , which has a crossover temperature of 341k .
rpmd is well - known to overestimate the quantum rate for this system at low temperatures.@xcite fig .
[ fig : hch4 ] shows that above the crossover temperature ( 500k ) the friction parameter has a negligible effect on the rate .
as the temperature is decreased below the crossover temperature ( 300k and 250k ) , the friction induces more recrossings of the dividing surface and , as a result , the trpmd rate approaches the exact quantum rate with increasing the friction parameter .
reaction with a crossover temperature of 264k.,scaledwidth=50.0% ] [ fig : fhh ] thus far , kramers theory has been surprisingly successful at qualitatively explaining the behaviour of the trpmd rate with increasing friction .
present results would suggest that trpmd would therefore improve upon rpmd for all asymmetric reactions , where rpmd generally overestimates the rate beneath crossover@xcite .
we then examine another prototypical asymmetric reaction , f+h@xmath0 , with a low crossover temperature of 264k .
[ fig : fhh ] shows that at 500k and 300k , the trpmd rate is in good agreement with the quantum result , but increases very slightly with @xmath57 causing a spurious small negative value of @xmath96 in table [ tab : tar ] . beneath crossover , at 200k the rate is virtually independent of lambda , apart from a very slight increase around @xmath59 .
consequently , trpmd fares no better than rpmd for this system , contrary to the h+ch@xmath1 results and the predictions of kramers theory .
this is likely attributable to a highly anharmonic and exothermic energy profile , and a very flat saddle point in ring - polymer space@xcite . as can be seen from the graphs ,
the simple kramers prediction is in surprisingly good qualitative agreement with the numerical results ( apart from f+h@xmath0 beneath crossover ) , even for the multidimensional cases , which is probably attributable to those reactions being dominated by a significant thermal barrier which appears parabolic on the ring - polymer potential energy surface , meaning that the one - dimensional kramers model is adequate for capturing the friction - induced recrossing . in table [
tab : tar ] the @xmath96 values , fitted to the numerical data , show that for a given reaction @xmath146 above the crossover temperature , and beneath the crossover temperature @xmath96 increases as the temperature is decreased .
this can be qualitatively explained as the optimal dividing surface becoming more dependent on the thermostatted higher normal modes as the temperature is lowered@xcite .
not surprisingly , the highest value of @xmath96 is observed for the highly quantum mechanical d+muh reaction at 200k with @xmath147 .
this is beneath one quarter of the crossover temperature , and one would therefore expect that friction would have a very significant effect on the rate .
in this paper we have , for the first time , applied thermostatted ring polymer molecular dynamics ( trpmd ) to reaction rate theory .
regardless of the applied friction , the long - time limit of the trpmd flux - side time - correlation function ( and therefore the trpmd rate ) is independent of the location of the dividing surface , and its short - time limit is equal to rigorous qtst@xcite . in section [ ssec : kram ]
we use kramers theory @xcite to predict that , above the crossover temperature , the rpmd and trpmd rates will be similar , and beneath crossover the trpmd rate for asymmetric systems will decrease with @xmath57 , and the same effect should be observed for symmetric systems beneath half the crossover temperature .
trpmd rate theory has then been applied to the standard one - dimensional model systems of the symmetric and asymmetric eckart barriers , followed by the bimolecular reactions h+h@xmath0 , d+muh , h+ch@xmath1 and f+h@xmath0 . for all reactions considered , above the crossover temperature
the trpmd rate is virtually invariant to the value of @xmath57 and therefore almost equal to rpmd , as predicted by kramers theory . beneath the crossover temperature ,
most asymmetric reactions show a decrease in the trpmd rate as @xmath57 is increased , and in qualitative agreement with the kramers prediction in eq . .
a similar trend is observed for symmetric reactions , which also show some diminution in the rate with increasing friction above half the crossover temperature ( @xmath148 ) , probably due to the skewed ring - polymer pes causing a breakdown in the one - dimensional assumption of kramers theory . for the asymmetric and anharmonic case of f+h@xmath0 , beneath the crossover temperature
there is no significant decrease in the rate with increased friction , illustrating the limitations of kramers theory .
these results mean that beneath the crossover temperature trpmd will be a worse approximation to the quantum result than rpmd for symmetric and quasisymmetrical systems ( where rpmd underestimates the rate@xcite ) , and trpmd will be closer to the quantum rate for asymmetric potentials ( where rpmd overestimates the rate ) .
however , the apparent increase in accuracy for asymmetric systems appears to be a cancellation of errors from the overestimation of the quantum rate by rpmd which is then decreased by the friction in the non - centroid normal modes of trpmd , and there is no a priori reason to suppose that one effect should equal the other for any given value of @xmath57 .
although the above results do not advocate the use of trpmd rate theory as generally being more accurate than rpmd , trpmd rate calculation above the crossover temperature may be computationally advantageous in complex systems due to more efficient sampling of the ring - polymer phase space by trpmd trajectories than rpmd trajectories .
trpmd may therefore provide the same accuracy as rpmd rate calculation at a lower computational cost , and testing this in high - dimensional systems where rpmd has been successful , such as complex - forming reactions@xcite , surface dynamics@xcite , and enzyme catalysis@xcite would be a useful avenue of future research . future work could also include non - adiabatic systems@xcite , applying a thermostat to the centroid to model a bath system@xcite , and generalizations to non - markovian friction using grote - hynes theory@xcite . in closing , present results suggest that trpmd can be used above the crossover temperature for thermally activated reactions , and beneath crossover further testing is required to assess its utility for asymmetric systems .
tjhh acknowledges a research fellowship from jesus college , cambridge , and helpful comments on the manuscript from stuart althorpe .
yvs acknowledges support via the newton international alumni scheme from the royal society .
yvs also thanks the european regional development fund and the republic of cyprus for support through the research promotion foundation ( project cy - tera nea @xmath149o@xmath150omh/@xmath151tpath/0308/31 ) .
for a homogeneous markov process such as trpmd for which negative time is not defined@xcite , detailed balance is defined as@xcite @xmath152 where @xmath153 is the stationary distribution and @xmath154 is the conditional probability that a ring polymer will be found at point @xmath155 at time @xmath36 , given that is was at @xmath35 at time @xmath43 . to demonstrate that eq .
is statisfied , we rewrite the fokker - planck operator eq . as @xmath156 where the vectors @xmath157 , @xmath158 and the matrix @xmath159 .
note that the derivatives in eq .
act on @xmath160 , @xmath161 or @xmath162 and whatever follows them which is acted upon by @xmath163 . the necessary and sufficient conditions for detailed balance [ eq .
] to hold , in addition to @xmath164 being a stationary distribution , are then given by@xcite @xmath165 condition eq .
is trivially satisfied .
provided that the friction matrix is even w.r.t .
momenta ( satisfied here as @xmath17 is not a function of @xmath14 ) eq .
will be satisfied .
. becomes @xmath166 which is satisfied with @xmath167 and the friction matrix used here .
given that eq .
is satisfied , for an arbitrary correlation function one can then show @xmath168 and for the langevin trajectories considered here , which are continuous but not differentiable , this means @xmath169 where @xmath170 is the vector of positions stochastically time - evolved according to eqs . and , and @xmath171 with @xmath172 similarly defined .
we use a similar methodology to that which craig and manolopoulos employed for rpmd@xcite , and give the main steps here .
we firstly differentiate the side - side correlation function in eq .
the location of the dividing surface @xmath84 ( or any other parameter specifying the nature of the dividing surface ) , giving @xmath173}h[f ( { { \bf q } } _ t ) ] + h [ { f({\bf q } ) } ] \delta[f ( { { \bf q } } _ t ) ] \right\}. { \label{eq : dcssdq}}\end{aligned}\ ] ] since trpmd dynamics obeys detailed balance ( as shown in appendix [ ap : db ] ) , and the dividing surface is only a function of position , the second term on the rhs of eq . is identical to the first , @xmath174}h[f ( { { \bf q } } _ t ) ] .
{ \label{eq : dcssdq2}}\end{aligned}\ ] ] differentiation of eq .
time using eq . , and relating the side - side and flux - side functions using eq . , yields @xmath175}\delta[f ( { { \bf q } } _ t ) ] s_n ( { { \bf p } } _ t , { { \bf q } } _ t ) .
{ \label{eq : dcfsdqdd}}\end{aligned}\ ] ] equation corresponds to a trajectory commencing at the dividing surface at time zero and returning to it at time @xmath36 with non - zero velocity @xmath176 . at finite times
while there is recrossing of the barrier , there will be trajectories satisfying these conditions , but after the plateau time when no trajectories recross the barrier [ cf .
eq . ] , these conditions are clearly not satisfied , and the rate will be independent of the location of the dividing surface.@xcite this proof is valid for any friction matrix which satisfies the detailed balance conditions of appendix [ ap : db ] , and does not require the presence of ring - polymer springs in the potential , so is valid for _ any _ classical - like reaction rate calculation using langevin dynamics .
we note that the flux - side function in the gas phase was originally derived as an asymmetric - split trace@xcite and later as a symmetric - split trace@xcite , both of which are equivalent to the kubo transformed trace given here in the @xmath177 limit . in practice ,
both the location of the dividing surface and the degree of recrossing through it are very difficult to determine numerically , though richardson and althorpe@xcite found that for the symmetric and asymmetric eckart barrier at almost twice the inverse crossover temperature ( @xmath178 ) fewer than 20% of trajectories recrossed .
this proof@xcite was originally for a gas - phase system and a centroid dividing surface , but this can easily be extended to the condensed phase provided there is the necessary separation of timescales@xcite . | we apply thermostatted ring polymer molecular dynamics ( trpmd ) , a recently - proposed approximate quantum dynamics method , to the computation of thermal reaction rates .
its short - time transition - state theory ( tst ) limit is identical to rigorous quantum transition - state theory , and we find that its long - time limit is independent of the location of the dividing surface .
trpmd rate theory is then applied to one - dimensional model systems , the atom - diatom bimolecular reactions h+h@xmath0 , d+muh and f+h@xmath0 , and the prototypical polyatomic reaction h+ch@xmath1 . above the crossover temperature ,
the trpmd rate is virtually invariant to the strength of the friction applied to the internal ring - polymer normal modes , and beneath the crossover temperature the trpmd rate generally decreases with increasing friction , in agreement with the predictions of kramers theory .
we therefore find that trpmd is approximately equal to , or less accurate than , ring polymer molecular dynamics ( rpmd ) for symmetric reactions , and for certain asymmetric systems and friction parameters closer to the quantum result , providing a basis for further assessment of the accuracy of this method . _
copyright ( 2015 ) american institute of physics .
this article may be downloaded for personal use only .
any other use requires prior permission of the author and the american institute of physics .
the following article appeared in j. chem .
phys .
, * 143 * ( 2015 ) , 074107 , and may be found at http://dx.doi.org/10.1063/1.4928599._ |
quantum error correcting codes are a central weapon in the battle to overcome the effects of environmental noise associated with attempts to control quantum mechanical systems as they evolve in time @xcite .
it is thus important to develop techniques that assist in determining whether one code is better than another for a given noise model . in this paper
we make a contribution to this study by introducing a notion of entropy for quantum error correcting codes .
no single quantity can be expected to hold all information on a code , its entropy included .
nevertheless , the entropy of a code is one way in which the amount of effort required to recover a code can be quantified . in the extremal case ,
a code has zero entropy if and only if it can be recovered with a single unitary operation .
this is the simplest of all correction operations in that a measurement is not required as part of the correction process .
these codes have been recently coined _ unitarily correctable _
@xcite , and include _ decoherence - free subspaces _
@xcite in the case that recovery is the trivial identity operation .
thus more generally the entropy can be regarded as a measure of how close a code is to being unitarily correctable , or decoherence - free in some cases . in the next section
we introduce the entropy of a code , along with required nomenclature .
we also consider an example motivated by the stabilizer formalism @xcite and discuss an extension of code entropy to operator quantum error correcting subsystem codes @xcite .
we then consider in detail an illustrative class of quantum operations for which the code structure has recently been characterised , the class of _ binary unitary channels _ @xcite .
let @xmath0 denote a quantum state : a hermitian , positive operator , satisfying the trace normalisation condition tr@xmath1 . a linear quantum operation ( or channel ) @xmath2 , which sends a density operator @xmath0 of size @xmath3 into its image @xmath4 of the same size may be described in the choi - kraus form @xcite @xmath5 the kraus operators @xmath6 can be chosen to be orthogonal @xmath7 , so that the non - negative weights @xmath8 become eigenvalues of the dynamical ( choi ) matrix , @xmath9 .
we refer to the rank of the choi matrix as the _ choi rank _ of @xmath2 .
observe that the choi rank of @xmath2 is equal to the minimal number of kraus operators required to describe @xmath2 as in ( [ kraus ] ) .
hence in the canonical form the number @xmath10 of non - zero kraus operators does not exceed @xmath11 . due to the theorem of choi the map
@xmath2 is completely positive ( cp ) if and only if the corresponding dynamical matrix is positive if and only if @xmath2 has a form as in ( [ kraus ] ) .
the map @xmath2 is _ trace preserving _
, @xmath12 , if and only if @xmath13 where we assume some @xmath14 are zero if @xmath10 is less than @xmath15 .
the family of operators @xmath6 is not unique .
however , if @xmath16 is another family of operators that determine @xmath2 as in ( [ kraus ] ) , then there is a scalar unitary matrix @xmath17 such that @xmath18 for all @xmath19 .
we refer to this as the _ unitary invariance _ of choi - kraus decompositions .
to characterise the information missing in a quantum state one uses its _ von neumann _ entropy , @xmath20 we will use the convention that @xmath21 refers to logarithm base two as this provides a cleaner operational qubit definition in the context of quantum information . in order to describe the action of a cp map @xmath2 , represented in the canonical choi - kraus form ( [ kraus ] ) , for an initial state @xmath0 we may compare its entropy with the entropy of the image @xmath22 . to obtain a bound for such an entropy change we can define an operator @xmath23 acting on an extended hilbert space @xmath24 , @xmath25 if the map @xmath2 is stochastic , the operator @xmath26 is positive definite and normalised so it represents a density operator in its own right , @xmath27 ( specifically , it is an initially pure environment evolved by the unitary dilation of @xmath2 ) .
observe that for any unitary map , @xmath28 , the form ( [ kraus ] ) consists of a single term only .
hence in this case the operator @xmath26 reduces to a single number equal to unity and its entropy vanishes , @xmath29 .
the auxiliary state @xmath26 acting in an extended hilbert space was used by lindblad to derive bounds for the entropy of an image @xmath30 of any initial state .
the bounds of lindblad @xcite , @xmath31 are obtained by defining another density matrix in the composite hilbert space @xmath32 , @xmath33 where @xmath34 and @xmath35 forms an orthonormal basis in @xmath36 .
thus , @xmath37 is simply the system and an initially pure environment evolved by the unitary dilation of @xmath2 .
computing partial traces one finds that tr@xmath38 and tr@xmath39 .
it is possible to verify that @xmath40 , so making use of the subadditivity of entropy and the triangle inequality @xcite one arrives at ( [ boundent ] ) .
if the initial state is pure , that is if @xmath41 , we find from ( [ boundent ] ) that the final state @xmath4 has entropy @xmath42 . for this reason @xmath42
was called the _ entropy exchange _ of the operation by shumacher @xcite . in that work
an alternative formula for the entropy exchange was proven , @xmath43 where @xmath44 is an _
arbitrary _ purification of the mixed state , tr@xmath45 .
thus , the entropy exchange is invariant under purification of the initial state and remains a function only of the initial density operator @xmath0 and the map @xmath2 . a quantum operation @xmath2 allows for an error correction scheme in the standard framework for quantum error correction @xcite if there exists a subspace @xmath46 such that for some set of complex scalars @xmath47 the corresponding projection operator @xmath48 satisfies @xmath49 specifically , this is equivalent to the existence of a quantum _ recovery operation _ @xmath50 such that @xmath51 where @xmath52 is the map @xmath53 .
the subspace related to @xmath48 determines a _ quantum error correcting code _ ( qecc ) for the map @xmath2 .
a special class of codes are the _ unitarily correctable codes _ ( ucc ) , which are characterised by the existence of a unitary recovery operation @xmath54 .
these codes include _ decoherence - free subspaces _ ( dfs ) in the case that @xmath50 is the identity map , @xmath55 .
it can be shown that the matrix @xmath56 is hermitian and positive , and is in fact a density matrix , so it can be considered as an auxiliary state acting in an extended hilbert space of size at most @xmath11 .
it is easy to obtain a more refined global upper bound on the rank of @xmath57 in terms of the map @xmath2 .
[ upperbound ]
let @xmath57 be the matrix determined by a code @xmath58 for a quantum map @xmath2 .
then the rank of @xmath57 is bounded above by the choi rank of @xmath2 .
* proof . * without loss of generality assume the choi matrix @xmath59 is diagonal .
we have for all @xmath19 , @xmath60 and the result follows from the positivity of @xmath59 and @xmath57 . ' '' ''
unitarily correctable codes are typically highly _ degenerate _ codes , as the map @xmath2 collapses to a single unitary operation when restricted to the code subspace .
in particular , the unitary invariance of choi - kraus representations implies the restricted operators @xmath61 are all scalar multiples of a single unitary .
more generally , one can see that a code @xmath62 is degenerate for @xmath2 precisely when the choi rank of @xmath2 is strictly greater than the rank of @xmath57 .
indeed , the choi rank counts the minimal number of kraus operators required to implement @xmath2 via ( 2.1 ) , and satisfaction of this strict inequality means there is redundancy in the description of @xmath63 by the operators @xmath64 . thus , for these reasons we shall refer to codes as _ non - degenerate _ if the choi rank of @xmath2 coincides with the rank of @xmath57 and if the spectrum of @xmath57 is equally balanced that is to say that non - degenerate codes correspond to the maximally degenerate error correction matrix , @xmath57 proportional to the identity matrix assume now that an error correcting code @xmath46 exists and all conditions ( [ compression ] ) are satisfied .
if a quantum state @xmath0 is supported on the code then @xmath65 and calculation of the entropy exchange ( [ osigma ] ) simplifies , @xmath66 in this way we have shown that the error correction matrix @xmath57 is _ equal _ to the auxiliary matrix @xmath26 of lindblad , provided the initial state belongs to the code subspace . from another direction ,
given an error correcting code @xmath67 for a map @xmath2 , in @xcite it was shown that there is a quantum state @xmath68 and an isometry @xmath69 such that for all @xmath70 , @xmath71 the result , which can be seen as a consequence of the decoupling condition of @xcite , gives an explicit way to `` see '' a code at the output stage of a quantum process for which the code is correctable . the result ( and its subsystem generalization
see below ) may also be viewed as a formalisation of the _ subsystem principle _ for preserving quantum information @xcite . from the proof of this result
one can see directly that the entropy of @xmath68 satisfies @xmath72 .
this equality follows also from the fact that @xmath68 and @xmath57 can be interpreted as the states obtained by partial trace of an initially pure state with respect to two different subsystems .
thus , from multiple perspectives we find motivation for the following : [ entdefn ] given a quantum operation @xmath2 with kraus operators @xmath73 and a code @xmath67 with matrix @xmath57 given by ( [ compression ] ) , we call the von neumann entropy @xmath74 the _ entropy of _
@xmath75 _ relative to _ @xmath2 .
the entropy of a code depends only on the map and the subspace defined by @xmath76 , not on any particular state in the code subspace .
thus , the entropy exchange will be the same for all initial states supported on the code subspace and is therefore a property of the code itself . in the following result
we determine what possible values the code entropy can take , and we derive a characterisation of the extremal cases in terms of both the code and the map .
[ ucccase ] let @xmath2 be a quantum operation and let @xmath77 be a code with matrix @xmath57 given by ( [ compression ] ) .
then @xmath78 belongs to the closed interval @xmath79 $ ] , where @xmath80 is the choi rank of @xmath2 .
furthermore , the extremal cases are characterised as follows : * @xmath81 if and only if @xmath77 is a unitarily correctable code for @xmath2 . *
@xmath82 if and only if @xmath83 is a non - degenerate code for @xmath2 .
* by lemma [ upperbound ] and the subsequent discussion , the maximal entropy case occurs when the rank of @xmath57 and @xmath59 coincide and the spectrum of @xmath57 is equally balanced ; that is , the code is non - degenerate .
this occurs ( by a standard spectral majorization argument ) precisely when the code entropy satisfies @xmath82 . for the minimal entropy case
, first suppose that @xmath67 is a ucc for @xmath2 .
then by ( [ superoperator ] ) there is a unitary operation @xmath84 such that @xmath85 .
thus by the unitary invariance of choi - kraus decompositions , it follows that @xmath86 for some scalars @xmath87 .
hence we have @xmath88 , where @xmath89 is the vector state with coordinates @xmath87 , and so @xmath90 .
on the other hand , suppose @xmath91 is rank one .
let @xmath69 be a scalar unitary that diagonalises @xmath57 .
it follows that @xmath69 induces a unitary change of representation for @xmath2 via ( [ compression ] ) from @xmath92 to @xmath16 .
but since @xmath93 is diagonal , only one @xmath94 , say @xmath95 , is non - zero , and hence by unitary invariance we have @xmath96 for some unitary @xmath97 .
thus @xmath98 , and the result follows . ' '' '' from an operational perspective , the numerical value of the entropy of a code allows us to quantify the number of ancilla qubits needed to perform a recovery operation .
specifically , the choi rank , @xmath80 , of a map @xmath2 gives the minimum number of kraus operators necessary to describe the map or , by stinespring s dilation theorem @xcite , the dimension of the ancilla required to implement @xmath2 as a unitary .
the rank of @xmath57 then gives the number of kraus operators necessary to describe the action of the map restricted to @xmath99 and thus the number of kraus operators , @xmath10 , necessary for a recovery operation in the usual measurement cum reversal picture of recovery .
again by stinespring s dilation theorem we need an @xmath10-dimensional ancilla to implement the recovery as a unitary , and thus this requires @xmath100 qubits .
hence the entropy is equal to zero for a unitarily correctable code , in which the action of the noise is unitary and thus requires no ancilla to implement the recovery operation .
if the code entropy is positive , then any state of the code can potentially evolve to any one of multiple locations in the system hilbert space under the action of the noise @xmath2 .
this fact has to be compensated by the recovery operation @xmath50 .
the maximal entropy case for a particular @xmath2 is characterised by evolution to each of these locations ( @xmath101 by lemma [ upperbound ] ) with equal probability ( by theorem [ ucccase ] ) . here the entropy and thus the number of qubits in the ancilla will be @xmath102 .
as an example from the stabilizer formalism , consider a three - qubit system with the usual notation @xmath103 , @xmath104 , @xmath105 , for pauli operators @xcite .
the single - qubit stabilizer code with generators @xmath106 , is spanned by @xmath107 and @xmath108 .
the set of operators @xmath109 form a correctable set of errors for this stabilizer thus we can consider a channel , a three - qubit bit - flip channel , comprised of these operators for example , the channel with kraus operators @xmath110 , @xmath111 , @xmath112 and @xmath113 .
using @xmath114 , ( [ compression ] ) tells us that the error correction matrix is @xmath115 the entropy of this code is therefore @xmath116 as we would expect , the minimum entropy is achieved when @xmath117 .
the maximum entropy occurs when , which we might also expect since this puts the auxiliary density matrix @xmath57 into the maximally mixed state . here , the choi rank of our map is @xmath118 as is the rank of @xmath57 , and the spectrum of @xmath57 is equally balanced .
so @xmath99 is non - degenerate for @xmath2 .
further , as we expect from theorem [ ucccase ] , the entropy is . since the rank of @xmath57 is the number of kraus operators required for a recovery operation we find that in order to dilate the recovery operation to a unitary process , we need a @xmath118-dimensional space , or equivalently a @xmath119-qubit ancilla , which matches the value of the entropy .
we next consider a variant of this noise model , in which the third qubit undergoes no error and the first and second qubits are flipped with probability equal to that of no error .
thus the map @xmath2 has noise operators @xmath120 , each weighted by @xmath121 .
the entropies for a trio of correctable codes for @xmath2 are given in table [ enttable ] ( vectors are assumed to be normalised ) .
the first code is non - degenerate , and thus yields the maximal entropy for this noise model .
the second code has positive entropy less than the maximum since it is partially degenerate for @xmath2 .
indeed , one can check that @xmath122 ( and @xmath123 ) act trivially on the code , whereas @xmath124 maps the code to an orthogonal subspace .
the final code shows zero entropy since it is fully degenerate for @xmath2 .
in fact , as can be directly verified it is a decoherence - free subspace for @xmath2 .
.codes for the noise model @xmath125 . [ cols="^,^",options="header " , ] we next consider an extension of code entropy to the case of operator quantum error correcting subsystem codes .
we shall only introduce this notion here and leave a potential further investigation for elsewhere .
subsystem codes were formally introduced under the umbrella of _ operator quantum error correction _
@xcite , a framework that unifies the active and passive approaches to quantum error correction .
these codes now play a central role in fault tolerant quantum computing .
let @xmath126 be a hilbert space of finite dimension @xmath3 .
any decomposition @xmath127 determines subsystems @xmath128 and @xmath129 of @xmath126 .
given a quantum operation @xmath2 acting on @xmath126 , we say that a subsystem @xmath129 of @xmath126 is _ correctable _ for @xmath2 if there exists maps @xmath50 on @xmath126 and @xmath130 on @xmath128 such that @xmath131 where @xmath132 and @xmath133 is the projection onto the subspace @xmath134 .
subsystem codes generalize standard ( subspace ) codes ( [ superoperator ] ) in the sense that subspaces may be regarded as subsystems with trivial ancilla ( @xmath135 ) .
thus , it is natural to suggest that a notion of entropy for subsystem codes should generalize the subspace definition .
we find motivation for such a notion through the main result of @xcite alluded to above . to every correctable subsystem for @xmath2
, there are subsystems @xmath136 and @xmath137 , a map @xmath138 from @xmath128 to @xmath136 and a unitary map @xmath139 from @xmath129 to @xmath140 such that @xmath141 recall that the _ maximal output entropy _ of a channel @xmath50 is given by @xmath142 .
this motivates the following .
let @xmath2 be a quantum operation with correctable subsystem @xmath129 that satisfies ( [ subsystemcode ] ) .
then we define the entropy of @xmath129 relative to @xmath2 as the maximal output entropy of the associated ancilla channel , @xmath143 from ( [ ksresult ] ) .
observe that this generalizes definition [ entdefn ] , since a density operator may be regarded as a channel from a one - dimensional input space .
the minimal entropy case is characterized by the ancilla channel @xmath138 having range supported on a one - dimensional subspace .
such codes are unitarily correctable , in fact as subspaces , but the converse is not true . instead , in the more general subsystem setting
, the minimal entropy case is described by the associated ancilla subsystem @xmath128 undergoing `` cooling '' to a fixed state . in principle one
should be able to conduct a deeper analysis of subsystem code entropy .
we leave this as an open investigation for elsewhere .
a binary unitary channel has the form @xmath144 where @xmath145 and @xmath146 denote two arbitrary unitary operators and the probability @xmath147 belongs to @xmath148 $ ] .
it is clear that the problem of finding an error correcting code subspace @xmath46 for the above map is equivalent to the case @xmath149 where @xmath150 .
the number @xmath10 of kraus operators is equal to @xmath119 , with @xmath151 and @xmath152 .
thus the error correction matrix @xmath57 is of size two and reads @xmath153 where @xmath154 is a solution of the _ compression problem _ for @xmath97 @xmath155 the set of solutions to this problem can be phrased in terms of the _ higher - rank numerical range _ of the matrix @xmath97 .
the rank-@xmath156 numerical range of @xmath97 is defined as @xmath157 given a dimension-@xmath156 that defines the size of the desired correctable code , each @xmath154 in @xmath158 corresponds to a particular correctable code defined by the associated projection @xmath159 that solves ( [ comp1 ] ) .
the following is a straightforward application of ( [ compression ] ) .
[ correctable - if - numrange ] given a binary unitary channel @xmath2 , there exists a rank-@xmath156 correctable code for @xmath2 if and only if the rank-@xmath156 numerical range of @xmath97 is non - empty .
thus , the problem of finding the correctable codes for a given binary unitary channel can be reduced to the problem of finding the higher - rank numerical range of @xmath97 .
this problem has recently been solved in its entirety @xcite .
most succinctly , in terms of the eigenvalues @xmath160 of @xmath97 , the @xmath156th numerical range of @xmath97 is the convex subset of the unit disk given by @xmath161 where @xmath162 is the set of linear combinations @xmath163 such that @xmath164 and @xmath165 .
figures 1.a and 1.b depict the case of a generic two - qubit unitary ( @xmath166 ) with @xmath167 , while figure 1.c shows the case of a generic two - qutrit unitary noise ( @xmath168 ) with @xmath169 . for unitary matrices
@xmath97 describing a bi - unitary channel : two qubit system , a ) example 2 with @xmath170 , b ) case with @xmath171 for which code entropy is smaller , c ) two qutrit case with @xmath172 chosen to maximize its modulus @xmath173 and to minimize the code entropy @xmath174 . ] with a particular @xmath154 in - hand , straightforward algebra provides us with the spectrum of the matrix ( [ lambda1 ] ) , @xmath175 which allows us to calculate the entropy of the code , @xmath176 we are then led to the following : [ entropy - biunitary ] a minimum entropy rank-@xmath156 code for a binary unitary channel @xmath177 corresponds to any code for which the magnitude @xmath178 of the compression values @xmath179 is closest to unity , while the maximum entropy corresponds to @xmath178 closest to zero
. moreover , a code with minimal entropy can be constructively obtained .
* proof . *
the first statement follows directly from an application of first and second derivative tests on ( [ lambda5 ] ) , constrained to the unit disk .
a minimal entropy code can be explicitly constructed based on the analysis of higher - rank numerical ranges @xcite . ' '' '' as an illustration of the code construction in the simplest possible case ( @xmath180 ) , let @xmath97 be the unitary with spectrum depicted in figure 1.a , @xmath181 .
let @xmath182 be the associated eigenstates , @xmath183 .
in this case we have @xmath184 so @xmath170 , and one can check directly that a single qubit correctable code for @xmath2 is given by @xmath185 , where @xmath186 for a concrete example , in the case that @xmath187 , ( [ lambda5 ] ) yields a code entropy of .
the general case requires a more delicate construction , nevertheless it can be done .
the `` eigenstate grouping '' procedure used above can be applied whenever @xmath156 divides @xmath3 .
for instance , in the generic @xmath188 and @xmath169 case depicted in figure 1.c , a single qutrit code can be constructed for all @xmath154 in the region @xmath189 .
the states @xmath190 , @xmath105 , can be constructed in an analogous manner by grouping the nine eigenstates for @xmath97 into three groups of three , and writing @xmath154 in three different ways as a linear combination of the associated unimodular eigenvalues @xmath191 , @xmath192 .
however , without going into the details of this construction we can still analyze the corresponding code entropies . for simplicity assume the nine eigenvalues are distributed evenly around the unit circle with @xmath193 . by theorem [ entropy - biunitary ]
we know that the entropy will be minimized for any @xmath154 that gives the minimum distance from @xmath189 to the unit circle .
an elementary calculation shows that one such @xmath154 , given by the intersection of the lines through the first and seventh , and sixth and ninth eigenvalues ( counting counterclockwise ) is approximately @xmath194 . with the probability @xmath187
, the corresponding error correction matrix @xmath57 has spectrum @xmath195 .
thus , ( [ lambda3 ] ) and ( [ lambda5 ] ) yield the minimal qutrit code entropy for this channel as @xmath196 on the other hand , as @xmath197 belongs to @xmath189 , by theorem [ entropy - biunitary ] and ( [ lambda1 ] ) we also see the maximal entropy for @xmath187 occurs for any code with @xmath170 . in such cases we have @xmath57 spectrum @xmath198 , and
hence the maximal entropy is @xmath199 .
changing focus briefly , if we fix an arbitrary unitary @xmath97 , then we could consider the family of channels determined by varying the probability @xmath147 .
it follows from ( [ lambda5 ] ) that the channel with the correctable codes of maximal entropy corresponds to the @xmath200 channel , and the channels whose correctable codes possess minimal entropy correspond to @xmath201 and @xmath202 .
indeed , the value of @xmath154 depends on @xmath97 but not @xmath147 , thus a given @xmath154 will solve ( [ comp1 ] ) for any @xmath147 and so @xmath154 can be chosen independently using the above theorem .
the result once again follows from application of first and second derivative tests with @xmath147 between @xmath203 and @xmath204 .
the following results show that the entropy of a code for a binary unitary channel can be regarded as a measure of how close the code is to a decoherence - free subspace .
[ uccdfs ] if @xmath2 is a binary unitary channel , then the sets of unitarily correctable subspaces and decoherence - free subspaces coincide .
* proof . *
as proved in @xcite , for a bistochastic ( unital ) map @xmath2 , a condition satisfied by every binary unitary channel , the unitarily correctable codes ( respectively the decoherence - free subspaces ) for @xmath2 are imbedded in the fixed point algebra for the map @xmath205
( respectively @xmath2 ) , where @xmath206 is the hilbert - schmidt dual map of @xmath2 . in particular , it follows from this fact that the former set is given by the set of operators that commute with @xmath97 and @xmath207 , whereas the latter is the set of operators that commute with @xmath97 . by the spectral theorem
these two sets coincide . ' '' '' [ dfsbuc ] let @xmath2 be a binary unitary channel .
then there is a rank-@xmath156 code @xmath67 of zero entropy , @xmath81 , for @xmath2 if and only if there is a @xmath156-dimensional decoherence - free subspace for @xmath2 if and only if there exists @xmath208 .
* a @xmath156-dimensional decoherence - free subspace for @xmath2 corresponds to an eigenvalue @xmath154 of @xmath97 with multiplicity at least @xmath156 ( see @xcite and the references therein ) ; that is , @xmath208 .
the rest follows from the lemma and previous theorem . ' '' '' in order to further illustrate these results , consider again the case of an arbitrary two - qubit system ( @xmath166 ) .
the correctable codes with largest entropy are those with @xmath200 and so the spectrum of @xmath57 reads @xmath209 in the two - qubit case , the complex number @xmath154 is given by the point inside the unit circle at which two diagonals of the quadrangle formed by the spectrum of @xmath97 cross ( see figures 1.a and 1.b ) .
consider a special case of the problem where @xmath97 has a doubly degenerated eigenvalue , so that @xmath210 . for example , @xmath97 could be any ( non - identity ) element of the two - qubit pauli group .
then the spectrum of @xmath57 consists of @xmath211 which implies @xmath212 ( despite @xmath147 having been chosen for the largest entropy correctable codes ) .
hence @xmath57 is pure and there exists a decoherence free subspace the one spanned by the degenerated eigenvalues of @xmath97 . in general , for binary unitary channels one may use the entropy ( [ lambda5 ] ) as a measure quantifying to what extend a given error correction code is close to a decoherence - free subspace .
for instance , any channel ( [ biunitary2 ] ) acting on a two qubit system and described by unitary matrix @xmath150 of size @xmath118 may be characterized by the radius @xmath213 of the point in which two diagonals of the quadrangle of the spectrum cross .
the larger @xmath173 , the smaller entropy @xmath214 , and the closer the error correction code is to a decoherence - free space .
the code entropy can be also used to classify codes designed for a binary unitary channel acting on larger systems .
for instance in the case of two qutrits , @xmath215 , one can find a subspace supported on @xmath169 dimensional subspace .
the solution is by far not unique and can be parametrized by complex numbers @xmath154 belonging to an intersection of @xmath216 triangles , which forms a convex set of a positive measure . from this set
one can thus select a concrete solution providing a code @xmath46 , such that @xmath213 is the largest , which implies that the code entropy , @xmath214 , is the smallest
see figure 1.c .
such an error correction code is distinguished by being as close to a decoherence - free subspace as possible .
we have investigated a notion of entropy for quantum error correcting codes and quantum operations .
the entropy has multiple natural realisations through fundamental results in the theory of quantum error correction .
we showed how the extremal cases are characterised by unitarily correctable codes and decoherence - free subspaces on the one hand , and the non - degenerate case determined by the choi matrix of the map on the other .
we considered examples from the stabilizer formalism , and conducted a detailed analysis in the illustrative case of binary unitary channels .
recently developed techniques on higher - rank numerical ranges were used to give a complete geometrical description of code entropies for binary unitary channels ; in particular , the structure of these subsets of the complex plane can be used to visually determine how close a code is to a decoherence - free subspace .
we also introduced an extension of code entropy to subsystem codes , and left a deeper investigation of this notion for elsewhere .
it could be interesting to explore further applications of the code entropy in quantum error correction .
for instance , although quantum error correction codes were originally designed for models of discrete time evolution in the form of a quantum operation , generalizations to the case of continuous evolution in time @xcite have been investigated .
further , we have investigated perfect correction codes only , for which the error recovery operation brings the quantum state corrupted by the noise back to the initial state with fidelity equal one .
such perfect correction codes may be treated as a special case of more general approximate error correction codes @xcite .
another recent investigation @xcite includes analysis that suggests the measurement component of recovery may prove to be problematic in quantum error correction , and hence may motivate further investigation of unitarily correctable codes .
we thank the referee for helpful comments .
d.w.k . was partially supported by nserc grant 400160 , by nserc discovery accelerator supplement 400233 , and by ontario early researcher award 48142 .
was partially supported by an ontario graduate scholarship . k. .
acknowledges support of an european research project scala and the special grant number dfg - sfb/38/2007 of polish ministry of science .
d. gottesman .
quantum error correction and fault tolerance . in j .-
francoise , g.l .
naber , and s.t .
tsou , editors , _ encyclopedia of mathematical physics _ , volume 4 , page 196 .
oxford , elsevier , 2006 , quant - ph/0507174 . | we define and investigate a notion of entropy for quantum error correcting codes .
the entropy of a code for a given quantum channel has a number of equivalent realisations , such as through the coefficients associated with the knill - laflamme conditions and the entropy exchange computed with respect to any initial state supported on the code . in general the entropy of a code
can be viewed as a measure of how close it is to the minimal entropy case , which is given by unitarily correctable codes ( including decoherence - free subspaces ) , or the maximal entropy case , which from dynamical choi matrix considerations corresponds to non - degenerate codes .
we consider several examples , including a detailed analysis in the case of binary unitary channels , and we discuss an extension of the entropy to operator quantum error correcting subsystem codes . |
high gain fel amplifiers are of interest for a variety of potential applications that range from x - ray lasers @xcite to ultraviolet mw - scale industrial lasers @xcite .
there are various versions of the high gain fel amplifier .
a number of high gain fel amplifier concepts may prove useful for xfel applications .
two especially noteworthy ones are the fel amplifier with a single uniform undulator @xcite ( see fig .
[ fig : a1 ] ) and the distributed optical klystron @xcite ( see fig . [
fig : a2 ] ) .
the high gain cascade klystron amplifier described in this paper is an attractive alternative to other configurations for operation in the x - ray wavelength range ( see fig .
[ fig : a ] ) .
let us first investigate qualitatively the processes that take place in any fel amplifier ; this investigation should also enable us to distinguish somewhat more accurately between the so - called distributed optical klystron concept and the new high gain klystron amplifier concept . in any fel , several processes must take place , either separately or simultaneously . the first process that must occur in any fel is energy modulation .
energy modulation means the application of radiation fields to the electrons as they move in their trajectories under the action of undulator magnetic fields , in such a way that the energies of these electrons are varied in some periodic fashion .
it is then necessary to take advantage of this phase - dependent motion . the second process , then , that must occure in a fel is the conversion of this variation in the motion of the electrons into a usable form .
it is necessary to turn the variations in the electron energy into a radiation frequency current , which can deliver energy to the electromagnetic field ; i.e. , the variations in velocity must be converted into density variations .
the ways in which this conversion is brought about are numerous , and fel amplifier types are distinguished by their conversion mechanism
. it may be useful to illustrate processes at this point to see how energy modulation and conversion into density modulation take place in some of the types of fel amplifier . in a conventional high gain fel amplifier with a single uniform undulator energy modulation and bunching
occurs both simultaneously and continuously along the undulator .
energy modulation occurring at any one point continues to affect bunching at points beyond , with additional modulation being added as the beam moves along .
the problem of electromagnetic wave amplification in the uniform undulator refers to a class of self - consistent problem . to solve the problem ,
the field equations and equations of particle motion should be solved simultaneously .
when the fel amplifier operates in linear regime the transverse distribution of the radiation field remains fixed , while its amplitude grows exponentially with undulator length .
perhaps the simplest fel amplifier , from pedagogical point of view , is the high gain klystron amplifier .
this simplicity arises from the fact that the important processes of energy modulation of the beam and subsequent conversion of the energy modulation to density modulation occur in distinctly separate regions of the fel and thus may be considered in a sequential manner .
this separability of functions aids materially in developing an understanding of modulation and bunching phenomena . in its simplest configuration
the klystron amplifier consists of an input undulator ( energy modulator ) , and an output undulator ( radiator ) separated by a dispersion section . in the input undulator an incident electron beam is subjected to a radiation field , which produces an energy modulation , a modulation which after dispersion section , is converted into an density modulation in the beam . in the output undulator
the bunched beam radiates coherently .
this process , of conversion of energy modulation to bunching by the action of a dispersion section , is one of the characteristics of the klystron and is one of the reasons why this device is so useful in many connections .
we have considered the two - undulator klystron amplifier . in some experimental situations
this simplest configuration is not optimal .
for example , this study has shown that in the soft x - ray wavelength range the maximal ( intensity ) gain of this amplifier does not exceed 30 - 40 db .
the obvious and technically possible solution of the problem of gain increase might be to use a klystron with three or more undulators .
suppose that we have a three undulator ( or so - called two - cascade ) klystron , and that we introduce a very small signal into the first undulator .
this will result in a small value of the amplitude of energy modulation , and a resulting rather small power output from the second undulator .
nevertheless , the radiation field amplitude in the second undulator will still be much greater than that in the first undulator , on account of the considerable amplification of the first cascade .
we now disregard the power produced by the second undulator , but consider its effect on the electron beam passing through its magnetic system .
the considerable radiation field in the second undulator will produce a further energy modulation of the beam and , when it emerges on the far side of the second undulator , it will have a large energy modulation , which , with a second dispersion section , will result in optimum bunching .
this bunched beam then enters the third ( radiator ) undulator , and produces a large amount of power . in other words
, we have essentially combined two stages of amplification by incorporating essentially two klystrons in a single unit .
the final power is no greater than could be produced by a two - undulator klystron with the same output undulator , but we can secure this power with a much weaker input signal .
let us consider now the problem of a periodically spaced sequence of undulators and dispersion sections , with the electron beam traversing each in succession .
the larger the number of cascades the longer is the amplifier length .
suppose that the increase of total length ( gain ) of amplifier is tempered by an accompanying decrease in the length ( gain ) of cascade .
from this approach at very large number of cascades and small gain per cascade pass we are led to the other device , which was the subject of the papers @xcite : the `` distributed optical klystron '' . at first glance
this is essentially the extension klystron amplifier , extended to a very large number of cascades .
however , as noted earlier , this amplifier is considerably more complicated in principle than a klystron amplifier with high gain per cascade pass . when the gain per cascade pass is small , the cascades can not be considered independent .
it is important to realize that the cascades are connected optically , so that the radiation produced , for example , into the first undulator passes on into the second and produces additional energy modulation of the beam .
it is clear that we can not conveniently treat the theory of distributed klystron by the same method we used in the klystron case .
because of the low gain per cascade pass used for the distributed klystron , it can not be approximated as scheme with separate functions , and the klystron amplifier theory has all based on that assumption . a quite different approach to the problem , which proves to be more suited to mathematical development of distributed klystron theory ,
is more along the lines of an analogy with the theory of conventional fel amplifier with uniform undulator .
the problem of electromagnetic wave amplification in the distributed klystron refers to a class of self - consistent problems .
for instance , when the distributed klystron operates in linear regime its amplitude gain grows exponentially with cascade numbers . from a study of the rate of exponential build - up of the electromagnetic wave we can find the gain of the amplifier .
our theory of klystron amplifier is based on the assumption that beam density modulation does not appreciably change as the beam propagates through the undulator .
this approximation means that only the contributions to the radiation field arising from the initial density modulation are taken into account , and not those from the induced bunching . at very long klystron undulators
we are led to device , which is the subject of the papers @xcite .
two - segment high gain fel amplifier is variant of the standard high - gain fel amplifier with uniform undulator .
because of the high gain per undulator pass used for this modification , it can not be approximated as scheme with separate functions .
the klystron amplifier analysis is not only relevant from a pedagogical point of view .
there are reasons to expect that the new concept of fel amplifier discussed here can be of interest for short wavelength applications such as x - ray sase fels .
generation of x - ray radiation using klystron amplifiers has many advantages , primarily because of the separation of electronic processes , gain , and potentially low sensitivity to alignment . in principle
, the gain of the klystron amplifier can be increased indefinitely by adding additional cascades along the beam . in practice , with three - cascade klystron amplifier , a gain in excess of 80 db may be obtained in the soft x - ray wavelength range . at this level , the shot noise of the beam is amplified up to saturation and a cascade klystron amplifier operates as soft x - ray sase fel .
now our intention is to develop certain physical ideas about the operation of the klystron amplifier , the characteristics of its radiation , and those features that set it apart from other fel sources in the short wavelength region of the spectrum .
the resonance properties of the radiation of the electron beam in the klystron can be understood as follows . since
a dispersion section is a nonresonant device , the klystron bandwidth is restricted by the resonance properties of the cascade undulator , that is , by its number of periods @xmath0 .
the typical amplification bandwidth of the klystron in the soft x - ray wavelength range is on the order of one percent @xmath1 .
since we study the start - up from shot noise , we assume the input current to have a homogeneous spectral distribution . at large number of undulator periods the spectrum of transversely coherent fraction of radiation
is concentrated within the narrow band , @xmath2 .
the actual physical picture of start - up from noise should take into account that the fluctuations of current density in the electron beam are uncorrelated not only in time but in space too .
thus a large number of transverse radiation modes are excited when the electron beam enters the undulator .
these radiation modes have different gain .
obviously , as cascade number progresses , the high gain modes will predominate more and more and we can regard the klystron as a filter , in the sense that it filters from arbitrary radiation field those components corresponding to the high gain modes . if we consider the undulator radiation from the point of view of paraxial optics , then we immediately see that the high gain modes are associated with radiation propagating along the axis of the undulator , as opposed to radiation propagating at an angle to the axis which has a low gain .
consider an electron beam of radius @xmath3 moving in the ( cascade ) undulator with length @xmath4 .
the parameter @xmath5 can be referred to as the fresnel number . here
@xmath6 is the radiation wavelength .
the requirement for the mode selection to be quickly holds at small value of the fresnel number . hence ,
for thin electron beam with fresnel number of @xmath7 , the emission will emerge in a single ( fundamental ) transverse mode and the degree of transverse coherence of output radiation will approach unity . as mentioned above , amplification in each cascade
is a two - step process .
since we study the start - up from shot noise there is no electromagnetic field at the undulator entrance and the modulation of the beam density serves as the input signal for the klystron amplifier .
the radiation field produced an ponderomotive potential across the undulator ; as is now known from general fel theory , this voltage modulates the energy of the electron beam .
then the energy modulation causes modulation of the beam density in the dispersion section .
if we can neglect beam bunching in the cascade undulator , the amplitude of energy modulation at the undulator exit , @xmath8 , is proportional to the resonance fourier component of beam current , @xmath9 .
here @xmath10 is the initial beam microbunching at the resonant wavelength , and @xmath11 is the peak beam current .
we now examine the behaviour of the electrons as they travel in a dispersion section .
like the situation with usual microwave klystron , the bessel function factor @xmath12 represents the microbunching at the dispersion section exit .
the parameter @xmath13 in our case is proportional to the product of the amplitude of energy modulation @xmath8 and the compaction factor of the dispersion section @xmath14 . in the high gain linear regime
@xmath13 is much smaller then unity and we make the approximation @xmath15 .
thus , microbunching at the cascade exit is proportional to input signal amplitude , @xmath16 .
when a cascade operates in the high gain linear regime the gain per cascade pass is independent of input signal amplitude and proportional to the product of the compaction factor and the peak beam current , @xmath17 .
if we build a klystron , we want to have small number of cascades and high gain .
one might think that all we have to do is to get enough gain per cascade pass - we can always increase compaction factor @xmath14 and we can always increase @xmath18 , and there is no reason why we can not use the simple two - undulator klystron scheme .
so the limitations of a gain per cascade pass are not that it is impossible to build a dispersion section that has a very large compaction factor .
in fact , one of the main problems in klystron amplifier operation is preventing spread of microbunching due to energy spread into the electron beam .
this energy spread is unimportant when the compaction factor of dispersion section is relatively small , but we shell see that energy spread is of fundamental importance in the case of very large compaction factor .
to get a rough idea of the spread of the microbunching , the position of the particles within the electron bunch at the dispersion section exit has a spread which is equal to @xmath19 , where @xmath20 is the local energy spread into the electron beam . and
how big is @xmath21 in the spread - out microbunching ?
we know that uncertainty in the phase of the particles is about @xmath22 .
therefore , the largest compaction factor that we can have is approximately @xmath23 .
this is what we wanted to deduce - that the gain maximum per cascade pass varies as ratio @xmath24 .
it is of course desirable that the klystron gain be maximum .
an experimenter can easily tune the gain of the klystron by tuning the magnetic fields in the dispersion sections .
it should be pointed out that the local energy spread plays a very important role in the operation of the fel klystron amplifier , and this characteristic distinguishes the klystron from conventional fel amplifier .
electron bunches with very small transverse emittance and high peak current are needed for the operation of conventional xfels .
this is achieved using a two - step strategy : first generate beams with small transverse emittance using an rf photocathode and , second , apply longitudinal compression at high energy using a magnetic chicane .
although simple in first - order theory , the physics of bunch compression becomes very challenging if collective effects like space charge forces and coherent synchrotron radiation forces ( csr ) are taken into account .
wakefields of bunches as short as 0.1 mm have never been measured and are challenging to predict .
the situation is quite different for klystron amplifier scheme described in our paper .
an distinguishing feature of the klystron amplifier is the absence of apparent limitations which would prevent operation without bunch compression in the injector linac . according to our discussion above
, the gain maximum per cascade pass is proportional to the peak current and inversely proportional to the energy spread of the beam .
since the bunch length and energy spread are related to each other through liouville s theorem , the peak current and energy spread can not vary independently of each other in the injector linac . to extent that local energy spead is proportional to the peak current , which is usually the case for bunch compression
, the gain will be independent of the actual peak current .
we see , therefore , that klystron gain in linear regime depends only on the actual photoinjector parameters .
this incipient proportionality between gain and @xmath25 is a temptation , in designing an xfel , to go to very high values of @xmath25 and very long values of bunch length .
let us present a specific numerical example for the case of a 4th generation soft x - ray light source based on the use rf photoinjector and superconducting linac with duty factor 1 % .
the average brilliance of a klystron facility operating without bunch compression in the injector linac surpasses the spontaneous undulator radiation from 3rd generation synchrotron radiation facilities by 4 or more orders of magnitude . decreasing
the peak current also decreases the peak brilliance of the sase fel radiation by about of factor 100 , but this is still 6 order of magnitude higher than that of 3rd generation synchrotron radiation sources .
our studies have shown that the soft x - ray cascade klystron holds great promise as a source of radiation for generating high power single femtosecond pulses .
the obvious temporal limitation of the visible pump / x - ray probe technique is the duration of the x - ray probe . at a klystron facility
operating without bunch compression , the x - ray pulse duration is about 10 ps .
this is longer than the timescale of many interesting physical phenomena .
the new principle of pump - probe techniques described in section 6 offers a way around this difficulty .
section 6 also deals with the design strategy for the multi - user distribution system for an x - ray laboratory .
an x - ray laboratory should serve several , may be up to ten experimental stations which can be operated independently according to the needs of the user community . on the other hand , the prefered layout of a conventional sase fel is a linear arrangement in which the injector , accelerator , bunch compressors and undulators are nearly collinear , and in which the electron beam does not change direction between accelerator and undulators .
the situation is quite different for the klystron amplifier scheme proposed in our paper .
since it operates without bunch compression in the injector linac , the problem of emittance dilution in the bending magnets does not exist .
an electron beam distribution system based on unbunched electron beam can provide efficient ways to generate a multi - user facility - very similar to present day synchrotron radiation facilities .
it may be wondered , after hearing of all of the wonderful properties of the high gain klystron amplifier , why this advance in xfel techniques occurred only this year .
it should be note that the xfel which was proposed in 1982 @xcite has had nearly 20 years of development .
one of the explanations is as follows .
for years we were led to believe that local energy spread into electron beam which delivered from a photoinjector is relatively large .
recent analysis of experimental results obtained at ttf sase fel @xcite shows that the value of the local energy spread should be revised .
after bunch compression , it is expected to be about 0.2 mev ( for @xmath26 ) or smaller which is significantly less than the previously projected value of 1 mev .
this decrease significantly improves operation of klystron amplifier and extends the safety margin for x - ray klystron facility operation .
this paper is intended primarily for the experimental scientist or xfel program manager who has to design xfels and who wishes to undestand the operating principles of xfels .
the material to be discussed provides the necessary background information required for making strategic xfel design decisions . in this paper
the electronic processes that occur in an xfel will be discussed in elementary fashion .
the xfel theory has a , not entirely undeserved , reputation for difficulty and obscurity .
this is due partly to the complexity of systems in which ( shot ) noise is input signal and partly to the unfamiliar nature of some of the mathematical tools involved ( integrodifferential equations ) .
klystron amplifiers are one of the simplest examples of high gain fel amplifiers and therefore we shall use them as an introduction to the techniques of xfel theory . in the klystron amplifier case little demand
is made on the reader s mathematical ability .
the mathematics that is involved is particularly simple , involving very simple differential equations and algebraic operations and no integrodifferential equations .
such introduction to the operating principles of xfels has never been done before , to our knowledge .
klystron amplifiers amplify the input signal . to describe the klystron amplifier operation , we first define the initial conditions .
important practical kind of initial condition refers to the case when there is no radiation field at the undulator entrance and the modulation of the beam current density serves as the input signal for the amplifier . as stated in the introductory section , in undulator two physical processes
take place simultaneously : \1 ) the modulated electron beam excites electromagnetic waves which are propagated along the undulator ; \2 ) the electric field components of these waves produce a modulation of the energy of the electrons .
the approach discussed in this section consists of deriving for these two process separate expressions which , when joined , give quantitative information about the composite phenomena taking place in the undulator .
we assume that the density modulation effect is negligible .
hence , electrons passing through the first undulator receive only energy modulation , but no density modulation .
density modulation present when the electrons enter the second undulator , must have developed while the electrons were traveling in the dispersion section .
this effect will be discussed later .
the simplest form of bunched beam is found if we have swarm of electrons , all moving in the same direction ( say the @xmath27-direction ) with the same velocity ( say @xmath28 ) , but in which the number of electrons per unit volume ( say @xmath29 ) depends on @xmath27
. then if the charge on the electron @xmath30 we can easily find the current density at any value of @xmath27 .
the quantity @xmath29 must really be a function of @xmath31 , on account of the velocity @xmath28 of the electrons .
thus , the current density at point @xmath27 , at time @xmath32 , is @xmath33 , where @xmath34 indicates the function @xmath29 of the argument @xmath31 .
unless @xmath29 is a constant independent of its argument , this will lead to a current density which varies with time .
the simplest variation of @xmath29 for our present purposes is a superposition of a constant , or d - c component , and a periodic function of @xmath31 , with an angular frequency @xmath35 .
first we consider the case where @xmath29 is simply a constant plus a single cosine term : @xmath36 $ ] .
we must note one important fact : @xmath29 can never become negative , and hence @xmath10 can not be greater than unity . to understand the basic principles of klystron amplifier operation , let us consider the helical undulator .
the magnetic field on the axis of the helical undulator is given by @xmath37 where @xmath38 is the undulator wavenumber and @xmath39 are unit vectors directed along the @xmath40 and @xmath41 axes of the cartesian coordinate system @xmath42 .
the lorentz force @xmath43 is used to derive the equations of motion of electrons with charge @xmath30 and mass @xmath44 in the presence of the magnetic field . the explicit expression for the electron velocity in the field of the helical undulator
has the form : @xmath45 \ , \ ] ] which means that the electron in the undulator moves along a constrained helical trajectory parallel to the @xmath27 axis .
the angle of rotation is given by the relation @xmath46 , where @xmath47 is the relativistic factor and @xmath48 .
as a rule , the electron rotation angle @xmath49 is small and the longitudinal electron velocity @xmath50 is close to the velocity of light , @xmath51 .
let us consider a bunched relativistic electron beam moving along the @xmath27 axis in the field of a helical undulator . in what follows we use the following assumptions : i ) the electrons move along constrained helical trajectories in parallel with the @xmath27 axis ; ii ) the radius of the electron rotation in the undulator , @xmath52 , is much less than the transverse size of the electron beam .
next let us assume that electron beam density at the undulator entrance is simply @xmath53 $ ] , where @xmath54 in other words we consider the case in which there are no variations in amplitude and phase of the density modulation in the transverse plane . under these assumptions
the transverse current density may be written in the form @xmath55 \ , \ ] ] where we calibrated the time in such a way that current density has its maximum at time @xmath56 , at point @xmath57 . even through the measured quantities
are real , it is generally more convenient to use complex representation . for this reason , starting with real @xmath58 , one defines the complex transverse current density : @xmath59 \ .
\label{eq : c}\ ] ] transverse current have the angular frequency @xmath35 and two waves traveling in the same directions with variations @xmath60 and @xmath61 will add to give a total current proportional to @xmath62 .
the factor @xmath63 indicates a fast wave , while the factor @xmath64 indicates a slow wave .
the use of the word `` fast '' ( `` slow '' ) here implies a wave with a phase velocity faster ( slower ) than the beam velocity .
now we should consider the electrodynamic problem .
using maxwell s equations , we can write the equation for the electric field : @xmath65 then make use of the identity @xmath66 where @xmath67 can be found from the poisson equation . finally , we come to the inhomogeneous wave equation for @xmath68 : @xmath69 this equation allows one to calculate the electric field @xmath70 for given charge and current sources , @xmath71 and @xmath72 .
thus , equation ( [ eq : u1 ] ) is the complete and correct formula for radiation .
however we want to apply it to a still simpler circumstance in which the second term ( or , the current term ) in the right - hand side of ( [ eq : u1 ] ) provides the main contribution to the value of the radiation field . since in the paraxial approximation
the radiation field has only transverse components , we are interested in the transverse component of ( [ eq : u1 ] ) .
thus we consider the wave equation @xmath73 which relates the transverse component of the electric field to the transverse component of current density .
we wish to examine the case when the phase velocity of the current wave is close to the velocity of light .
this requirement may be met under resonance condition @xmath74 first we may point out that the statement of ( [ eq : s ] ) , the condition for the relation between @xmath75 and @xmath50 , is the condition for synchronism between the transverse electromagnetic wave and the fast transverse current wave with the propagating constant @xmath76 . with a current wave traveling with the same phase speed as the electromagnetic wave , we have the possibility of ( space ) resonance between electromagnetic wave and electrons .
if this is the case cumulative interaction between bunched electron beam and transverse electromagnetic wave takes place .
we are therefore justified in considering the contributions of all the waves except the synchronous one to be negligible .
it would be nice to find an explanation of resonance condition ( [ eq : s ] ) which is elementary in the sense that we can see what is happening physically .
let us call the plane electromagnetic wave velocity @xmath77 , which is of course , related to the phase constant by @xmath78 .
then the velocity of the beam relative to the electromagnetic wave is merely @xmath79 . with this relative velocity , the frequency of the electromagnetic waves as seen by the electrons
is give by @xmath80 , where we have used the obvious relation between the phase constant @xmath81 and the wavelength of the electromagnetic wave , @xmath82 .
relation for @xmath83 can be written as @xmath84 . under the condition
when ( [ eq : s ] ) holds , we see by comparison with that equation that this would be merely the undulator frequency @xmath85 .
accordingly , the condition of ( [ eq : s ] ) that we have imposed is merely that the electron velocity is such that as the electron run past the electromagnetic waves , the apparent frequency they see is the undulator frequency .
this is the basic effect involved in this interaction .
if the apparent frequency is the undulator frequency , a cumulative effect on the electromagnetic wave results , since the waves are being driven at a frequency corresponding to natural rotation frequency of the electrons in the magnetic field .
another way of saying the same thing is that in the frame of reference moving with the electrons , the electromagnetic waves have a doppler - shifted frequency equal to undulator frequency .
any state of transverse electromagnetic wave can always be written as a linear combination of the two base states ( polarizations ) . by giving the amplitudes and phases of these base states
we completely describe the electromagnetic wave state .
it is usually best to start with the form which is physically clearest .
we choose the cartesian base states and seek the solution for @xmath86 in the form @xmath87 + { \mathrm{c.c . } } \label{eq : f}\ ] ] here and in what follows , complex amplitudes related to the field are written with a tilde .
the description of the field given by ( [ eq : f ] ) is quite general .
however , the usefulness of the concept of carrier wave number is limited to the case where the amplitude is slowly varying function of @xmath27 . to determine the form of @xmath88 we substitute ( [ eq : c ] ) and ( [ eq : f ] ) into ( [ eq : u3 ] ) .
we have @xmath89\left\{\vec{\nabla}^{2}_{\perp } + \frac{2{\mathop{\mathrm{i}}}\omega}{c}\frac{\partial}{\partial z } + \frac{\partial^{2}}{\partial z^{2}}\right\}{\tilde{e}_{x } \choose \tilde{e}_{y } } + { \mathrm{c.c . } } \nonumber\\ & \mbox { } & = - 4\pi\frac{\omega}{c } { \cos(k_{\mathrm{w}}z ) \choose -\sin(k_{\mathrm{w } } ) } e\theta_{\mathrm{w}}a_{\mathrm{in } } n_{0}(\vec{r}_{\perp})\sin\omega(z / v_{z}-t ) \ .
\label{eq : u4a}\end{aligned}\ ] ] here @xmath90 is the laplace operator in transverse coordinates .
now we have an apparently simple pair of equations - and they are still exact , of course .
the question is , how to solve these equations ?
first we simplify the equations obtained by noting that for a radiation field it is reasonable to assume that @xmath91 are slowly varying functions of @xmath27 ( because of the radiation directivity ) so that @xmath92 may be neglected .
the corresponding requirement for the complex amplitude is @xmath93 . in other words ,
the radiation pulse must not change significantly while traveling through a distance comparable with the wavelength @xmath94 .
this assumption is not a restriction .
such is the case in all practical cases of interest .
differential equations becomes @xmath89\left\{\vec{\nabla}^{2}_{\perp } + \frac{2{\mathop{\mathrm{i}}}\omega}{c}\frac{\partial}{\partial z}\right\}{\tilde{e}_{x } \choose \tilde{e}_{y } } + { \mathrm{c.c . } } \nonumber\\ & \mbox { } & = 4\pi\frac{\omega}{c } { \cos(k_{\mathrm{w}}z ) \choose -\sin(k_{\mathrm{w } } ) } e\theta_{\mathrm{w}}a_{\mathrm{in } } n_{0}(\vec{r}_{\perp})\sin\omega(t - z / v_{z } ) \ .
\label{eq : u4}\end{aligned}\ ] ] the reader may well wonder why such a transformation is useful ; therefore we digress temporarily to address this question .
such a transformation immediately modifies the hyperbolic wave equations to the parabolic equations .
we should stress that the equations ( [ eq : u4 ] ) are preferable for an analytical solution , since the mathematical techniques are always connected with more conventional parabolic equations .
although equations ( [ eq : u4 ] ) can not be solved in general , we will solve them for some special cases .
these equations can be simplified further by noting that the complex amplitudes @xmath95 will not vary much with @xmath27 , especially in comparison with the exponential terms @xmath96 . the slow wave of transverse current oscillates very rapidly about an average value of zero and , therefore
, does not contribute very much to the rate of change of @xmath95 .
so we can make a reasonably good approximation by replacing these terms by their average value , namely , zero . we will leave them out , and take as our approximation :
@xmath97 even the remaining terms , with exponents proportional to @xmath98 will also vary rapidly unless @xmath99 is near zero .
only then will the right - hand side vary slowly enough that any appreciable amount will accumulate when we integrate the equations with respect to @xmath27 .
the required conditions will be met if @xmath100 in other words , we use the resonance approximation here and assume that complex amplitudes @xmath95 are slowly varying in the longitudinal coordinate . by `` slowly varying ''
we mean that @xmath101 . for this inequality
to be satistied , the spatial variation of @xmath95 within an undulator period @xmath102 has to be small .
equations ( [ eq : u5 ] ) are simple enough and can be solved in any number of ways .
one convenient way is the following . taking the sum and the difference of the two we get @xmath103 @xmath104
these equations describe the general case of electromagnetic wave radiation by the bunched electron beam in the helical undulator .
equation ( [ eq : u6 ] ) and ( [ eq : u7 ] ) refer to the right- and left - helicity components of the wave respectively . the solutions for the right- and
left - helicity waves are linearly independent .
it follows from ( [ eq : u6 ] ) and ( [ eq : u7 ] ) that only those waves are radiated that have the same helicity as the undulator field itself .
of course we could predict such a result . the electric field
, @xmath86 , of the wave radiated in the helical undulator in resonance approximation is circularly polarized and may be represent in the complex form : @xmath105 \ .
\label{eq : d}\ ] ] finally , the equation for @xmath106 can be written in the form @xmath107 equation ( [ eq : u8 ] ) is an inhomogeneous parabolic equation .
its solution can be expressed in terms of a convolution of the free - space green s function ( impulse response ) @xmath108 \ \label{eq : ir}\ ] ] with the source term .
when the right - hand side of ( [ eq : u8 ] ) is equal to zero , it transforms to the well - known paraxial wave equation in optics .
the radiation process displays resonance behaviour and the amplitude of electric field depends stronly on the value of the detuning parameter @xmath99 . with the approximation made in getting ( [ eq : u8 ] ) the equation can be solved exactly , but the work is a little elaborate , so we wo nt do that until later when we take up the problem of output radiation characteristics .
now we will find an exact solution for the case of perfect resonance .
when the parameters are tuned to perfect resonance , such that @xmath109 , the solution of the equation ( [ eq : u8 ] ) has the form @xmath110 \ , \label{eq : r}\ ] ] where @xmath111 and @xmath112 are the coordinates of the observation and the source point , respectively .
let us consider an axisymmetric electron beam with gradient profile of the current density .
in this case we have @xmath113 , where @xmath114 is the radial coordinate of the cylindrical system @xmath115 and @xmath116 describes the transverse profile of the electron beam . to be specific ,
we write down all the expressions for the case of a gaussian transverse distribution : @xmath117 where @xmath11 is the total beam current . at this point
we find it convenient to impose the following restriction : we focus only on the radiation see by an observer lying on the electron beam axis .
when @xmath118 and the beam profile is gaussian , we can write ( [ eq : r ] ) in the form @xmath119 \exp\left[\frac{{\mathop{\mathrm{i}}}\omega ( r^{\prime})^{2}}{2c(z- z^{\prime})}\right ] { \mathop{\mathrm{d}}}r^{\prime } \ .\ ] ] integrating first with respect to @xmath120 , we have @xmath121 where @xmath122 is the radiation wavenumber . the physical implications of this result are best understood by considering some limiting cases .
it is convenient to rewrite this expression in a dimensionless form .
after appropriate normalization it is a function of one dimensionless parameter only : @xmath123 where @xmath124 is the dimensionless coordinate along the undulator , @xmath4 is the total undulator length , @xmath125 is the diffraction parameter and @xmath126 is the normalized field amplitude .
the changes of scale performed during the normalization process , mean that we are measuring distance along the undulator and electron beam size as multiples of `` natural '' undulator radiation units .
let us study the asymptotic behaviour of the field amplitude at large values of the diffraction parameter @xmath127 .
in this case @xmath128 and we have asymptotically : @xmath129 now let us study the asymptote of a thin electron beam . in this case @xmath130 and
the function @xmath131 can be evaluated in the following way .
first we remark that the integral can be expressed as @xmath132 : @xmath133 the first integral should go from @xmath134 to @xmath135 , but @xmath134 is so far from @xmath135 that the curve is all finished by that time , so we go instead to @xmath136 - it makes no difference and it is much easier to do the integral . the integral is an inverse tangent function .
if we look it up in a book we see that it is equal to @xmath137 .
the second integral can be expressed as logarithm function .
thus we have @xmath138 it is interesting to note that in the case of a thin electron beam , the radiation field , acting on the electrons is almost constant along the undulator axis , which is a little strange , because the modulated electron beam radiates electromagnetic waves along the undulator .
but that is the way it comes out - fortunately a rather simple formula . ) and the logarithmic growth rate asymptote for conventional fel amplifier at small diffraction parameter ( see @xcite ) are ultimately connected ] special attention is called to the fact that in the thin electron beam case , at @xmath130 , amplitude @xmath139 is a complex function .
one immediately recognizes the physical meaning of the complex @xmath140 .
note that electric field ( response ) is given by fast wave of transverse current ( `` force '' ) times a certain factor .
this factor can either be written as @xmath132 , or as a certain magnitude @xmath141 times @xmath142 .
if it is written as a certain amplitude @xmath141 times @xmath142 , let us see what it means .
this tells us that electric field is not oscillating in phase with the fast wave of transverse current , which has ( at @xmath109 ) the phase @xmath143 , but is shifted by an extra amount @xmath144 .
therefore @xmath144 represent the phase shift of the response . to get an intuitive picture on what happens with the radiation beam inside the near zone according to equation ( [ eq : r ] ) ,
let us first choose thin beam asymptotic .
this is an example in which diffraction effects play an important role .
simple physical consideration can lead directly to a crude approximation for the radiation beam cross - section .
there is a complete analogy between the radiation effects of the bunched electron beam in the undulator and the radiation effects of a sequence of periodically spaced oscillators .
the radiation of these oscillators always interferes coherently at zero angle with respect to the undulator axis .
when all the oscillators are in phase there is a strong intensity in the direction @xmath145 .
an interesting question is , where is the minimum ? if we have a triangle with a small altitude @xmath146 and a long base @xmath27 , then the diagonal @xmath147 is longer than the base .
the difference is @xmath148 .
when @xmath149 is equal to one wavelength , we get a minimum . now why do we get a minimum when @xmath150 ? because the contributions of various oscillators are then uniformly distributed in phase from @xmath134 to @xmath151 . in the limit of small size of the electron beam interference will be constructive within an angle of about @xmath152 . in the limit of large electron beam size
, the spread of the angle distribution can be found from a two - dimensional fourier transform of the field .
the radiation field across the electron beam may be presented as a superposition of plane waves , all with the same wavenumber @xmath122 .
the value of @xmath153 gives the sine of the angle between the @xmath27 axis and the direction of propagation of the plane wave . in the paraxial approximation @xmath154 .
we can expect that the typical width of the angular spectrum should be of the order @xmath155 , a consequence of the reciprocal width relations of the fourier transform pair @xmath156 .
the boundary between these two asymptotes is about @xmath157 or ( another way to write it ) @xmath158 .
a rough estimate for the diffraction effects to be small is @xmath159 , which simply means that the diffraction expansion of the radiation at undulator length must be much less than the size of the beam .
another way to write this condition is @xmath160 .
the parameter @xmath127 can be referred to as the electron beam fresnel number .
our estimates show ( see fig .
[ fig : aps ] ) that in the case of a thin beam the square of the radiation beam grows linearly with the undulator length @xmath161 in the case of a wide electron beam the most of the radiation overlaps with the electron beam @xmath162 up to this point we only talked about electric field amplitude on the electron beam axis .
ultimately we want to consider the distribution of the electric field over the full electron beam cross - section .
the full analysis is a difficult analytical problem .
the calculations can be performed without great difficulty in two limiting cases , namely , the cases of diffraction parameter very large and very small comared with unity . to calculate equation ( [ eq : r ] )
we note that the behaviour of green s function ( [ eq : ir ] ) for @xmath163 approaches the behaviour of the delta function .
the source function @xmath164 does not vary very much across the region @xmath165 in the case of a wide electron beam : therefore we can replace it by a constant .
in other words , we simply take @xmath164 outside the integral sign and call it @xmath166 . in this case
the integral over source coordinates @xmath167 in ( [ eq : r ] ) is calculated analytically @xmath168 we emphasize the following features of this result . at exact resonance @xmath169 amplitude
@xmath106 is a real function .
this tells us that electric field is oscillating in phase with the fast wave of transverse current .
let us turn next to the second case where @xmath170 .
it should be noted that dealing with klystron amplifier this situation is more preferable than the previous one . in thin electron beam case
there is only a small variation in electric field amplitude over electron beam cross section .
this feature follows naturally from the fact that the radiation expands out and electron beam becomes thin with respect to the radiation beam . using ( [ eq : r ] ) , we find that the field amplitude inside the thin electron beam is given by the formula @xmath171 \qquad { \mathrm{as } } \quad \beta \ll 1 \ .
\label{eq : t1}\ ] ] the nonzero imaginary term is a manifestation of the fact that electric field is not oscillating in phase with the fast wave of transverse current . in ( [ eq : r ] ) we have the general solution of the problem of finding the field produced by modulated current , and this is the first part of the problem of the operation of the high gain klystron amplifier .
the next question which we must take up is the converse one : given an electric field of the type we have just considered , what is the electronic motion in such a field ?
since , the wavelength of the radiation pulse is much less than either the transverse electron beam size or the pulse length it can be accurately represented by a wave traveling at @xmath172 .
further , since the electron also travels at @xmath51 , the electron and the radiation remain essentially overlapped during the transit through the undulator .
although the radiation fields alone do not affect the electron trajectory , the interaction of the transverse radiation electric field with the transverse oscillations of the electron induced by the static undulator magnetic field causes the electron to either gain or lose energy . as simple example of the interaction between electrons and electromagnetic wave in an undulator we shall consider a circularly polarized plane electromagnetic wave propagating parallel to the electron beam .
the field of the electromagnetic wave has only a transverse component , so the energy exchange between the electron and the electromagnetic wave is due to the transverse component of the electron velocity .
the rate of electron energy change is @xmath173 where @xmath86 is the vector of the electric field of the wave : @xmath174 + \vec{e}_{y}\sin\left[\omega(z / c - t)\right]\right\ } \ .\ ] ] remembering that @xmath175 we find @xmath176 - \sin(k_{\mathrm{w}}z)\sin\left[\omega(z / c - t)\right]\right\ } \nonumber\\ & \mbox { } & = - e\theta_{\mathrm{w}}e\cos\left[k_{\mathrm{w}}z + \omega(z / c - t)\right ] \nonumber\\ & \mbox { } & = - e\theta_{\mathrm{w}}e\cos\psi \ . \label{eq:1a}\end{aligned}\ ] ] the phase @xmath177 has a simple physical interpretation and is equal to the angle between the transverse velocity of the particle , @xmath178 , and the vector of the electric field . for effective energy exchange between the electron and the wave ,
the scalar product @xmath179 should be kept nearly constant along the whole undulator length , i.e. a synchronism should be provided .
this resonance condition may be written as @xmath180 remembering that @xmath175 , we have @xmath181 . or ,
since @xmath182 , @xmath183 thus , we see that synchronism takes place when the wave advances the electron beam by one wavelength at one undulator period .
this resonance condition is exactly equation ( [ eq : s ] ) , so we have come full circle .
electromagnetic wave excitation and production of the electron beam energy modulation are ultimately connected . from this simple analysis
it is apparent that the electron s relative position within a radiation wavelength will determine whether it consistently gains or loses energy as it travels through the undulator magnet .
a convenient variable for the description of the interaction between electrons and electromagnetic wave in an undulator is the phase @xmath177 .
the relevant value of the phase @xmath177 is that at the location of the particle .
hence , the total derivative of @xmath177 is given by @xmath184 thus , the phase of the particle changes when the resonance condition is not satisfied exactly .
since discrete electrons are considered , their entrance into the interaction region is conveniently described in terms of their entrance phase positions relative to one cycle of the modulating electromagnetic wave at the input , i.e. @xmath57 .
an entrance phase variable is defined by @xmath185 , where @xmath186 is the entrance time .
let the initial electron energy be @xmath187 .
thus , at a given displacement plane the electron will have a kinetic energy deviation @xmath188 if @xmath189 is the arrival phase of this particular electron at the undulator entrance .
we should remark that our analysis of the interaction between electrons and radiation in an undulator gives a result that is somewhat simpler than we would actually find in nature .
all of our calculations have been made for the plane electromagnetic wave with phase velocity equal to the velocity of light .
the actual radiation wave in an undulator will experience an amplitude change and a phase shift due to the radiation process .
the electric field of the wave radiated in the helical undulator can be represented by equation ( [ eq : d ] ) .
we can also write this expression as @xmath190 \ , \ ] ] which says that radiation wave is not ordinary transverse plane wave .
our final result for the kinetic energy deviation is therefore @xmath191 \ , \label{eq : m1}\ ] ] this expression deserves some remarks . in dealing with start - up from the electron beam density modulation the appropriate `` frame of reference '' is moving with the fast wave rather than at the velocity of light .
the angle between the transverse velocity of the particle and the vector of the electric field is given by @xmath192 where @xmath193 represent the phase shift of the radiation wave ( `` response '' ) relative to the fast wave ( `` force '' ) .
what do we use for the entrance phase in our formula ( [ eq : m1 ] ) ?
we use the entrance phase position relative to one cycle of the fast wave at the input . for simplicity of analysis
, we consider the cases in which there are no variations in phase of the density modulation in the transverse plane at the undulator entrance . in those cases an entrance phase variable
is defined by @xmath185 , where @xmath186 is the entrance time .
we have what we need to know - the electron beam energy modulation at the undulator exit ( [ eq : m1 ] ) .
we are ready to find energy modulation , because we have already worked out what field is produced by a bunched electron beam inside the near zone ( [ eq : r ] ) .
these equations are not well suited for quick calculation of the energy modulation at a particular diffraction parameter .
we may , however , express ( [ eq : r ] ) in much simpler form for very small and very large diffraction parameters , making use of limiting expressions ( [ eq : w1 ] ) and ( [ eq : t1 ] ) .
let us see what happens if the diffraction parameter @xmath127 is large . at exact resonance @xmath169 using ( [ eq : w1 ] )
the energy modulation amplitude achieved at a given displacement plane can be written as : @xmath194 where @xmath195 is the alfven current .
now let us analyze this expression and study some of its consequences . to develop a useful mathematical formalism for the description of interaction between electrons and fields in an undulator we must choose the appropriate coordinate system to maximize physical clarity .
we chose at the beginning of this section to use the stationary coordinate system @xmath27 .
for example , the beam density at point @xmath27 , at time @xmath32 is @xmath196 a convenient way of description the particle beam dynamics is to use a coordinate system @xmath197 that moves at a velocity @xmath50 with respect to the stationary coordinate system @xmath27 .
when a particular electron enters the undulator , its initial position is @xmath198 in the moving coordinate system .
the electron would remain stationary at @xmath198 in the moving coordinate system ; in the stationary coordinate system , at a time @xmath32 , it would have moved a distance from the entrance @xmath199 , where @xmath186 is the entrance time .
let us write an expression for beam density modulation in terms of the moving coordinate system introducing the wave number @xmath200 defined by @xmath201 , where @xmath202 is a wavelength of the modulation frequency , i.e. the distance through which a particle travels in a cycle of modulation frequency .
we may thus write @xmath203 . note that @xmath204 and this expression may be rewritten in the terms of arrival phase @xmath205 where we calibrated time in such a way that when a particular electron enters the undulator at time @xmath56 , its initial position is @xmath206 in the moving coordinate system .
just as we expected , in the case of wide electron beam , the beam energy modulation is oscillating in phase with beam density modulation . in the case of thin electron beam ,
things are quite different . for a small diffraction parameter @xmath170 we may apply an asymptotic approximation for the field amplitude ( [ eq : t1 ] ) and get @xmath207\sin\psi_{0 } \right\ } \ .
\label{eq : me2}\ ] ] thus , we see that , in the case of thin electron beam , the beam energy modulation is not oscillating in phase with beam density modulation
. equations ( [ eq : me1 ] ) and ( [ eq : me2 ] ) give us an idea of what we should expect . generally we can try to calculate the energy modulation precisely by using the ( [ eq : r ] ) and ( [ eq : m1 ] ) . that is the end of our calculations of the electron beam energy modulation , but there is one physically interesting thing to check , and that is the conservation of energy . as any oscillating charge radiates energy , so must a modulated electron beam moving along an undulator radiate energy .
if the system radiates energy , then in order to preserve conservation of energy we must find that electron beam energy is being lost .
an interesting question is , what forces are electrons working against ? that is interesting question
which can be completely and satisfactorily answered for modulated electron beam in an undulator .
what happens in this : in bunched beam , the fields produced by the moving charges in one part of the undulator react on the moving charges in another part of the undulator .
we can calculate these forces and find out how much work they do , and so find the right rule for the radiation force .
we can make this calculation because at short distance we know the electric field .
above we calculated the radiation field inside the near zone ( [ eq : r ] ) . in our system energy
is transported both in the form of electromagnetic waves and in kinetic energy .
the well - known poynting vector represents the electromagnetic power flow ; the kinetic - power flow is merely the number of electrons crossing per unit area per unit time multiplied by the kinetic energy per electron .
consider first the electromagnetic power . in the paraxial approximation the diffraction angles are small , the vectors of the electric and magnetic field are equal in absolute value and are perpendicular to each other .
thus , the expression for the radiation power , @xmath208 , can be written in the form : @xmath209 where @xmath210 denotes averaging over a cycle of oscillation of the carrier wave .
if we consider a system with fields and bunched electron beam in an undulator , the energy stored in any volume fluctuates sinusoidally with time .
but on the average there is no increase or decrease in the energy stored in any portion of the volume , so that the below conservation theorem holds if averaged over time , although the integrals might not be zero instantaneously because of the time variations in the energy storage .
since the radiation field inside the near zone has the form of equation ( [ eq : r ] ) , than @xmath208 is given by : @xmath211\right\ } \nonumber\\ & \mbox { } & \times \left\{\int\limits^{z}_{0}\frac{{\mathop{\mathrm{d}}}z^{\prime\prime}}{z - z^{\prime\prime } } \int{\mathop{\mathrm{d}}}\vec{r}_{\perp}^{\prime\prime}n_{0}(\vec{r}^{\prime\prime}_{\perp } ) \exp\left[- \frac{{\mathop{\mathrm{i}}}\omega\mid\vec{r}_{\perp}-\vec{r}^{\prime\prime}_{\perp } \mid^{2}}{2c(z - z^{\prime\prime})}\right]\right\ } \ .
\label{eq : b1}\end{aligned}\ ] ] the product of integrals over @xmath197 and @xmath212 can be represent as @xmath213 the integral over transverse coordinate @xmath214 is equal to @xmath215 \nonumber\\ & \mbox { } & = \int\limits^{\infty}_{-\infty}{\mathop{\mathrm{d}}}x\int\limits^{\infty}_{-\infty}{\mathop{\mathrm{d}}}y \exp\left\{\frac{{\mathop{\mathrm{i}}}\omega}{2c}\frac{[(x - x^{\prime})^{2 } + ( y - y^{\prime})^{2}]}{z - z^{\prime } } \right . \nonumber\\ & \mbox { } & \left
. - \frac{{\mathop{\mathrm{i}}}\omega}{2c}\frac{[(x - x^{\prime\prime})^{2 } + ( y - y^{\prime\prime})^{2}]}{z - z^{\prime\prime}}\right\ } \nonumber\\ & \mbox { } & = \frac{2\pi{\mathop{\mathrm{i}}}c}{\omega}\frac{(z - z^{\prime})(z - z^{\prime\prime } ) } { z^{\prime}-z^{\prime\prime}}\exp\left[-\frac{{\mathop{\mathrm{i}}}\omega\mid\vec{r}^{\prime } _
{ \perp } - \vec{r}^{\prime\prime}_{\perp}\mid^{2}}{2c(z^{\prime } - z^{\prime\prime})}\right ] \ .
\label{eq : b2}\end{aligned}\ ] ] as a result , expression ( [ eq : b1 ] ) can be written in the form : @xmath216 + { \mathrm{c.c . } } \label{eq : b3}\end{aligned}\ ] ] latter expression can be written in terms of the complex amplitude of the radiated field @xmath217 . using ( [ eq : r ] ) and ( [ eq : b3 ] )
we obtain the expression for @xmath208 : @xmath218 the radiation power , @xmath208 , must be equal to the difference of the electron beam power at the exit and the entrance of the undulator .
the rate of the energy change for a single electron is given by @xmath219 \ .\ ] ] to obtain the mean power loss by the electron beam , we should multiply this value by the particle flux density @xmath220 average over time , and integrate over the beam cross - section and the undulator length .
finally , we obtain the following result : @xmath221\right\ } \nonumber\\ & \mbox { } & = - \frac{{\mathop{\mathrm{i}}}e\theta_{\mathrm{w}}ca_{\mathrm{in}}}{4}\int\limits^{z}_{0}{\mathop{\mathrm{d}}}z^{\prime}\int{\mathop{\mathrm{d}}}\vec{r}^{\prime}_{\perp } n_{0}(\vec{r}^{\prime}_{\perp } ) \tilde{e}(z^{\prime},\vec{r}^{\prime}_{\perp } ) + { \mathrm{c.c . } } \label{eq : b5}\end{aligned}\ ] ] comparing this expression with ( [ eq : b4 ] ) , we see that the radiation power and the change in the electron beam power have equal absolute values and are opposite in sign , i.e. @xmath222 .
so , power balance takes place .
our theory of klystron amplifier is based on the assumption that beam density modulation does not appreciably change as the beam propagates through the undulator .
when the resonance condition takes place , the electrons with different arrival phases acquire different values of the energy increments ( positive or negative ) , which results in the modulation of the longitudinal velocity of the electrons @xmath50 with the radiation frequency @xmath35 .
since this velocity modulation is transformed into density modulation of the electron beam during the undulator pass , an additional radiation field exists because of variation in amplitude density modulation . instead
, we assume that the amplitude of the electron beam density modulation has the same value at all points in the undulator .
this approximation means that only the contributions to the radiation field arising from the initial density modulation are taken into account , and not those arising from the induced bunching .
it is interesting to estimate the amount of bunching produced during the undulator pass .
the velocity @xmath50 of the electron at given displacement plane @xmath27 is given by @xmath223\ ] ] where @xmath224 is the injected velocity , and @xmath225 is a small amplitude of high frequency oscillation . from ( [ eq : me2 ] ) we can find @xmath225 remembering that @xmath51 and @xmath226 .
we have @xmath227 we now examine the behaviour of these electrons as they travel a certain distance in an undulator .
since successive electrons are not traveling with the same velocity , bunching takes place . to calculate a numerical value for the current
, we must compute the actual arrival time of the various electrons at the undulator exit .
the appropriate expression is formulated by means of the law of conservation of charge .
quantitatively , we can say that during an infinitesimal interval of time @xmath228 , a total charge @xmath229 passes through plane @xmath230 .
we follow the same particles ( i.e. the same charge ) through the plane @xmath231 at some later time @xmath232 .
these particles will pass through the @xmath231 plane during the an infinitesimal time interval @xmath233 which may be shorter or longer than @xmath228 . in either case , the instantaneous current @xmath234 due to the charge passing through @xmath231 plane in the interval @xmath233 is obtained from @xmath235 .
obviously , the two charges are the same since we are following the same set of particles .
hence due to this charge , there is a current through @xmath231 plane @xmath236 . if an electron has velocity @xmath50 when it leaves point @xmath230 and maintains this velocity , then its transit time to point @xmath231 is given by the expression @xmath237 \ .
\label{eq : ret}\ ] ] this equation provides us with the relation between the arrival time @xmath232 at point @xmath231 for the particular group of charges which left point @xmath230 at the departure time @xmath238 .
it is possible to calculate current at point @xmath231 for small values of @xmath239 by differentiating the last equation and substituting in equation @xmath240 .
this procedure leads to @xmath241 where @xmath242 .
this equation is obviously still not in a useful form , since it also only gives the current at point @xmath231 in terms of departure times of the charges from point @xmath230 .
however , by using ( [ eq : ret ] ) again ( to eliminate @xmath238 ) and using the fact that @xmath243 , one can write approximately @xmath244\right\ } \ , \ ] ] where the terms that have been neglected in @xmath238 would have contributed only terms of the order of @xmath245 to the current @xmath246 . thus the induced bunching that we want at the point @xmath231 due to the velocity modulation at the point @xmath230 is equal to @xmath247 .
now we find the total induced bunching at the point @xmath231 by finding the induced bunching there from each point @xmath248 , and then adding the contributions from all the points .
to calculate this sum we need to use ( [ eq : dv ] ) .
we have , then , @xmath249 thus , the requirement for the induced bunching to be small can be written as @xmath250 .
the result of a more careful analysis shows that the last condition can be written as @xmath251 during the passage through an undulator the electron density modulation can be suppressed by the longitudinal velocity spread in the electron beam . for effective operation of klystron amplifier
the value of suppression factor should be close to unity .
if there is an ( local ) electron energy spread into the electron bunch , @xmath252 , there will be a corresponding longitudinal velocity spread given by @xmath253 . to get a rough idea of the spread of electron density modulation
, the position of the particles within the electron beam at the undulator exit has a spread which is equal to @xmath254 we know that uncertainty in the phase of the particles is about @xmath255 .
therefore , a rough estimate for the microbunching spread to be small is @xmath256 it is clear that we have made several rough approximations . besides various factors of two we have left out , we have used @xmath257 , where we should have used variance @xmath258 .
all of these refinements can be made ; the result of a more careful analysis shows that for the gaussian distribution function the suppression factor is equal to @xmath259 $ ] .
so a better answer is @xmath260 remembering that @xmath261 and @xmath262 , we have @xmath263 where @xmath0 is the number of undulator periods .
this condition is not a restriction .
let us present a specific numerical example .
for unbunched electron beam the local energy spread is about @xmath264 .
if the nominal energy is @xmath265 , the last condition can be written as @xmath266 .
such is the case in all practical cases of interest . for a klystron amplifier
it is of great interest to maximize the gain per cascade pass which is proportional to the amplitude of electric field inside the electron beam .
it is usually desired to optimize the amplifier to achieve the highest possible gain coefficient for a given total current @xmath11 .
reducing the particle beam cross - section by diminishing the betatron function reduces also the size of the radiation beam and increases , according to ( [ eq : w1 ] ) , the amplitude of the electric field .
this process of reducing the beam cross section is , however , effective only up to some point .
further reduction of the particle beam size would have practically no effect on the radiation beam size because of diffraction effects ( [ eq : t1 ] ) .
it is therefore prudent not to push the particle beam size to values much less than the difraction limited radiation beam size . from the preceding discussion we may want to optimize the beam geometry as follows . for maximum amplification
, the transverse size of the electron beam has to be matched to the diffraction limited radiation beam size @xmath267 the wavelength and undulator length dictates the choice of the optimal transverse size of the electron beam .
one might think that all we have to do is to get electric field amplitude to a maximum - we can always reduce electron beam cross section and we can always increase gain .
so it is not impossible to build an electron focusing system that has a small betatron function .
in fact , one of the main problems in klystron amplifier operation is preventing the spread of microbunching due to angle spread into the electron beam .
some electrons traverse the undulator not along or parallel to the @xmath268axis , but at a small angle @xmath269 .
if there is an rms angular divergence @xmath270 within the electron bunch there will be a corresponding longitudinal velocity spread given by @xmath271 .
we assume that the amplitude of beam modulation does not appreciably change as the beam propagates through the undulator . to avoid an decrease in the amplitude of the beam modulation
, the longitudinal velocity spread must be small enough @xmath272 .
this gives the condition @xmath273 the last two conditions can be written as @xmath274 where @xmath275 is the transverse electron beam emittance .
the last condition can be seen as `` phase matching '' of the electrons and radiation . to produce maximal gain
the particle beam emittance must be less than the diffraction limited radiation emittance .
obviously , this condition is easier to achieve for long wavelengths . for optimal gain for a 30 nm radiation ,
for example , the electron beam emittance must be smaller than about @xmath276 this condition may be easily satisfied in practice .
in two - undulator klystron , energy modulation occurs in the first undulator and the conversion to density modulation in the dispersion section .
the dispersion section is designed to introduce the energy dependence of a particle s path length , @xmath277 .
several designs are possible , but the simplicity of a four - dipole magnet chicane is attractive because it adds no net beamline bend angle or offset and allows simple tuning of the momentum compaction , @xmath14 , with a single power supply .
the trajectory of the electron beam in the chicane has the shape of an isosceles triangle with base length @xmath278 .
the angle adjacent to the base , @xmath279 , is considered to be small . for ultra - relativistic electrons and small bend angles , the net @xmath14 of the chicane is given by ( see fig .
[ fig : ads ] ) @xmath280 optimal parameters of the dispersion section can be calculated in the following way .
the phase space distribution of the particles in the first undulator is described in terms of the distribution function @xmath281 written in `` energy - phase '' variables @xmath282 and @xmath283 , where @xmath284 is the nominal energy of the particle and @xmath35 is the angular frequency . before entering the first undulator ,
the electron energy distribution is assumed to be gaussian : @xmath285 the present study assumes the density modulation at the end of first undulator to be very small and there is an energy modulation @xmath286 only .
then the distribution function at the entrance to the dispersion section is @xmath287 after passing through the dispersion section with dispersion strength @xmath288 , the electrons of phase @xmath189 and energy deviation @xmath289 will come to a new phase @xmath290 .
hence , the distribution function becomes @xmath291 the dispersion strength parameter and compaction factor are connected by the relation @xmath292 the integration of the phase space distribution over energy provides the beam density distribution , and the fourier expansion of this function gives the harmonic components of the density modulation converted from the energy modulation @xcite . at the dispersion section exit , we may express current @xmath293 in the form @xmath294 \nonumber\\ & \mbox { } & \times j_{n}\left(np_{0}\frac{{\mathop{\mathrm{d}}}\psi_{0}}{{\mathop{\mathrm{d}}}p}\right)\cos(n\psi_{0 } ) \ . \label{eq:1}\end{aligned}\ ] ] we find a set of harmonic waves , of which the fundamental term , with angular frequency @xmath35 , is the one of importance in an amplifier .
this fundamental involves the phase variation @xmath295 , but it is multiplied by the amplitude term @xmath296 \ .\ ] ] for small input signal we may assume that the argument of the bessel function is small .
the function @xmath12 approaches @xmath297 for small @xmath13 ; thus the microbunching approaches @xmath298 \ .\ ] ] we see that microbunching depends greatly on the choice of the dispersion section strength .
an attempt to increase of the amplitude of the fundamental harmonic , by increasing the strength of dispersion section , is countered by a decrease of the exponential factor .
the microbunching @xmath299 has clearly a maximum @xmath300 and the optimum strength of the dispersion section is @xmath301 consider the situation at @xmath302 , @xmath265 , @xmath303 .
appropriate substitution in ( [ eq : dso ] ) show that the optimal compaction factor is equal to @xmath304 .
we can now piece together the information we have obtained in the preceding sections , and discuss the behaviour of the klystron as an amplifier of the electron beam intensity modulation .
the present section is concerned with cascade amplifiers designed for conditions approximating maximum gain per cascade pass . in the section 2
we learned that when a modulated electron beam passes through an undulator , the radiation field modulates the energy of the electrons . in order to calculate the energy modulation we can use ( [ eq : r ] ) and ( [ eq : m1 ] ) and note that energy modulation at the undulator exit generally can be written as @xmath305 , or as @xmath306 .
when the gain per cascade pass is high enough @xmath307 , the cascades can be considered independent . in this case
we do not see the effect of phase shift @xmath308 , but see only total amplitude of modulation equal to @xmath309 .
we are going to apply ( [ eq : r ] ) to our analysis of energy modulation taking advantage of similarity techniques we used in section 2.1 .
the beam , of peak current @xmath11 , is modulated in energy by the input undulator , and acquires an energy modulation amplitude on the electron beam axis , of amount @xmath310 where @xmath311 and universal function @xmath312 should be calculated numerically by using ( [ eq : r ] ) .
the physical implication of this result are best understood by considering some limiting cases .
equations ( [ eq : me1 ] ) and ( [ eq : me2 ] ) give us an idea what we should expect in the case of wide and thin electron beam .
we have asymptotically : @xmath313 @xmath314^{2 } } \qquad { \mathrm{as}}\qquad \beta \to 0 \ .\ ] ] a natural and interesting choice is to calculate the gain in thin beam case which we can identify with maximal gain .
note that the dependence of the factor @xmath312 on the exact size of the bunch in the thin beam asymptote is rather weak .
thus @xmath315 is a reasonable approximation .
finally , the energy modulation amplitude can be estimated simply as : @xmath316 the energy modulated beam then enters the dispersion section , in which density modulation is developed . from equations ( [ eq : dm1 ] ) and ( [ eq : me3 ] ) the over - all gain per cascade pass at resonance may be readily be obtained if optimum dispersion section strength is assumed ; thus @xmath317 where the following notation has been introduced : @xmath318 .
putting @xmath319 , @xmath320 , @xmath321 and @xmath322 in ( [ eq : g1 ] ) gives @xmath323 the equation ( [ eq : g2 ] ) tells us the maximal gain of cascade as a function of number of undulator periods @xmath0 , undulator parameter @xmath324 , peak current @xmath11 and rms local energy spread @xmath325 . for conventional xfels the following problem exists .
the required very small transverse beam emittance can be obtained only in a photoinjector .
the bunch length must be shortened .
the distribution of particles in phase space is given either by the injector characteristics and the injection process .
longitudinal phase space can be exchanged by special application of magnetic and rf fields .
this is done in a specially designed beam transport line consisting of a nonisochronous transport line ( magnetic chicane ) and an accelerating section installed at the beginning of the bunch compressor .
liouville s theorem is not violated because the energy spread in the beam has been increased through the phase dependent acceleration in the bunch compression system .
the cascade gain with continually optimized strength of the dispersion section would be directly proportional to @xmath326 .
we have a rather surprising result . we know that ratio @xmath326 depends on the parameters of photoinjector but not on the bunch compression .
we expect this ratio to depend only on the longitudinal emittance .
so the simple result says that the klystron gain is independent of absolute value of peak current .
the change in peak current with change of bunch length is compensated by the larger energy spread .
since @xmath25 is independent of bunch length , the gain would then be independent of bunch compression .
the formula for the gain which we derived ( [ eq : g2 ] ) refer to the case of the helical undulator . for somewhat wider generality , although we are still making some special assumption about undulator magnetic structure , we shall calculate the characteristics of the klystron amplifier with a planar undulator .
the magnetic field on the axis of the planar undulator is given by @xmath327 .
the explicit expression for the electron velocity in the field of the planar undulator has the form : @xmath328 , where @xmath329 .
it is not hard to go through the derivation of electron beam energy modulation again . if we do that , and calculate the gain the same way , we get @xmath330 where @xmath331 \ , \ ] ] @xmath332 is a bessel function of @xmath29th order , @xmath333 when we simplified the expression for @xmath334 , we used the resonance condition for the planar undulator @xmath335 $ ] .
it is convenient to rewrite the expression ( [ eq : g3 ] ) in the form : @xmath336 the significance of the proposed scheme can not be fully appreciated until we determine typical values of the gain per cascade pass that can be expected in practice .
let us present a specific numerical example for the case of a klystron amplifier with a planar undulator . with the numerical values @xmath337 , @xmath338 , @xmath339 , the resonance value of wavelength is @xmath303 .
if the number of the undulator period is @xmath340 , normalized transverse emittance @xmath341 , and betatron function is equal to the undulator length , the diffraction parameter is about @xmath342 . remembering the results of section 2 , we come to the conclusion that we can treat this situation as an klystron amplifier with thin electron beam .
it is shown that the residual energy spread in the ttf fel injector is on the order of a few kev only , even at a peak current of 100 a. if @xmath343 , and @xmath344 , appropriate substitution in ( [ eq : g4 ] ) shows that the intensity gain per cascade pass is about @xmath345 .
it has been seen above that there are definite upper limits to the gain of a two undulator klystron with a given current .
it is obvious that a gain much higher than that allowed by these limits may be obtained by using more than one stage of amplification . in the three - undulator type of klystron ,
called the cascade - amplifier klystron , the additional undulator ( and dispersion section ) , which lies between the input and output undulators , has no need of radiation phase matching .
second cascade is excited by the bunched beam that emerges from the first cascade , and it produces further bunching of the beam .
cascade bunching in a high - gain cascade amplifier results in current components just like those of simple bunching ; the equivalent intensity gain , which determines the cascade bunching , is given by @xmath346 , where @xmath347 is the number of cascades . in practice , with a two - cascade klystron amplifier , an intensity gain in excess of 80 db may be obtained at wavelengths around 30 nm .
this simple description is valid when beam modulation amplitude at the exit of the last cascade is much smaller than unity .
what happens when this condition is not met is discussed later .
fluctuations of the electron beam current serve as an input signal to the xfel . the initial modulation of the electron beam is defined by the shot nose and has a white spectrum .
when the electron beam enters the first undulator , the presence of the beam modulation at frequencies close to resonance frequency initiates the process of radiation .
the study has shown ( see section 5 ) that the mean square value of input is about @xmath348 at @xmath349 , @xmath340 and @xmath303 .
it is seen that saturation in the cascade klystron should occur at an exit of the second cascade in this situation .
thus , we conclude that a klystron amplifier which consists of a succession of 2 - 3 cascades can operate as a soft x - ray sase fel .
a disadvantage of using conventional fel amplifier at small peak current is the enormously long undulator that is required .
let us calculate the parameters of the conventional sase fel for the value @xmath349 of the peak current .
suppose @xmath344 , @xmath350 , @xmath351 , @xmath337 , @xmath352 , focusing beta function is equal to 3 m. numerical solution of the corresponding 3-d eigenvalue equation ( see @xcite ) demonstrates that the field gain length should be about @xmath353 .
well - known that saturation in a sase fel with uniform undulator occurs after approximately 10 exponential field gain lengths . remembering the total undulator length of the two - cascade klystron amplifier , we come to the conclusion that a significant impruvement in this parameter for klystron amplifier can be achieved . all of the foregoing discussion of klystron amplifier gain has been concerned solely with the gain at resonance - that is @xmath354 .
nothing has been said about amplification bandwidth .
now , we would like to find out how the gain varies in the circumstance that the seed signal frequency @xmath35 is nearly , but not exactly , equal to @xmath355 .
it is not hard to go through the derivation of electron beam energy modulation again .
if we take @xmath356 , the solution of the wave equation ( [ eq : u8 ] ) has the form @xmath357 \ .
\label{eq : r21}\end{aligned}\ ] ] using ( [ eq : m1 ] ) the energy modulation achieved at nonzero detuning parameter can be written as @xmath358 + { \mathrm{c.c . } } \label{eq : me21}\end{aligned}\ ] ] now we shall calculate the energy modulation on the electron beam axis . when @xmath118 and beam profile is gaussian , we have @xmath359 let us see what happens if the diffraction parameter @xmath127 is large .
in this case we have asymptotically : @xmath360 the integrals over @xmath197 and @xmath212 in the latter equation are calculated analytically @xmath361 taking into account the definition of the detuning parameter we find that @xmath99 is connected by a simple relation with the frequency deviation : @xmath362 .
if we follow through the simple algebra we find that @xmath363 } { [ \pi n_{\mathrm{w}}\delta\omega/\omega_{0}]^{2}}\cos\psi_{0 } \right .
\nonumber\\ & \mbox { } & \left .
+ \frac{[\pi n_{\mathrm{w}}\delta\omega/\omega_{0 } ] - \sin[\pi n_{\mathrm{w}}\delta\omega/\omega_{0 } ] \cos[\pi n_{\mathrm{w}}\delta\omega/\omega_{0 } ] } { [ \pi n_{\mathrm{w}}\delta\omega/\omega_{0}]^{2 } } \sin\psi_{0 } \right\ } \ .
\label{eq : int2}\end{aligned}\ ] ] the energy modulation at the undulator exit can be written as @xmath364 .
the intensity gain @xmath365 is proportional to @xmath366 .
it is interesting to plot this gain as a function of input signal frequency in order to see how sensitive it is to frequencies near the resonant frequency @xmath367 .
we show such a plot in fig .
[ fig : fx ] , where the following notation has been introduced : @xmath368 , @xmath369^{2 } = f(x ) = \frac{\sin^{4}(x)}{x^{4 } } + \frac{[x-\sin(x)\cos(x)]^{2}}{x^{4 } } \ .
\label{eq : df}\ ] ] the curve falls rather abruptly to zero for @xmath370 and never regains significant size for large frequency deviations .
so , for large value of diffraction parameter , the frequency response curve is @xmath371 .
this is also true for the practically important case in which the diffraction parameter is close to unity .
let s examine the implication of this results for a real klystron amplifier .
suppose that the electron beam is in the undulator for a reasonable length , say for 100 periods .
then we can calculate that the fwhm of the amplification bandwidth is equal to @xmath372 .
we have , then , a clear picture of the nature of the operation of the klystron amplifier with nonzero detuning parameter , and a calculation of the bandwidth to be obtained from it .
a detail analytical theory of the amplification at @xmath356 and finite transverse electron beam size is rather complicated , but the discussion which we have given here is enough for ordinary purposes . the amplification process in the klystron amplifier can be divided into two stages , linear and nonlinear . during the linear stage of amplification ,
significant growth of the beam modulation amplitude and of the electromagnetic field amplitude takes place .
nevertheless , the beam modulation is much less than unity in the cascade undulators , and the largest fraction of the radiation power is produced in the output undulator , when the beam modulation becomes about unity .
suppose we have a modulated electron beam at the exit of the last cascade .
the problem of electromagnetic wave radiation in the output undulator refers to a class of self - consistent problem .
there is considerable analogy between the nonlinear mode operation of the conventional fel amplifier and klystron amplifier . in the high gain linear regime of a conventional fel amplifier
the typical length of the change in the radiation field amplitude is about the gain length @xmath373 , where for the one dimensional model @xmath374^{1/3 } \ .\ ] ] in the case of a uniform output undulator , the bunched beam effectively interacts with the electromagnetic wave along a length which is of the order of the gain length @xmath375 . at this stage of the self - interaction electrons
lose the visual fraction of their energy which results in the violation of the resonance condition . as a result ,
the beam is overmodulated , most electrons fall into the accelerating phase of the ponderomotive potential and the electron beam starts to absorb power from the electromagnetic wave .
this is analogous to the situation with a conventional fel amplifier . like the efficiency of the conventional fel amplifier , the efficiency @xmath376 of the klystron amplifier with a long enough uniform output undulator is limited by the value of the efficiency parameter @xmath141 : @xmath377 it should be note that the saturation efficiency of the klystron amplifier does not depend on the amplitude of the input signal when the klystron amplifier operates in the high - gain regime , i.e. when @xmath378 . for some purposes it is convenient to express @xmath141 in a different form @xmath379^{2/3 } \left[\frac{i_{0}}{\gamma i_{\mathrm{a}}}\right]^{1/3 } \left[\frac{k^{2}}{1+k^{2}}\right]^{1/3 } \ .
\label{eq : rho}\ ] ] thus the efficiency of the klystron amplifier , defined in this particular way , depends on only the one - third power of peak current , a very weak dependence .
let us make a calculation of @xmath376 for some cases .
suppose @xmath349 , @xmath350 , @xmath351 , @xmath337 , @xmath338 , focusing beta function is equal to 3 m ; then by equation ( [ eq : rho ] ) it follows that @xmath380 .
returning to ( [ eq : rho ] ) , we could imaging the injector linac running at a @xmath381 .
as we mentioned above , the dependence of @xmath376 on the peak current is rather weak . as a result
, the efficiency is increased by only a factor three ( @xmath382 ) for @xmath381 .
let us present a specific numerical example for the case of a fourth - generation light source .
one of the key parameters to compare different radiation sources is their brilliance , which is simply given by the spectral flux divided by the transverse photon phase space .
we illustrate with numerical example the potential of the proposed klystron amplifier scheme for the sase fel at the tesla test facility accelerator .
the average brilliance of the conventional ttf sase fel at a wavelength of @xmath383 is about @xmath384 for the case when @xmath385 @xcite .
average photon flux and average brilliance varies simply as efficiency @xmath376 .
since @xmath386 , the dependence of @xmath387 on the electron peak current is contained completely in the term @xmath141 .
if @xmath349 , appropriate substitution in ( [ eq : rho ] ) shows that the average brilliance of the cascade klystron amplifier without bunch compression at a wavelength of @xmath383 is about @xmath388 on the other hand , at a wavelength of 30 nm brilliance of a third - generation synchrotron light source is equal to @xmath389 a comparison to synchrotron light sources shows the drastic improvement of average brilliance at the klystron sources .
the average brilliance of the klystron facility operating without bunch compression in the injector linac surpasses the spontaneous undulator radiation from third - generation synchrotron radiation facilities by four or more orders of magnitude .
however , it is not only the high average brilliance which makes a klystron amplifier facility a unique research tool in the vuv and soft x - ray regime , the radiation from this facility shows a very high peak brilliance too .
average brilliance is normalized to seconds at the highest possible repetition rate while the peak brilliance is the brilliance scaled to the length of a single pulse .
another type of current dependence are expected for peak brilliance , a 4/3 power dependence @xmath390 .
the peak brilliance of the conventional ttf sase fel at a wavelength of @xmath391 is about @xmath392 on the other hand , the peak brilliance of the cascade klystron amplifier without bunch compression at a wavelength of @xmath383 is about @xmath393 decreasing the peak current also decreases the peak brilliance of the sase fel radiation by about of factor 100 , but this is still 6 orders of magnitude higher than that of 3rd generation synchrotron radiation sources . a cascade klystron amplifier can be modified to increase significantly the spectral brightness of output radiation .
a reliable method to increase the klystron amplifier efficiency consists in an adiabatic change of the output undulator parameters ( or , in other words , by the use of so - called undulator tapering ) . when the electron bunch passes through the last dispersion section the energy modulation leads to an effective ( nonlinear ) compression of the particles
. then the bunched electron beam enters a tapered output undulator , and from the very beginning produces strong radiation because of the large spatial bunching .
the strong radiation field produces a ponderomotive well which is deep enough to trap the particles , since the original beam is relatively cold .
the radiation produced by these captured particles increases the depth of the ponderomotive well , and they are effectively decelerated . as a result ,
much higher power can be achieved than for the case of a uniform undulator . in @xcite
we have described the tapering undulator scheme for a sase fel .
simulations using the code fast provide a `` full physics '' description of the process . despite the original spiking seeding the process of the electron density modulation
, we effectively trap a significant fraction of the particles , and can achieve much higher power than for the case of an untapered output undulator .
another important feature of the radiation from a tapered undulator is the significant suppression of the sideband growth in the nonlinear regime .
this means that in the proposed scheme the spectral brightness of the radiation is increased proportionally to the radiation power . in the case of a uniform undulator
the peak brightness is reached at the saturation point and is then reduced due to sideband growth .
till now we have considered the electron beam as a continuous medium when describing the theory of klystron amplifier . to some extent this is idealization , since in reality the beam current is produced by a large number of moving electrons .
if we consider the microstructure of the electron current , we find that electrons enter the undulator randomly in time and space .
so , we can expect that the fel amplifier should possess intrinsic noise properties .
here it is relevant to remember that fels form a separate class of vacuum - tube devices .
the analysis of the noise properties of traditional vacuum - tube amplifiers has always been an important problem .
this has been mainly connected with the practical need for reducing the intrinsic noise of the amplifier .
the result of the experience obtained during these investigations shows that there always exists the fundamental effect of shot noise originating from the random emission of the electrons from the cathode .
when we analyze this effect for the parameters of traditional microwave amplifiers , we find that it is complicated .
in particular , suppression of the shot noise in some frequency band can occur due to space charge effects .
besides the shot noise effect , there are a number of different sources of noise which influence the operation of traditional vacuum - tube amplifiers .
fluctuations of the electron beam current serve as the input signal in the xfel . as for the fel amplifier operating in short wavelength range , its noise properties are defined only by the shot noise . at optical frequencies
the quieting effect of space - charge limitation seems negligible . an fel amplifier which starts up from shot noise
is frequently known as a self - amplified spontaneous emission ( sase ) fel .
however , it is worth mentioning that such an essentially quantum terminology does not reflect the actual physics of the process .
the amplification process in the sase fel has its origin in the density fluctuations in the electron beam .
the latter effect is completely classical .
the shot noise is the pure fluctuations in number which correspond to the fact that photoemission is a random process .
the rate at which electrons are emitted from a photocathode is not constant in time .
the emission process is a random one and it is impossible to predict the time dependence of the instantaneous current .
any such random fluctuations in the beam current correspond to an intensity modulation of the beam current at all frequencies simultaneously - including , of course , the frequency to which the undulator is tuned .
when the electron beam enters the first undulator , the presence of the beam modulation at frequencies close to the resonance frequency initiates the process of radiation .
the emission of electrons from the cathode is believed to be a poisson process , and from this assumption alone the total fluctuation in current can be deduced .
why is this the right rule , what is the fundamental reason for it , and how is it connected to anything else ?
the explanation in deep down in quantum mechanics .
when electromagnetic fields are incident on a photosurface , a complex set of events can transpire .
the major steps in this process can be identified as absorption of a quantum light energy ( i.e. a photon ) and the transfer of that energy to an excited electron , transport of the excited electron to the surface , and finally , release of the electron from the surface . when laser light having a deterministic variation of intensity over space and time is incident on a photosurface , the fluctuations of the photons obey poisson statistics .
this explains the relation between fundamental photon `` shot noise '' and shot noise in the electron beam . when describing the physical principles , it is always important to find a model which provides the possibility of an analytical description without loss of essential information about the features of the random process . in the first analysis of the problem , we adopt some rather simplifying assumptions that are only occasionally met in practice .
we will investigate the klystron amplifier start - up from shot noise in the framework of the one - dimensional model which assumes the input shot noise and output radiation have a full transverse coherence .
this assumption allow us to assume that the input shot noise signal is defined by the value of total beam current . in reality
the fluctuations of the electron beam current density are uncorrelated in the transverse dimension . using the notion of the beam radiated modes
, we can say that many transverse radiation modes are radiated when the electron beam enters the undulator .
the one - dimensional model can be used for the calculations of statistical properties of transversely coherent output current . in practice
such an assumption is valid for the current at the exit of klystron with a thin electron beam . during the amplification process in a klystron , the number of transverse modes decreases , and the contribution of the coherent radiation to the total radiation power
is increased up to full coherence . with this assumption
, attention can be concentrated completely on temporal coherence effects .
we study the case when the initial modulation of the electron beam is defined by the shot noise and has a white spectrum .
since in the linear regime all the harmonics are amplified independently , we can use the results of the steady - state theory for each harmonic and calculated the corresponding fourier harmonics of the output current .
the electric current in the time domain , @xmath394 , and its fourier transform , @xmath395 , are connected @xmath396 the klystron amplifier has a response which can be economically expressed as a frequency response curve @xmath397 .
thus , the fourier harmonic of the current at the dispersion section exit and the fourier harmonic of the input current are connected by the relation : @xmath398 when @xmath399 the fourier harmonic is defined by the relation @xmath400 .
it is convenient to isolate explicitly the slowly varying complex amplitude @xmath401 where @xmath402 the real output current consists of a carrier modulation of frequency @xmath367 subjected to random amplitude and phase modulation .
the fourier decomposition of the output current contains frequencies spread about @xmath367 .
it is not possible in practice to resolve the oscillations in @xmath394 that occur at frequency of the carrier modulation .
a good experimental resolving time is of order of picoseconds , at least four orders of magnitude too long to detect oscillations at the frequency @xmath367 .
it is therefore appropriate for comparison with experiment to average the theoretical results over a cycle of oscillation of the carrier modulation .
the deviation @xmath403 has a zero cycle - average . according to ( [ eq : sn1 ] ) ,
the cycle average of the square of the current deviation is @xmath404 shot noise is a random statistical process and statements about shot noise are probability statements .
these are most easily handled using the concept of a statistical ensemble , drawn from statistical mechanics .
if the physical construction of the injector and set of the external parameters ( e.g supply voltage , photoinjector laser beam parameters ) which specify its state are known , then these parameters define a definite statistical ensemble to which the injector belongs.this ensemble consists of identical copies of the injector subject to identical macroscopic external conditions .
any external property of the injector , e.g. its laser light on the photosurface , will depend , however , not only on these given parameters but also on the exact microscopic internal state of the photocathode , and each copy of the injector in the ensemble will lead to a different output .
we shall always denote ensemble averages by @xmath405 . in practice
the xfel injector is required to produce a single electron bunch at a repetition rate of 100 - 100 000 bunch / s and averaging symbol @xmath405 simply means the ensemble average over electron bunches .
the output fluctuations , expressed as current fluctuations , have a mean square value @xmath406 to calculate correlation of the spectral components of the output beam current , we should consider a microscopic picture of the electron beam current at the undulator entrance .
we start the analysis for the case of a rectangular electron pulse of finite duration @xmath407 and then we go over to the limit of an infinitely long pulse .
the electron beam current is made up of moving electrons randomly arriving at the entrance of the undulator : @xmath408 where @xmath409 is the delta function , @xmath30 is the charge of the electron , @xmath410 is the number of electrons in a bunch and @xmath411 is the random arrival time of the electron at the undulator entrance .
the beam current averaged over an ensemble of the bunches can be written in the form : @xmath412 the electron beam current @xmath394 and its fourier transform @xmath395 are connected by : @xmath413 @xmath414 now we can calculate the first - order correlation of the complex fourier harmonics @xmath395 and @xmath415 : @xmath416 expanding this relation , we can write : @xmath417\right\rangle + e^{2}\left\langle\sum_{k\ne n}\exp ( { \mathop{\mathrm{i}}}\omega t_{k } - { \mathop{\mathrm{i}}}\omega^{\prime}t_{n})\right\rangle \nonumber\\ & \mbox { } & = e^{2}\sum^{n}_{k=1}\langle\exp [ { \mathop{\mathrm{i}}}(\omega - \omega^{\prime})t_{k}]\rangle + e^{2}\sum_{k\ne n}\langle\exp ( { \mathop{\mathrm{i}}}\omega t_{k})\rangle\langle\exp(- { \mathop{\mathrm{i}}}\omega^{\prime}t_{n})\rangle \ .
\label{eq : s1}\end{aligned}\ ] ] the probability of the arrival of an electron during a time interval @xmath418 is equal to @xmath419 .
it is easy to find that @xmath420 where @xmath421 substituting ( [ eq : s2 ] ) to the ( [ eq : s1 ] ) ,
we obtain : @xmath422 when @xmath423 we can write the following expression for the first - order spectral correlation : @xmath424 let us consider the case of an infinitely long electron pulse with the homogeneous linear density and current @xmath11 .
in this limit @xmath425 the first - order correlation of the complex fourier harmonics has the form : @xmath426 = 2\pi ei_{0}\delta(\omega-\omega^{\prime } ) \ .
\label{eq : s4}\ ] ] the following representation of the delta function has been used here : @xmath427 the definition given in ( [ eq : df1 ] ) can be used to prove the basic property of the delta function .
let @xmath428 be any function of @xmath35 that is non - singular at @xmath354 , and consider the integral @xmath429}{(\omega_{0}-\omega)}{\mathop{\mathrm{d}}}\omega \ .\ ] ] changing the variable of integration to @xmath430 gives @xmath431 inclusion of the delta function in an integral therefore picks out the value of the integrand at the point specified by the constant in the delta function .
thus , if a current of mean value @xmath11 flows in the input of an amplifier whose frequency response curve is @xmath397 , the output fluctuations are @xmath432 we can also write ( [ eq : s5 ] ) in the form @xmath433 where the effective amplification bandwidth of the klystron is defined as @xmath434 we can further express this result by saying that the mean square value of the output is equivalent to a fluctuating current of mean square value @xmath435 at the input . note that it has not been necessary to make any assumptions about the amplification process .
the result is completely general ; that is , it applies for any type of linear vacuum - tube amplifier @xcite . with the preceding results in hand
, it should now be possible to estimate , for example , the amplitude of initial beam density modulation @xmath10 .
for this purpose it is convenient to express the result ( [ eq : s5 ] ) in a different form . noting that @xmath436 and @xmath437 we can write @xmath438 where @xmath439 is the number of cooperating electrons , and @xmath440 is the number of electrons per radiation wavelength . if @xmath349 , @xmath441 , and @xmath340 , appropriate substitutions in ( [ eq : s10 ] ) show that the mean square value of input is about @xmath442 .
now we return to the question about the electron bunch profile .
equation ( [ eq : s3 ] ) is not only true at the arbitrary chosen rectangular bunch profile , but of course it is true at any other bunch profile , and thus under the condition ( [ eq : s8 ] ) in the long electron bunch limit the output intensity fluctuations are all the same .
( the final statement ( [ eq : s5 ] ) does not involve the bunch profile function , which appeared only in the intermediate arguments . )
let us discuss the region of validity of the approximation ( [ eq : s8 ] ) for different bunch profile functions .
if we consider the case of rectangular bunch profile , we find that the region of applicability of condition ( [ eq : s8 ] ) is less than that for the case of a gaussian bunch of the same duration .
this is due to the fact that the bunch form factor @xmath443 , decreas more slowly with the increase in frequency . on the other hand , in a realistic situation
there is no sharp boundary of the bunch and the beam current falls to zero during some time interval @xmath444 .
when the beam current at the edge falls in accordance with a gaussian law , @xmath445 must obey the following conditions : @xmath446 when @xmath447 and @xmath448 the value @xmath449 is equal to @xmath450 . as a rule ,
the number of particles in the bunch , @xmath410 , is less than or about @xmath451 , so condition ( [ eq : s8 ] ) is fulfilled .
the shot noise in the electron beam is a gaussian random process .
the klystron amplifier operated in linear regime can be considered as a linear filter which does not change the statistics of the signal . as a result
, we can define general properties of the beam density modulation after the dispersion section without any calculations .
for instance , the square of instantaneous amplitude of density modulation fluctuates in accordance with the negative exponential distribution .
we can also state that the spectral density of the electron beam modulation and first - order time correlation function should form a fourier transform pair ( this is the wiener khintchine theorem ) .
the correlation between the output currents at times @xmath32 and @xmath452 has the form : @xmath453 + { \mathrm{c.c.}}\ ] ] using ( [ eq : s7 ] ) and ( [ eq : s4 ] ) we calculate the correlation between the complex amplitude @xmath454 and @xmath455 : @xmath456 \nonumber\\ & \mbox { } & = \frac{1}{4\pi^{2}}\int\limits^{\infty}_{0}{\mathop{\mathrm{d}}}\omega \int\limits^{\infty}_{0}{\mathop{\mathrm{d}}}\omega^{\prime } \exp(-{\mathop{\mathrm{i}}}\omega t + { \mathop{\mathrm{i}}}\omega^{\prime}t^{\prime } ) g(\omega)g^{*}(\omega^{\prime } ) \langle\bar{i}(\omega)\bar{i}^{*}(\omega^{\prime})\rangle \nonumber\\ & \mbox { } & = \frac{ei_{0}}{2\pi } \int\limits^{\infty}_{0}{\mathop{\mathrm{d}}}\omega \exp[-{\mathop{\mathrm{i}}}\omega(t - t^{\prime})]\mid g(\omega)\mid^{2 } \ . \label{eq : cf}\end{aligned}\ ] ] we define the first - order time correlation function as follows : @xmath457 using ( [ eq : d1 ] ) , ( [ eq : s5 ] ) and ( [ eq : cf ] ) we can write @xmath458 } { \int\limits^{\infty}_{0}{\mathop{\mathrm{d}}}\omega\mid g(\omega)\mid^{2 } } \ .\ ] ] the integral should go from @xmath134 to @xmath136 , but @xmath134 is so far from @xmath367 that the curve is all finished by that time , so we go instead to minus @xmath136 - it makes no difference . if we do that we get @xmath459 } { \int\limits^{\infty}_{-\infty}{\mathop{\mathrm{d}}}(\delta\omega)\mid g(\delta\omega)\mid^{2 } } \ .
\label{eq : cf1}\ ] ] where @xmath460 .
therefore , the slowly varying correlation function and the normalized spectrum of the narrow band - signal are a fourier transform pair .
it is seen from latter expression that the first - order correlation function possess the property @xmath461 . when the klystron gain curve is symmetrical with respect to the resonance frequency , @xmath367 , the function @xmath462 is real . in our approximation ,
the klystron gain function @xmath463 is symmetric with respect to @xmath464 and the integration therefore needs to be performed only from @xmath134 to @xmath136 with the result being doubled . inserting ( [ eq : df ] ) into the integrand ( [ eq : cf1 ] ) we get @xmath465 here we use the following notation @xmath466 , @xmath467 .
the integral is a fourier transform of @xmath371 . in fig .
[ fig : g1 ] we present a plot of first - order time correlation function . first , let us notice a remarkable feature of that plot .
the correlation function is zero for @xmath468 .
of course we could predict such a result . remembering that the synchronism takes place when the electromagnetic wave advance the electron beam by one wavelength at one undulator period ( see section 3 )
, we see that @xmath469 is the total relative slippage of the wave with respect to the electron beam . in many applications
it is desirable to have a precise and definite meaning for the term `` coherence time . ''
such a definition can be made in terms of the complex correlation function , but there are a multitude of definitions in terms of @xmath470 that can be imagined . however , in future discussions there is one definition that arises most naturally and most frequently . following the approach of mandel , we define the coherence time , @xmath471 , as @xmath472 its magnitude is of the order of the inverse of the output frequency spread . in all theory that follows ,
attention is restricted to modulated electron beams whose frequency spreads are small compared with the mean frequency , that is , where @xmath473 is very much larger than unity .
figure [ fig : fl ] illustrates the type of fluctuations that occur in the cycle - averaged intensity modulation .
the figure has been constructed by a computer simulation of the first cascade output in which the summation is carried out explicitly for the real number of electrons in the electron bunch .
the instantaneous intensity modulation varies in an unpredictable manner about the average value , and it is this kind of variation that characterizes the phenomenon as noise .
it is seen that substantial changes in the square value of amplitude occur over a time span @xmath471 , but this quantity is reasonably constant over time intervals @xmath474 .
the value of the coherence time for the case of klystron is given by @xmath475 .
with the numerical values @xmath340 and @xmath441 the value of coherence time is @xmath476 .
our discussion of the current fluctuations would be incomplete if we did not refer the reader to another method of average calculation .
the klystron output coherence time , @xmath471 , is in femtosecond range and @xmath477 over the time span @xmath471 is a good assumption in all practical problems .
clearly we can assume that the current in question is adequately modelled as a ergodic and hence stationary random process .
a stationary process obeys statistical laws and is subject to conditions that do not vary with time . for a stationary process time and
ensemble averages are equivalent . in this case
averaging symbol @xmath405 simply means the average over a period much longer compare with @xmath471 .
in the preceding sections we have given a simple discussion of a soft x - ray cascade klystron amplifier .
we have not only neglected many features of the klystron amplifier , but we have also completely omitted some important applications .
we shall mention three here : the frequency multiplier , a scheme for femtosecond experiments , and multi - user soft x - ray facility .
the remaining application of the klystron which we shall mention is the frequency multiplier .
we have seen , in section 3 , that the bunched beam at large values of the bunching parameter has not only a fundamental radiation frequency component , but also has considerable intensity in its harmonics .
it is then possible to have an input undulator operating at one frequency , and an output undulator operating at a multiple of this frequency .
the radiation in the output undulator will then be excited by the harmonic component in the electron beam , and the klystron will operate as a combination of frequency multiplier and amplifier .
a possible application would be to the production of very high frequencies .
a klystron gets more and more difficult to construct , as the frequency becomes higher and higher , on account of the reduced period of the undulators .
if we use a frequency multiplier klystron to reach the highest frequencies , it is only the output undulator which must be made with the smallest period ; the cascade undulators , operating at a lower frequency , can be made with larger period , and this can accommodate an injector with larger emittance and make the source easier to construct .
the idea of using two undulators , with the second undulator resonant to one of the harmonics of the first one ( the so - called `` after - burner '' scheme ) , was considered in @xcite .
the technique of using an conventional fel amplifier as a frequency multiplier relies on the fact that spatial bunching in the exponential gain regime can be very strong .
the first uniform undulator is long enough to reach saturation and produce strong spatial bunching in harmonics .
the bunched beam generates coherent radiation in the second undulator which immediately follows the first one .
the main problem with this approach is the large induced ( correlated ) energy spread which will be generated by the bunching of the electron beam at the fundamental frequency .
while this energy spread is necessary for the bunching , it degrades the performance of the radiation section at the harmonic frequency .
the problem to be solved is how to prolong the interaction of the bunched electron beam with the electromagnetic wave in the harmonic radiator undulator .
the present section is concerned with frequency doubler , designed for conditions approximating maximum efficiency . a reliable method to increase
the frequency doubler efficiency is to use a cascade klystron in which the input undulator operates at the fundamental resonance , the penultimate undulator operates at the fundamental harmonic , and the output undulator operates at the second harmonic .
in particular it can be shown that a so - called 1 - 1 - 2 sase scheme may operate with rather high efficiency in the soft x - ray wavelength range .
( the numbers 1,2 here designate the resonant harmonic number for each undulator . )
the klystron gain is controlled in such a way that the maximum energy modulation of the electron beam at the penultimate undulator exit is about equal to the local energy spread .
this is much smaller than the induced energy spread of the electrons at the exit of a saturated conventional sase fel . as a result
, the operation of a frequency doubler is based on prolonged interaction of the bunched electron beam with the electromagnetic wave in the harmonic radiator .
in contrast to the after - burner method , in a two - cascade 1 - 1 - 2 scheme frequency doubler operation is based on prolonged interaction of the bunched electron beam with the electromagnetic wave in the output undulator and copious coherent emission at double frequency is produced .
the doubler efficiency can be optimized by monitoring the output radiation intensity at double frequency as a function of the compaction factor of the first and second dispersion section . in @xcite
we have described an effective frequency doubler for sase fels .
for the first time the frequency multiplication scheme has been analyzed for sase fels .
computer modeling with time - dependent code fast has demonstrated that the final output power of a 1 - 1 - 2 klystron doubler is no smaller than that which could be produced by a 2 - 2 - 2 klystron with the same electron beam parameters .
time - resolved experiments are used to monitor time - dependent phenomena .
the standard technique for high - resolution time - resolved experiments is the pump - probe scheme in which a process is started by a short pulse of radiation ( pump ) and the evolution of the process is then observed ( probed ) after a delay by means of a second pulse of radiation , generally at another photon energy .
the obvious temporal limitation of the visible pump / x - ray probe technique is the duration of the x - ray probe . here
we will concentrate on the performance which can be obtained by a klystron xfel operating without bunch compression in the injector linac . at these sources ,
the x - ray pulse duration is about 10 ps .
this is longer than the timescale of many interesting physical phenomena .
the new principle of pump - probe techniques described below offers a way around this difficulty .
our studies have shown that the soft x - ray cascade klystron holds great promise as a source of radiation for generating high power single femtosecond pulses .
our femtosecond soft x - ray facility concept is based on the use of an x - ray sase fel combined with a femtosecond quantum laser @xcite .
the operation of a femtosecond soft x - ray sase fel is illustrated in fig .
[ fig : app ] .
an ultrashort laser pulse is used to modulate the density of electrons within the femtosecond slice of the electron bunch at a frequency @xmath478 .
the seed laser pulse will be timed to overlap with the central area of the electron bunch .
this ultrashort laser pulse serves as a seed for a modulator which consists of an uniform undulator and a dispersion section .
the interaction of seed pulse with the electron beam produces an energy modulation at @xmath478 .
this energy modulation is converted into spatial bunching in the dispersion section .
density modulation at the modulator exit is about 20 % .
the energy modulation introduced by the modulator is smaller than the local energy spread . following the modulator
the beam enters the first undulator of the klystron amplifier which is resonant with soft x - ray radiation at frequency @xmath367 .
the process of amplification in the soft x - ray klystron amplifier develops in the same way as was explained in previous sections : fluctuations of the electron beam current density serve as the input signal . at the chosen level of density ( energy ) modulation the sase process develops nearly in the same way as with a non - modulated electron beam because of the large ratio of cooperation length to optical wavelength . by the time the beam is bunched in the klystron cascades , at frequency @xmath367
, the density modulation amplitude has reached saturation .
this leads to amplitude modulation of the density at the sidebands @xmath479 .
the sideband density modulation takes place at the part of the electron pulse defined by the duration of the seed laser pulse , which is much shorter than the electron pulse .
following the last cascade the beam enters the output undulator which is resonant at the frequency @xmath480 . because the beam has a large component of bunching at the sideband
, coherent emission is copiously produced within the femtosecond slice of the electron bunch .
analytical methods are of limited use in to the study of the sideband - seeded sase fel and numerical simulations must be used .
simulations with time - dependent code fast provide a `` full physics '' description of the process in a sideband - seeded sase fel .
the results of numerical simulations confirm our simple physical considerations @xcite .
we assume that the sase bandwidth is much less than the separation of the sidebands from the main peak .
this requirement is of critical importance to the overall performance of the femtosecond sase fel facility . in this case
, the ( output ) undulator can be used to distinguish the sideband density modulation from the density modulation at the central frequency .
obviously , this requirement is easier to achieve for long radiation wavelengths . for 300 nm laser radiation and for 30 nm output radiation ,
for example , the amplification bandwidth of the cascade klystron must be much smaller than 10 % .
this condition may be easily satisfied in practice .
the long ( unbunched ) electron pulse is one of the advantages of the adopted fel design .
since the electron pulse duration ( about 10 ps ) is much longer than the time jitter of the electron and seed laser pulses @xmath481 , the synchronization of the optical laser with the electron pulses is not a problem .
pump - probe techniques which are commonly used with optical lasers , are highly desirable in order to make full use of the femtosecond soft x - ray pulses . since , in this case , precise timing is needed with a jitter of less than 10 fs , in @xcite we suggested to combine the femtosecond soft x - ray pulses generated by the sideband - seeded sase fel with optical pulses generated by the seed laser system .
it should be emphasized that in the proposed scheme a femtosecond soft x - ray pulse is naturally synchronized with his femtosecond optical pulse and cancels jitter .
this development of xfel based pump - probe experiments allow us to investigate phenomena at timescale down to 100 fs .
an x - ray laboratory should serve several , may be up to ten experimental stations which can be operated independently according to the needs of the user community .
on the other hand , the multi - user distribution system has to satisfy an additional requirement . passing the electron bunch through the bending magnets
must avoid emittance dilution due to coherent synchrotron radiation ( csr ) effects .
for very short bunches and very high peak current , csr can generated energy spread in the bending magnets and thus dilute the horizontal emittance .
as a result , the prefered layout of a conventional sase fel is a linear arrangement in which the injector , accelerator , bunch compressors and undulators are nearly collinear , and in which the electron beam does not change direction between accelerator and undulators .
the situation is quite different for the klystron amplifier scheme proposed in our paper .
since it operates without bunch compression in the injector linac , the problem of emittance dilution in the bending magnets does not exist .
an electron beam distribution system based on unbunched electron beam can provide efficient ways to generate a multi - user facility - very similar to present day synchrotron radiation facilities .
a possible layout of a soft x - ray fel laboratory based on an electron ring distribution system is shown in fig .
[ fig : a4gf ] .
the layout of the laboratory follows a similar approach as that used for synchrotron light sources .
the sase fel user facility consists of an injector linac and electron beam distribution system .
the injector is composed of a rf gun with photocathode and a main ( superconducting ) linac . in order to make efficient use of the new source
it is proposed to segment the full circumference of a distribution system into arcs which are repeated a number of times to form a complete ring .
each cell includes a two - cascade klystron and a bending magnet .
a specific realization of the electron ring cell is sketched in fig .
[ fig : acu ] .
the electron beam transport line guiding electrons from the injector linac to the experimental hall is connected tangentially to one of the straight sections of the electron ring . in order to obtain a useful separation between the experimental areas behind the photon beam lines ,
an angle of 36 degrees between two neighboring lines would be desirable .
thus , ten beam lines can be installed on a complete electron ring .
using klystron ( electromagnetic ) dispersion sections in each cell as switching elements it is possible to quickly switch the fel photon beam from one experiment to the other , thus providing multi - user capability .
users can define the radiation wavelength for their experiment independently of each other to a very large extent , since they use different undulators .
injector linac and electron beam transport lines operate at fixed parameters . at a fixed electron energy
the magnet gap of the klystron undulators can be varied mechanically for wavelength tuning .
this design makes it possible to make various wavelengths of sase radiation available in the xfel laboratory quasi - simultaneously .
it is a great advantage that injector and electron beam transport lines in the new scheme of multi - user facility operate at fixed parameters and that an `` electron switchard '' is not required .
when a relativistic electron beam passes through the undulator , it emits incoherent radiation .
this process leads to an increase of the energy spread in the beam due to quantum fluctuations of the undulator radiation .
this effect grows significantly with an increase in electron beam energy , strength field and length of the undulator .
this effect should be carefully taken into account when designing multi - user distribution system .
for the case of planar undulator the expression for the rate of energy diffusion can be written in the following form@xcite : @xmath482 where @xmath483 is the compton wavelength , @xmath484 is the classical electron radius , @xmath485 for @xmath486 .
the next numerical example illustrates the amplitude of the effect .
if @xmath339 , @xmath338 , @xmath487 , and the total undulator length is equal to 100 m , appropriate substitution in ( [ eq : cfz ] ) shows that the energy spread increase due to quantum fluctuations is about @xmath488 at the end of the 10th klystron amplifier , and has a negligible effect on the klystron performance .
the high - gain klystron amplifier described in this paper is an attractive alternative to other fel configurations for operation in short wavelength range .
an distinguishing feature of the klystron amplifier is the absence of apparent limitations which would prevent operation without bunch compression in the injector linac .
we have illustrated the proposed cascade klystron scheme using the parameters of the tesla test facility .
although the present work is concerned primarily for use in the soft x - ray spectrum , its applicability is not restricted to that range , for example an x - ray sase fel is a suitable candidate for application of cascade klystron scheme described here .
we thank r. brinkmann , b. faatz , j. feldhaus , g. geloni , m. koerfer , j. pflueger , j. rossbach for many useful discussions .
we thank c. pagani , j.r .
schneider and d. trines for interest in this work . | the simplest high gain free electron laser ( fel ) amplifier concept is proposed .
a klystron amplifier has the useful property that the various electronic processes take place in separate portions of the amplifier , rather than overlapping as in fel amplifier with an uniform undulator .
the klystron consists of two fundamental parts : succession of 2 - 3 cascades ( modulator ) , and an output undulator ( radiator ) in which the modulated electron beam coherently radiates .
each cascade consists of uniform undulator and dispersion section .
unlike distributed optical klystrons , we have a high gain per cascade pass .
this has a few consequences .
first , klystron gain does not depend on the bunch compression in the injector linac , i.e. maximum gain per cascade pass at high peak beam current is the same as at low peak beam current , without compression .
conventional , short - wavelength fel amplifier and distributed optical klystron require electron beam peak current of a few ka .
second , there are no requirements on the alignment of the cascade undulators and dispersion sections , because , in our ( high gain ) case , there is no need for radiation phase matching .
there are applications , like xfels , where the unique properties of high gain klystron fel amplifier are very desirable .
such a scheme allows one to decrease the total length of the magnetic system . in this paper
we discuss also implementation of the proposed sase fel scheme for the frequency multiplier , for femtosecond experiments , and multi - user soft x - ray facility .
desy 03 - 108 august 2003 , , and |
blazars are believed to be a sub - class of radio - loud agns with the relativistic jets pointing close to the line of sight , such that the entire electromagnetic spectra are dominated by non - thermal radiation produced in the jets .
these spectra usually reveal two broad components , the low energy one - peaking in the ir - to - x ray range , and the high energy one - peaking in the mev tev range .
both spectral components are variable , particularly strongly in their high energy parts @xcite .
variability time scales range from years to a fraction of a day , where the rapid variability often takes the form of high amplitude flares .
the most extreme events ( with the largest amplitude and shortest time scales ) have been recorded in @xmath0ray bands @xcite .
flares recorded in the gev and optical bands in high luminosity blazars @xcite and in the tev and x - ray bands in low luminosity blazars @xcite appear to be correlated .
these are most likely produced by the high energy ends of the distributions of radiating particles , and this behavior suggests co - spatial production of high energy and low energy components .
the flares are presumably produced in shocks propagating down the jet with relativistic speeds and formed by colliding inhomogeneities @xcite .
however , present data do not allow us to exclude other possibilities , e.g. , production of flares by inhomogeneities flowing through and shocked in the reconfinement regions @xcite or via collisions of jets with clouds @xcite . in any case
, the rapid , high amplitude variability events indicate that a significant fraction of the blazar radiation is produced on sub - parsec scales and this provides exceptional opportunity to explore jet properties in the vicinity of the central engine . in order to advance such a study
, one must first understand the dominant radiation processes operating in the sub - parsec jets . until about 10 years ago , the most viable radiation model for relativistic jets was the synchrotron - self - compton ( ssc ) model @xcite . according to this model , the smooth , polarized and variable low energy component of blazar spectra is produced by the synchrotron mechanism , while the high energy component results from comptonization of synchrotron radiation by the same population of electrons which produce the synchrotron component .
however , this rather simple picture changed after the discovery of very high and rapidly variable mev - gev fluxes in many ovv ( optically violent variable ) and hp ( highly polarized ) quasars by the egret instrument on the compton gamma - ray observatory ( cgro ) : during states of high activity , those were measured to exceed the synchrotron fluxes by a factor 10 or more @xcite , and it was quickly realized that other processes besides ssc can be more important for @xmath1-ray production @xcite .
two families of new models emerged , describing the @xmath0ray production : these were erc ( external - radiation - compton ) and hadronic models . regarding the external sources of seed soft photons for the inverse - compton process in the erc models the following
have been suggested : direct disc radiation @xcite ; belr ( broad - emission - line - region ) @xcite ; disc radiation scattered by the gas surrounding the jet @xcite ; and jet synchrotron radiation scattered / reprocessed back to the jet by the external gas @xcite . in hadronic models , the @xmath0-rays are produced via synchrotron mechanism by secondary electrons which are much more energetic than the electrons accelerated directly .
they are injected following hadronic processes and accompanying pair cascades .
the primary reaction in hadronic models involves collision of ultra - relativistic protons with soft photons @xcite or with the ambient cold protons / ions @xcite .
the hadronic scenarios , as alternative for the ssc model , have been proposed also to explain tev radiation , detected by atmospheric cherenkov detectors in a number of low luminosity bl lac objects @xcite . in our review
we discuss : environment of sub - parsec jets in agns ; dissipation of jet energy and particle acceleration mechanisms ; production of non - thermal radiation and blazar models .
typical electromagnetic spectrum of luminous `` non - blazar '' agns / quasars consists of three distinctive spectral components : the uv bump and two broad ( but less luminous ) components , the ir and the x - ray one .
in addition , in the optical - uv range , strong , broad emission lines ( bel ) are present .
uv bumps are produced by central parts of the optically thick accretion disc ; x - rays come from hot disc coronae , with possible additions from the base of a jet in radio - loud agns @xcite ; near / mid ir radiation presumably comes from dust , located in geometrically thick molecular torus and heated by radiation from the center ; and bels very likely originate in the disc winds , photoionized by uv radiation of an accretion disc @xcite .
these radiation components form dense radiative environment , through which jets presumably formed in the vicinity of the central black hole must pave the way out . because of doppler enhancement , the relativistic jets interact most effectively with radiation incident from the front and from side of the jet . at distances
where most of blazar non - thermal radiation is produced ( @xmath2 cm ) , such radiation is provided by the bel region and by the hot dust .
energy density of bels on the jet axis is @xmath3 where @xmath4 pc @xcite and @xmath5 is the correction function which depends on geometry of the bel region .
if the bel region forms a narrow torus lying in the disc plane , then @xmath6 energy density of dust radiation is @xmath7 where @xmath8 and @xmath9 is the fraction of the central engine luminosity @xmath10 reprocessed by hot dust . if a luminous agn possess a powerful jet , it is a radio - loud quasar .
observationally , it is classified as ovv / hp quasar if viewed along the jet ; as a radio - lobe dominated quasar if viewed at intermediate angles to the jet axis ; and as a radio - galaxy with fr ii type of large scale radio - morphology ( edge brightened lobes with hotspots ) if viewed at large angles to the jet axis @xcite . much less is known about the sub - parsec environment in very low luminosity agns .
in particular , the accretion process is less - well understood ; we do not know about the radiation efficiency of such objects , and to what extent optically thick plasma participates in the radiation processes .
the low luminosity radio - loud agns are often accompanied by strong jet activity .
the objects with jets pointing away from the line of sight are most likely fr i type radio galaxies with radio morphology characterized by edge darkened extended radio structures .
fr i radio galaxies probably form the parent population for bl lac objects @xcite , which , in similarity to the ovv / hp quasars , are viewed at small angles to the jet axis .
bl lac objects , together with the ovv / hp quasars , form the _ blazar _ category of agns .
it should be noted , however , that the existence of two distinct radio morphologies , fr i and fr ii ( and related bimodal classification of all radio loud objects ) does nt necessary imply bimodal distribution of agn properties .
specifically , that there is a large luminosity overlap between fr ii and fr i objects as well as between bl lac objects and ovv / hp ; that there is no evidence for bimodal distribution of equivalent widths of emission lines in blazars ; that there are some radio galaxies which on one side have fr i type radio structure and on another side fr ii type radio structure @xcite ; and , finally , the global spectral properties of blazars seem to form a continuous pattern parameterized by the total blazar luminosity @xcite .
rapid variability and the extent of blazar spectra to gev and tev energies imply particle acceleration _ in situ_. the ultimate source of particle energy is the kinetic and intrinsic energy of a jet .
the simplest and most popular model of production of high amplitude short term flares involves collisions of inhomogeneities moving with different velocities down the jet @xcite .
if inhomogeneities are ejected by the central engine with bulk lorentz factors @xmath11 , separated by a distance @xmath12 , they start to collide at a distance @xmath13 the collisions are followed by a formation of forward - reverse shock structures , with the shocked plasma enclosed between the shock fronts and moving with the bulk lorentz factor of the contact discontinuity surface , which for equal rest densities of inhomogeneities is @xmath14 ( for more general cases see @xcite ) . as the collision proceeds ,
relativistic particles are accelerated and produce non - thermal radiation . for symmetrical inhomogeneities , collisions last @xmath15 where @xmath16 is the radial width of the colliding inhomogeneities and @xmath17 is the velocity of the faster moving inhomogeneity , both as measured in the contact surface frame .
after that time the shocks and the injection of relativistic particles terminate .
thus , in case of a point source , the observer located at @xmath18 will see the flare which lasts @xmath19 provided electrons cool faster than the collision lasts . for the flow which is modulated such that @xmath20 , and @xmath21few , eqs.(5 - 6 ) give @xmath22 i.e. during the collision the source of radiation doubles its distance from the central object . for sources with finite size , light - travel effects
must be included , and then the flare time is @xmath23 where @xmath24 is the cross - sectional radius of the source and @xmath25 is its radial width ( the width of the layer of the shocked gas enclosed between the forward and reverse shock fronts ) .
however , because @xmath26 , where @xmath27 is the shock down - stream velocity , and because effectively in the conical jets @xmath28 ( no larger area than this can contribute to the observed radiation due to doppler beaming ) , the time scale for a source with a finite dimension is of the same order as for the point source .
efficiency of energy dissipation via collisions of symmetric inhomogeneities is @xcite @xmath29 some portion of dissipated energy is converted to the thermal energy of the shocked plasma , and the remaining part is used to accelerate protons and electrons up to relativistic energies .
protons are quite efficiently accelerated by fermi mechanism in shocks ( see reviews by @xcite ) .
time scale of the acceleration is @xmath30 where @xmath31 is the ratio of the mean free path for fermi scatterings to the gyro - radius , @xmath32 is the velocity of the upstream flow , and @xmath33 is the gyro - time .
the value of @xmath31 is equal to @xmath34 , where @xmath35 is the magnetic energy density per unit wavenumber @xmath36 in the turbulent magnetic field , and @xmath37 is the total magnetic energy density . for kolmogorov turbulence spectrum , @xmath38 , i.e. the smallest values of @xmath31 are predicted for most energetic protons @xcite .
protons accelerated by the fermi mechanism in strong but non - relativistic shocks have a power - law energy distribution with a slope corresponding to the equal energy per decade of energy .
electrons can be accelerated by shocks as well , but first they must be preheated to energies @xmath39 , in order to be scattered freely over the shock front and energized between diverging flows , where @xmath40 is the upstream flow velocity in the shock front frame .
there are several processes which have been suggested to preheat electrons @xcite .
the most promising scenario is , where electrons are heated by waves induced in the upstream flow by protons reflected from the shock front .
in general , the slope of energy distribution of preheated electrons can be different than the slope of accelerated protons , but those accelerated further up to @xmath41 via fermi scatterings will have similar slope to that of the protons .
the large difference is predicted , however , for maximum energies , which for electrons is much stronger limited by radiative energy losses than for protons .
radiative cooling of relativistic electrons is dominated by synchrotron mechanism and inverse - compton process .
the rate of radiative cooling in the source comoving frame , in which electrons are assumed to have isotropic momentum distribution , is @xmath42 where @xmath43 , @xmath44 is the magnetic energy density , @xmath45 $ ] is the radiation energy density within the frequency range for which scatterings off electrons with energy @xmath0 are within the thomson limit , and @xmath46 .
comparing time scale of electron radiative cooling , @xmath47 , with the collision time scale , @xmath48 , one can find that energy , above which energy distribution of electrons steepens due to radiative losses ( by unity with respect to the injection slope ) is @xmath49 radiation energy density is contributed by both local emission and by external radiation fields .
the dominant local contribution to @xmath50 for @xmath51 is provided by synchrotron radiation : @xmath52 where @xmath53 is the apparent luminosity of synchrotron radiation within the thomson limit ( @xmath54 ) .
contribution from the external radiation fields to @xmath55 is @xmath56 where @xmath57 is the intensity of the incoming external radiation , @xmath58 , and @xmath59 is the angle between the jet axis and an incoming ray .
provided that the dependence of @xmath57 on @xmath59 is much weaker than the dependence of @xmath60 on @xmath59 , the @xmath61 is dominated by those external radiation fields which at a distance of the flare production contribute significantly from the side and the front of the jet @xcite .
such radiation fields are provided by bel region and by infrared radiation by hot dust , and from eq .
( 15 ) @xmath62 where for the external sources located in the narrow torus at @xmath63 @xmath64 noting that distance of the flare production , @xmath65 is on the order of @xmath66 , the value of @xmath67 can be sometimes significantly lower than unity . for scatterings of ir radiation , because @xmath68 , we have @xmath69 . noting that @xmath70 hz , @xmath71 hz , and that @xmath72 , one can find that bels are scattered within the thomson limit by electrons with energies @xmath73 while the dust radiation is scattered within the thomson limit by electrons with energies @xmath74 protons lose energy via synchrotron and inverse compton radiation at a rate @xmath75 where @xmath76 . comparing the time scale of proton radiative cooling , @xmath77 , with the flare time scale , one can find that only protons with energies larger than @xmath78 can radiate efficiently .
protons can also lose energy via inelastic collisions with photons and cold ions .
collisions with photons can lead to meson production ( photo - meson process ) and to direct e@xmath79e@xmath80 pair production .
time scales of proton energy losses in these processes are @xcite : @xmath81 @xmath82 @xmath83 for photo - meson production , pair production , and @xmath84 collisions , respectively , where @xmath85 , @xmath86 , and the threshold photon energies for @xmath87 collisions are @xmath88 , and @xmath89 .
depending on the dominant mechanism considered for the @xmath0-ray production , blazar models can be divided into three groups : - ssc models , where production of @xmath0-rays is dominated by comptonization of locally emitted synchrotron radiation ; - erc models , where production of @xmath0-rays is dominated by comptonization of external radiation fields ; - hadronic models , where production of @xmath0-rays is dominated by synchrotron radiation of proton - initiated - cascades ( pic ) , and/or by synchrotron radiation of protons and muons .
these models are discussed below as applied to ovv / hp quasars and tev - bl lac objects .
the simplest ssc model is the one - component version , where the radiation observed at any given moment is produced by a source / shock moving along a jet . in the ssc model , the frequency of the @xmath0-ray luminosity peak is located at @xmath90 where @xmath91 is the location of the synchrotron peak , @xmath92 is the location of the break in the electron energy distribution , and @xmath93 . these equations together with the ratio of the luminosity peaks ,
@xmath94 allow to derive @xmath95 , @xmath96 , and @xmath92 .
the ssc process can dominate @xmath0-ray production in ovv / hp quasars if @xmath97 this can be satisfied in ovv / hp quasars only if geometry of the bel region is such that @xmath98 and if temperature of dust is lower than @xmath99 k , or that @xmath100 .
first condition is difficult to satisfy , because from reverberation mapping , @xmath101 ; second condition is in contradiction with observations of near / mid ir radiation in agns of radio - lobe dominated quasars .
finally , @xmath100 contradicts with vlbi data .
furthermore , one - component ssc model predicts x - ray spectra which are too soft .
much more promising are ssc models as applied to bl lac objects , in particular to the bl lac objects which are tev sources @xcite . in these objects ,
klein - nishina effects are important . using eqs .
( 26 - 28 ) , one can find that typically @xmath102 gauss and @xmath103 @xcite .
using these numbers , eq .
( 13 ) gives @xmath104 .
this suggests that spectral breaks at peaks are due to effects of cooling ; furthermore , the fact that @xmath92 is only by a factor few lower than @xmath105 explains why spectral components in tev blazars are hard almost up to highest frequencies . finally , noting that maximum electron energies presumably are determined by the balance of the acceleration time scale and radiative cooling time scale , having @xmath106 provides natural explanation for existence of `` histeretic '' variability patterns observed with both signs in the x - ray band in the spectral index vs. flux diagram @xcite . in the erc one - component model
the high energy peak is located at @xmath107 while ratio of the peak luminosities is @xmath108 when applying erc model to q - blazars , it is better not to rely on relation ( 27 ) .
this is not only because @xmath109 is located in these objects in the spectral range which observationally is poorly covered ( far - ir band ) , but also because there are some observational indications that far - infrared radiation can be strongly contaminated by radiation produced at larger distance in a jet than the high energy flares .
these indications are : the low energy break of the synchrotron component is at lower frequencies than the break due to synchrotron - self - absorption on sub - parsec scales @xcite ; and there is clear trend of decreasing amplitude of the flare with decreasing frequency in the synchrotron component @xcite . however , in order to close a system of equations describing the model , one has to replace one equation by another . such relationship can be derived by assuming that @xmath110 @xcite .
another useful approximation is to treat as a known observable @xmath96 , instead of @xmath111 .
this is because due to unknown geometry of the bel region and unknown maximum temperature of dust , model errors which can result from using uncertain values of @xmath67 and @xmath112 can be larger , than resulting from uncertain value of @xmath96 . alternatively , instead of treating @xmath96 as known , one can make an assumption about equipartition of magnetic fields with relativistic particles .
our preliminary results show that for @xmath113 both approaches give similar results . using observables typical of ovv / hp blazars , namely : @xmath114 hz ; @xmath115 ; @xmath116 day ; and @xmath117 , and assuming that the high energy luminosity is dominated by erc(bel ) , we obtain @xmath118 gauss ; @xmath119 ; @xmath120 hz ( which actually is below the synchrotron - self - absorption frequency ) ; @xmath121 ; and @xmath122 .
model parameters derived in this manner can be used as `` first approximation '' parameters for dynamical models intended for a study of time evolution of electron energy distribution .
such models have been recently developed using two kinds of approximations : one is a spherical source approximation by @xcite and another is a thin shell approximation by @xcite .
results of both demonstrate the ability of erc models to explain the observed spectra and flares of ovv / hp quasars .
in addition , these models support the earlier predictions that the ssc radiation component , despite its lower luminosity than the erc component , can dominate the soft / mid x - ray bands @xcite .
very little is known about radiative environment in these objects and , so , energy density of external radiation fields must be treated as a free parameter .
but there are interesting constraints regarding the level of the ir radiation from dust
. this radiation must be very weak , otherwise tev @xmath0rays would be absorbed in the pair production process @xcite . however , production of tev component by comptonization of radiation of dust is still feasible , provided @xmath123 .
hadronic models , which are motivated by theories of particle acceleration , suffer difficulties in explaining the observed electromagnetic spectra and/or short term variability . proposed for the ovv / hp quasars , models involving
proton induced cascades ( pic ) @xcite predict x - ray spectra which are softer than observed .
they also require fine - tuning of model parameters in order to have the location of the radiation deficiency ( the `` dip '' between the low energy and high energy peaks in the broad - band spectra ) located at the right frequency in all ovv / hp blazars .
furthermore , there are already several examples of ovv / hp blazars , where x - ray spectra have slopes @xmath124 @xcite , which can not be explained in terms of the synchrotron radiation mechanism .
this is because synchrotron radiation in the x - ray band requires such energetic electrons , that they must cool efficiently and , therefore , produce spectra with @xmath125 , even if injected mono - energetically . from unknown us reasons ,
no contribution of pp@xmath126 pe@xmath79e@xmath80 was taken into account in pic models .
the process is about 100 times less efficient per photon than photo - meson one , but this is compensated by a much larger number of photons above the threshold energy , which for pair production is @xmath127 lower than for photo - meson process . another hadronic model proposed for ovv / hp blazars is based on the process of inelastic collisions of ultrarelativistic protons with the cold ions .
such process in order to be efficient requires very dense targets and huge bulk lorentz factor ( @xmath128 ) of streaming protons . in the paper @xcite it is proposed that the dense target is provided by the walls of the funnel formed by geometrically thick disc .
the main problem with this model we can envisage is overproduction of soft x - rays , which would be produced due to comptonization of accretion disk uv photons by cold electrons in a jet this effect can be avoided if in the vicinity of the black hole , up to at least 100 gravitational radii , jet is not yet collimated @xcite . but then beamed blazar radiation must be produced at larger distances , probably at @xmath129 cm as the flare time scales suggest .
hadronic models have been proposed also for tev - blazars . in these objects
the soft spectra of pic models contrast with the very hard observed spectra even more .
this problem is avoided in proposed recently the proton - synchrotron model @xcite .
in such a model high energy spectra are produced directly by protons , just via synchrotron radiation .
the model predicts hard spectra and is also attractive , because can explain constant slope of high energy tail of tev component during high amplitude flux changes @xcite . in the model , maximum proton energies
@xmath130 are required in order to satisfy condition @xmath131 .
another version of the hadronic model has been suggested by rachen & mszros @xcite . in their model
the bulk lorentz factor is assumed to be rather low ( @xmath132 ) and protons with @xmath133 lose comparable energies via photo - meson process as via synchrotron mechanism . following the former ,
the muons are produced with such energies , that before decay they undergo significant synchrotron energy losses , dominating production of tev peak .
the weak point of both above hadronic models is that in order to accelerate protons up to @xmath133 energies , the quantities @xmath31 and @xmath134 ( see eq . [ 10 ] ) must be pushed to extremes , i.e. , should have values 1 .
as analysis of jet energetics and x - ray observations of blazars indicate , plasma in sub - parsec jets can contain from few up to tens of e@xmath79e@xmath80 pairs per proton @xcite . with such a number of pairs energy flux of sub - parsec jets is still dominated by protons , and , therefore , the structure of shocks is determined by proton plasma .
then , one should expect very efficient shock acceleration of protons , with maximum energies as high as @xmath135 , as limited by size of the source or balanced by energy losses .
furthermore , very large values of @xmath31 ( ratio of the mean free path for fermi scatterings to gyro - radius ) for electrons , @xmath136 , as derived for maximum electron energies assuming that for them @xmath137 , seem to be consistent with the presence of ultrarelativistic protons , provided the turbulent magnetic field has a kolmogorov spectrum @xcite .
on the other hand , very hard x - ray spectra and short variability time scales put severe constraints on radiative role of protons .
does this contradict with theoretical predictions about very efficient acceleration of protons ?
not necessary .
the average proton can not reach more random energy in the shock than @xmath138 , where @xmath139 is the bulk lorentz factor of the upstream flow as measured in the rest frame of the shock front .
intrinsic shocks are at most mildly relativistic , and therefore @xmath140 is expected . comparing this with the average energy of electrons which
, as deduced from hard x - ray spectra of some ovv / hp quasars , is in the range @xmath141 mev , and noting that limited energetics of jets combined with electron emissivity implies @xmath142 to be at least of the order of 10 , one can find that fraction of energy dissipated in the shock and used to accelerate electrons does not have to be much lower than energy channeled to protons .
this , combined with the fact that radiative efficiency of electrons is much larger than of protons , can explain negligible contribution of the proton related processes to the observed spectra .
aharonian f.a . ,
_ new astronomy _ * 5 * , 377 ( 2000 ) .
bednarek w. , _ apj _ * 402 * , l29 ( 1993 ) .
bednarek w. , and protheroe r.j . ,
_ mnras _ * 310 * , 577 ( 1999 ) .
begelman m.c . , rudak b. , and sikora m. , _ apj _ * 362 * , 38 ( 1990 ) .
biermann p.l . , and strittmater p.a . , _ apj _ , 643 ( 1987 ) .
blackman e.g. , _ apj _ * 456 * , l87 ( 1996 ) .
blandford r.d . , and
eichler d. , _ physics reports _ , 1 ( 1987 ) .
blandford r.d . , and levinson a. , _ apj _ * 441 * , 79 ( 1995 ) . baejowski m. , sikora m. , moderski r. , and madejski g.m . ,
_ apj _ * 545 * , 107 ( 2000 ) .
bottorff m. , korista k.t .
, shlosman i. , and blandford r.d .
, _ apj _ * 479 * , 200 ( 1997 ) .
celotti a. , fabian a.c . , and rees m.j . ,
_ mnras _ * 293 * , 239 ( 1998 ) .
chiaberge m. , and ghisellini g. , _ mnras _ * 306 * , 551 ( 1999 ) .
daigne f. , and mochkovitch r. , _ mnras _ * 296 * , 275 ( 1998 ) .
dar a. , and laor a. , _ apj _ * 478 * , l5 ( 1997 ) .
dermer c.d . , and schlickeiser r. , _ apj _ * 416 * , 458 ( 1993 ) .
dermer c.d . , and chiang j. , _ new astr . _ * 3 * , 157 ( 1998 ) .
dieckmann m.e .
, mcclements k.g .
, chapmann s.c .
, dendy r.o . , and drury l.oc . , _
* 356 * , 377 ( 2000 ) .
drury l. oc .
_ * 46 * , 973 ( 1983 ) .
edelson r. , _ apj _ * 401 * , 516 ( 1992 ) .
edelson r. , and malkan m.a .
, _ apj _ * 323 * , 516 ( 1987 ) .
emmering r.t .
, blandford r.d . , and shlosman i. , _ apj _ , 460 ( 1992 ) .
fabian a.c .
, iwasawa k. , celotti a. , brandt w.n . ,
mcmahon r.g . , and
, _ mnras _ * 295 * , l25 ( 1998 ) .
fossati g. , celotti a. , chiaberge m. , zhang y.h .
, chiapetti l. , ghisellini g. , maraschi l. , tavecchio f. , pian e. , and treves a. , _ apj _ * 541 * , 166 ( 2000 ) .
fossati g. , maraschi l. , celotti a. , comastri a. , and ghisellini g. , _ mnras _ * 299 * , 433 ( 1998 ) .
gaidos j.a .
et al . , _ nature _
* 383 * , 319 ( 1996 ) .
ghisellini g. , _ blazar demographics _
asp conference series , eds . c.m .
urry and p. padovani , in press ( astro - ph/0011356 ) .
ghisellini g. , and madau p. , _ mnras _ * 280 * , 67 ( 1996 )
. ghisellini g. , and maraschi l. , _ apj _ * 340 * , 181 ( 1989 ) .
ghisellini g. , padovani p. , celotti a. , and maraschi l. , _ apj _ * 407 * , 65 ( 1993 ) .
gopal - krishma , and wiita p.j . , _ a&a _ * 363 * , 507 ( 2000 ) .
hardcastle m.j . , and worrall d.m . , _ mnras _ * 309 * , 969 ( 1999 ) . hoshino m. , arons j. , gallant y.a . , and langdon a.b .
, _ apj _ * 390 * , 454 ( 1992 ) .
inoue s. , and takahara f. , _ apj _ * 463 * , 555 ( 1996 ) .
komissarov s.s . , and falle s.a.e.g . , _ mnras _ * 288 * , 833 ( 1997 ) .
knigl a. , _ apj _ * 243 * , 700 ( 1981 ) .
kirk j.g . ,
rieger f.m . , and mastichiadis a. , _ a&a _ * 333 * , 452 ( 1998 ) .
kubo h. et al .
, _ apj _ * 504 * , 693(1998 ) .
kuijpers j. , _ plasma astrophysics _ , european astrophysical doctoral network vii , eds . c. chiuderi , g. einaudi ( 1995 ) .
lazzati d. , ghisellini g. , and celotti a. , _ mnras _ * 309 * , l13 ( 1999 ) .
levinson a. , _ mnras _ * 278 * , 1018 ( 1996 ) .
macomb d.j . , et al . , _ apj _ * 449 * , l99 ( 1995 ) . mannheim k. , and biermann p.l .
_ a&a _ * 253 * , l21 ( 1992 ) .
marscher a.p . , and gear w.k . , _ apj _ * 298 * , 114 ( 1985 ) .
mattox j.r . , et al .
, _ apj _ * 476 * , 692 ( 1997 ) .
mcclements k.g . ,
dendy r.o . , bingham r. , kirk j.g . , and drury l.oc . ,
_ mnras _ * 291 * , 241 ( 1997 ) .
murray n. , and chiang j. , _ apj _ * 474 * , 91 ( 1997 ) .
mcke a. , and protheroe r.j .
, _ astropart .
_ in press ( astro - ph/0004052 ) .
nicastro f. , _ apj _ * 530 * , l65 ( 2000 ) .
peterson b.m . , _ pasp _ * 105 * , 247 ( 1993 ) .
petry d. et al .
, _ apj _ * 536 * , 742 ( 2000 ) .
rachen j.p . , and mszros p. , _ phys .
_ , 123005 ( 1998 ) .
rees m.j . , _ mnras _ * 184 * , p61 ( 1978 ) .
reeves j.n . , et al . , _
a&a _ * 365 * , l116 ( 2000 ) .
romanova m.m . , and lovelace r.v.e . , _ a&a _ * 262 * , 26 ( 1992 ) .
sambruna r.m . , et al . , _ apj _ * 538 * , 127 ( 2000 ) .
shimada n. , and hoshino m. , _ apj _ * 543 * , l67 ( 2000 ) .
sikora m. , begelman m.c .
, rees m.j .
, _ apj _ * 421 * , 153 ( 1994 ) .
sikora m. , and madejski g.m .
, _ apj _ * 534 * , 109 ( 2000 ) .
sikora m. , madejski g.m .
, moderski r. , and poutanen j. , _ apj _ * 484 * , 108 ( 1997 ) .
sikora m , sol h. , begelman m.c . , and madejski g.m .
, _ mnras _ * 280 * , 781 ( 1996 ) .
takahashi t. , et al .
, _ apj _ * 470 * , l89 ( 1996 ) .
takahashi t. , et al .
, _ apj _ * 543 * , l124 ( 2000 ) .
tavecchio f. , maraschi l. , and ghisellini .
g. , _ apj _ , 608 ( 1998 ) .
tavecchio f. , et al .
, _ apj _ * 543 * , 535 ( 2000 ) .
ulrich m .- h .
, maraschi l. , and urry c.m .
, _ araa _ , 445 ( 1997 ) .
urry c.m . , and padovani p. , _ pasp _ * 107 * , 803 ( 1995 ) .
von montigny c. , et al .
, _ apj _ * 440 * , 525 ( 1995 ) .
wagner s. , et al .
, _ apj _ * 454 * , l97 ( 1995 ) .
wehrle a.e . , et al . , _ apj _ * 497 * , 178 ( 1998 ) .
woniak p.r . ,
zdziarski a.a .
, smith d. , madejski g.m . , and johnson w.n . ,
_ mnras _ * 299 * , 449 ( 1998 ) . | in our review of the blazar phenomenon , we discuss blazar models , with a focus on the following issues : sub - parsec jets and their environment ; energy dissipation and particle acceleration ; and radiative processes . |
the @xmath0 meson is an important probe for studying relativistic heavy ion collisions .
since the mass of the @xmath0 meson is close to twice the kaon mass , any medium modification of its spectral shape ( mass and/or width ) @xcite as chiral symmetry restoration is approached may induce a change in its branching ratio in the kaon channel .
the simultaneous measurement of @xmath0 decay into @xmath7 and @xmath8 is a very powerful tool in the search for such in - medium modifications .
consisting of @xmath9 , the @xmath0 meson is also a sensitive probe of strangeness production @xcite .
the phenix experiment at rhic , with its excellent mass resolution and particle identification capability , ( comparable to or better than the natural width of the @xmath0 meson ) has the unique capability to measure the @xmath0 meson through both @xmath10 and @xmath2 decay channels at mid - rapidity .
we present the preliminary results of the @xmath0 meson measurement made during the 2001 rhic run .
the results presented here were obtained using the two central arms of the phenix spectrometer @xcite . the @xmath11 analysis used the drift chamber ( dc ) , two sets of multiwire proportional chambers with pixel - pad readout ( pc1 and pc3 ) , and the time of flight ( tof ) module of the east arm .
the acceptance is defined by the tof module which covers the pseudorapidity range @xmath12 and an azimuthal range of @xmath13 .
the kaons were identified via reconstructed momentum combined with a tof measurement with a time resolution @xmath14 ps . with a 2@xmath15 momentum dependent cut in
the mass squared distributions , @xmath16 were well separated up to @xmath17 gev / c .
the @xmath18 measurements were performed with the dc , pc1 , pc3 , the ring imaging cerenkov detector ( rich ) and the electromagnetic calorimeter ( emcal ) of both the east and west arm , each one covering @xmath19 and @xmath12 .
electrons were identified primarily by the rich .
further identification was provided by requiring the energy in the emcal to match the measured momentum of the tracks . charged particle tracks and their momenta were reconstructed using the dc and pc1 .
tracks were confirmed by a matching hit in pc3 and tof in the case of kaons , and in the emcal for electrons . the beam - beam counters ( bbc ) and
the zero - degree calorimeters ( zdc ) provided the trigger and were used to determine the event centrality .
the bbc were also used to determine the z - coordinate of the collision vertex ( @xmath20 ) .
the analysis used @xmath21 ( for kaons ) and @xmath22 ( for electrons ) minimum bias events with a vertex position within @xmath23 cm .
all identified kaon tracks in a given event were combined to form the invariant mass distributions of the like sign ( @xmath24 , @xmath25 ) and unlike sign pairs .
the large combinatorial background inherent to this procedure was estimated by an event mixing method in which all @xmath26 tracks from one event were combined with @xmath27 tracks of ten other events of the same centrality and vertex class .
the mixed event invariant mass distribution was then normalized to the measured @xmath28 .
the validity of the method was tested by constructing , in a similar way , a combinatorial like - sign spectrum and comparing it to the measured like - sign pair distribution .
finally , the signal was obtained by subtracting the mixed event spectrum from the measured one .
this gives the uncorrected @xmath0-yield ( @xmath29 ) of the @xmath30 decay channel .
an identical procedure was used for the measurement of the @xmath31 decay channel . in order to obtain the yield @xmath4 from the uncorrected signal , we correct our invariant mass distributions for detector acceptance and reconstruction efficiency .
the correction was determined using monte carlo simulations in two steps . in the first step ,
we generated single @xmath0 mesons with an exponential transverse momentum distribution ( @xmath32 ) , assuming a temperature of _ t _ = 380 mev which is consistent with the measurement by the star experiment at @xmath33 gev @xcite .
the generated @xmath0 s were then propagated through the phenix detector simulation and the pair acceptance and reconstruction efficiency was calculated by : @xmath34 in the second step , the multiplicity dependent efficiencies ( @xmath35 ) were estimated by embedding single tracks into the real data events .
the final corrected yield is given by @xmath36 where br represents the branching ratio for the specific decay channel ( @xmath37 or @xmath38 ) .
1 shows the minimum bias invariant mass distributions for @xmath11 ( panel ( a ) ) and @xmath18 ( panel ( c ) ) .
the kaon channel results for the 40%-80% centrality range are shown in fig .
1(b ) , where we see a considerable improvement in the signal to background ratio compared with minimum bias spectrum . since the combinatorial background increases quadratically with centrality
, this means that the signal increases at a much lower rate with centrality .
[ cols= " < , < , < " , ] 0.8 cm figure 2 : centrality dependence of @xmath0 yields ( solid lines are statistical and brackets represent systematic errors ) .
phenix has measured the yield of @xmath0 mesons via the @xmath10 and @xmath2 channels .
the values of @xmath4 for @xmath39 and @xmath40 are consistent within the large statistical and systematic errors of the present preliminary analysis .
the centrality dependence of the @xmath0 yields in the kaon decay channel is also presented .
7 r. rapp , nucl - th/0204003 ; s. pal , c.m .
ko , zi - wei lin , nucl .
a707 ( 2002 ) 525 .
p. koch , b. muller and j. rafelski , phys .
rep . 142 ( 1986 ) 167 .
t. chujo for the phenix collaboration , these proceedings .
c. adler _ et al _ ( star collaboration ) , phys .
c 65 ( 2002 ) 041901(r ) .
d. rhrich , j. phys .
g 27 ( 2001 ) 355 .
a. bazilevsky for the phenix collaboration , these proceedings . | a simultaneous measurement of the @xmath0 meson via its @xmath1 and @xmath2 decay channels was performed in au + au collisions at @xmath3 = 200 gev at mid - rapidity by the phenix experiment .
the preliminary minimum bias yields @xmath4 in the kaon and electron channels are @xmath5 and @xmath6 , respectively .
the centrality dependence of the yield in the @xmath1 channel is presented . |
one of the most challenging problems in quantum computation has been the design of quantum algorithms which outperform their classical counterparts in meaningful tasks .
few quantum codes in this category have been discovered after the well known examples by shor and grover @xcite . since the classical codes based on the random walk process
have been extremely well succeeded in certain tasks @xcite , the hope that the quantum random walk @xcite may provide a similar insight for quantum coding has generated a great interest in this model .
however , the dynamical properties of the quantum random walk are rich enough so that this system is physically interesting in its own right @xcite . for instance , due to quantum coherence effects , the position distribution of the quantum random walker spreads out linearly in time .
in contrast , the classical random walker spreads out only as the square root of time .
several systems have been proposed as candidates to implement quantum random walks .
these proposals include atoms trapped in optical lattices @xcite , cavity quantum electrodynamics ( cqed ) @xcite and nuclear magnetic resonance ( nmr ) in solid substrates @xcite . in liquid - state nmr systems
@xcite , time resolved observations of spin waves has been done @xcite .
it has also been pointed out that a quantum walk can be simulated using classical waves instead of matter waves @xcite .
all these proposed implementations face the obstacle of decoherence due to environmental noise and imperfections .
decoherence in the quantum walk on the line has been considered recently by several authors .
numerical simulations of the effect of different kinds of measurements have shown that the quantum walk properties are highly sensitive to decoherent events and in particular that the quadratic increase of the variance is eventually suppressed @xcite .
other studies focused on the effect of measurements in chirality and reached similar conclusions @xcite . a different decoherent mechanism , unitary noise , also leads to a crossover from a quantum behavior at short times to a classical - like behavior at longer times @xcite .
it is clear that the quadratic increase in the variance of the quantum walk with time is a direct consequence of the coherence of the quantum evolution @xcite .
this can also be visualized through the separation of the evolution equation for a quantum system into a markovian term and a quantum interference term as proposed in @xcite . in this work , besides the study of the effect of measurements on the evolution of quantum random walkers , we also consider the decoherences generated by a different process , namely the influence of randomly broken links on the dynamics of the quantum walk .
these mechanisms may be relevant in experimental realizations of quantum computers based on ising spin1/2 chains in solid - state substrates @xcite .
the paper is organized as follows . in the next section ,
we briefly introduce the basic notions and notation relative to the discrete time quantum walk on the line .
then , in section [ sec : medidas ] , we consider joint periodic measurements of chirality and position and show how the resulting evolution can be described in terms of a master equation . in section [ sec :
blinks ] decoherence is introduced through random failures in the links between neighboring sites . in section [ sec : brownian ] , we show how the stochastic classical model of brownian motion can be used to describe the time dependence of the variance of the decoherent quantum walk . finally , in section [ sec : conclusions ] , we summarize our conclusions .
let us consider a particle that can move freely over a series of interconnected sites .
the discrete quantum walk on the line may be implemented by introducing an additional degree of freedom , the chirality , which can take two values : `` left '' or `` right '' , @xmath2 or @xmath3 , respectively .
this is the quantum analog of the coin flipping decision procedure for the classical random walker . at every time step ,
a rotation ( or , more generally , a unitary transformation ) of the chirality takes place and the particle moves according to its final chirality state .
the global hilbert space of the system is the tensor product @xmath4 where @xmath5 is the hilbert subspace associated to the motion on the line and the chirality hilbert subspace is @xmath6 .
if one is only interested in the properties of the probability distribution , it suffices to consider unitary transformations which can be expressed in terms of a single real angular parameter @xmath7 @xcite .
let us call the operators that translate the walker one site to the left ( right ) on the line in @xmath5 as @xmath8 ( @xmath9 ) , respectively , and let @xmath10 and @xmath11 be the chirality projector operators in @xmath6 .
we consider transformations of the form @xcite , @xmath12 where @xmath13 is an unitary operator acting on @xmath6 , @xmath14 and @xmath15 being the pauli matrices , and @xmath16 is the identity operator in @xmath5 .
the unitary operator @xmath17 evolves the state @xmath18 by one time step , @xmath19 the wavevector @xmath18 is expressed as the spinor @xmath20{|n\rangle } , \label{spinor}\ ] ] where the qubit @xmath21 has an upper ( lower ) component associated to the left ( right ) chirality and the states @xmath22 are eigenstates of the position operator corresponding to the site @xmath23 on the line . in this work , for simplicity we set @xmath24 in eq .
( [ ugen ] ) so that the unitary operation on chirality reduces to a hadamard operation , @xmath25 the evolution corresponding to eq .
( [ evol1 ] ) , can then be written as the map @xmath26 \label{map } \\
b_{n}(t+1 ) & = { \frac{1}{\sqrt{2}}}\left[a_{n-1}(t)-b_{n-1}(t)\right ] . \notag\end{aligned}\ ] ] the probability distribution for the walker position at time @xmath27 is given by @xmath28 note that the unitary evolution described by eq .
( [ map ] ) has the property that after each time step two new sites are occupied .
therefore , if initialized at site @xmath29 , after @xmath27 steps the wavefunction is non - zero , at most , in the @xmath30 sites around the origin . from the map in eq .
( [ map ] ) it may be shown that the hadamard walk has a characteristic interference profile with two wavefronts which propagate away from the origin with constant speed @xmath31 @xcite . in what follows
we consider the effect of performing measurements on the evolution of this system .
let us take as an initial condition the random walker starting from the position eigenstate @xmath32 with chirality @xmath33 .
this choice results in a symmetric evolution with @xmath34 @xcite . the position and chirality of the walker are jointly measured every @xmath35 steps . among the several alternatives for measuring chirality
, we choose to measure it in such a way that the chirality is projected on the @xmath36 direction by the @xmath37 pauli operator .
then , the qubit states @xmath33 and @xmath38 are eigenstates of the measurement operator and the symmetric character of the evolution is preserved .
the wavefunction evolves according to the unitary map ( [ map ] ) for the first @xmath35 steps . at time @xmath39 position and chirality
are measured for the first time .
the probability that the result of the position measurement is @xmath23 , _
i.e. _ that the wavefunction collapses into the eigenstate @xmath22 of position , is @xmath40 the probability distribution @xmath41 depends on the initial qubit state .
however , for the chirality measurement procedure described above , this distribution repeats itself after each measurement , although centered at different position eigenstates , as can be seen from ( [ map ] ) .
after this first measurement , the unitary evolution continues and new measurements are performed at times @xmath42 .
the discrete variable @xmath43 indicates the number of measurements performed on the system up to a given time . to illustrate this procedure , in figure [ fig : dia1 ] we consider an example for the evolution in the case @xmath44 .
time steps .
the evolution shown assumes that the initial state in position of the system is @xmath45 . ] for arbitrary intervals between consecutive measurements @xmath35 , the probability distribution @xmath46 satisfies the master equation @xmath47 where the markovian transition probabilities from site @xmath48 to site @xmath23 , @xmath49 are defined in eq .
( [ position ] ) .
we calculate now , using eq .
( [ markov1 ] ) , the first moment as @xmath50 and the second moment as @xmath51 . after some manipulation
we have @xmath52 where @xmath53 and @xmath54 are the first and second moments associated to the unitary evolution between measurements .
therefore the variance , @xmath55 , verifies @xmath56 where @xmath57 is the variance associated to the unitary evolution between measurements . and @xmath58 .
the parabolic segments correspond to the free quantum evolution between measurements .
the slope just before a measurement is proportional to the diffusion coefficient according to eq .
( [ dqrm0 ] ) .
the initial condition is as mentioned in the text . ] from eq .
( [ varia ] ) the diffusion coefficient in the case of repeated measurements , @xmath59 over a time scale @xmath60 including several measurement processes , is @xmath61 since the variance of the free evolution of the quantum system increases quadratically with time , we have @xmath62 where , for @xmath63 , @xmath64 is a constant determined by the initial conditions .
then , the diffusion coefficient depends linearly on the time interval between measurements , @xmath65 _ i.e. _ inversely with the frequency @xmath66 of decoherent events .
we show in fig .
[ fig : dia2 ] the variance @xmath67 calculated through a computer simulation of the time evolution of an ensemble of @xmath68 trajectories .
these calculations confirm the linear dependence of the diffusion coefficient with the time interval @xmath35 in eq .
( [ drm ] ) .
let us now consider a different mechanism for introducing decoherence in the quantum walk .
suppose that , at time @xmath27 , a given site @xmath23 has one or both of the links connecting it to its neighboring sites broken .
if site @xmath23 has no broken links , as in figure [ fig : lnk ] ( a ) , the evolution law ( [ map ] ) , which implies a hadamard operation in chirality space followed by a conditional translation ( upper spinor component to the left and the lower component to the right ) is applied . of the line when there are : ( a ) no broken links , ( b )
the link to the left of the site is broken , and ( c ) both links are broken .
the upper ( lower ) arrows indicate the direction of the probability flux associated to the upper ( lower ) spinor component . ]
when one or both links at site @xmath23 are opened there can be no translation across the broken link and the evolution must be modified accordingly . if the link to the left of site @xmath23 is broken , as shown in fig .
[ fig : lnk ] ( b ) .
the upper component of the spinor at @xmath23 receives probability flux from @xmath69 . in order to conserve flux , the outgoing probability flux from the upper component at @xmath23
must be diverted to the lower component at the same site and the corresponding transformation on the spinor components is , @xmath70\nonumber\\ b_n(t+1)&=&{\frac{1}{\sqrt{2}}}\left[a_{n}(t)+b_{n}(t)\right].\label{left_blink}\end{aligned}\ ] ] if the broken link is to the right of @xmath23 , the situation is similar , and the transformation is @xmath71\nonumber\\ b_n(t+1)&=&{\frac{1}{\sqrt{2}}}\left[a_{n-1}(t)-b_{n-1}(t)\right ] .
\label{right_blink}\end{aligned}\ ] ] finally , if site @xmath23 is isolated , as in fig .
[ fig : lnk ] ( c ) , the hadamard operation is followed by a chirality exchange , @xmath71\nonumber\\ b_n(t+1)&=&{\frac{1}{\sqrt{2}}}\left[a_{n}(t)+b_{n}(t)\right ] .
\label{2blink}\end{aligned}\ ] ] the evolution consists of a sequence of unitary transformations applied to the initial wavefunction @xmath72 and the form of each @xmath73 depends on the ( random ) topology of the line at time step @xmath74 .
it follows that eqs .
( [ map ] ) and ( [ left_blink])([2blink ] ) preserve the norm of the wavefunction , @xmath75 , in the presence of an arbitrary number of broken links . in this sense , this model of decoherence is similar to the case of random unitary noise discussed in @xcite and is essentially different from repeated measurements , which of course can not be described as a sequence of unitary transformations .
we choose the same initial condition as in the previous section , namely + @xmath76 , which leads to a symmetric evolution in the coherent case .
the algorithm for implementing the quantum walk with broken links proceeds as follows . at each time
step @xmath27 , the state of the links in the line is defined .
each link has a probability @xmath0 of breaking in a given time step , @xmath0 being the only parameter in the model and the hadamard walk is recovered if @xmath77 .
then , the corresponding transformation , either of eqs .
( [ map ] ) , ( [ left_blink ] ) , ( [ right_blink ] ) , or ( [ 2blink ] ) , is applied at each site .
we focused on two quantities : the probability distribution , eq .
( [ prob ] ) , at a fixed time and the time evolution of its variance . as a result of decoherence ,
the distributions for @xmath78 are qualitatively different from the distribution for the unitary hadamard walk .
the probability distributions for @xmath79 approach a gaussian shape in the long time limit . at fixed time , the width of the gaussian becomes narrower as the probability of breaking links increases . for values of @xmath0 larger than about @xmath80 the high frequency of broken links prevents the spread of the wavefunction beyond a certain region around the origin . in order to find out how does the transition between the coherent quantum walk distribution and the gaussian like distributions take place we investigate the low probability region of the model . in fig .
[ fig : distav ] , the average profile @xmath81 for @xmath82 is shown at different times . at early times ( upper panel )
the distribution is still close to the coherent one while at long times ( lower panel ) the characteristic gaussian profile has emerged .
the corresponding profiles for the case @xmath77 , are shown in the background as a reference . as disruptive events ( in this model ,
the breaking of the links ) take place , the distribution gradually shifts from the typical shape for the coherent quantum random walker to the one characteristic of a classical walker .
there is a characteristic time , @xmath83 , associated to this transition .
a simple argument can be used to understand the dependence of @xmath83 on the frequency @xmath0 . at early times , @xmath84 , there are no disruptive events under the initially localized wavefunction and it spreads coherently in both directions with speed @xmath31 , covering a range @xmath85 in time @xmath27 .
the mean number of broken links per time step under the wave function increases as @xmath86 .
when this number becomes of order 1 , the effect of the disruptive events becomes relevant .
this happens at a coherence time @xmath87 as mentioned before , for longer times @xmath88 , the probability profile approaches a gaussian distribution .
the transition is reflected in the variance of the distribution as well . for early times , @xmath84 , the variance shows the quadratic increase characteristic of the unitary quantum walk .
however , for non zero @xmath0 , a transition to a linear increase regime takes place .
[ fig : mdv001 ] shows the time evolution of @xmath89 for @xmath82 , for which eq .
( [ tcoherence ] ) implies a coherence time @xmath90 .
it is clear that the transition takes place gradually in time . in the case of broken links ,
disruptive events take place at each time step so the diffusion coefficient , @xmath91 , in fig .
[ fig : mdv2av ] , the average time evolution of @xmath89 for several @xmath0 values is shown in a log - log scale . from these results , an estimate for the diffusion coefficient
can be obtained as a function of @xmath0 .
the results , shown in fig .
[ fig : kpfit ] , indicate a linear dependence with @xmath92 , @xmath93 a linear regression gives @xmath94 and this implies that for @xmath95 the diffusion coefficient is @xmath80 as in the unbiased classical random walk . for higher values of @xmath0 , links are broken too frequently and the wavefunction is confined to a region about the origin .
this prevents full diffusion . for smaller link breaking frequencies
the diffusion rates are higher than the classical rate , suggesting that correlations persist , as we show in section [ sec : brownian ] .
this fact has been observed in the context of decoherence induced by non - unitary noise @xcite .
therefore , it should be kept in mind that the estimate for @xmath96 , based on finite time data , is not exact because of residual correlations . as a consequence ,
it depends ( weakly ) on the initial conditions . in the next section , where we consider the decoherent quantum walk from the perspective of the brownian motion model
, we provide a simple analytical argument showing that the assumption of complete correlation decay leads to incorrect results .
there is an interesting similarity between eqs ( [ drm ] ) and ( [ dbl ] ) for the dependence of the diffusion coefficients in the case of repeated measurements and in the evolution with broken links . in both cases ,
the dominant contribution to the diffusion coefficient depends inversely on the frequency of disruptive events .
a similar dependence appears in other decoherent quantum walks @xcite .
this dependence is due to the persistence of quantum correlations . in order to investigate the role of quantum correlations in the diffusion coefficient , and in particular to see how the corresponding classical diffusion scales with @xmath0
, we consider the quantum evolution equations , eqs .
( [ map ] ) and ( [ left_blink][2blink ] ) , written in terms of the occupation probabilities defined in ( [ prob ] ) , @xmath97+\beta_{n+1}(t)-\beta_{n-1}(t)\nonumber\\ p_n(t+1)&=&\frac12\left[p_{n-1}(t)+p_{n}(t)\right]-\left[\beta_{n-1}(t)+\beta_{n}(t)\right]\label{pev}\\p_{n}(t+1)&=&\frac12\left[p_{n}(t)+p_{n+1}(t)\right ] + \beta_{n}(t)+\beta_{n+1}(t)\nonumber\\ p_{n}(t+1)&=&p_{n}(t).\nonumber\end{aligned}\ ] ] in these equations @xmath98 $ ] and @xmath99 is the real part of @xmath100 .
these terms take into account quantum coherence effects and are responsible for the essential differences between the classical and quantum random walks . under the assumption of a complete decay of correlations
their contribution is negligible and a classical description emerges @xcite . at a given time and
site only one among eqs .
( [ pev ] ) applies , depending on the number of broken links for the site .
a statistical description can be obtained by combining them into a single evolution equation with the appropriate statistical weights . in the model described in section [ sec :
blinks ] , the probability that a given site has no adjacent broken links is @xmath101 , that it has a left or right broken link is @xmath102 and that it is isolated is @xmath103 .
these probabilities are appropriate statistical weights since they satisfy @xmath104 .
the resulting classical evolution equation for @xmath105 is , @xmath106 .
\label{clas}\ ] ] this evolution describes a genuine diffusive process with a diffusion coefficient @xmath107 this can be easily seen by changing to continuous time and position variables @xmath108 , in terms of which eq .
( [ clas ] ) becomes the diffusion equation @xmath109 thus the classical version of the broken link model , with complete correlation decay , implies a diffusion coefficient which starts at the unbiased random walker value @xmath80 for @xmath77 and decreases as the frequency of broken links increases .
the fact that this classical diffusion coefficient does not scale as @xmath92 as in the quantum random walk with broken links , is due to persistent quantum correlations . as a result of these persistent quantum correlations the diffusion coefficient of the decoherent quantum walk , eq .
( [ dbl ] ) , is always larger than the corresponding classical one , eq .
( [ dcbl ] ) .
this characteristic of the quantum walk has also been mentioned by other authors which have considered other sources of decoherence @xcite and is also apparent in the case of repeated measurements considered in this work .
the periodic measurement model discussed in section [ sec : medidas ] can be easily generalized to the case where the time intervals between consecutive measurements are randomly distributed . in this case the diffusion coefficient is given by @xmath110 where @xmath111 @xmath112 is the average of the ( squared ) time intervals between measurements .
the evolution of the average variance when the distribution of time intervals between measurements is uniform in @xmath113 $ ] is shown in fig .
[ fig : dia3 ] . in this case ,
@xmath114 , @xmath115 and the diffusion coefficient from eq . (
[ ran ] ) agrees with that of the periodic case , eq .
( [ drm ] ) , for @xmath116 .
this diffusion with random intervals between measurements is analogous to the one found in the case of simple classical brownian motion , which describes a heavy particle immersed in a fluid of light molecules which collide with it randomly .
the collisions may be associated to the quantum measurement processes , while the free motion between collisions plays the role of the unitary quantum evolution between measurements . .
the dashed line corresponds to the periodic measurement case with @xmath116 ] to explore further this analogy , let us consider the langevin equation for the one - dimensional motion of a brownian particle @xcite @xmath118 where @xmath119 is the velocity of the particle , @xmath120 a viscosity coefficient and @xmath121 the impulsive force per unit mass due to the random collisions with the surrounding molecules .
it is assumed that @xmath121 has a zero average and that its correlations satisfy the fluctuation - dissipation theorem . under conditions of thermal equilibrium with the surrounding fluid and
if initially @xmath122 , the time evolution of the variance for the position of the brownian particle is @xmath123,\label{browniano}\ ] ] where @xmath64 is a constant , fixed by the initial conditions , which is related to the temperature in the brownian motion model .
we now consider how this model applies to the decoherent quantum walks described in this work .
the inverse of the dissipation coefficient , @xmath124 , defines a characteristic time in which the variance has a transition between a quadratic and a linear increase . for short times
there is a small chance of a collision and the variance increases quadratically @xmath125 this expression is analogous to eq .
( [ sigmaq ] ) , obtained in section [ sec : medidas ] for the periodic measurement case . for long times , there are many collisions , the dissipative effects become dominant and the variance increases linearly with time , @xmath126 since the particle is slowed down and suffers a diffusive process with diffusion coefficient @xmath127 this can be compared to the quantum random walker at long times , _
i.e. _ after many position and chirality measurements have been performed . from eq .
( [ drm ] ) we obtain the relation between the period @xmath35 of the measurements for the quantum walker of section [ sec : medidas ] and the characteristic time of brownian motion as @xmath128 thus , repeated measurements can be viewed as a dissipative process and the quantum random walker with decoherence brought about by these measurements can be described in terms of a fully classical model .
this suggests that the particular mechanism by which the decoherence is introduced in the dynamics does not seem to be important as long as the proper time scale is considered .
in fact , if decoherence is introduced through broken links , as described in section [ sec : blinks ] , the brownian motion model also applies and both parameters of the brownian model , @xmath64 and @xmath120 , can be read out from the numerical results . for early times , @xmath84 , the variance of the quantum walker increases quadratically because the chance of decoherent events under the wave function , proportional to @xmath129 , is initially very low . for the initial conditions used in this work , we have @xmath130 .
the open links under the wavefunction act as dissipative events , so the dissipation rate @xmath120 is related to the frequency of broken links @xmath0 . a comparison of the diffusion coefficients , eqs .
( [ difusion ] ) and ( [ dbl ] ) , implies the relation @xmath131 as shown in fig .
[ fig : langevin ] , this simple stochastic model describes well the time evolution of the variance over a broad range of @xmath0 values , including the smooth transition that takes place between a quadratic and a linear increase regime in the variance .
two instances of decoherent quantum walks have been considered . in one of them ,
decoherence is introduced through frequent measurements of position and chirality . in the other , decoherence results from randomly breaking a few links in the line . through comparison of the results obtained with these two models and their classical counterparts ,
we have drawn some interesting conclusions about the diffusion rates and the decay of correlations in decoherent quantum walks . in the case of the quantum
walk on the line with periodic position and chirality measurements , it has been shown that the evolution , in a timescale involving many measurement events , can be described as a markovian process and the probability distribution satisfies a master equation .
the associated diffusion coefficient has been analytically obtained and it varies inversely with the frequency of measurements .
this results from the fact that the measurements introduce decoherence in the quantum dynamics , partially breaking correlations and providing a route to a classical - like behavior .
the periodic measurement has been extended to random time intervals between consecutive measurements and in this case the diffusion coefficient depends on a characteristic time given by the ratio of the second to the first moment of the time interval distribution . a comparison with the well known markovian process of brownian motion allowed us to relate the average time between measurements with the characteristic time for brownian motion , @xmath124 .
furthermore , the initial condition of the quantum random walker determines a constant @xmath64 which is directly related to the temperature in the brownian motion model . when the decoherence is introduced by randomly breaking a some links in the line , a crossover from a coherent regime with the characteristic quadratic increase of the variance to a diffusive regime with a linear increase of the variance results .
we find numerically that in this case the diffusion coefficient scales with the link breaking probability @xmath0 as @xmath92 and that coherence effects are limited to a timescale @xmath132 .
the simple brownian motion model adequately describes the time evolution of the variance , including the transition from quadratic to linear growth , for a broad range of decoherence rates .
these results suggest that in noisy quantum systems the coherence of the quantum evolution persists below a characteristic time of order @xmath133 where @xmath134 is the frequency of single decoherent events .
after this characteristic time a quantum diffusion regime results in which the quantum diffusion rate is faster by a factor @xmath133 than the classical rate .
this is due to quantum correlations present between consecutive decoherent events . for low decoherence rates , even at long times ,
the decoherent quantum evolution diffuses faster than the corresponding classical evolution . on the other hand , at high decoherence rates , the role of quantum coherences
is minimized and the quantum diffusion approaches the classical diffusion rate .
acknowledgements : + g.a . and r.d .
thank h. pastawski for useful discussions .
we acknowledge support of _ pedeciba _ and _ conicyt _ ( uruguay ) .
r.d . acknowledges partial financial support from the brazilian research council ( cnpq ) .
a.r , g.a . and r.d .
acknowledge financial support from the _
brazilian millennium institute for quantum information_cnpq .
a. sinclair , _
algorithms for random generation and counting : a markov chain approach _ , birkhauser press , 1993 ; m. dyer , a. frieze and r. kannan , j. of the acm , * 38 * , 1 ( 1991 ) ; u. schning , _ a probabilistic algorithm for k - sat and constraint satisfaction problems _ , @xmath136 annual symp . on foundations of computer science , ieee , 1999 ; m. jerrum , a. sinclair and e. vigoda , proc .
@xmath137 stoc , 712 , ( 2001 ) . | we investigate the quantum walk on the line when decoherences are introduced either through simultaneous measurements of the chirality and particle position , or as a result of broken links .
both mechanisms drive the system to a classical diffusive behavior . in the case of measurements , we show that the diffusion coefficient is proportional to the variance of the initially localized quantum random walker just before the first measurement . when links between neighboring sites are randomly broken with probability @xmath0 per unit time , the evolution becomes decoherent after a characteristic time that scales as @xmath1 . the fact that the quadratic increase of the variance is eventually lost even for very small frequencies of disrupting events , suggests that the implementation of a quantum walk on a real physical system may be severely limited by thermal noise and lattice imperfections .
hadamard walk ; quantum information ; random walk ; markov process ; brownian motion |
the purpose of this paper is to address the flatness issue in some class of analytic trigonometric polynomials , and to give an alternative proof to the proof given by the author in @xcite .
therefore , this paper can be seen as a revised version of that article with an alternative proof .
but , as in @xcite , the combinatorial singer s construction and marcinkiewicz - zygmund interpolation inequalities ( * ? ? ?
x ) lies in the heart of the proof .
we thus construct , in the same spirit as in @xcite , a class of polynomials that are flat in the almost everywhere sense using the singer s construction combined with the refinement of the interpolation methods initiated by marcinkiewicz & zygmund .
it follows that the hardy spaces and the carleson interpolation theory play an important role in the proof . for a nice account on the interpolation theory and the @xmath3 theory ,
we refer the reader to ( * ? ? ?
9 ) and ( * ? ? ?
* , chap . 7 ) , @xcite , @xcite .
+ our construction further benefited from ideas of ben green and gowers related to the flatness problem in connection with singer and sidon sets @xcite , @xcite .
we also take advantage from the recent investigations on the marcinkiewicz - zygmund inequalities and its refinements @xcite , @xcite , @xcite , @xcite .
+ we stress that this paper is deeply indebted to the investigation started in @xcite , @xcite and @xcite .
so , it is may seen as a companion to those papers .
+ we further stress that the proof given here is completely different than the proof given in the previous version of this work @xcite .
although , it was stated in @xcite that one can give an alternative proof of the main results using a carlson interpolation theory combined with the methods of disturbed root of unit due to c. chui and zhong @xcite .
so , here , our main task is to present this alternative proof .
+ we remind that the flatness problem was initiated by littlewood @xcite and erds @xcite , and it has a long history . in the beginning , littlewood asked on the existence of the sequence of the polynomials on the circle @xmath4 with @xmath5 such that @xmath6 where @xmath7 are positive absolute constants and uniformly on @xmath8 of modulus @xmath9 . nowadays the analytic polynomials on the circle with @xmath10 coefficients are called littlewood polynomials .
+ erds and newman @xcite considered the problem of the existence of the positive absolute constant @xmath11 such that @xmath12 where @xmath13 .
beller and newman established that the answer is affirmative if one asked for the polynomials with coefficients bounded by 1 @xcite . in the opposite , j - p .
kahane disproved the conjecture @xcite .
+ besides , using a construction of byrnes @xcite , t. krner @xcite disproved erds - newman s conjecture .
but , it is turn out that the main ingredient form @xcite used by krner is not valid @xcite .
nevertheless , kahane s proof does not used this argument .
+ kahane s result has been strengthened by j. beck who proves that the ultraflat polynomials exist from the class of polynomials of degree @xmath14 whose coefficients are @xmath15th roots of unity @xcite .
j. beck s construction is essentially based on the random construction of kahane .
+ since then , it was a long standing problem to obtain effective construction of ultraflat polynomials until solved very recently by bombieri and bourgain @xcite . for a deeper treatment on the kahane ultraflat polynomials ,
we refer the reader to @xcite .
+ the third extremal problem in the class of analytic trigonometric polynomials concern @xmath0-flatness . this problem seems to be mentioned first in @xcite .
therein , newman wrote that it has been conjectured : for any littlewood polynomial @xmath16 of degree @xmath14 , @xmath17 , where @xmath18 .
+ in @xcite , newman solved the problem of @xmath0-flatness in the class of analytic trigonometric polynomials with coefficients of modulus 1 .
he proved that the gauss - fresnel polynomials are @xmath0-flat .
we refer to @xcite for a simple proof .
this result has been strengthened by beller @xcite , and beller & newman in @xcite by proving that the sequence of the mahler measure of the @xmath1 normalized gauss - fresnel polynomials converge to one .
+ for the polynomials with random coefficients @xmath19 , salem and zygmund @xcite proved that for all but @xmath20 choices of @xmath21 , @xmath22 for some absolute constant @xmath23 .
halz @xcite strengthened this result by proving @xmath24 for some absolute constant @xmath25 .
byrnes and newman computed @xmath26-norm of those polynomials @xcite .
later , browein & lokhart @xcite , and choi & erdlyi @xcite used the central limit theorem to compute the limit of the @xmath27-norm and the mahler measure of the polynomials with random coefficients @xmath10 .
their results can be linked to the recent results of peligrad & wu @xcite , barrera & peligrad , cohen & conze @xcite and thouvenot & weiss @xcite .
therein , the authors investigated a dynamical approach with dynamical coefficients , that is , @xmath28 , where @xmath29 is a measure - preserving transformation on some probability space and @xmath30 is a square - integrable function .
+ we remind that the polynomials with coefficients @xmath31 and the constant term equal to 1 are nowadays called newman polynomials .
we further notice that the polynomials with coefficients @xmath31 are known as idempotent polynomials , and since we are concern with @xmath0-flatness , we may assume that the constant term is 1 .
the connection between the banach problem in ergodic theory and the @xmath0-flatness problem in the class of littlewood polynomials or newman polynomials was established by bourgain @xcite , guenais @xcite , and downarowich & lacroix @xcite .
m. guenais proved that the littlewood problem and the banach problem are equivalent in some class of dynamical system @xcite .
she further constructed a generalized fekete polynomials on some torsion groups , and proved that those polynomials are @xmath0-flat .
as a consequence , m. guenais obtained that there exist a group action with simple lebesgue component .
subsequently , el abdalaoui and lemaczyk proved that the generalized fekete polynomials constructed by guenais are ultraflat @xcite .
very recently , el abdalaoui and nadkarni strengthened guenais s result @xcite by proving that there exist an ergodic non - singular dynamical system with simple lebesgue component . + here
, we exhibit a class of @xmath0-flat polynomials with coefficients @xmath32 and @xmath9 .
this allow us to produce a dynamical system with simple lebesgue spectrum .
we thus get an affirmative answer to the banach question .
+ furthermore , combining our result with that of @xcite , we provide a positive answer to the mahler s problem in the class of newman polynomials @xcite , @xcite,@xcite .
+ our methods breaks down for the polynomials with coefficients @xmath10 .
thus , we are not able to answer the weaker form of littlewood question on the existence of @xmath0-flat polynomials with coefficients @xmath10
. + we notice that the flatness problem is connected to the number theory and to some practical issues arising in the design of a mobile cellular wireless ofdm system @xcite .
consequently , it is related to some engineering issues @xcite , @xcite , @xcite .
+ for the convenience of the reader , we repeat the relevant material from @xcite,@xcite and @xcite , without proofs , thus making our exposition self - contained .
+ the paper is organized as follows . in section 2 ,
we state our main results . in section 3 , we remind the notion of generalized riesz products and its connection to ergodic theory . in section 4 ,
we present several definitions of flatness in the class of analytic trigonometric polynomials and the fundamental characterization of @xmath0-flatness . in section 5 ,
we remind the notion of singer and sidon sets in the number theory , and we establish that the @xmath26-norm strategy can not be apply to the case of the newman polynomials . finally , we prove our main results in section 6 .
consider the torus @xmath33 equipped with the normalized lebesgue measure @xmath34 .
let @xmath35 be a positive sequence of integers and put @xmath36 with @xmath37 and @xmath38 is a probability vector .
such polynomials are raised in the study of the spectral type of some class of dynamical systems in ergodic theory . for more details
we refer to @xcite .
+ here , we restrict ourself to the case @xmath39 and @xmath40 .
we thus concentred our investigations on the flatness problem in the class of polynomials of the from @xmath41 following @xcite , this class is called class b. + we state our main results as follows . [ main1 ]
there exist a sequence of analytic trigonometric polynomials @xmath42 with coefficients @xmath32 and @xmath9 such that the polynomials @xmath43 are flat in almost everywhere sense , that is , @xmath44{}}1,\ ] ] for almost all @xmath8 with respect to the lebesgue measure @xmath34 . as a consequence , we obtain the following theorem . [
main2]there exist a dynamical system @xmath45 with @xmath46 and simple lebesgue spectrum .
we remind that @xmath45 is a dynamical system if @xmath47 is a measure space with @xmath48 is finite or @xmath49-finite measure , and @xmath29 is a measure - preserving transformation , that is , for any measurable set @xmath11 , we have @xmath50 theorem [ main2 ] gives an affirmative answer to the long - standing problem attributed to banach on the existence of dynamical system which simple lebesgue spectrum and with no - atomic measure . let us remind that ulam in his book @xcite stated the banach problem as follows .
does there exist a square integrable function @xmath51 and a measure preserving transformation @xmath52 , @xmath53 , such that the sequence of functions @xmath54 forms a complete orthogonal set in hilbert space ?
the most famous banach problem in ergodic theory asks if there is a measure preserving transformation on a probability space which has simple lebesgue spectrum .
a similar problem is mentioned by rokhlin in @xcite . precisely , rokhlin asked on the existence of an ergodic measure preserving transformation on a finite measure space whose spectrum is lebesgue type with finite multiplicity .
later , kirillov in his 1966 s paper @xcite wrote there are grounds for thinking that such examples do not exist " .
however he has described a measure preserving action ( due to m. novodvorskii ) of the group @xmath55 on the compact dual of discrete rationals whose unitary group has haar spectrum of multiplicity 2 .
similar group actions with higher finite even multiplicities are also given .
+ subsequently , finite measure preserving transformation having lebesgue component of finite even multiplicity have been constructed by j. mathew and m. g. nadkarni @xcite , kamea @xcite , m. queffelec @xcite , and o. ageev @xcite .
fifteen years later , m. guenais produce a torsion group action with lebesgue component of multiplicity one @xcite .
+ our methods is far from making any contribution to this problem . at know ,
it is seems that this problem is a dark continent " for the ergodic theory and for the spectral theory of dynamical systems .
+ nevertheless , it is turn out that our results allows us to answer a bourgain s question @xcite on the supremum of the @xmath0-norm over all polynomials from class b. indeed , theorem [ main1 ] assert the following .
@xmath56 in @xcite , @xcite and @xcite , the authors established already that @xmath57 , and , it easy to see that the simple case @xmath58 , gives @xmath59 + we further have .
there exist a sequence of analytic trigonometric polynomials @xmath60 with coefficients @xmath32 and @xmath9 such the mahler measure of the polynomials @xmath61 converge to 1 .
we remind that the mahler measure of analytic trigonometric polynomials @xmath62 is given by @xmath63 using jensen s formula @xcite , it can be shown that @xmath64 where , @xmath65 denoted the zero of the polynomial @xmath66 . in this definition , an empty product is assumed to be @xmath9 so the mahler measure of the non - zero constant polynomial @xmath67 is @xmath68 . a nice account on the subject
may be founded in @xcite , @xcite .
+ the next proposition list some elementary properties of the mahler measure . for the reader s convenience , we provide its proof .
[ basic]let @xmath69 be a probability space .
then , for any two positive functions @xmath70 , we have a. @xmath71 is a limit of the norms @xmath72 as @xmath73 goes to @xmath32 , that is , @xmath74 { } } m_{\rho}(f),\ ] ] provided that @xmath75 is integrable .
b. if @xmath76 then @xmath77 .
c. if @xmath78 , then @xmath79 .
d. if @xmath80 , then @xmath81 .
e. @xmath82 f. @xmath83 . g. @xmath84 .
we start by proving ( ii ) . without loss of generality , assume that @xmath85 and put @xmath86 let @xmath87 be in @xmath880,1[$ ] , @xmath89
. then @xmath90 .
hence , by hlder inequality , we have @xmath91 thus we have proved @xmath92{}}0.\end{aligned}\ ] ] to prove ( i ) , apply the mean value theorem to the following functions @xmath930,1[;$ } \\ t \longmapsto t^\delta , & \hbox{if $ x>1,$ } \end{array } \right.\ ] ] hence , for any @xmath940,1[$ ] and for any @xmath95 , we have @xmath96 furthermore ,
it is easy to see that @xmath97{}}\log(f),\ ] ] and , by lebesgue dominated convergence theorem , we get that @xmath98 { } } \int \log(f ) d\rho.\ ] ] on the other hand , for any @xmath940,1[$ ] , we have @xmath99 and for a sufficiently small @xmath73 , we can write @xmath100 since @xmath101 as @xmath102 . summarizing we have proved @xmath103 for the proof of ( iii ) and ( iv ) , notice that the function @xmath104 is a convex function and @xmath105 is a concave function .
applying jensen s inequality to @xmath106 we get @xmath107 and this finishes the proof , the rest of the proof is left to the reader . using the classical beurling s outer and inner decomposition , el abdalaoui and nadkarni in @xcite computed the mahler measure of @xmath62 .
they further apply the @xmath3 theory to establish a formula for the mahler measure of the generalized riesz product .
we remind a part of this in the next section .
the classical notion of riesz products is based on the notion of dissociation , which can be defined as follows .
+ consider the polynomial @xmath108 .
then , we have @xmath109 for any integer @xmath110 , we can write @xmath111 in the first case we group terms with the same power of @xmath8 , while in the second case all the powers of @xmath8 in the formal expansion are distinct . in the second case
we say that the polynomials @xmath112 and @xmath113 are dissociated .
more generally , we have if @xmath114 , @xmath115 , are two trigonometric polynomials then for some @xmath116 , @xmath112 and @xmath117 are dissociated
. it is well know that if the sequence of polynomials @xmath118 is dissociated ( each finite product has dissociation property ) with constant term equal to 1 .
then , the sequence of probability measures @xmath119 converge to some probability measure called a riesz product and denoted by @xmath120 + more generally , we have the following definition : [ def1 ] let @xmath121 be a sequence of trigonometric polynomials such that a. for any finite sequence @xmath122 of natural numbers @xmath123 where @xmath124 denotes the circle group and @xmath34 the normalized lebesgue measure on @xmath124 , b. for any infinite sequence @xmath125 of natural numbers the weak limit of the measures @xmath126 as @xmath127 exists .
then the measure @xmath48 given by the weak limit of @xmath128 as @xmath127 is called generalized riesz product of the polynomials @xmath129 and denoted by @xmath130 [ [ connection - to - ergodic - theory - and - rank - one - transformations ] ] * connection to ergodic theory and rank one transformations * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + using the cut and stack procedure described in @xcite , @xcite , one can construct inductively a family of measure - preserving transformations , called rank one transformations or rank one maps , as follows .
0.1 cm let @xmath131 be the unit interval equipped with lebesgue measure .
at stage one we divide @xmath131 into @xmath132 equal parts , add spacers and form a stack of height @xmath133 in the usual fashion . at the @xmath134 stage
we divide the stack obtained at the @xmath135 stage into @xmath136 equal columns , add spacers and obtain a new stack of height @xmath137 .
if during the @xmath134 stage of our construction the number of spacers put above the @xmath138 column of the @xmath135 stack is @xmath139 , @xmath140 , @xmath141 , then we have @xmath142 proceeding in this way , we get a rank one map @xmath29 on a certain measure space @xmath143 which may be finite or @xmath144finite depending on the number of spacers added .
+ the construction of a rank one map thus needs two parameters , @xmath145 ( cutting parameter ) , and @xmath146 ( spacers parameter ) .
put @xmath147 in @xcite and @xcite it is proved that the spectral type of this map is given ( up to possibly some discrete measure ) by @xmath148 where @xmath149 @xmath150 denotes weak star limit in the space of bounded borel measures on @xmath151 .
as mentioned by nadkarni in @xcite , the infinite product @xmath152 taken over a subsequence @xmath153 also represents the maximal spectral type ( up to discrete measure ) of some rank one maps . in case @xmath154 for infinitely many @xmath155 , the maps acts on an infinite measure space . + the spectrum of any rank
one map is simple and using a random procedure , d. s. ornstein produced a family of mixing rank one maps @xcite .
it follows that ornstein s class of maps may possibly contain a candidate for banach s problem .
unfortunately , in 1993 , j. bourgain proved that almost surely ornstein s maps have singular spectrum @xcite .
subsequently , using the same methods , i. klemes @xcite showed that the subclass of staircase maps has singular maximal spectral type . in particular , this subclass contains the mixing staircase maps of adams - smorodinsky @xcite . using a refinement of peyrire
criterium @xcite , i. klemes & k. reinhold proved that the rank one maps have a singular spectrum if the inverse of the cutting parameter is not in @xmath156 ( that is , @xmath157 , where @xmath158 is the cutting parameter ) @xcite .
this class contains the mixing staircase maps of adams & friedman @xcite . in 1996 , h. dooley and s. eigen adapted the brown - moran methods @xcite and proved that the spectrum of a subclass of ornstein maps is almost surely singular @xcite .
+ later , el abdalaoui - parreau and prikhodko extended bourgain theorem @xcite by proving that for any family of probability measures in ornstein type constructions , the corresponding maps have almost surely a singular spectrum @xcite .
they obtained the same result for rudolph s construction @xcite . in 2007 ,
el abdalaoui showed that the spectrum of the rank one map is singular provided that the spacers @xmath159 , are lacunary for all @xmath160 @xcite .
the author used the salem - zygmund central limit theorem methods . as a consequence ,
the author presented a simple proof of bourgain s theorem @xcite .
+ recently , by appealing to the martingale approximation technique , c. aistleitner and m. hofer @xcite proved a counterpart of the result of @xcite . precisely , they proved that the spectrum of the rank one maps is singular provided that the cutting parameter @xmath161 and the spacers @xmath159 satisfy a. @xmath162 converge to @xmath32 ; b. the proportion of equal terms in the spacers is at least @xmath163 for some fixed constant @xmath164 and some subsequence @xmath165 .
+ we further recall that i. klemes & k. reinhold in @xcite conjectured that all rank one maps have singular spectrum , and in the same spirit , c. aistleitner and m. hofer wrote in the end of their paper `` several authors believe that all rank one transformations could have singular maximal spectral type . '' .
it seems that this conjecture was formulated since baxter result @xcite , @xcite .
we remind that the cutting and stacking rank one construction may goes back to ornstein s paper in 1960 @xcite . indeed ,
therein , ornstein constructed a non - singular map for which there is non @xmath49-finite measure equivalent to lebesgue measure .
of course , this example is connected to the example of non - singular map with simple lebesgue component obtain by el abdalaoui and nadkarni @xcite .
notice that in @xcite , the rank one maps are called transformations of class one . +
it follows from bourgain s observation ( ( eq 2.15)@xcite ) that if the spectral type of any rank one map acting on infinite measure is singular then the spectral type of any rank one is singular . unfortunately , by our main result , this strategy fails .
therefore , the new approaches are needed to tackle this conjecture .
+ we remind that in @xcite , it is proved that if @xmath166 , then the absolutely continuous part @xmath167 verify @xmath168 { } } 0.\ ] ] furthermore , the mahler measure of @xmath48 on the circle is given by @xmath169 where @xmath16 ranges over all analytic trigonometric polynomials with zero constant term .
] satisfy @xmath170 we further remind from @xcite the following notion of generalized riesz products from dynamical origin . [ def1 ] a generalized riesz product @xmath171 , where @xmath172 , @xmath173 , is said to be of dynamical origin if with @xmath174 it is true that for @xmath175 , @xmath176 if , in addition , the coefficients @xmath177 are all positive , then we say that @xmath48 is of purely dynamical origin .
+ the following is proved in @xcite .
[ lem1 ] given a sequence @xmath178 of analytic trigonometric polynomials in @xmath179 with non - zero constant terms and @xmath179 norm 1 , @xmath180 .
then there exist a sequence of positive integers @xmath181 such that @xmath182 is a generalized riesz product of dynamical origin . applying carefully the previous lemma ,
the following theorem is proved in @xcite .
[ th7 ] let @xmath183 be a sequence of non - constant polynomials of @xmath179 norm 1 such that @xmath184 a.e .
@xmath185 then there exists a subsequence @xmath186 and natural numbers @xmath187 such that the polynomials @xmath188 are dissociated and the infinite product @xmath189 has finite nonzero value a.e @xmath185 .
a sequence @xmath183 of trigonometric polynomials is said to be @xmath27-flat if the sequence @xmath190 @xmath175 converge to the constant function 1 in the @xmath27-norm , @xmath191,~~p\neq 2 $ ] .
if @xmath192 the sequence @xmath193 is said to be ultraflat .
+ the flatness issue can be considered for three class of analytic trigonometric polynomials .
the polynomials with non - negative coefficients , the littlewood polynomials which correspond to the polynomials with coefficients @xmath10 , and the newman polynomials which correspond to the polynomials with coefficients @xmath32 or @xmath9 and the constant term 1 .
for all those polynomials , it seems that the existence of @xmath27-flat polynomials is unknown .
+ the following notion of almost everywhere flatness is introduced in @xcite .
a sequence @xmath183 of trigonometric polynomials with @xmath1-norm one is said to be flat almost everywhere , if @xmath194 converge almost everywhere to 1 with respect to @xmath34 . applying vitali convergence theorem @xcite
one can see that if @xmath193 is almost everywhere flat then @xmath193 is @xmath0-flat . in the opposite direction ,
if @xmath193 is @xmath0-flat then one can drop a subsequence over which @xmath193 is almost everywhere flat .
+ we further have the following . [
mainp]let @xmath195 be a sequence of analytic trigonometric polynomials with @xmath1-norm one .
then , the following are equivalent 1 .
@xmath196 is @xmath0-flat , 2 .
@xmath197 converge to 1 .
moreover , if @xmath196 is almost everywhere flat then @xmath198{}}0.\ ] ] \(1 ) implies ( 2 ) is straightforward . for ( 2 )
implies ( 1 ) , notice that @xmath199 for the last fact , by cauchy - schwarz inequality , we have @xmath200 the last inequality is due to @xmath201 combined with the triangle inequality .
thus , it is suffice to see that @xmath202 { } } 0.\ ] ] but , by , this equivalent to @xmath203 converge to 1 which follows from the vitali convergence theorem .
we remind that the classical strategy introduced by newman and beller to produce the @xmath0-flat polynomials is based on the reduction of the problem of @xmath0-flatness to @xmath26-flatness problem . for the polynomials form class @xmath204
this strategy fails .
this is proved in @xcite . for sake of completeness
, we give the proof in the next section .
let @xmath205 be a positive integer and let @xmath206 be a subset of @xmath207 . put @xmath208^{+}=\big\{s_j - s_i,~~i < j\big\}=\big\{r_1,r_2,\cdots , r_{n(n)}\big\}.\ ] ] evidently , @xmath209^{+}$ ] is a subset of @xmath207 since @xmath210 it can be useful to see @xmath209^{+}$ ] as a upper part of the following matrix @xmath211 we will denote by @xmath212 the multiplicity of @xmath213 which correspond to the number of the pair @xmath214 such that @xmath215 , and we set @xmath216 following chidambaraswamy and kurtz - shah @xcite , @xcite , the sequence @xmath217 is @xmath73-admissible if @xmath218 if @xmath219 , then @xmath220 is called sidon set .
we remind that the classical definition of sidon sets goes back to sidon how introduce the notion in 1932 or 1933 according to erds @xcite .
this definition can be stated as follows .
a set @xmath220 is called a sidon set if all the sums @xmath221 , @xmath222 are distinct .
more generally , one can defined @xmath223 $ ] sets , where @xmath224 and @xmath225 are a positive integers .
a subset @xmath11 of @xmath226 $ ] is said to be @xmath223 $ ] set if the linear equation @xmath227 , has at most @xmath225 solutions up to permutations .
$ ] correspond to the sidon sets .
it is easy to see that a subset is sidon set if and only if all the difference are distinct .
this last property is not shared with the @xmath223 $ ] sets , @xmath229 @xcite .
+ sidon asked on the maximal size of the sidon set subset of @xmath230 .
erds and turn @xcite proved that if @xmath231 $ ] is a sidon set then @xmath232{h}+1.\ ] ] lindstrm strengthened this result and proved @xcite @xmath233{h}+1.\ ] ] in the other direction it has been shown by chowla @xcite and erds using a theorem of singer @xcite that @xmath234 nowadays , it is customary to use algorithmically a singer s theorem to produce a sidon subset of the given set @xmath235 . furthermore , the construction can be used to produce a sidon subset with some desired proprieties of its sumsets @xcite , @xcite .
we notice that singer established his theorem in the finite projective geometry setting . in the number theoretic
setting , the theorem can be stated as follows .
let @xmath236 be a prime and let @xmath237 .
then , there exist @xmath238 with @xmath239 such that for all @xmath240 there exist @xmath241 such that @xmath242 .
such set , in which every non - zero difference @xmath243 arises exactly one is called a perfect difference set or singer set . for the construction of singer set , we refer the reader to @xcite . using the singer set erds - srkzy - ss @xcite , @xcite and rusza @xcite , @xcite constructed a sidon set @xmath220 subset of @xmath235 such that @xmath244 with some desired properties .
+ we notice that singer s construction is based on the nice properties of finite fields @xcite .
+ let us further mention that the lower bound and the upper bound of the quantities @xmath245 and @xmath246 can be obtained by considering the following toy examples .
+ for the first example we take @xmath247 .
this gives @xmath248 and @xmath249 therefore @xmath250 and @xmath251 .
we thus have @xmath245 is maximal and @xmath246 is minimal . indeed , for any @xmath252
, the number of solution of the equation @xmath253 is less than @xmath254 since any solution @xmath255 , when it exists , satisfy @xmath256 . + the second example correspond to the case for which the support of the fourier transform is a sidon subset @xmath220 of @xmath257 $ ] . in this case @xmath258 and @xmath259 . indeed , by definition of the sidon set
all @xmath260 are distinct .
hence , the first row of the matrix @xmath261 contain @xmath262 elements , the second row @xmath263 , and the last row one element . by adding , we get @xmath264 whence @xmath245 is minimal and @xmath246 is maximal .
it is seems that the quantity @xmath265 is balanced .
+ [ [ on - l4-norm - strategy - and - newman - polynomials . ] ] * on @xmath26-norm strategy and newman polynomials . *
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + it is hidden in the proof given by chidambaraswamy @xcite that the @xmath1-norm of the polynomials @xmath266 does not converge to @xmath32 .
indeed , @xmath267 where @xmath212 is the multiplicity of @xmath213 given by @xmath268 and @xmath213 is defined by @xmath269 whence @xmath270 since @xmath271 therefore @xmath272 which complete the proof of the claim . from this
it easy to see that @xmath273 , for any @xmath14 .
hence , @xmath274 never converge to 1 .
thus the @xmath26-norm strategy of beller - newman can not be used .
for any finite subset of integer @xmath11 , we put @xmath275 where @xmath276 is the number of elements in @xmath11 .
the @xmath1-norm of @xmath277 is one since @xmath278 where @xmath279 is the set of difference of @xmath11 .
+ if @xmath280 then @xmath11 is a sidon set .
the nice properties of singer s set allows us to prove the following .
the second main ingredient of our proof is based on the classical marcinkiewicz - zygmund inequalities ( see ( * ? ? ?
* theorem 7.5 , chapter x , p.28 ) ) and some ideas linked to its recent refinement obtained by chui - shen - zhong @xcite and many others authors .
therefore , by appealing to some classical results form the @xmath3 theory and interpolation theory of carleson , we will gives an alternative proof to the proof given in @xcite .
+ as is customary , @xmath288 $ ] is the integer part of @xmath289 , @xmath290 is the dirichlet kernel , @xmath291 is the fejr kernel and @xmath292 is the de la vall de poussin kernel .
we remind that @xmath293 @xmath294 and @xmath295 we remind that the poisson kernel @xmath296 , @xmath297 , is given by @xmath298 this kernel is related to the cauchy kernel @xmath299 by the following relation @xmath300 the imaginary part of @xmath301 is called the conjugate poisson kernel and denoted by @xmath302 let us also remind that if @xmath303 is analytic in the closed disc with @xmath304 is real then @xmath305 and @xmath306 we notice that @xmath307 is the harmonic conjugate to @xmath308 , which vanishes at the origin , and of course , @xmath309 is the the harmonic conjugate to @xmath296 . for @xmath310 , the harmonic conjugate of @xmath30
is given by @xmath311 it is well known that the radial limit of @xmath312 exist almost everywhere , and this radial limit denoted by @xmath313 is the conjugate function of @xmath30 .
+ we will use often the following classical property : if @xmath314 , where @xmath315 is an analytic function .
then @xmath316 given a continuous function @xmath30 on the torus @xmath317 and a triangular family of equidistant points @xmath318 , @xmath319 , that is , @xmath320 we define the lagrange polynomial interpolation of @xmath30 at @xmath321 by @xmath322 where @xmath323 is a function defined by @xmath324 @xmath323 is a step function with jump @xmath325 at the points @xmath326 and @xmath327 its riemann - stieltjes integral . in the same manner , we define the step function @xmath328 , for any @xmath329 and we denote its riemann - stieltjes integral by @xmath330 . + we will need the following classical inequality due to s. bernstein and a. zygmund . for its proof , we refer to ( * ? ? ?
* theorem 3.13 , chapter x , p. 11 ) .
mt , nevai and arestov extended bernstein - zygmund inequality by proving that it is valid for @xmath335 .
@xcite . for @xmath336 ,
the result is due to mahler , a simple proof can be found in @xcite .
although we will not need this result in such generality .
+ the marcinkiewicz - zygmund interpolation inequalities assert that for @xmath337 , @xmath338 , and polynomial @xmath16 of degree @xmath339 , @xmath340 where @xmath341 and @xmath342 are independent of @xmath14 and @xmath16 .
+ the left hand inequality in is valid for any non - negative non - decreasing convex function and in the more general form ( * ? ? ?
* remark , chapter x , p. 30 ) . for sake of completeness
, we will state and present a sketch of the proof of it .
[ key]let @xmath343 , @xmath344 .
then , for any non - negative , non - decreasing and convex function @xmath345 , for any trigonometric polynomial @xmath346 of degree @xmath14 , we have @xmath347 where @xmath348 the proof is the same as in @xcite .
one only needs to substitute the de vall de poussin kernel @xmath349 by @xmath350 where @xmath351 + 1 $ ] and @xmath352 by @xmath353 .
@xmath354 is the de vall de poussin kernel of order @xmath224 . applying bernstein theorem ( theorem [ bernstein ] ) ,
we get @xmath362{}}0.\end{aligned}\ ] ] this combined with the standard triangle inequalities gives @xmath363{}}0,\end{aligned}\ ] ] and the proof of the lemma is complete . in the previous version of this work @xcite , theorem [ main1 ]
was proved by showing that the @xmath0-flatness holds by appealing to lemma [ lem3 ] to construct a disturbed triangle family of nodal points @xmath364 of @xmath365 that satisfy and with @xmath366 , @xmath367 , where @xmath343 is given . as a consequence , by theorem [ key ] , we obtain that @xmath368 and letting @xmath369 , we get @xmath370 we notice that there is many way to obtain such sequence of disturbed points . here , we follows the spirit of the kadets @xmath371 theorem for polynomials due to marzo - seip @xcite .
this new strategy is based on the carleson interpolation theory and the very recent refinement of the marcinkiewicz - zygmund inequalities .
+ we start by proving theorem [ main1 ] . + let @xmath220 be a fixed singer set in @xmath281 with @xmath237 , @xmath236 prime number and put @xmath372 define @xmath373 and for a given @xmath374 @xmath375 , we put @xmath376 let @xmath377 .
we define @xmath378 with @xmath379 and @xmath380 .
we thus have @xmath381 and we set @xmath382 we remind that the family of nodal points @xmath383 is said to be an @xmath384 marcinkiewicz - zygmund family if the @xmath384 marcinkiewicz - zygmund inequalities holds for the nodal points @xmath385 , for every @xmath386 .
+ we associate to any family of nodal points @xmath383 the function @xmath387 defined by @xmath388 the family @xmath389 is said to be uniformly separated if there is a positive number @xmath164 such that @xmath390 this notion is related to the notion of carleson measures and following @xcite the sequence @xmath391 of points in the open unit disk @xmath392 is said to satisfy carleson s condition if , @xmath393 of course this condition is connected to the well - known carleson s interpolation theorem @xcite . for the proof of carleson s interpolation theorem ,
we refer the reader to @xcite , @xcite , @xcite .
+ we remind that a finite measure @xmath48 is a carleson measure if the injection mapping from @xmath394 , @xmath395 to the space @xmath396 is bounded .
these measures were described geometrically in carlson s theorem @xcite , @xcite , which assert that the finite measure @xmath48 is a carleson measure if and only if there exist a constant @xmath397 such that @xmath398 for any @xmath395 .
furthermore , by carleson s interpolation theorem , we have that the discrete measure @xmath48 given by @xmath399 where @xmath400 is the dirac measure on @xmath401 , is a carlson measure if the family @xmath402 is uniformly separated .
this result was strengthened in [ 15 ] by mcdonald and sundberg @xcite , who proved that the sequence @xmath402 of points in @xmath392 generates a discrete carleson measure @xmath48 if and only if @xmath402 is a finite union of uniformly separated sequences . for a simple proof ,
we refer to @xcite .
we notice that if the sequence is uniformly separated then the constant @xmath403 depend uniquely on @xmath404 . in this setting
, we have the following lemma notice that @xmath412-\pi,\pi[$ ] , and the function @xmath413 is an even function .
we further have , for any @xmath414 , @xmath415 therefore , we can reduce our study to the case of @xmath4120,\pi/2]$ ] , and if @xmath416 we substitute @xmath417 by @xmath418 .
+ now , assuming @xmath4120,\pi/2]$ ] , it follows that @xmath419 since , for any @xmath4200,\pi/2]$ ] , we have @xmath421 whence @xmath422 by the monotonicity of the function @xmath423 .
+ we further have @xmath424 since , for any @xmath338 , @xmath425 .
this combined with and gives @xmath426 we thus get @xmath427 by the convergence of @xmath428 it follows from lemma [ carlsonsomme ] that the union of the families @xmath389 and @xmath429 generates a carleson measure since the sum of two carleson measures is a carlson measure .
we further deduce the following + put @xmath433 and let @xmath434 @xmath435 .
then , either @xmath436 or @xmath437 .
assuming that @xmath436 , it follows that @xmath438 by the same arguments as in lemma [ carlsonsomme ] ( see also @xcite ) combined with .
the same conclusion can be drawn for the case @xmath437 since the two sets plays symmetric roles .
the proof of the lemma is complete . according to chui - zhong s theorem @xcite
the family @xmath439@xmath440 of the points on the unit circle is an @xmath384 marcinkiewicz - zygmund family if and only if it is uniformly separated and there exist a constant @xmath441 such that @xmath442 for every subarc @xmath443 of the unit circle and every @xmath444 . +
we notice that the fact that the family is uniformly separated insure that this family generates a carlson measure , and it is turn out that the second condition is well - known as @xmath445 condition in the setting of the bmo spaces ( bounded mean oscillation ) @xcite .
we remind that locally integrable positive function @xmath401
satisfy @xmath445 condition if @xmath446 for the proof of the equivalence of when @xmath447 and @xmath452 , we refer to @xcite .
therein , the reader can found also the proof of the prediction helson - szeg s theorem related to ( hs ) @xcite .
+ now , according to the equivalence of @xmath453 when @xmath447 and @xmath452 , marzo and seip @xcite observe that in order to prove that the condition holds it suffices to establish that the following uniform helson - szeg condition holds : we are going to prove that the uniform helson - szeg condition holds .
let @xmath343 and @xmath459 .
we claim first that we have @xmath460 where @xmath461 indeed , the mahler measure of the fonction @xmath462 verify @xmath463 by proposition [ basic ] combined with the well - know jensen formula .
we further have @xmath464 therefore @xmath465 is in @xmath466 and @xmath467 is integrable .
put @xmath468 then @xmath469 is an analytic function in the unit disc @xmath392 and @xmath470 , where @xmath471 is the poisson integral of @xmath472 , that is , @xmath473 where @xmath296 is the poisson kernel and @xmath474 is the convolution operator . by fatou theorem @xcite ,
@xmath475 almost everywhere on the unit circle @xmath317 .
we further have @xmath476 where @xmath477 is _ the principal value of the logarithm_. writing @xmath478 where @xmath479 it follows that @xmath480 since @xmath481 we can thus write @xmath482 since @xmath483 for any @xmath95 , and @xmath484 for @xmath485 .
we thus conclude that @xmath486 and the proof of the claim is complete .
+ we move now to construct the functions @xmath455 . for that , we start by proving the following lemma let @xmath487 , with @xmath297 and @xmath488
. then @xmath489 , where @xmath450 is the conjugate function of the function @xmath451 given by @xmath490}(\theta_0)-\theta - c,\end{aligned}\ ] ] and @xmath164 is any suitable constant .
obviously @xmath491 is an outer function since the zeros of @xmath491 are out of the disc @xmath392 .
we further have @xmath492 .
whence @xmath493 , where @xmath450 is the conjugate function of the function @xmath451 given by . indeed , for any @xmath494
, we have @xmath495 where @xmath477 is _ the principal value of the logarithm_. we further have @xmath496}(\theta_0)-\theta - c=\int_{0}^{\theta}p_r(\theta_0-t ) dt-\theta - c\\ & = & \int_{0}^{\theta}p_r(t-\theta_0 ) dt -\theta - c,\end{aligned}\ ] ] since @xmath296 is an even function .
but @xmath497 by .
consequently @xmath498 and @xmath499 combining with , we obtain @xmath500 which gives @xmath501 since for any analytic function @xmath225 , we have @xmath502 and the proof of the lemma is complete .
we now apply lemma [ flem ] to write @xmath503 where @xmath504 @xmath505 and @xmath164 is any suitable constant . taking @xmath506 with @xmath507 we can rewrite as @xmath508 since , for any odd @xmath509 , we have @xmath510 again writing @xmath511 where @xmath512 we thus need to estimate @xmath513 and @xmath514 .
but , by the same reasoning as above , it is easy to check that @xmath515 consequently , we get @xmath516 whence @xmath517 it is still to estimate @xmath514 . in the same manner it can be seen that @xmath518 and by choosing @xmath519 sufficiently large and @xmath520 , we obtain @xmath521 from this we conclude that the uniform helson - szeg condition holds for @xmath522 . + for the case @xmath523 .
assuming @xmath524 where @xmath525 , one may apply the standard argument from the @xmath3 theory combined with the hlder inequality and lemma 2 from @xcite to conclude that the uniform helson - szeg condition holds .
+ now , let @xmath526 and @xmath527 , with @xmath528 .
by lemma [ sulem ] the family @xmath430 is uniformly separated . we can thus write @xmath529 where @xmath530 the computation of constant @xmath531 can be found in @xcite .
therein , by appealing to the duality argument , it is shown that for any @xmath532 , there exist @xmath533 such that for some @xmath534 , with @xmath535 , and @xmath536 is the conjugate of @xmath65 , one can assert @xmath537 where @xmath538 but the measures @xmath539 and @xmath540 are a carleson measures .
therefore , @xmath541 where @xmath542 an alternative proof can be found in @xcite .
the reader may notice that the proof of theorem f in @xcite can be drawn from the above proof .
we further have @xmath543 the second inequality is due to the simple fact that @xmath544 and @xmath545 , and the third inequality follows by the same arguments as in the proof of lemma 12 and theorem 9 from @xcite .
applying bernstein - zygmund inequality ( theorem [ bernstein ] ) combined with the classical marcinkiewicz - zygmund inequalities , we obtain @xmath546 the reader should notice that the constant @xmath547 depend only on @xmath65 .
if follows that @xmath548 combined these inequalities with lemma [ lem2 ] , we can rewrite as @xmath549 where @xmath550 letting @xmath551 , we obtain @xmath552 and by letting @xmath553 , we conclude that @xmath554 since @xmath555 hence the sequence of polynomials @xmath556 is @xmath0-flat .
+ therefore , by appealing to proposition [ mainp ] , we deduce that the sequence of polynomials @xmath557 is almost everywhere flat over some subsequence .
thus the proof of theorem [ main1 ] is complete .
+ theorem [ main2 ] follows from theorem [ main1 ] combined with proposition [ mainp ] and lemma [ flatf ] .
finally , by , we deduce that the spectral type @xmath49 of the rank one map constructed in theorem [ main2 ] verify @xmath558 whence , @xmath559{}}1.\ ] ] this finishes the proof . for more details on the construction of rank one map in theorem [ main2 ] , we refer the reader to @xcite and @xcite . obviously , as in the proof given by zygmund in (
* , chap x ) , in our proof of theorem [ key ] , we take advantage of the following classical identity ( * ? ? ?
* , chap ii ) @xmath560 for any @xmath561 .
+ the reader may notice that there is some analogies between our proof and the fast fourier transform algorithm ( fft ) .
we refer the reader to @xcite for more details on the fft . applying carleson interpolation theory
, one can prove that for any @xmath562 , there is a constant @xmath563 such that , for any polynomial @xmath16 of degree less than @xmath14 , @xmath564 an alternative proof can be found in @xcite . besides this , marcinkiewicz and zygmund proved @xcite that for any @xmath331 and for any polynomial @xmath16 of degree less or equal than @xmath14 , we have @xmath565 to the best of this author s knowledge , the explicit constant for the case @xmath336 seems not to be known . nevertheless , in the case of the classical riesz product , if we consider the polynomial @xmath566 where @xmath65 is non - negative number less than @xmath9 .
then @xmath567 this can be proved as follows .
+ following @xcite , we put @xmath568 where @xmath569 it is easily seen that @xmath570 does not vanish on the disc @xmath571 .
we thus get that the function @xmath572 is harmonic .
applying the mean property , we obtain @xmath573 rewriting , we see that @xmath574 now , any easy computation shows that @xmath575 for @xmath576 + whence @xmath577 obviously @xmath578 we conclude that holds .
+ this leads us to ask .
+ 1 .
can one prove or disapprove that @xmath579 converge to @xmath9 as @xmath580 .
2 . let @xmath581 be a family of singer sets , @xmath236 is a prime number and consider the sequence of polynomials @xmath582 can one prove or disapprove that the sequence of the mahler measure of @xmath583 converge to 1 .
let @xmath584 be an ergodic dynamical system where @xmath48 is a finite measure .
can one prove or disapprove that there exist a borel set @xmath11 with @xmath585 such that for @xmath48-almost all @xmath586 , @xmath587{}}1.\ ] ] 4 .
in the same setting as in the previous question , can one prove or disapprove that for any measurable @xmath30 with values @xmath10 , for @xmath48-almost all @xmath586 , we have + @xmath588 as mentioned in introduction , this problem can be linked to the annealed and quenched business .
the author wishes to express his thanks to franois parreau who introduced him to the riesz products business and to jean - paul thouvenot who introduced him to the world of rank one maps .
he would like further to express his thanks to mahendra nadkarni for many stimulating e - conversations and to jean - franois mla for his stimulating questions on the rank one maps .
the author is deeply indebted to jean - paul thouvenot and mahendra nadkarni for their suggestions and remarks on the previous version .
e. h. el abdalaoui and m. nadkarni _ a non - singular transformation whose spectrum has lebesgue component of multiplicity one , _ ergodic theory and dynamical systems , available on cjo2014 .
doi:10.1017/etds.2014.85 . http://fr.arxiv.org/pdf/1402.1312 .
byrnes and d. j. newman , the @xmath26 norm of a polynomial with coefficients @xmath10 .
monthly 97 ( 1990 ) , no .
1 , 4245 . l. carleson , interpolations by bounded analytic functions and the corona problem , ann . of math . , 76 ( 1962 ) , 547 - 559 .
l. carleson , an interpolation problem for bounded analytic functions , amer .
, 80 ( 1958 ) , 921 - 930 . j. chidambaraswamy , _ on the mean modulus of trigonometric polynomials and a conjecture of s. chowla _ , proc .
volume 36 , number i , november 197 .
a. chmelnitzki , a. h. dooley & s. j. eigen , _ a family of generalized riesz products , _ canad .
, * 48 * ( 1996 ) , no . 2 , 302 - 315 . t. downarowich , y. lacroix , _ merit factors and morse sequences _ theoretical computer science , * * ( * * 209 ) ( 1998 ) , 377 - 387 .
p. erds , an inequality for the maximum of trigonometric polynomials , annales polonica math .
12 ( 1962 ) , 151154 .
p. erds , a. srkzy and v. t. ss , on sum sets of sidon sets i , j. number theory 47 ( 1994 ) , 329347 .
p. erds , a. srkzy and v. t. ss , on sum sets of sidon sets ii , israel j. math . 90 ( 1995 ) , 221234 . c. c. graham ,
o. c. mcgehee , _ essays in commutative harmonic analysis , _ grundlehren der mathematischen wissenschaften [ fundamental principles of mathematical science ] , 238 .
springer - verlag , new york - berlin , 1979 .
h. helson & g. szeg , a problem in prediction theory , ann . mat .
pura appl .
( 4 ) 51 1960 107 - 138 .
k. hoffman , banach spaces of analytic functions , dover publications inc .
, new york , 1988 , reprint of the 1962 original .
kahane , _ sur les polynmes coefficients unimodulaires , _ bull .
london math .
soc . 12 , 1980 , 321 - 342 .
p.koosis , introduction to @xmath3 spaces , second edition , with two appendices by v. p. havin [ viktor petrovich khavin ] , cambridge tracts in mathematics , 115 .
cambridge university press , cambridge , 1998 .
d. j. newman , norms of polynomials , amer .
monthly , vol .
67 , no . 8 ( oct . , 1960 ) ,
778 - 779 d. j. newman , _ an @xmath594 extremal problem for polynomials , _ proc .
16 1965 1287 - 1290 .
d. ornstein , on invariant measures , bull .
66 ( 1960 ) , 297300 .
d. s. ornstein , _ on the root problem in ergodic theory , _ proc .
sixth berkeley symposium in math .
statistics and probability , university of california press , 1971 , 347 - 356 .
j. singer , a theorem in finite projective geometry and some applications to number theory , trans .
soc . , 43:377385 , 1938 .
e. m. stein & r. shakarchi , fourier analysis , an introduction , princeton university press , princeton and oxford , 2007 .
thouvenot , _ private communication .
thouvenot , b. weiss , limit laws for ergodic processes .
12 ( 2012 ) , no . 1 , 1150012 , 9 pp . s. m. ulam , _ problems in modern mathematics , _ science editions john wiley & sons , inc .
, new york 1964 .
a. zygmund , _ trigonometric series vol .
, second ed .
, cambridge univ . press ,
cambridge , 1959 . | we construct a sequence of polynomials that are flat in the almost everywhere sense .
the construction is done by appealing to the nice combinatorial properties of the singer s sets and sidon sets . as a consequence ,
we get a positive answer to littlewood s flatness problem in the class of the newman polynomials .
we further obtain that there exist a rank one map acting on space of infinite measure with simple lebesgue spectrum .
this answer an old question attributed to banach .
it is turn out that our construction answer also positively the mahler s problem in the class of newman polynomials .
moreover , we get an answer to the bourgain s question on the @xmath0-norm of @xmath1 normalized idempotent polynomials . @xmath2{mafigure.eps}\ ] ] |
let @xmath2 be an associative unital algebra over a field @xmath3 , an element @xmath4 is said to satisfy the quantum pentagon equation if @xmath5 where @xmath6 acts as @xmath7 on @xmath8-th and @xmath9-th factors in the tensor product and leaves unchanged elements in the remaining factor .
equation looks like a degenerate version of the quantum yang baxter equation , well known in the theory of exactly solvable models of statistical mechanics and quantum field theory @xcite , and quantum groups @xcite .
there are also some similarities in constructing solutions of both equations , for example the role of the drinfeld double construction @xcite of solutions of the quantum yang
baxter equation is replaced by the heisenberg double @xcite . however , in modern theory of quantum groups @xcite ( see also @xcite for a review , and @xcite for discussion of the finite dimensional case ) the quantum pentagon equation seems to play more profound role .
remarkably , given a solution of satisfying some additional non - degeneracy conditions , it allows to construct all the remaining structure maps of a quantum group and of its pontrjagin dual simultaneously .
let @xmath10 be a set , @xmath11 be a map from its square into itself .
we call @xmath12 _ pentagon map _ if it satisfies the functional ( or set - theoretical ) pentagon equation @xcite @xmath13 regarded as an equality of composite maps ; here again @xmath14 acts as @xmath12 in @xmath8-th and @xmath9-th factors of the cartesian product
. one can consider a parameter ( it may be functional ) dependent version , then the parameters of the five maps in may be constrained by relations involving also the dynamical variables . the corresponding functional yang
baxter equation @xcite has been studied recently @xcite in connection to integrability ( understood as the multidimensional consistency @xcite ) of two dimensional lattice equations .
a generalization of the quantum yang - baxter equation for three dimensional models is the tetrahedron equation proposed by zamolodchikov @xcite .
it is related @xcite to four dimensional consistency of the discrete darboux equations @xcite , or equivalently , to four dimensional compatibility of the geometric construction scheme of the quadrilateral lattice @xcite .
it turns out @xcite that all presently known solutions of the quantum tetrahedron equation can be obtained , by a canonical quantization , from classical solutions of the functional tetrahedron equation derived from the quadrilateral lattices .
recently it has been observed @xcite that the quadrilateral lattice theory can be considered as a part of the theory of the desargues maps , which describe in geometric terms hirota s discrete kadomtsev
petviashvili ( kp ) equation @xcite and its integrability properties .
it is known that the hirota equation encodes @xcite the kp hierarchy of integrable equations @xcite .
it plays also an important role in many branches of mathematics and theoretical physics related to integrability ; see @xcite for a recent review of some of its application .
there is a natural question what should replace zamolodchikov s tetrahedron equation in the transition from quadrilateral lattices do desargues maps .
we demonstrate that the answer is provided by the pentagon equation . to make presentation more transparent we start in section [ sec : geom ] from the geometric essence of the construction based on elementary considerations on the veblen and desargues configurations @xcite and their symmetry groups .
given five points of the veblen configuration we are ( almost ) uniquely given the last one , what we call the _ veblen flip_. the presence of five veblen configurations within the desargues configuration gives the pentagon property of the veblen flip . by parametrizing the veblen flip using homogeneous coordinates from a division ring @xmath1
we obtain in section [ sec : alg ] a birational map of @xmath15 into itself , which contains a functional gauge parameter .
we check that the veblen map satisfies the functional pentagon equation , provided the functional parameters are restricted by some additional relations . in section [ sec : norm ]
we discuss another solution , without parameters , of the pentagon map together with its geometric interpretation .
section [ sec : hir ] is devoted to presentation of the equivalence to the above solutions of the pentagon equation with the desargues maps and the non - commutative hirota ( discrete kp ) equation . the quantization procedure , which we present in section [ sec : quant ] ,
can be understood in two ways .
first , under appropriate choice of the gauge in the commutative case the veblen map leaves invariant a natural poisson structure .
such a map preserves also the ultra - local weyl commutation relations which quantize that structure . from other point of view
the procedure can be considered as integrable reduction of the generic division ring solution of the functional pentagon equation to a particular division algebra .
this allows to rewrite the map as inner automorphism , and to obtain the corresponding solution of the dynamical quantum pentagon equation , which can be expressed using the quantum dilogarithm function .
finally , in addition to the concluding remarks we present several open problems and research directions .
below we present elementary considerations , which form however a geometric core of the paper with far reaching consequences . and its tetrahedron representation .
faces of the tetrahedron represent lines of the configuration , and the arrows denote position of the @xmath16 and @xmath17 coefficients in the normalization of the corresponding linear relations discussed in section [ sec : alg].,width=529 ] consider the veblen configuration @xmath18 of six points and four lines , each point / line is incident with exactly two / three lines / points ; see fig .
[ fig : veblen - flip - tetr ] . to elucidate the @xmath19 symmetry group of the configuration is convenient to label its points by two - element subsets of the four - element set @xmath20 , and lines by three - element subsets ; the incidence relation is defined by containment .
this combinatorial description of the veblen configuration has a geometric origin @xcite , which associates ( in a non - unique way ) with the configuration four points @xmath21 in general position in @xmath22 .
six lines of edges of the simplex and its four planes intersected by a generic plane form six points and four lines the veblen configuration ( on the intersection plane ) .
the desargues configuration is of the type @xmath23 , i.e. it consists of ten points and ten lines , each line / point is incident with exactly three points / lines .
it is known that combinatorially there are ten such distinct configurations @xcite .
the desargues configuration is selected by the property that it contains five veblen configurations .
again , it is possible to label points of the configuration by two - element subsets of the five - element set @xmath24 , and lines by three - element subsets ; the incidence relation is defined by containment .
similarly , like for the veblen configuration , given five points in general position in @xmath25 , consider lines joining pairs of points , and planes defined by the triples .
a section of such a system of ten lines and ten planes by a generic three - dimensional hyperplane gives a desargues configuration , see fig .
[ fig : desargues - simplex ] .
five four - element subsets give rise to five veblen configurations .
given two ordered pairs @xmath26
, @xmath27 of distinct and non - collinear points of a projective space such that the lines @xmath28 and @xmath29 are coplanar , and thus intersect in the point @xmath30 .
we denote such five points by @xmath31 and say that they satisfy the veblen configuration condition .
we can complete these five points ( and two lines ) , by the intersection point @xmath32 of the two lines @xmath33 and @xmath34 ( and the lines ) , to the veblen configuration ; notice that the ordering in pairs is important , because @xmath35 .
if we remove the old intersection point @xmath30 , which does not belong to two new intersecting lines we obtain new system @xmath36 satisfying the veblen configuration condition .
such an involutory transition we call _ a veblen flip_. given five points satisfying the veblen configuration condition we can label them , in the way described above in section [ sec : gc - vd ] , by five edges of the three - simplex or , equivalently ( up to an action of @xmath19 ) , by five two - element subsets of a four element set .
the edge representing the intersection point @xmath30 contains two vertices of valence three , the sixth edge will represent the point @xmath32 ; see fig .
[ fig : veblen - flip - tetr ] which illustrates the veblen flip @xmath37 . to understand the geometric origin of the pentagon relation property of the veblen flip
let us start from two sets of points satisfying the veblen configuration property with three points and one line in common , see fig .
[ fig : desargues - simplex ] . without loss of generality (
the symmetry group of desargues configuration is the permutation group @xmath38 ) we consider points @xmath39 , @xmath40 , @xmath41 of the line @xmath42 , two points @xmath43 and @xmath44 on the line @xmath45 , and two points @xmath46 and @xmath47 on the line @xmath48 . by a sequence of veblen
flips we can recover all the other points of the desargues configuration . the desargues configuration is considered usually in relation to the celebrated desargues theorem valid in projective spaces over division rings @xcite .
it states that two triangles ( e.g. @xmath49 and @xmath50 on fig .
[ fig : desargues - simplex ] ) are in perspective from a point ( @xmath40 in our case ) if and only if they are in perspective from a line ( that passing through the remaining three points @xmath44 , @xmath51 , @xmath52 of the configuration , which are constructed as intersections of the corresponding sides of the triangles ) .
if we want to apply the veblen flip , starting from the initial configuration described above , it can be either the flip @xmath53 ( in the configuration labelled by the tetrahedron with vertices @xmath54 ) or the flip @xmath55 .
after the first flip we have a similar choice of two flips , but we exclude that related to the veblen configuration just used ( in order not to go back immediately to the initial configuration )
. therefore the choice of the first flip determines the next ones . by inspection of fig .
[ fig : pentagon - f ] we obtain that superposition @xmath56 of transformations when applied to the initial configuration , gives the same result as @xmath57 .
in other words , starting from the data described above and going around the diagram we obtain the sequence of veblen flips of period five .
the present section is devoted to algebraization of the geometric considerations presented above . in doing that we introduce numerical coefficients ( in the non - commutative hirota equation interpretation they will serve as soliton fields , see section [ sec : hir ] ) attached to vertices of the veblen configuration .
these describe positions of points on the corresponding lines , and the veblen flip gives a map form a set of old coefficients to the corresponding set of new ones .
the pentagon property of the veblen flip implies that the map satisfies the functional dynamical pentagon equation .
consider three distinct collinear points @xmath43 , @xmath39 , @xmath44 of the ( right ) projective space @xmath58 over division ring @xmath1 .
we label them in the spirit of the combinatorics of the veblen and desargues configurations , by edges of a triangle @xmath45 .
their collinearity , expressed in terms of the homogeneous coordinates @xmath59 , @xmath60 , @xmath61 , takes the form of a linear dependence relation @xmath62 with non - vanishing coefficients @xmath63 ( we add the minus sign for convenience ) .
the above linear dependence relation is normalized at @xmath64 by putting corresponding coefficient equal to one .
we can incorporate this new information to the graphic representation on the triangle @xmath45 by adding an arrow from the @xmath16-edge ( i.e. the edge which represents the point with homogeneous coordinates multiplied by the coefficient @xmath16 ) to the @xmath17-edge , as visualized on fig .
[ fig : veblen - flip - tetr ] . consider five distinct points of projective space @xmath58 , which satisfy the veblen configuration condition and are labeled by five edges of a three - simplex , as explained in section [ sec : geom ] .
two triplets of collinear points give rise to two linear dependence relations . in order to write them down
we have to fix enumeration of the lines , and to normalize the corresponding linear relations .
it turns out that in discussing the pentagon property of the veblen map it is convenient to choice the normalization in relation to labelling of straight lines as follows : 1 .
pick up a point ( @xmath44 on fig .
[ fig : veblen - flip - tetr ] ) different from the initial intersection point ( @xmath39 on fig .
[ fig : veblen - flip - tetr ] ) and declare it to belong to the lines @xmath65 and @xmath66 , moreover the point is the normalization point on the both lines @xmath65 and @xmath66 .
the second point on line @xmath65 ( @xmath43 on fig .
[ fig : veblen - flip - tetr ] ) different from the intersection point is attributed the @xmath16 coefficients on both lines @xmath65 and @xmath67 .
3 . the old intersection point is attributed the @xmath16 coefficient on line @xmath68 , and the new intersection point ( @xmath69 on fig .
[ fig : veblen - flip - tetr ] ) is the @xmath16-point on line @xmath66 .
the point on the lines @xmath68 and @xmath67 ( @xmath40 on fig .
[ fig : veblen - flip - tetr ] ) is attributed the @xmath17 coefficients on both the lines . in algebraic terms
we have , as the starting point , homogeneous coordinates of five initial points and four coefficients @xmath70 , @xmath71 , @xmath72 , @xmath73 , in the linear relations @xmath74 in writing down similar linear dependence relations for coordinates of points on two new lines of the veblen configuration @xmath75 we are looking for coordinates @xmath76 of the new intersection point , and for the corresponding coefficients @xmath77 , @xmath78 , @xmath79 , @xmath80 . in consequence
we obtain a map @xmath81 @xmath82 with a free parameter @xmath83 which expresses possibility of multiplying @xmath60 by a non - zero factor .
indeed , equations , give @xmath84 the inverse transformation is given as follows @xmath85 for algebraic purposes one may consider the gauge parameter as an arbitrary rational function @xmath86 of the variables @xmath70 , @xmath71 , @xmath72 , @xmath73 .
then the corresponding function @xmath87 of the variables @xmath77 , @xmath78 , @xmath79 , @xmath80 which should be inserted in equations - in order to make the map @xmath88 birational , should satisfy the functional relation @xmath89 where @xmath77 , @xmath78 , @xmath79 , @xmath80 are given by equations - . [ ex : gauge ] fix the gauge function @xmath90 by requiring @xmath91 , which gives @xmath92 such a choice is a solution of condition .
the algebraic counterpart of the pentagon relation satisfied by the veblen flips @xmath93 will be an analogous relation on the level of the map @xmath88 .
[ th : pentagon - s ] the veblen map @xmath94 , given by equations - , satisfies the functional dynamical pentagon relation on @xmath95 @xmath96 provided the parameters @xmath97 , @xmath98 on the right hand side of are related to the left hand parameters by the conditions @xmath99 the result can be verified by direct calculation , which we sketch below .
the right hand side of equation gives @xmath100 while application of the transformations described by the left hand side gives @xmath101 comparison of both results gives identification of the gauge parameters .
alternatively , notice that equation is just an algebraic reformulation of the geometric consistency of the series of veblen flips explained in section [ sec : geom ] . to check the new ingredient , which are the @xmath102 coefficients ,
the main step is to check consistency of arrows on fig .
[ fig : pentagon - f ] in relation to the rule provided by the veblen flip in the normalization visualized on fig .
[ fig : veblen - flip - tetr ] .
then the only point which needs some calculations is the adjustment of the gauge parameters .
this can be done by comparing expressions for homogeneous coordinates @xmath76 and @xmath103 of the new points , which can be found with the help of equation , @xmath104 thus relations between parameters of the transformations follow from equality of the homogeneous coordinates of the new points generated by the veblen flips . for algebraic purposes equations should be understood as functional relations between functions @xmath105 present in the transformation formulas , i.e. in such a context we write @xmath106 [ ex : gauge-2 ] if the gauge functions @xmath107 are of the same form as @xmath90 described in example [ ex : gauge ] then this choice solves also equations , thus giving a solution to the functional pentagon equation .
in this section we present another map with the pentagon property .
we give also its simple geometric meaning related to four collinear points .
this map will be used in the next section , where we discuss its role in establishing the relation of the veblen map to the hirota equation .
first we discuss another auxiliary map of period three .
consider a map of vertices @xmath108 of a triangle into a projective space @xmath58 such that all the vertices are mapped into collinear ( but different ) points .
the corresponding linear constraint involving the homogeneous coordinates of the points @xmath109 we call normalized at @xmath110 , while @xmath111/@xmath112 is called its @xmath113/@xmath114-point . graphically , we put an arrow at the vertex of the triangle representing the normalization point .
the arrow is directed from the edge joining the ( vertex representing the ) normalization point and the @xmath113-point , see fig .
[ fig : triangle - lin - abcd ] . notice that this is the different normalization rule than that used in section [ sec : v - pent ] .
notice that by changing the normalization point `` by @xmath115 rotation '' , where the orientation is given by the arrow , we arrive at an equivalent linear relation @xmath116 where @xmath117 the birational map @xmath118 , which can be called the non - commutative newton map @xcite , is of order three .
let us consider four collinear ( but distinct ) points of a projective space @xmath58 .
we label them by vertices @xmath54 of a three - simplex .
four triangular faces of the simplex give four linear relations ( up to possible normalizations ) . by analogy with the veblen map presented in section [ sec : v - pent ] we write first two of them @xmath119 and transfer them into the remaining equations @xmath120 the relation between old and new coefficients reads @xmath121 and defines a birational map @xmath122 .
the inverse transformation is given by @xmath123 the map @xmath124 , given by equations - is called _ the ( change of the ) normalization _ map .
this time homogeneous coordinates of all points are fixed , which results in the absence of gauge parameters in the map .
equations can be rewritten in the form @xmath125 which will be relevant in section [ sec : hir ] .
finally , let us consider five collinear points and label them by vertices of the four - simplex .
the normalization map @xmath124 , given by equations - satisfies the functional pentagon relation . the result can be verified by direct calculation .
notice however that , even having much simpler geometric interpretation then the veblen map @xmath88 , the new map @xmath126 has the same graphic representation ( compare fig . [
fig : veblen - flip - tetr ] with fig .
[ fig : triangle - lin - abcd ] ) .
therefore the statement follows from consistency of arrows on fig .
[ fig : pentagon - f ] .
in @xcite it was shown ( see also an earlier related work @xcite ) that the non - commutative version of hirota s discrete kp equation @xcite can be derived from the veblen configuration . moreover ,
its four dimensional compatibility follows from the desargues theorem .
the important observation that the four dimensional consistency of the discrete kp equation in the schwarzian form has combinatorics of the desargues configuration is due to wolfgang schief , see @xcite and remarks in @xcite . in view of the previous sections it is clear that the incidence geometry structures are the same for both the non - commutative hirota system and the veblen map .
the goal of this section is to establish a dictionary between both subjects .
the reader not interested in the theory of integrable discrete equations may go directly to the next part where we discuss the quantum pentagon equation .
we collect first some facts on incidence geometric interpretation of the hirota equation .
the desargues maps , as defined in @xcite , are maps @xmath127 of multidimensional integer lattice into projective space of dimension @xmath128 over a division ring @xmath1 , such that for any pair of indices @xmath129 the points @xmath130 , @xmath131 and @xmath132 are collinear ; here @xmath133 is the @xmath8-th element of the canonical basis of @xmath134 .
we will write @xmath135 instead of @xmath136 for any function @xmath137 on @xmath138
. moreover we will often skip the argument @xmath139 .
the desargues maps can be defined starting from the root lattice @xmath140 , instead of the @xmath138 lattice .
such an approach , proposed in @xcite , makes transparent the @xmath141 affine weyl group symmetry of desargues maps and the discrete kp system .
its `` local '' version is the pentagon property of the veblen map . in the homogeneous
coordinates @xmath142 the map can be described in terms of the linear system @xmath143 where @xmath144 are certain non - vanishing functions .
the compatibility of the linear system turns out to be equivalent to equations @xmath145 where the indices @xmath146 are distinct .
equations and are equivalent , in an appropriate gauge given in @xcite , to the non - commutative hirota system proposed in @xcite .
the precise relation with the hirota equation is as follows ( see @xcite for details and other gauge forms ) .
equations and imply existence of a non - vanishing function @xmath147 satisfying @xmath148 after rescaling the homogeneous coordinates as @xmath149 we obtain that @xmath150 satisfies the linear problem @xcite @xmath151 with the coefficients @xmath152 moreover , equations - reduce to the following systems for distinct triples @xmath146 @xmath153 equations allow to introduce potentials @xmath154 such that @xmath155 when @xmath1 is commutative then the functions @xmath156 can be parametrized in terms of a single potential @xmath157 @xmath158 and equation reduces to the hirota equation @xmath159 as it was discussed in @xcite the three dimensional compatibility of the desargues maps ( or equivalently , the compatibility of the linear problem of the non - commutative hirota equation ) can be stated as the veblen theorem in the form : given four distinct points @xmath160 , @xmath161 , @xmath162 , @xmath163 ; if the lines @xmath164 and @xmath165 intersect , then the lines @xmath166 and @xmath167 intersect as well , see fig .
[ fig : desargues - veblen-3d ] , in the point @xmath168 .
the compatibility of the linear problem is the non - commutative hirota equation .
as the geometry of both the hirota system and the veblen map is the same , we have only to identify the functions @xmath169 entering into the equation with the coefficients @xmath114 , @xmath113 , @xmath16 and @xmath17 .
notice the presence of the additional point @xmath170 outside of the veblen configuration on fig .
[ fig : desargues - veblen-3d ] , which results in the incorporation of the normalization map into the picture .
also different normalizations of equations for three points on the line imply the presence of the newton map relations . in the euclidean geometry approach the relation of the schwarzian form of the discrete kadomtsev petviashvili equation to the menelaus theorem , closely related to the veblen configuration ,
was given in @xcite .
the non - commutative hirota system - is equivalent to the system composed of the normalization map - and of the veblen map - .
the first part of the non - commutative hirota is composed of three equations for different permutations of indices @xmath146 . as it was shown in ( corollary 3.2 in @xcite ) any two of them imply the third one .
we first show that equations can be identified with equations which form the second pair of the normalization map @xmath126 .
in the linear problem identify @xmath171 with @xmath110 of equation , and @xmath172 with @xmath111 , then @xmath173 with @xmath112 , and @xmath174 with @xmath175 .
this gives the following identification of the coefficients : @xmath176 then equation of the non - commutative hirota system is identified with the second of equations , while the first of equations is with @xmath8 and @xmath9 reversed .
the third equation of follows from the first two , as explained in @xcite .
equations describe coefficients @xmath177 and @xmath178 of the linear equation involving @xmath172 , @xmath174 , and @xmath173 , which will be needed in the sequel . in order to find the precise relation of the veblen map to ( three ) equations of the second part of the hirota system we first label the lines and points on fig .
[ fig : desargues - veblen-3d ] in a way described in section [ sec : v - pent ] . by identifying the lines @xmath179 and @xmath180 with lines @xmath65 and @xmath68 , correspondingly , and lines @xmath181 and @xmath182 with @xmath66 and @xmath67
we identify : @xmath183 with @xmath184 , @xmath160 with @xmath64 , @xmath161 with @xmath185 , @xmath162 with @xmath186 , @xmath163 with @xmath187 , and @xmath168 with @xmath188 .
this gives the following pairings : @xmath189 where the newton map relations have been used as well .
the coefficients @xmath70 and @xmath71 , which enter the linear equation involving @xmath172 , @xmath174 , and @xmath173 , can be obtained from the relation of the normalization map and the first part of the hirota system , as described above . due to equations we have @xmath190 then
the second of equations is equation with the same indices @xmath146 , while the first one defines the gauge coefficient @xmath191 the first equation of turns out to be equation with indices @xmath8 and @xmath192 exchanged . finally , in order to check that the second equation of is equation with indices @xmath9 and @xmath192 exchanged ( the last possibility ) , it is convenient to use the following consequence of equations proved in @xcite @xmath193
in this section we discuss an example of the transition from solutions of the functional pentagon equation to its quantum version .
this will be motivated by the quantization of the commutative ( the division ring @xmath1 is replaced by the field @xmath194 of complex numbers ) veblen map .
we show that , under appropriate choice of the gauge , the veblen map preserves a natural poisson structure , which is a limit of the weyl commutation relations . after showing that the veblen map preserves the commutation relations we construct a corresponding solution of the quantum pentagon equation .
analogous results for zamolodchikov s tetrahedron equation @xcite in relation to the four dimensional consistency of the quadrilateral lattices @xcite and discrete darboux equations @xcite were discussed in @xcite .
notice that because our evolution rule given by the veblen map is valid for every division ring , from that point of view integrable quantization can be understood as an integrable reduction , where we specify a particular division ring as a ring of fractions of an algebra , and we demand the evolution preserves the structure of the algebra .
this approach resolves also the problem of ordering of non - commutative factors , often present in the quantization procedures .
such a point of view is analogous to the idea of integrable geometric reductions of the quadrilateral lattices , as initiated in @xcite . in this section
we investigate the veblen map in the commutative case , when the division ring @xmath1 is the field @xmath194 of complex numbers .
we will require however that the map is poisson with respect to the brackets @xmath195 this can be achieved by a suitable choice of the gauge parameter / function @xmath90 in formulas - of the map @xmath12 . in other words
we are looking for an automorphism of the poisson algebra @xmath196 of rational functions of four variables with the bracket defined by .
[ prop : s - g - p ] if the gauge function @xmath197 in the birational map @xmath198 is of the form @xmath199 then the map @xmath88 preserves the poisson bracket . the result can be checked by direct calculation , which is however simpler when we make the ansatz @xmath200 .
then the poisson map condition gives relations @xmath201 and the homogeneity of degree one condition @xmath202 [ cor : inv - p ] the inverse map - in the poisson case with @xmath90 given by takes the form @xmath203 the function @xmath204 mentioned in remark [ ex : gauge ] reads @xmath205 and satisfies the functional relation . from the proof it follows that the function @xmath137 can be of much more general form @xmath206 with arbitrary function @xmath93 .
in order to formulate the poisson veblen map analogue of theorem [ th : pentagon - s ] we should translate the condition into the language of the parameters @xmath207 and @xmath208 .
the following result can be checked by direct calculation .
the condition in the case of the gauge functions / parameters restricted by equations is equivalent to the constraints @xmath209 as the set of independent parameters one can take @xmath210 and @xmath211 . [ prop : pentagon - s - p ] the poisson veblen map @xmath212 given by @xmath213 satisfies the functional pentagon relation @xmath214 the solution @xmath215 , @xmath216 of equations was discussed in the general non - commutative situation in examples [ ex : gauge ] and [ ex : gauge-2 ] .
[ cor : locality ] formulas - and - which define the map @xmath217 and its inverse have been obtained from general formulas - , - for @xmath88 and its inverse assuming only the form of @xmath90 ( with @xmath207 and @xmath208 in the center of @xmath1 ) and the ultra - locality , i.e. variables with different indices commute .
[ cor : norm - p ] under the ultra - locality assumption the normalization map @xmath126 given by equations - , with @xmath102 replaced by @xmath218 , coincides with the local veblen map - with @xmath219 .
because such values of the parameters are solutions of equations then also the normalization map preserves in the commutative case the poisson structure and satisfies the functional pentagon equation .
motivated by proposition [ prop : pentagon - s - p ] we formulate its quantum analogue by replacing the poisson brackets by the corresponding weyl commutation relations @xmath220 where we assume that @xmath221 is a non - zero element of the center of @xmath1 .
because formulas - and - were derived under the locality assumption only we may ask if they preserve the commutation relations .
it turns out that such a simple choice works , which can be checked by direct calculation .
[ prop : s - weyl ] the local veblen map @xmath222 , given by equations - with parameters @xmath223 , @xmath224 , from the center of @xmath1 , preserves the weyl commutation relations .
we have shown in particular , that the map - with entries satisfying relations preserves ultra - locality of the variables .
it turns out that assuming only the ultra - locality of the input it is not enough to prove the ultra - locality of the output .
we have however the following remarkable fact .
let @xmath225 be a subfield of the center of the division ring @xmath1 considered then as a @xmath3-algebra . denote by @xmath226 , @xmath227 , @xmath3-subalgebras of @xmath1 generated by elements @xmath228 , and by @xmath229 denote division hulls of @xmath226 .
assume : 1 .
ultra - locality of the algebras , i.e. @xmath230 , @xmath231 is a subset of the centralizer of @xmath232 in @xmath1 ; 2 . the general position condition , i.e. @xmath233 .
if the algebras @xmath234 , @xmath227 , generated by @xmath235 defined by equations - with @xmath236 satisfy also conditions ( 1 ) and ( 2 ) then there exists @xmath237 such that the remaining part of the commutation relations is satisfied as well . the condition @xmath238 and ultra - locality of the input variables in equations - leads to relation @xmath239 which due to the general position assumption concludes the proof ( other ultra - locality relations for the output variables are identically satisfied or give equations equivalent to the above one ) .
the analogous problem for the tetrahedron equation has been solved in @xcite .
another question which should be answered before considering the pentagon condition in the case under consideration is if in the quantization of the poisson structure of the map the relations remain unchanged . also here
the answer is positive .
the condition when applied to the gauge parameters of the form gives , in the case of weyl commutation relations , the same constraints on the parameters @xmath207 , @xmath208 like in the poisson case . by direct verification ,
notice however that now the order of factors in equations and - matters .
[ prop : pentagon - s - q ] the veblen map @xmath222 given by - when applied to the pairs @xmath240 constrained by the weyl commutation relations with @xmath241 , satisfies the functional pentagon relation .
corollary [ cor : norm - p ] applies also when the poisson structure is replaced by the weyl commutation relations . in this section
we briefly recapitulate known properties of the @xmath221-plane , which are relevant from the point of view of the paper .
moreover we sketch the idea how to get the solution of the quantum pentagon equation from the quantum veblen map .
let @xmath3 be a field .
given @xmath242 , by @xmath243 denote the standard manin s quantum plane ( see for example @xcite ) defined as a quotient of the free associative @xmath3-algebra with generators @xmath102 by the two sided ideal @xmath244 generated by @xmath245 .
its @xmath246-th tensor power @xmath247 is isomorphic to the quotient algebra @xmath248 , where the ideal @xmath249 is generated by relations for @xmath250 .
the isomorphism follows from identification @xmath251 with @xmath252 , and @xmath253 with @xmath254 , where @xmath255 .
it is well known ( see for example @xcite ) that @xmath256 is a noetherian domain , and thus it has a division ring of fractions ( quantum rational functions or quantum weyl division algebra ) denoted here by @xmath257 . in some theoretical physics papers @xmath257
is referred to as the weyl algebra . for @xmath258
the @xmath221-plane is rationally isomorphic to the ( first ) @xmath221-weyl algebra @xmath259 , where two sided ideal @xmath260 generated by @xmath261 ( recall that two noetherian domains are called rationally isomorphic if their division rings of fractions are isomorphic ) .
the isomorphism is given by relations ( see @xcite ) @xmath262 an intermediate object between @xmath263 and @xmath257 is the corresponding @xmath221-torus @xmath264 .
it is known ( see proposition 1.3 of @xcite ) that for @xmath221 not being a root of unity , which we assume in what follows , @xmath265 is simple , and its centre is @xmath3 .
elements of @xmath265 are laurent polynomials in @xmath266 , i.e. finite sums of @xmath267 , @xmath268 with coefficients in @xmath3 .
another close related object is @xcite the hahn malcev
neumann completion @xcite of @xmath265 , which will be denoted by @xmath269 .
its elements are laurent series with support bounded from below ( we consider the lexicographic order in @xmath270 ) .
it turns out that @xmath269 is a division ring containing @xmath257 as a subring @xcite . by proposition [ prop : s - weyl ]
we have : the following elements of @xmath271 @xmath272 with @xmath273 , generate a subalgebra @xmath274 of @xmath271 isomorphic to @xmath275 .
the above veblen map is an automorphism of the division ring @xmath271 .
the @xmath221-torus @xmath276 obtained from @xmath274 is a subalgebra of @xmath271 isomorphic to @xmath277 . in the finite dimensional case , by the skolem noether theorem @xcite ,
any isomorphism of two simple subalgebras of a central simple algebra is given by an inner automorphism of the algebra .
motivated by that we will be looking for an element @xmath278 of a suitable completion of @xmath271 , which generates the transformation - as an inner automorphism . then for any rational function @xmath137 of @xmath279 we would have @xmath280 which , together with the functional pentagon relation , implies that , after eventual rescaling , that @xmath278 satisfies the quantum dynamical pentagon equation @xmath281 notice that even in the root of unity case we are not in the finite dimensional situation because the map in general is not the identity map on the center .
see the relevant discussion in @xcite for the quantum tetrahedron equation .
let us consider the case @xmath282 and equip the algebra @xmath275 with a @xmath283-structure , i.e. an involutive antiautomorphism @xmath284 .
assume that @xmath285 and @xmath286 , for @xmath287 , then it is easy to show that @xmath221 , @xmath288 and @xmath289 must have modulus @xmath65
. we will be interested when the weyl reduction of the veblen map preserves that additional hermitean condition . by direct calculation
we obtain the following result .
[ prop : hermit ] the weyl veblen map is a @xmath283-map if and only if @xmath290 , and @xmath291 , @xmath292 .
the subsequent construction uses a special function which appeared first in works of barnes @xcite , see also @xcite . in the context of quantum integrable systems it was introduced in @xcite under the name of quantum dilogarithm .
essentially the same function was used in the theory of non - compact quantum groups in @xcite as the quantum exponential function .
this function ( in various equivalent forms ) was exploited in other works related to quantum integrable models @xcite .
the ( non - compact ) quantum dilogarithm is a meromorphic function defined by the integral representation @xmath293 where @xmath294 is a parameter .
it satisfies the difference equation @xmath295 which is used to extended it , from the initial convergence strip of the integral representation , to a meromorphic function on the whole complex plane @xmath194 .
its poles ( zeros ) are at points @xmath296 if @xmath297 is real ( or a pure phase @xmath298 , which is not our case ) the function @xmath299 is unitary in the sense that @xmath300 for detailed discussion of its other remarkable properties we refer to the cited works , see also @xcite for a review .
let @xmath113 and @xmath301 provide the schrdinger representation in @xmath302 of the heisenberg commutation relations
@xmath303 = \frac{i}{2\pi}.\ ] ] we have then the corresponding representation of the weyl pair which respects the hermitean restriction described in proposition [ prop : hermit ] @xmath304 the additional requirement @xmath305 , which corresponds to @xmath306 being root of unity is not essential in this representation . in the class of functions which can be analytically extended around the strip @xmath307
the action of @xmath308 reads @xmath309 there is remarkable five term relation @xcite connecting the non - compact quantum dilogarithm function and heisenberg pairs @xmath310 which will be used in the proof of the final result of the paper . for real @xmath207 , @xmath208 let @xmath278 be the unitary operator on @xmath311 given by @xmath312 where @xmath313 , and @xmath314 . then : + 1 ) @xmath278 is the intertwiner satisfying , in particular we have @xmath315 2 ) it satisfies the quantum dynamical pentagon equation . to show the first point we use the property of the function @xmath316 , the representation of the shift action of the exponentiated momentum operator , and the baker
campbell hausdorff ( bch ) formula .
details are known and can be found , for example , in @xcite . in order to conclude the proof of the second point (
the transition from the functional to quantum pentagon equation has been discussed earlier ) we have to check the scalar factor in front of the operator part of @xmath278
. we will do it however in a way independent of the previous part by showing directly that the intertwiner satisfies the quantum dynamical pentagon equation .
denote by @xmath317 the third part in the structure of @xmath7 apart the normalization and dilogarithmic parts .
after shifting all @xmath318 in the pentagon equation to the right and cancelling the dilogarithmic part of the equation using the five term relation with @xmath319 we are left with the equation @xmath320 which can be proven by bch exchanges .
the weyl reduction of the normalization map ( see remark after proposition [ prop : pentagon - s - q ] and corollary [ cor : norm - p ] ) corresponds to the case @xmath321 in formula . in an equivalent form
such a solution of the quantum pentagon equation was used by kashaev @xcite in the quantum teichmller theory .
essentially the same formula as giving a solution of the quantum pentagon equation was obtained in @xcite as a limit of a solution of zamolodchikov s tetrahedron equation .
the same elementary incidence geometric structures , the veblen and desargues configurations , which are behind the non - commutative discrete kp equation and its integrability , provide solutions to the pentagon equation .
there is a natural question how large is the class of the geometric ( in the above sense ) solutions of the functional or quantum pentagon equation within all the solutions .
we note that in the commutative case , as it was demonstrated recently @xcite , the discrete kp equation ( in various equivalent forms ) is the only multidimensionally consistent equation of octahedron type .
this corresponds to the already mentioned observation @xcite that all presently known solutions of the quantum tetrahedron equation can be obtained , by a canonical quantization , from classical solutions of the functional tetrahedron equation describing @xcite four dimensional consistency of quadrilateral lattices @xcite .
it would be interesting to study relation of the quantum pentagon and tetrahedron equations , as initiated in @xcite .
as it was shown in @xcite , on the division ring level desargues maps are equivalent to quadrilateral lattices , provided we take also their laplace transforms @xcite into consideration . on the level of commutative discrete three dimensional multidimensionally consistent systems there
are not known other then the geometric ones .
these are the hirota system , describing desargues maps , and its reductions : * discrete darboux equations @xcite describing multidimensional quadrilateral lattices @xcite , whose relation to desargues maps was given in @xcite ; * the miwa ( discrete b - kp equation ) @xcite describing the b - quadrilateral lattices @xcite ; * discrete c - kp equation @xcite describing the c- ( or symmetric ) quadrilateral lattices @xcite .
as far as two dimensional integrable systems and integrable maps are concerned , there is abundance of examples of their derivation as reductions of the hirota equation , directly or via the three dimensional systems listed above , see the very incomplete list @xcite which starts from the original paper @xcite of hirota .
in fact , the possibility of writing of a given system in hirota - like bilinear form is a paradigm of the integrable system theory .
it is remarkable that the the veblen flip , which turned out to be so fundamental in the domain of integrable discrete systems , supplemented by the very physical ultra - locality condition singles out the weyl commutation relations , modulo the possible different choice of the gauge function @xmath90 , which however does not affect the geometric structure .
a. d. would like to thank to piotr stachura for his criticism , which helped to improve presentation .
the work of a. d. was supported in part by the ministry of science and higher education grant no .
n n202 174739 .
v. v. bazhanov , v. v. mangazeev , and s. m. sergeev , _ quantum geometry of 3-dimensional lattices and tetrahedron equation _ ,
[ in : ] xvith international congress on mathematical physics , p. exner ( ed . ) , pp .
23 - 44 , world scientific publishing , 2010 .
a. i. bobenko , _ from discrete differential geometry to classification of discrete integrable systems _ , talk given at the workshop _ quantum integrable discrete systems _ , 2327 march 2009 , isaak newton institute for mathematical sciences , cambridge uk , ` http://www.newton.ac.uk/programmes/dis/seminars/032610006.html ` .
e. date , m. kashiwara , m. jimbo , t. miwa , _ transformation groups for soliton equations _ , [ in : ] proceedings of rims symposium on non - linear integrable systems classical theory and quantum theory ( m. jimbo and t. miwa , eds . ) , pp .
39119 , world science publishing co. , singapore , 1983 .
s. kharchev , d. lebedev , m. semenov - tian - shansky , _ unitary representations of @xmath324 , the modular double and the multiparticle @xmath221-deformed toda chains _ , commun .
* 225 * ( 2002 ) 573609 .
f. w. nijhoff , h. w. capel , _ the direct linearization approach to hierarchies of integrable pdes in @xmath327 dimensions : i. lattice equations and the differential - difference hierarchies _ , inverse problems * 6 * ( 1990 ) 567590 .
s. m. sergeev , _ quantum integrable models in discrete @xmath327-dimensional space - time : auxiliary linear problem on a lattice , zero - curvature representation , isospectral deformation of the zamolodchikov bazhanov baxter model _ , physics of particles and nuclei
* 35 * ( 2004 ) 131 .
s. m. sergeev , _ mathematics of quantum integrable systems in multidimensional discrete space - time ( a preliminary draft of a book , version 4 ) _ , ` http://ise.canberra.edu.au/mathphysics/files/2009/09/book5.pdf ` ( access 19.04.2011 ) .
t. timmermann , _ an invitation to quantum groups and duality . from hopf algebras to multiplicative unitaries and beyond
_ , ems textbooks in mathematics , european mathematical society publishing house , zrich , 2008 . | we define a map @xmath0 , where @xmath1 is an arbitrary division ring ( skew field ) , associated with the veblen configuration , and we show that such a map provides solutions to the functional dynamical pentagon equation .
we explain that fact in elementary geometric terms using the symmetry of the veblen and desargues configurations .
we introduce also another map of a geometric origin with the pentagon property .
we show equivalence of these maps with recently introduced desargues maps which provide geometric interpretation to a non - commutative version of hirota s discrete kadomtsev
petviashvili equation .
finally we demonstrate that in an appropriate gauge the ( commutative version of the ) maps preserves a natural poisson structure the quasiclassical limit of the weyl commutation relations . the corresponding quantum reduction is then studied .
in particular , we discuss uniqueness of the weyl relations for the ultra - local reduction of the map .
we give then the corresponding solution of the quantum pentagon equation in terms of the non - compact quantum dilogarithm function . |
measuring the evolution of the sfr density of the universe as a function of look - back time is essential to understand the formation and evolution of galaxies .
the h@xmath1 luminosity is one of the best optical estimators of the current sfr , modulo the initial mass function ( e.g. , kennicutt 1992 , 1998 ) .
unfortunately , h@xmath1 can only be observed with optical ccds out to @xmath7 . at higher redshifts ,
the @xmath03727 line is frequently used as a sfr estimator due to its availability in optical spectroscopic samples . however , the transformation between the number of ionising photons ( which is a direct consequence of the current sfr ) and the intensity or equivalent width ( @xmath8 ) of this line is strongly dependent on the excitation conditions and metallicity .
previous empirical calibrations of the sfr in terms of the line ( e.g. , gallagher , bushouse , & hunter 1989 ; kennicutt 1992 ; guzmn et al .
1997 ) vary within a factor of a few . among other problems
, these calibrations may suffer from small number statistics , incompleteness in luminosities , starburst ages and/or physical conditions of their samples , and uncertain extinction corrections .
we have analysed the emission line properties of a local sample of h@xmath1-selected star forming galaxies ( the ucm sample ) and derived reasonably precise empirical calibrations between the and h@xmath1 luminosities as a function of the galaxy population properties . in a recent paper , jansen ,
franx & fabricant ( 2001 ) carried out a similar study for the nearby field galaxy survey ( nfgs , jansen et al .
2000a , b ) and found , among other things , that the observed /h@xmath1 ratio varies by a factor of 7 at luminosities near @xmath9 , and that the /h@xmath1 ratio is inversely correlated with luminosity .
our study is , in many ways , complementary of that of jansen et al . , but we use an h@xmath1-selected ( thus sfr - selected ) sample of galaxies instead of a @xmath3-band selected one .
we confirm most of the findings of jansen et al . , but we also demonstrate that using an extra piece of information , namely the @xmath8 of , much of the scatter in the /h@xmath1 ratio can be removed , allowing a much more accurate calibration of vs.sfr .
our basic dataset comes from universidad complutense de madrid ( ucm ) survey of h@xmath1 selected galaxies .
optical spectroscopy ( gallego et al . 1996 ; 1997 ) and optical and near - infrared photometry ( prez - gonzlez et al . 2000 , 2002 ; gil de paz et al .
2000 ) are available for this sample . based on their spectroscopic properties ,
the ucm galaxies have been roughly divided into two main classes : hii - like and disk - like galaxies ( see gil de paz et al . 2000 ) .
we will compare the properties of the ucm galaxies with those of the nfgs galaxies of jansen et al .
a detailed description of our study can be found in aragn - salamanca et al .
we assume @xmath10km@xmath11s@xmath12mpc@xmath13 throughout .
in figure 1 we plot the @xmath14}}/{\rm h}\alpha$ ] ratio ( both observed and corrected from extinction using the balmer decrement}}^0 $ ] and @xmath15 indicate extinction - corrected emission - line fluxes . ] ) for the ucm and nfgs galaxies as a function of absolute @xmath3- and @xmath4-band magnitudes and @xmath16@xmath17 .
there is a large scatter in the @xmath2}}/{{\rm h}\alpha}$ ] and @xmath2}}/{{\rm h}\alpha^0}$ ] ratios for all the galaxies , but the scatter in the extinction - corrected @xmath2}}^0/{{\rm h}\alpha^0}$ ] ratio is considerably smaller , indicating that much of this scatter is due to differences in extinction .
figure 1 also reveals significant difference in the median @xmath2}}/{{\rm h}\alpha}$ ] and @xmath2}}/{{\rm h}\alpha^0}$ ] ratios for the ucm and nfgs galaxies .
these ratios are considerably smaller for the ucm galaxies . however , the extinction - corrected @xmath2}^0}/{{\rm h}\alpha^0}$ ] ratios for both samples agree quite well , indicating that the line ratio differences between both samples are mainly due to a difference in the mean extinction .
this is readily explained by the fact that the nfgs sample has been selected in the @xmath3-band , while the ucm sample was selected in the red ( h@xmath1 ) , and thus it is not surprising that galaxies with higher extinction are underrepresented in the blue selected sample .
therefore , differences in sample selection can yield very large differences in the line emission properties .
the third conclusion from these plots is that the @xmath2}}/{{\rm h}\alpha}$ ] and @xmath2}}/{{\rm h}\alpha^0}$ ] ratios for disk - like and hii - like galaxies are very different .
however , the average @xmath6}^0}/{{\rm h}\alpha^0}$ ] ratio is the same for both classes of galaxies .
again , the mixture of galaxy types in a given sample can have large effects in the line ratios .
we also find that , although with significant scatter , the @xmath2}}/{{\rm h}\alpha}$ ] and @xmath2}}/{{\rm h}\alpha^0}$ ] ratios are reasonably well correlated with the absolute @xmath3- and @xmath4-band magnitudes and with @xmath18}})$ ] .
however , the extinction - corrected @xmath2}^0}/{{\rm h}\alpha^0}$ ] ratio is largely independent of these quantities , indicating that extinction is the main driver of these correlations .
this is supported by figure 2 , which shows quite reasonable correlations between the extinction and the absolute magnitudes of the galaxies .
the extinction is also correlated with @xmath5}})$ ] .
this figure also shows that the typical extinction for the nfgs galaxies is significantly smaller than for the ucm galaxies .
these correlations could allow us to statistically predict with varying degrees of accuracy the observed and even extinction - corrected h@xmath1 fluxes from the observed flux using the information contained in the @xmath5}})$ ] and/or the absolute magnitudes , provided that the sample selection effects are correctly taken into account .
indeed , it is very important to stress that these correlations apply , strictly speaking , to samples with similar properties to the ones presented here .
if the selection of the sample is different , significant systematics could be present , as shown by the differences in behavior between the ucm and the nfgs galaxies .
moreover , there is no guarantee that high redshift galaxy samples would not show significant evolution in their line ratios ( e.g. , due to chemical evolution ) which could , in principle , invalidate any locally - derived calibration .
any calibration derived from these empirical correlations would have a significant intrinsic scatter , and thus the predictions for individual galaxies will be quite uncertain
. however , if one is only interested in statistical properties of large samples ( e.g. , determining the star - formation rate density of large regions of the universe ) , the problem of the large scatter would be alleviated .
but it is crucial to take into account the properties of the sample , in particular the selection criteria , luminosity range and mixture of galaxy types , since important systematic effects are also present on top of the large scatter .
obviously , just using large samples would not solve the systematic problems .
finally , there is another problem we have not discussed and needs to be considered : the effect of galaxies with no emission on the statistical determination of sfrs and sfr densities .
a significant fraction of the ucm galaxies ( 20% ) have h@xmath1 emission but no detectable emission ( @xmath5}}>-5\,$] ) .
this means that by only counting the emitting ucm galaxies , we would miss @xmath19% of the extinction - corrected h@xmath1 flux , i.e. , @xmath19% of the sfr .
this problem becomes negligible for a blue - selected sample such as the nfgs since , due mainly to its lower average extinction , virtually all nfgs galaxies with h@xmath1 emission have detectable .
this research was supported in part by the spanish programa nacional de astronoma y astrofsica .
aas acknowledges generous financial support from the royal society .
agdp acknowledges financial support from nasa .
pgpg acknowledges financial support from the spanish ministerio de educacin y cultura .
cegd acknowledges financial support from the ucm .
aragn - salamanca , a. , alonso - herrero , a. , gallego , j. , garca - dab , c.e .
, gil de paz , a. , prez - gonzlez , p.g . & zamorano , j. 2002 , , submitted gallagher , j.s . , bushouse , h. & hunter , d.a .
1989 , , 97 , 700 gallego j. , zamorano j. , alonso o. & vitores a.g .
1996 , , 120 , 323 gallego j. , zamorano j. , rego m. & vitores a.g .
1997 , , 475 , 502 gil de paz , a. , aragn - salamanca , a. , gallego , j. , alonso - herrero , a. , zamorano , j. & kauffmann , g. 2000 , , 316 , 357 guzmn , r. , gallego , j. , koo , d.c . ,
phillips , a.c . , lowenthal , j.d . ,
faber , s.m . ,
illingworth , g.d .
& vogt , n.p .
1997 , , 489 , 559 jansen , r.a . ,
franx , m. , fabricant , d.g . & caldwell , n. 2000a , , 126 , 271 jansen , r.a . , fabricant , d.g . ,
franx , m. & caldwell , n. 2000b , , 126 , 331 jansen , r.a . , franx , m. & fabricant , d. 2001 , , 551 , 825 kennicutt , r.c .
1992 , , 388 , 310 kennicutt , r.c .
1998 , , 36 , 189 perz - gonzlez , p.g . ,
zamorano , j. , gallego , j. & gil de paz , a. 2000 , , 141 , 409 perz - gonzlez , p.g .
, gil de paz , a. , zamorano , j. , gallego , j. , alonso - herrero , a. & aragn - salamanca , a. 2002 , , in press , astro - ph/0209396
_ kennicutt : _ i was surprised by the large shift in @xmath2}}/{{\rm h}\alpha}$ ] ratio between your ucm disk galaxy sample and the nfgs objects because my impression was that the types of galaxies were similar ( despite the different selection methods ) .
could part of the difference arise because your spectra sample only the central parts of the galaxies ?
+ _ aragn - salamanca : _
i do nt think so .
the ucm spectra were taken with slit widths that approximately match the size of the galaxies ( at the expense of spectral resolution ) .
thus the spectra used here are close to `` global '' spectra .
note that the majority of the ucm galaxies are significantly more distant than the nfgs galaxies , and therefore have much smaller angular sizes .
moreover , the fact that the extinction - corrected @xmath6}^0}/{{\rm h}\alpha^0}$ ] ratios of the ucm and nfgs galaxies match very well gives us confidence that extinction is the main driver of the observed differences , and that the intrinsic ratios are indeed the same .
r. terlevich : _ in a recent analysis i did with elena terlevich and daniel rosa we found that using the @xmath20 ratio without taking into account the the underlying balmer absorption you tend to overestimate the extinction .
could this be related to the relation found between the @xmath5}})$ ] and extinction ? + _ aragn - salamanca : _ it is true that the observed @xmath21 ratio can be affected by stellar absorption and thus bias the extinction measurements . but the nfgs balmer decrement measurements were corrected for stellar absorption , and the trend between @xmath18}})$ ] and extinction is similar to the one found for the ucm galaxies ( albeit , with higher average extinction , as discussed above ; cf.figure 2c ) . moreover , being aware of this problem , we only attempted to determine balmer - decrement extinctions for ucm galaxies with reasonably strong emission lines ( @xmath22 ) , where this effect should be smaller .
it is interesting to note that @xmath5}})$ ] anti - correlates with extinction _ and _ luminosity ( figure 2 ) .
because of the luminosity - metallicity relation , it is not unreasonable to expect metal - rich galaxies to have more dust and thus more extinction . + | the @xmath03727 emission line is frequently used as an indicator of the star formation rate ( sfr ) despite its complex dependence on metallicity and excitation conditions .
we have analysed the properties of the and h@xmath1 emission lines for a complete sample of local h@xmath1-selected galaxies , the universidad complutense de madrid ( ucm ) survey .
we find a large scatter in the @xmath2}}/{{\rm h}\alpha}$ ] line ratios , although the scatter in the extinction - corrected @xmath2}}^0/{{\rm h}\alpha^0}$ ] ratio is considerably smaller .
we also find that the @xmath2}}/{{\rm h}\alpha}$ ] ratios are reasonably well correlated with the absolute @xmath3- and @xmath4-band magnitudes and with @xmath5}})$ ] .
however , the extinction - corrected @xmath6}^0}/{{\rm h}\alpha^0}$ ] ratio is largely independent of these quantities , indicating that extinction is the main driver of the correlations .
these correlations allow us to statistically predict with varying degrees of accuracy the observed and extinction - corrected h@xmath1 fluxes from the observed flux using the information contained in @xmath5}})$ ] and/or the absolute magnitudes , but extreme caution is needed to make sure that the sample selection effects are correctly taken into account .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
the classification scheme for sne is based on the presence or absence of hydrogen lines in their optical spectra ( type ii and type i , respectively ) . while type ii
/ ib sne occur during the core collapse of a massive star , type ia sne are believed to originate from the deflagration or detonation of an accreting white dwarf star in a binary system .
the massive ( @xmath6 ) progenitor stars for the controversial type ic sn subclass are thought to have lost their outer hydrogen and helium layers either through stellar winds or by mass transfer to a companion , leaving behind a stripped carbon and oxygen ( c+o ) star .
such a ` naked ' c+o star will later explode when its iron core collapses ( e.g. nomoto et al .
the interaction of the outgoing sn shock wave with the ambient csm , deposited either by a pre - sn stellar wind or non - conservative mass transfer to a companion , produces hot gas with characteristic temperatures in the range @xmath7@xmath8 k ( chevalier & fransson 1994 ) .
gas heated to such high temperatures produces radiation predominantly in the x - ray range .
x - ray emission from this interaction is expected for all type ib / c and ii sne with substantial csm established by the massive progenitors . over the last 20 years , searches for
x - ray emitting sne have been successful for only a relatively small number of type ii s in the near aftermath ( days to months ) of the explosion ( see immler & lewin 2002 for a review article ) . by contrast , no type ia or ib / c sn has ever been firmly detected in x - rays , apart from a @xmath9 excess in an epic - pn image @xmath10 offset from the optical position of the ic sn 2002ap ( rodriguez - pascual et al .
hard ( 210 kev band ) x - ray emission was recorded with from the position of the unusual sn 1998bw , which might be associated with a @xmath11-ray burst event ( grb 980425 ; galama et al .
1998 , pian et al . 1998 ) . due to the large error box of the wide field camera observation ( between @xmath12 and @xmath13 , @xmath14 confidence limit ) and the rather large probability of @xmath15
that the source is a random chance coincidence , the association of the x - ray source with sn 1998bw and grb 980425 is still tentative ( galama et al . 1998 ) .
evidence for soft ( 0.12.4 kev band ) x - ray emission has been reported from the type ic sn 1994i , based on hri observations 82 days after the outburst ( immler , pietsch & aschenbach 1998 ) . however , the observations , with a spatial resolution of @xmath16 ( fwhm on - axis ) , were not conclusive since sn 1994i is located close to the x - ray bright nucleus of the host galaxy m51 ( distance @xmath17 ) and is embedded in a high level of extended x - ray emission from hot gas and unresolved point - like x - ray sources in the bulge of m51 .
we used the x - ray observatory to search for x - ray emission from sn 1994i in x - ray images taken between six ( june 6 , 2000 ) and seven years ( june 23 , 2001 ) after the explosion ( march 31 , 1994 ; chandler , phillips & rupen 1994 ) .
our 14.9 ks and 26.8 ks observations were carried out with the s3 chip of the advanced ccd imaging spectrometer ( acis - s ) at the focal plane of the telescope and the nucleus of the host galaxy m51 placed at the aim - point ( terashima & wilson 2001 ) .
the superb spatial resolution ( @xmath18 fwhm on - axis ) , together with the high sensitivity of the instrument and the long combined exposure ( 41.7 ks ) , allows us to separate point - like sources from the diffuse emission in the bulge of m51 and to carry out a sensitive search for x - ray emission from sn 1994i .
the co - added and adaptively smoothed image of the central region of m51 is presented in fig . 1 .
an x - ray source is detected at the position of sn 1994i in the merged data with a significance of @xmath19 in the 0.32 kev band using the ` wave detect ' algorithm implemented in the ciao data analysis package .
the position of the x - ray source ( @xmath20 ) is fully consistent with the radio position of the sn ( @xmath21 offset ; rupen et al .
astrometry of the m51 nucleus shows a similarly small offset between the ( 28 kev band ) x - ray and radio positions ( @xmath22 ) .
exposure corrected source counts were extracted within a radius of 3 image pixels ( 90% encircled energy radius at 1 kev ) and corrected for the background taken in an annulus with inner and outer radii of 3.5 and 9.5 pixels , respectively ( 1 pixel corresponds to @xmath23 ) .
a spectral model has to be assumed to convert source counts into energy fluxes .
we adopted an effective ( 0.32 kev band ) cooling function of @xmath24 for an optically thin raymond - smith thermal plasma with a temperature of @xmath25 k ( raymond , cox & smith 1976 ) .
although the temperature is unknown , this value is consistent with theoretical expectations ( chevalier & fransson 1994 ) , previous x - ray observations of other sne and the detection of sn 1994i in the soft ( 0.32 kev ) x - ray band where the peak of a @xmath25 k spectrum is located . the equivalent count rate to ( unabsorbed ) flux conversion factors are then @xmath26 and @xmath27 for the hri and acis - s3 , respectively , and a galactic foreground column density of @xmath28 ( dickey & lockman 1990 ) .
the uncertainty of the conversion factors for optically thin thermal spectra with temperatures in the range @xmath25@xmath8 k is @xmath29 . assuming a 0.86 kev thermal bremsstrahlung spectrum ( corresponding to a temperature of @xmath25 k ) instead of a raymond - smith thermal plasma increases the conversion factor by @xmath30 .
the results for the two individual observations of sn 1994i , together with previous hri results ( immler , pietsch & aschenbach 1998 ) , are summarized in table 1 .
we used the and data to construct a combined x - ray lightcurve of sn 1994i , which is presented in fig . 2
. a best - fit single power - law rate of decline of @xmath31 with index @xmath32 ( solid line , fig .
[ f2 ] ) is inferred for the measurements on days 82 , 2,271 and 2,639 .
assuming an initial exponential rise of the x - ray luminosity after the outburst ( at time @xmath33 ) of sn 1994i and a subsequent power - law decline with index @xmath34 , we also parameterized the x - ray evolution as @xmath35 with @xmath36 .
this model has been successfully used to describe the time dependence of the radio emission of sne ( weiler et al .
the external absorption of the emission is represented by the @xmath37 term ( ` optical depth ' ) and the time - dependence of the optical depth is parameterized by the exponent @xmath38 . in the x - ray regime
, the rise could represent either decreasing absorption by material along the line of sight to the hot gas or simply non - production of x - rays in the band at early times .
if we adopt this strictly heuristic description , we find @xmath3 and @xmath39 ( dashed line , fig .
[ f2 ] ) . in case
the detection on day 82 corresponds to a pre - maximum measurement , a steeper index of @xmath40 is inferred ( dotted line , fig .
however , given the large error range of the flux measurements , @xmath34 and @xmath38 are not very well constrained . while a @xmath41 rate of decline has been observed for the type iib sn 1999em with ( pooley et al .
2002 ) , faster rates of decline have been reported for the ic sn 1998bw ( @xmath42 ) , inferred from the tentative hard ( 210 kev ) band data ( pian et al .
1999 ) , the ic radio sn 1990b ( @xmath43 ) and the ib radio sne 1983n ( @xmath44 ) and 1984l ( @xmath45 ; van dyk et al .
by contrast , the long - term x - ray lightcurve of the iib sn 1993j is best described by a slow rate of decline with @xmath46 ( immler , aschenbach & wang 2001 ) .
the x - ray emitting material could be either shocked sn ejecta ( reverse shock ) or shocked csm ( forward shock ; chevalier & fransson 1994 ) .
the lack of time - dependent x - ray spectroscopy precludes a distinction but , as we shall see , the x - ray lightcurve is consistent with the latter model with a constant pre - sn mass loss rate ( @xmath5 ) .
alternatively , the csm might result from a non - conservative mass - transfer to the companion star , leading to the formation of a flattened or disk - like h and he - rich shell ( nomoto et al .
1994 ) . in this case a csm profile flatter than @xmath5 is expected due to the non - spherically symmetric geometry of the mass - transfer . in the stellar wind scenario
the continuum equation requires a mass - loss rate of @xmath47 through a sphere of radius @xmath48 .
after the sn shock plows through the csm , its density is @xmath49 ( fransson , lundqvist & chevalier 1996 ) .
the x - ray luminosity of the shock - heated csm is @xmath50 , where d@xmath51 is the volume , @xmath52 is the number density of the shocked csm and @xmath53 is the mean mass per particle ( @xmath54 kg for a h+he plasma ) .
we thus obtain @xmath55 . the observed x - ray rate of decline ( @xmath41 ) is consistent with this description if @xmath56 , @xmath57 and @xmath58 are constant .
we can hence use the observed x - ray luminosity at time @xmath59 after the outburst to measure the ratio @xmath57 assuming a constant shell expansion velocity @xmath58 . given our observed x - ray luminosities ( table 1 ) and assuming a shell expansion velocity of @xmath60 ( filippenko et al .
1995 ) we derive a mass - loss rate of @xmath61 consistent with all three detections and the upper limit on day 1,368 .
it should be noted , however , that a single power - law model can only be reconciled with the data obtained on days 822,639 , but not with the early upper limit on day 52 ( see fig .
this is indicative that the above model might be incomplete for early epochs .
the addition of an exponential term describing the rise during the early phase of the emission leads to a fit that is in agreement with all measurements in case the detection on day 82 was a post - maximum measurement ( @xmath3 ) .
by contrast , the model with a steeper rate of decline ( @xmath40 ) , which postulates that the early detection corresponds to a pre - maximum measurement , is in conflict with a @xmath5 profile .
table 1 summarizes the csm number density @xmath62 for the different radii @xmath63 corresponding to the dates of the observations . as was already claimed based on the early data ( immler , pietsch & aschenbach 1998 ) , the @xmath64 upper limit for the csm density at @xmath65 ( day 1,368 ) is lower than that at @xmath66 ( day 82 ) .
our results at @xmath67 and @xmath68 ( days 2,271 and 2,639 , respectively ) clearly support these early indications for a decreasing csm density profile ( see table 1 ) .
the combined and data of sn 1994i give a best - fit profile of @xmath4 .
the only sn for which a csm density profile has been constructed from x - ray measurements is sn 1993j ( type iib with a progenitor of @xmath69 ) , based on long - term monitoring with between six days and five years after the outburst ( immler , aschenbach & wang 2001 ) . a comparison between the csm profiles of the two sne is presented in fig .
although the csm density profile of sn 1993j is significantly flatter ( @xmath70 ) than that of sn 1994i , the two sne have a similar csm number density of @xmath71 at a radius of @xmath72 from the site of the explosion ( cf .
fig . 3 ) . given that both the x - ray rate of decline ( @xmath73 ) and the csm profile ( @xmath4 ) of sn 1994i are not in conflict with what is expected for a constant stellar wind speed and a constant mass - loss rate of the progenitor ( @xmath73 , @xmath5 ) , it is likely that the stellar wind of the massive progenitor dominates the csm .
we thank roger chevalier for helpful discussion and the referee ( bernd aschenbach ) for his comments that led to significant improvement of the manuscript .
this research was supported by nasa grants nag 5 - 8999 to the university of massachusetts and nag 81755 to the university of maryland .
. is supported by the japan society for the promotion of science postdoctoral fellowship for young scientists .
chandler , c.j . , phillips , j. a. , & rupen , m. p. 1994
, iau circ .
5978 chevalier , r.a . , & fransson , c. 1994 , , 420 , 268 dickey , j.m .
, & lockman , f.j .
1990 , ara&a , 28 , 215 feldmeier , j.j . , ciardullo , r. , & jacoby , g.h .
1997 , , 479 , 231 filippenko , a.v . ,
1995 , , 450 , l11 fransson , c. , lundqvist , p. , & chevalier , r.a .
1996 , , 461 , 993 galama , t.j .
, et al . 1998 , , 395 , 670 immler , s. , pietsch , w. , & aschenbach , b. 1998 , a&a , 336 , l1 immler , s. , aschenbach , b. , & wang , q.d . 2001 , , 561 , l107 immler , s. , & lewin , h.g .
2002 , in supernovae and gamma - ray bursts , ed .
k. w. weiler ( springer publisher ) , astro - ph/0202231 nomoto , k. , et al .
1994 , , 371 , 227 pian , e. , et al .
1999 , a&as , 138 , 463 pooley , d. , et al .
2002 , apj in press , astro - ph/0103196 raymond , j.c . ,
cox , d.p . , & smith , b.w .
1976 , , 204 , 290 rodriguez - pascual , p.m. , et al .
2002 , iau circ .
7821 rupen , m.p . , et al .
1994 , iau circ .
5963 terashima , y. , & wilson , a. s. 2001 , , 560 , 139 van dyk , s. , sramek , r.a . , weiler , k.w .
, & panagia , n. 1993 , , 419 , 162 weiler , k.w . , et al .
1986 , , 301 , 790 ccccccc 52 & hri & @xmath74 & @xmath75 & @xmath76 & @xmath77 & @xmath78 + 82 & hri & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 + 1,368 & hri & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 + @xmath89 & acis - s & @xmath90 & @xmath91 & @xmath92 & @xmath93 & @xmath94 + 2,639 & acis - s & @xmath95 & @xmath92 & @xmath96 & @xmath93 & @xmath97 | we present two high - resolution x - ray observations of supernova ( sn ) 1994i which show , for the first time , that the interaction of the blast wave from a type ic sn with its surrounding circumstellar material ( csm ) can give rise to soft x - ray emission . given a 0.32 kev band x - ray luminosity of @xmath0 between six and seven years after the outburst of sn 1994i , and assuming the x - ray emission arises from the shock - heated csm , we derive a pre - sn mass - loss rate of @xmath1 . combining the results with earlier observations ,
we construct the x - ray lightcurve of sn 1994i
. a best - fit x - ray rate of decline of @xmath2 with index @xmath3 and a csm density profile of @xmath4 are inferred , consistent with what is expected for a constant mass - loss rate and constant wind velocity profile for the sn progenitor ( @xmath5 ) . |
gamma - ray bursts ( grbs ) are the most luminous and intriguing explosions in the universe .
a number of physical interpretations for grbs have been proposed since they were discovered about four decades ago .
the most successful model for grbs and their afterglows is the fireball shock model ( * ? ? ?
* and references therein ) .
although the interpretation for the prompt gamma - ray emission is still controversial even in this standard scenario , the long - wavelength afterglows have been widely accepted to be the emission from relativistic shocks sweeping up an external medium ( * ? ? ?
* ; * ? ? ?
* and references therein ) . in spite of the success of the standard external shock model for interpreting most features of afterglows ,
there are still a few issues recently raised by the _ swift _ x - ray telescope ( xrt ) @xcite .
the first issue is the ubiquitously detected shallow decay of early x - ray afterglows @xcite . in general , to account for the shallow decay , a
refreshed shock " scenario is assumed , either by a continuous activity of the grb progenitor @xcite , or by a power - law distribution of the lorentz factors in the ejecta @xcite .
however , although the
refreshed shock " model is generally able to explain the flatter temporal x - ray slopes from current analysis , it may not be able to explain self - consistently both the x - ray and the optical afterglows of some well - monitored grbs , such as grb 050319 and grb 050401 @xcite .
the second issue is the lack of jet breaks that are supposed to be detected simultaneously in x - ray and optical light curves @xcite .
this raises a serious concern about our understanding of the afterglow emission .
currently , possible solutions to this problem are additional mechanisms taking place in the early x - ray emission @xcite , x - ray emission arising from different outflows with the optical emission @xcite , or alternatively , the temporal break in x - rays is masked by some additional source of x - ray emission @xcite , so that a jet break at the expected time could still be seen in the optical band .
the last interpretation may bring renaissance to the physical meaning of a jet break @xcite and @xmath0 @xcite , and therefore revive the relation between @xmath0 and @xmath1 @xcite .
recently @xcite realized that the impulsive prompt x - ray emission scattering off the surrounding medium could give rise to a long - term x - ray flux , which is comparable to that of most observed afterglows .
the scattered x - ray emission ( which is called echo emission herein ) could be an additional contributor to the early x - ray afterglow , which was originally put forward by @xcite and then further discussed by @xcite and @xcite .
@xcite also worked out detailed light curves of the echo emission that could nicely reproduce most of the temporal features of observed x - ray afterglows , i.e. , a shallow decay followed by a
normal " decay and a further rapid decay @xcite .
this scenario provides a self - consistent solution to both the shallow decay and the jet break problem . by fitting the x - ray light curves of grb 060813 and grb 060814 ,
@xcite concluded that the scattering medium , i.e. , a dusty shell should be at a distance of about tens of parsecs from the grb source .
coincidentally , this conjectured dusty shell by @xcite is consistent with the massive wind bubble around a wolf - rayet ( wr ) star , which has been proposed as the progenitor of a grb .
the study of stellar structure and evolution has provided a mass - loss history of a massive star , starting with the main sequence ( ms ) stage as an o star , which will evolve into a wr star and then produce a final supernova and/or a grb . by the computation of stellar wind and atmosphere models for a o - type star of 60 @xmath2 mass , @xcite concluded that the stellar mass lost to the wind bubble is about 32 @xmath2 in the ms stage , 8 @xmath2 in the luminous blue variable ( lbv ) stage , and 16 @xmath2 in the wr stage .
such a massive wind bubble will produce a significant shell mainly swept up by the ms wind at the time of the supernova or grb ( see fig.[fig : bubble ] ) , with a distance from the progenitor of about @xmath3 tens of parsecs @xcite .
if the clumps formed during the lbv and wr stages survive long enough , they will be hitting the ms shell roughly at the time of the supernova or grb @xcite . in this paper
, we apply the echo emission model to a sample of 36 grbs with well - monitored x - ray light curves by the xrt onboard _
swift_. we find that most of the light curves without significant flares can be reproduced by our model . in this sample ,
14 bursts have available redshifts by observations . by fitting their x - ray light curves
, we find the distance @xmath4 of the wind bubbles from grb progenitors .
a measurement of @xmath5 might bring us closer to finding the mass loss rate , density and wind velocity of the progenitor prior to the explosion .
our paper is structured as follows : in @xmath6[sec : bubble ] , we first discuss the existence of massive stellar wind bubbles around grbs when they explodes and the subsequent x - ray echo emission .
we formulate the theoretical light curves of x - ray echo emission in @xmath6[sec : theory ] , and we introduce the temporal break in echo emission and fit the observational light curves of 36 grbs in @xmath6[sec : break ] .
discussion and conclusions follow in @xmath6[sec : discuss ] .
it has been shown by @xcite with a detailed calculation following @xcite that , the fiducial flux of the echo emission could be comparable to an already observed x - ray afterglow at some certain time , given that the prompt x - rays from a grb are significantly scattered by the circumstellar dust . here
, we provide a more concise approach by order - of - magnitude estimation .
evidence accumulated in the last few years has verified the long - hypothesized origin of soft grbs in the deaths of wr stars ( * ? ? ? * and references therein ) .
it is generally believed that the progenitor wr star will blow out a hot bubble in the interstellar medium ( ism ) by massive stellar wind @xcite .
the typical radius of the dense shell of swept - up surrounding ism is of order tens of parsecs , as given by a self - similar solution @xcite @xmath7 where the mass - loss rate during the ms stage @xmath8 , the typical wind velocity @xmath9 km / s , and the lifetime of the ms stage @xmath10 yr are adopted for a 35 @xmath11 star @xcite , and the number density of interstellar medium @xmath12 is assumed .
correspondingly , the geometry and density profile of a massive wind bubble around a wr star are shown in fig.[fig : bubble ] .
wr stars characterized by high mass - loss rate are believed to form massive dust in their winds ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
interestingly , all dust shells that were observed were essentially associated with carbon - rich wr stars of the latest subtypes of the wc sequence only ( which are called wc stars herein , e.g. , * ? ? ?
* ; * ? ? ?
wc stars are the final evolutionary phase of the most massive ( @xmath13 ) stars , which , in some sense seem to be favored here as the progenitors of some grbs , since there is where echo emission would be most efficient . assuming that an x - ray burst ( i.e. , the x - ray counterpart of a prompt grb produced at the final stage of a wc star ) with a fluence @xmath14 is scattered by the dusty shell at a distance @xmath4 with a scattering optical depth @xmath15 , the echo emission is spread within a characteristic time scale @xmath16 , due to the scattering delay of x - rays with different scattering angles @xmath17 ( see fig.[fig : bubble ] for the geometry of scattering ) .
since the characteristic scattering angle @xmath17 is physically small @xcite , the time scale @xmath16 is consequently much smaller than @xmath18 , where @xmath19 is the speed of light .
therefore , we have a fiducial flux of the x - ray echo emission over a time @xmath20 @xmath21 where @xmath14 is the total fluence of the burst observed in x - ray band ( say , 0.3 - 10 kev ) , the echo flux and the source flux are assumed to suffer the same interstellar absorption on their way to the observer , and the cosmological @xmath22-correction is ignored .
@xmath23}=2.5\times 10^{-7}\,{\rm ergs\,cm^{-2}}$ ] is approximated here , assuming @xmath23}/s_{[15 - 150]}= 1/4 $ ] and @xmath24}=10^{-6}\,{\rm ergs\,cm^{-2}}$ ] , given that @xmath25}/s_{[25 - 50]}\sim 1 $ ] @xcite and most grbs have a flat spectrum in soft x - ray band @xcite .
note that , a realistic @xmath23}$ ] should be essentially larger than the value we adopt here , because the x - rays between [ 0.3 - 10 ] kev before and after the bat trigger interval ( @xmath26 * ? ? ?
* ) , especially those from the soft tail emission in most grbs , are not included in this approximation .
since the scattering angle @xmath17 is very small , we have @xmath27 where @xmath28 is the redshift of the grb , and @xmath17 is the typical scattering angle of x - ray echo emission @xcite .
hence , we have three interesting coincidences : 1 .
the dusty shell at a distance @xmath4 of about tens of parsecs assumed by @xcite is consistent with the massive wind bubble around a wr star at its final stage @xcite ; 2 .
the timescale of the x - ray echo emission on such a wind bubble is about @xmath29 s , which is the typical timescale when the shallow decay phase dominates in x - ray afterglow @xcite ; 3 .
the x - ray echo flux @xmath30 from such a wind bubble is consistent with most of both pre-_swift _ and _ swift _ x - ray afterglows around @xmath31 s ( e.g. , * ? ? ?
* ; * ? ? ?
these three coincidences motivate the study of echo emission as an additional contributor to the x - ray emission .
in what follows , we work out the light curve of the echo emission , and show the fourth and the most spectacular coincidence that the theoretical light curves of x - ray afterglows are consistent with the observed ones .
@xcite have derived the temporal profile of the echo emission .
they showed that the feature in the x - ray light curve of a shallow decay followed by a
normal " decay and a further steepening could be reproduced , given that the grains in the dusty shell have a distribution over size and that the scattering is dependent on the wavelength of x - rays through the rayleigh - gans approximation . according to @xcite
, this theoretical light curve can be formulated as @xmath32\}{1\over t } { \rm d}a{\rm d}e , \label{eq : lightcurve}\end{aligned}\ ] ] where @xmath33 is a constant , @xmath34 is the energy of an x - ray photon , @xmath35 is the size of a grain in the dusty shell , and we define the following functions @xmath36 , \label{eq : s}\ ] ] @xmath37^{-s}\left({a\over0.1\,\micron}\right)^{4-q},\ ] ] @xmath38 @xmath39 @xmath40 where @xmath41 and @xmath42 are the spectral index and the observed peak energy of the source @xcite , @xmath43 and @xmath44 are the power - law indices in the formula of scattering optical depth @xcite , @xmath45 and @xmath46 are the cutoff sizes of dust grains @xcite , @xmath47 $ ] is the energy band of the detector , and @xmath48 is the planck constant ( see the appendix for a detailed derivation ) .
thanks to the _ swift _ satellite @xcite , our interpretation of grbs has become more complex .
the feature of a shallow decay followed by a normal " decay and a further rapid decay of x - ray afterglows appears to be canonical @xcite .
coincidentally , we find that this feature can be naturally reproduced by the echo emission . well - monitored x - ray light curves of 36 grbs ( see fig . [
fig : lightcurve ] ) were fitted by our theoretical model above ( see eq .
[ eq : lightcurve ] ) , using the online data by @xcite .
a total of 14 events have known redshifts ( see tab . [
tab : redshift ] ) while 22 without known redshifts are fixed to be at @xmath49 ( see tab .
[ tab : noredshift ] ) .
we assume @xmath50 kev and @xmath51 kev for _ swift _ xrt @xcite , @xmath52 @xcite , @xmath53 , and @xmath54 kev @xcite .
the other parameters , @xmath43 , @xmath44 , @xmath46 , and @xmath4 are given in tab .
[ tab : redshift ] and tab .
[ tab : noredshift ] .
apparently , @xmath46 is almost a constant here ( i.e. , @xmath55 , except @xmath56 for grb 050505 ) and consistent with most results from optical observations @xcite . here @xmath33 is a numerical coefficient , which is equal to 2@xmath57 as shown in the appendix .
@xmath58 is the coefficient in the integrated grb spectrum between 0.3 and 10 kev , which is not available based on current working frequency of bat onboard _
swift_. similarly , @xmath59 is hardly determined by current observations about the x - ray optical depth .
a further analysis of the parameter @xmath58 and @xmath59 is required once the relevant observational data are sufficient . in fig .
[ fig : lightcurve ] , all the light curves approach to a steep decay ( @xmath60 ) , which is already predicted by eq .
( [ eq : lightcurve ] ) .
since this is a double integral about @xmath34 and @xmath35 , @xmath61 is an independent parameter . approximately , when @xmath62\gg1 $ ] , we have @xmath63\over x^2[\alpha(t)]}- { \cos x[\alpha(t)]\over x[\alpha(t)]}\right\}^2{1\over t } \nonumber \\ & \propto & { \cos^2 x[\alpha(t)]\over x^2[\alpha(t ) ] t } \nonumber \\ & \propto & t^{-2},\end{aligned}\ ] ] where @xmath64 $ ] is a high - frequency periodic function of @xmath61 when @xmath62\gg1 $ ] , and will be independent on @xmath61 after the integration over @xmath34 and @xmath35 .
therefore , letting @xmath62=1 $ ] , we have a critical time @xmath65 after which , the light curves will be approaching the steep decay ( @xmath60 ) . by fitting the light curves
, we can indirectly derive the distant @xmath4 for a given event at redshift @xmath28 ( see table [ tab : redshift ] ) .
these values of @xmath4 appear to be consistent with the observed radii of wind bubbles around wc stars @xcite .
for bursts without known redshifts we only list a pesudo value of the bubble radius @xmath66 assuming a fixed redshift @xmath49 ( see table [ tab : noredshift ] ) .
the larger scatter in @xmath66 values for the latter group appears to be a direct consequence of the arbitrary @xmath49 prescription .
recently , @xcite provided a comprehensive analysis of shallow decays and `` normal '' decay segments in grbs
. they concluded in their sample that most grbs in these two segments have a spectral index @xmath67 around @xmath68 ( as the flux density @xmath69 ) .
here we can also get a knowledge of the spectrum of echo emission based on our analysis ( see eq .
[ eq : fluxdensity ] ) . approximately , when @xmath62\simeq \pi$ ] at a given time @xmath70 , we have @xmath71 where @xmath70 is about an order of magnitude larger than @xmath72 in eq .
( [ eq : td ] ) , and roughly equals most of the @xmath73 s defined in @xcite .
therefore , we should have the spectral index @xmath74 before @xmath70 , since the spectrum of echo emission is softening in the long run . as shown in fig .
[ fig : hr ] , the hardness ratio of echo emission is monotonously decreasing .
interestingly , it reaches a platform ( instead of an extremum proposed by * ? ? ?
* ) almost around @xmath70 and then keeps decreasing very slowly .
if we assume @xmath75 @xcite and @xmath76 @xcite , we should have @xmath77 .
this seems to be inconsistent with the observational result by @xcite , since they ended up with @xmath78 .
however , we have two reasons that lead us to be content with our result based on current analysis . first
, since we are dealing with the prompt x - ray emission and the @xmath75 here is the result from hard x - rays by batse @xcite , we can imagine that the real average @xmath41 should be larger than 0 .
specifically , based on the standard synchrotron emission , @xmath41 should equal @xmath79 , and the synchrotron self - absorption leads to @xmath80 .
second , the rayleigh - gans approximation is adopted here to calculate the differential cross section of scattering .
this approximation overestimates the echo emission at softer x - ray band ( say , in 0.3 - 1.5 kev ) , where absorption is important , which will change the spectral shape in soft x - ray band and alleviate the dilemma of a dramatically small hardness ratio in fig .
[ fig : hr ] . based on the current analysis ,
it is safe to say that -2 is only a lower limit for @xmath67 .
therefore , our result here is generally consistent with the observations .
thus , the echo emission can not be completely ruled out for the reason as discussed by @xcite .
originally , the optical echo emission was used to explain the red bumps observed in some optical afterglows @xcite . however , since the typical optical / infrared scattering angle would be about @xmath81 @xcite , the timescale in eq.([eq : timescale ] ) would be longer , and the optical / infrared echo emission would be much weaker , and hard to be detected @xcite . nevertheless , if the wr wind within a much shorter distance ( @xmath82 cm ) is also taken into account , there might be detectable variations in the optical afterglows due to echo emission @xcite .
however , the average lbv or wr wind bubble around a massive star at its death would be much larger @xcite .
therefore , the features of echo emission in the optical band would be not significant enough , which is possibly the reason that the standard external shock model was more successful in the pre-_swift _ era , when the optical observations dominated .
furthermore , the dramatic variations in the lbv or / and wr winds would drive instabilities that produce radial filaments in the lbv or wr shells @xcite , which could be the origin of some flares in the x - rays afterglows @xcite .
therefore , there might be another timescale due to the angular variation of these filaments , which could be given by @xmath83 where @xmath84 is the distance of the filaments from the progenitor and @xmath85 is the typical angular size of the clumps and ripples in the filaments . to account for the mean ratio of the width and peak time of the x - ray flares @xcite , i.e. , @xmath86 , we may have @xmath87 .
@xcite claimed on a different scattering circumstance ( molecular and atomic matter instead of dust grains ) that the opening angle of jets might be an important factor to the echo light curve , based on the popular assumption that the grb prompt emission is beamed . since only small - angle scattering ( @xmath88 ) for dust grains
is considered here , the typical jet properly pointing at us with a much larger opening angle of about few degrees will not affect out results .
however , the grb jet might be pointing at us with its edge in some case , where our result here would not be valid . instead , a rapid drop that is much steeper than @xmath89 will emerge at a late time @xcite .
this is probably the case for some grbs , e.g. , grb 060526 @xcite . in this paper , we have presented the x - ray echo emission in grb phenomenon and we can draw the following conclusions : 1 .
the shallow decay phase can still be understood without additional contributions from the central engine @xcite .
interestingly , the feature of a shallow decay followed by a normal " decay and a further rapid decay can be reproduced by the echo from the prompt x - ray emission scattered by the massive wind bubble around the progenitor @xcite . in general
, dust formation seems to be prevailing in carbon - rich wr stars of the wc subtype and there is where echo emission would be most efficient .
echo emission can be an additional x - ray source that masked the x - ray jet breaks @xcite .
therefore , the lack of jet breaks @xcite might also be understood via competing echo emission .
in addition , some results in the pre-_swift _ era , say , some correlations relevant for optical breaks might still be valid @xcite .
3 . there is an extra temporal break in x - ray echo emission introduced by @xmath90 above . by fitting the light curves
, we can indirectly measure the locations of the massive wind bubbles , which will bring us closer to finding the mass loss rate , wind velocity , and the age of the progenitors prior to the grb explosions @xcite .
even though some features in x - ray afterglows would be better understood if the echo emission is taken into account , the external shock model is still the dominant scenario for optical / infrared afterglows , since echo emission is negligible in the later .
in addition , the echo emission model does not completely rule out other interpretations ( energy injection , etc . ) in the framework of the fireball - shock model .
technically , echo emission takes place at a distance of @xmath91 tens of parsecs , while the internal or / and external shocks are believed to take place within a distance of @xmath92 cm .
it is more likely that both of them are taking place in the x - ray afterglow , which potentially complicates the light curves .
however , this may raise the problem about grb radiative efficiencies again . even by assuming that the x - ray emission is the forward shock emission , the grb efficiency is already too high to be accommodated by the leading internal shock models ( * ? ? ? * and references therein ) .
assuming an echo - dominated x - ray emission would exacerbate the situation .
we are grateful to the referee for constructive comments and suggestions .
this work made use of data supplied by the uk _ swift _ science data centre at the university of leicester .
we would like to thank bing zhang for helpful discussions and comments .
l.s . would also like to thank rosalba perna for kindly hosting during his study in university of colorado at boulder .
this work is supported by the national natural science foundation of china ( grants 10221001 and 10640420144 ) and the national basic research program of china ( 973 program ) no . 2007cb815404 .
was also supported by the scientific research foundation of graduate school of nanjing university and the state scholarship fund by china scholarship council .
the single - scattering approximation is adopted , which is valid up to a scattering optical depth , @xmath15 , of @xmath910.5 at 1 kev @xcite . therefore , the echo intensity at the earth per unit photon energy per unit size of the grains is ( see fig
. 1 of * ? ? ?
* for the geometry and quantities defined in this appendix ) @xmath93\tau_a ( e){{\rm d}\sigma\over \sigma_{\rm sca } { \rm d}\omega } , \label{eq : intensity}\end{aligned}\ ] ] where @xmath94 $ ] is the flux density of the initial grb before a time delay @xmath95 , which is also dependent on the viewing angle @xmath96 , @xmath97 is the frequency - dependent scattering optical depth per unit size of the grains , and the last term is the fraction of the scattered photons , which go into the line of sight in a unit solid angle .
the zero time point is set as the trigger of a grb on the earth .
@xmath98 is the azimuthal angle along the line of sight .
since the initial grb is usually much shorter than the long - term afterglow , we can approximate the initial grb light curve as a pulse .
we also assume that the initial grb spectrum is given by a band spectrum in the x - ray band @xcite , so we have @xmath99&=&a_1\left({e\over{100\,{\rm kev } } } \right)^\delta { \rm exp}\left[-{(\delta+1)e\over e_{\rm p}}\right]\times \delta[t - t_{\rm d}(\theta)],\end{aligned}\ ] ] where we only need the segment of the band spectrum in the lower energy band .
@xmath58 is a constant , and the time delay from a certain viewing angle @xmath96 is given by @xcite @xmath100 here the dirac delta function of @xmath61 can be translated into a function of @xmath96 , using the special properties of its own , @xmath101 = { c d_{\rm ds}\over ( 1+z)d_{\rm d } d_{\rm s } \theta}\delta[\theta-\hat{\theta}(t)],\end{aligned}\ ] ] where the function @xmath102 is defined as @xmath103^{1/2}$ ] ( see the appendix of * ? ? ?
we also assume the frequency - dependent scattering optical depth per unit size of the grains is @xmath104^ { -s}\left({a\over0.1\,\micron}\right)^{4-q } , \end{aligned}\ ] ] where @xmath59 is a constant with a dimension of @xmath105 $ ] .
this is valid only in a range of the size of the grains , say , @xmath106 $ ] ( * ? ? ?
* and the references therein ) . for the last term in eq .
( [ eq : intensity ] ) , we have @xmath107 , since @xmath96 is very small
. we can do some rearrangement @xmath108 where @xmath109 is the total scattering cross - section , @xmath110 is defined since @xmath17 is also very small , and we have @xmath111 in geometry . neglecting the chemical composition or shape of the grains , the rayleigh - gans approximation is valid for the differential cross - section @xcite @xmath112 where @xmath113 , and @xmath114 is the first - order spherical bessel function .
since the wind bubble is very close to the grb , we have @xmath115 and @xmath116 .
therefore , the echo flux density at the earth per unit size of the grains is @xmath117\}{1\over t},\end{aligned}\ ] ] where a is a numerical constant , and @xmath118 , @xmath119 , @xmath120 , @xmath121 , and @xmath122 are given by eqs .
( [ eq : s])-([eq : alpha ] ) .
so , the flux density and the flux of echo emission at earth will be @xmath123 and @xmath124 respectively .
alcock , c. , & hatchett , s. 1978 , , 222 , 456 allen , d. a. , harvey , p. m. , & swings , j. p. 1972
, , 20 , 333 band , d. l. et al .
1993 , , 413 , 281 burrows , d. n. , & racusin , j. 2007 , procs . of the swfit and grbs : unveiling the relativistic universe " conf . , venice ( astro - ph/0702633 ) burrows , d. n. , et al .
2007 , philosophical transactions a , in press ( astro - ph/0701046 ) burrows , d. n. , et al . 2006 , proceedings of the x - ray universe conference , ( esa sp-604 ; noordwijk : esa ) , 877 castor , j. , mccray , r. , & weaver r. 1975 , , 200 , l107 chincarini , g. , et al .
2007 , , in press ( astro - ph/0702371 ) chu , y .- h .
, weis , k. , & garnett , d. r. 1999 , , 117 , 1433 costa , e. 1999 , a&as , 138 , 425 crowther , p. a. 2003 , , 45 , 177 crowther , p.
a. 2003 , , 285 , 677 dai , x. , et al .
2007 , , 658 , 509 dai , z. g. 2004 , , 606 , 1000 dai , z. g. , & lu , t. 1998 , , 333 , l87 dwarkadas , v. v. 2007 , , 667 , 226 evans , p. a. , et al .
2007 , , 469 , 379 esin , a. a. , & blandford , r. d. 2000 , , 534 , l151 fan , y. z. , & piran t. 2006 , , 369 , 197 frail , d. a. , et al .
2001 , , 562 , l55 garca - segura , g. , & mac low , m. m. 1995a , , 455 , 145 garca - segura , g. , & mac low , m. m. 1995b , , 455 , 160 garca - segura , g. , et al .
1996a , , 305 , 229 garca - segura , g. , et al .
1996b , , 316 , 133 gehrels , n. , et al .
2004 , , 611 , 1005 gehrz , r. d. , & hackwell , j. a. 1974 , , 194 , 619 ghirlanda , g. , et al .
2004 , , 616 , 331 heng , k. , lazzati , d. , & perna , r. 2007 , , 662 , 1119 klose , s. 1998 , , 507 , 300 langer , n. , et al .
1994 , , 290 , 819 liang , e. w. , zhang , b. b. , & zhang , b. 2007a , , in press ( arxiv : 0704.1373 ) liang , e. w. , et al .
2007b , , submitted ( arxiv : 0708.2942 ) liang , e. w. , & zhang , b. 2005 , , 633 , 611 marston , a. p. 1997
, , 475 , 188 mathis , j. s. , rumpl , w. , & nordsieck , k. h. 1977 , , 217 , 425 mauche , c. w. , & gorenstein , p. 1986
, , 302 , 371 mszros , p. , & gruzinov , a. 2000 , , 543 , l35 mszros , p. 2006
phys . , 69 , 2259 mirabal , n. , et al .
2003 , , 595 , 935 miralda - escud , j. 1999 , , 512 , 21 moran , j. a. , & reichart , d. e. 2005 , , 632 , 438 nousek , j. a. , et al .
2006 , , 642 , 389 overbeck , j. w. 1965 , , 141 , 864 panaitescu , a. , et al .
2006 , , 369 , 2059 panaitescu , a. 2007 , , 380 , 374 predehl , p. , & klose , s. 1996 , , 306 , 283 predehl , p. , & schmitt , j. h. m. m. 1995 , , 293 , 889 preece , r. d. , et al .
2000 , , 126 , 19 ramirez - ruiz , e. , et al .
2001 , , 327 , 829 rees , m. , & mszros , p. 1998 , , 496
, l1 reichart , d. e. 2001 , , 554 , 643 rhoads , j. e. 1999 , , 525 , 737 sari , r. , & mszros , p. 2000 , , 535
, l33 sakamoto , t. , et al .
2007 , , in press ( arxiv : 0707.4626 ) sato , g. , et al .
2007 , , 657 , 359 sazonov , s. y. , & sunyaev , r. a. 2003 , , 399 , 505 shao , l. , & dai , z. g. 2007 , , 660 , 1319 white , r. l. 1979 , , 229 , 954 williams , p. m. , van der hucht , k. a. , & th , p. s. 1987 , , 182 , 91 woosley , s. e. , & bloom , j. s. 2006 , , 44 , 507 zhang , b. 2007 , chjaa , 7 , 1 zhang , b. , et al .
2007 , , 655 , 989 zhang , b. , et al . 2006 , , 642 , 354 zhang , b. , & mszros , p. 2001
, , 552 , l35 lcccccccc grb 050319 & 3.240 & @xmath125 & @xmath126 & 2.5 & 4.4 & 0.25 & 75 & 1.72 + grb 050505 & 4.27 & @xmath127 & @xmath126 & 2.0 & 4.0 & 0.5 & 50 & 1.42 + grb 050814 & 5.3 & @xmath128 & @xmath129 & 3.0 & 4.5 & 0.25 & 55 & 3.53 + grb 051022 & 0.8 & @xmath130 & @xmath129 & 2.0 & 3.1 & 0.25 & 30 & 2.85 + grb 060210 & 3.91 & @xmath127 & @xmath131 & 2.0 & 3.0 & 0.25 & 200 & 1.78 + grb 060218 & 0.033 & @xmath130 & @xmath129 & 3.0 & 3.1 & 0.25&50 & 1.92 + grb 060502a & 1.51 & @xmath132 & @xmath126 & 3.0 & 4.3 & 0.25&65 & 1.26 + grb 060512 & 0.4428 & @xmath127 & @xmath133 & 2.0 & 3.0 & 0.25&45 & 2.63 + grb 060707 & 3.425 & @xmath128 & @xmath126 & 3.0 & 4.5 & 0.25&170 & 2.53 + grb 060714 & 2.711 & @xmath134 & @xmath129 & 2.0 & 4.0 & 0.25&95 & 2.61 + grb 060814 & 0.84 & @xmath135 & @xmath129 & 2.0 & 4.0 & 0.25 & 34 & 2.33 + grb 060729 & 0.54 & @xmath136 & @xmath137 & 3.0 & 5.5 & 0.25&45 & 2.93 + grb 061110a & 0.758 & @xmath138 & @xmath139 & 3.0 & 4.5 & 0.25 & 76 & 2.61 + grb 061121 & 1.314 & @xmath140 & @xmath126 & 2.0 & 4.0 & 0.25&40 & 2.30 + [ tab : redshift ] lcccccccc grb 050712 & @xmath132 & @xmath126 & 2.3 & 4.3 & 0.25 & 55 & 2.75 + grb 050713a & @xmath132 & @xmath126 & 2.0 & 4.0&0.25 & 45 & 2.67 + grb 050713b & @xmath136 & @xmath129 & 3.0 & 4.5 & 0.25 & 70 & 2.21 + grb 050802 & @xmath134 & @xmath129 & 2.0 & 3.1 & 0.25 & 30 & 4.64 + grb 050822 & @xmath128 & @xmath141 & 3.0 & 4.5 & 0.25 & 80 & 2.19 + grb 050915b & @xmath128 & @xmath142 & 3.0 & 4.8 & 0.25 & 70 & 1.60 + grb 051008 & @xmath127 & @xmath143 & 2.5 & 3.5 & 0.25 & 10 & 1.93 + grb 051016a & @xmath132 & @xmath144 & 2.3 & 4.0 & 0.25 & 80 & 6.44 + grb 060204b & @xmath125 & @xmath143 & 2.0 & 3.1 & 0.25 & 25 & 1.64 + grb 060219 & @xmath140 & @xmath143 & 2.5 & 3.5 & 0.25 & 45 & 1.04 + grb 060306 & @xmath134 & @xmath133 & 2.0 & 4.0 & 0.25 & 50 & 1.41 + grb 060428a & @xmath134 & @xmath145 & 2.7 & 4.2 & 0.25 & 95 & 2.26 + grb 060510a & @xmath146 & @xmath142 & 3.0 & 4.5 & 0.25 & 10 & 2.28 + grb 060604 & @xmath127 & @xmath129 & 2.0 & 4.0 & 0.25 & 80 & 1.86 + grb 060708 & @xmath140 & @xmath129 & 2.0 & 4.0 & 0.25 & 50 & 2.03 + grb 060813 & @xmath147 & @xmath148 & 2.0 & 3.1 & 0.25 & 10 & 2.68 + grb 061004 & @xmath134 & @xmath149 & 2.0 & 4.0 & 0.25 & 15 & 2.61 + grb 061222a & @xmath140 & @xmath126 & 2.1 & 4.2 & 0.25 & 30 & 1.90 + grb 070129 & @xmath128 & @xmath126 & 3.0 & 4.5 & 0.25 & 90 & 1.63 + grb 070328 & @xmath146 & @xmath144 & 1.9 & 4.0 & 0.25 & 5 & 4.75 + grb 070419b & @xmath132 & @xmath143 & 2.0 & 3.1&0.25 & 20 & 2.74 + grb 070420 & @xmath134 & @xmath143 & 3.0 & 3.1 & 0.25 & 15 & 3.07 + [ tab : noredshift ] | we investigate the effect of x - ray echo emission in gamma - ray bursts ( grbs ) . we find that the echo emission can provide an alternative way of understanding x - ray shallow decays and jet breaks . in particular , a shallow decay followed by a
normal " decay and a further rapid decay of x - ray afterglows can be together explained as being due to the echo from prompt x - ray emission scattered by dust grains in a massive wind bubble around a grb progenitor .
we also introduce an extra temporal break in the x - ray echo emission . by fitting the afterglow light curves
, we can measure the locations of the massive wind bubbles , which will bring us closer to finding the mass loss rate , wind velocity , and the age of the progenitors prior to the grb explosions . |
a catalan problem is any enumerative problem that produces the catalan sequence of numbers or one of its many @xmath3-analogs .
stanley @xcite provides a catalog of @xmath4 catalan problems .
interestingly , many of the generating functions that arise from these problems can be given as a continued fraction with a simple yet elegant form .
two of these generating functions are reproduced below and a third we derive anew .
our intent is to show that the first two continued fractions are special instances of the third with the implication that many others are as well .
we begin with the three catalan problems and their corresponding generating functions .
* problem 1*. a @xmath0 pattern ( respectively , a @xmath5 pattern ) in a permutation @xmath6 of length @xmath7 is a triple @xmath8 of indices for which @xmath9 ( respectively , @xmath10 ) . let @xmath11 denote the number of permutations @xmath6 of length @xmath7 that have no @xmath0 patterns and exactly @xmath12 @xmath5 patterns .
recently , robertson , wilf and zeilberger ( @xcite ) derived the generating function @xmath13 in which the @xmath14th numerator is @xmath15 ( @xmath14 is for _ level _ in anticipation of problem 3 ) . *
problem 2*. the number of lattice paths from @xmath16 to @xmath17 with steps @xmath18 and @xmath19 that never rise above the line @xmath20 is a catalan number .
let @xmath21 be such a path , @xmath22 the area under the path ( and above the @xmath23-axis ) , and let @xmath24 .
then a generating function is given by ( see exercise 6.34 in stanley @xcite and replace the @xmath23 therein with @xmath25 ) @xmath26 in which the @xmath14th numerator is @xmath27 .
* problem 3*. the number of ordered trees ( also known as plane trees ) on @xmath7 edges is a catalan number .
the _ level _ of a vertex is the number of edges on the unique path from the root to the vertex .
thus , the root is the unique vertex at level zero and the vertices at level one are adjacent to the root .
let @xmath28 be the number of ordered trees that have @xmath29 vertices at level @xmath30 and let @xmath31 , @xmath32 , be indeterminates .
the generating function @xmath33 that enumerates ordered trees by the number of vertices at each level is defined as @xmath34 the first few terms ( number of edges @xmath35 ) of @xmath33 are @xmath36 that @xmath33 can be written as a continued fraction that subsumes the previous continued fractions is our main result .
it is simple , yet has some interesting applications .
[ t : main ] the generating function that enumerates ordered trees by the number of vertices at each level is @xmath37 we exploit the natural recursive property of ordered trees to obtain a recursion for @xmath33 .
the recursion immediately leads to the continued fraction .
any ordered tree on more than one vertex can be constructed from a collection of others ( the subtrees ) by joining the roots of these subtrees to a new vertex .
the new vertex becomes the root of the tree under construction .
note that the level of a vertex in a subtree _ increases by one _ after the new root is inserted .
the function @xmath38 enumerates the choices for a subtree and each of these choices contributes a factor of @xmath39 because of the level changes .
the factor of @xmath40 is present because the root of the subtree becomes a vertex at level one .
thus , the trees with @xmath1 subtrees ( of the root ) are enumerated by @xmath41 . the generating function
satisfies @xmath42 iteration of the last functional recursion produces the continued fraction .
an immediate application is obtained by replacing each indeterminate @xmath31 with @xmath43 denoted simply as @xmath44 .
the resulting function enumerates ordered trees by the number of edges and is @xmath45 the well - known solution of the above generates the catalan numbers and is @xmath46 .
the more challenging applications are the evaluations needed to produce the continued fractions of permutations ( equation [ e : gen1 ] of problem 1 ) and lattice paths ( equation [ e : gen2 ] of problem 2 ) .
both applications require that we map their respective problems to an ordered - tree problem .
these mappings are of interest in their own right and we explore them now .
we begin with the lattice path problem because it is simpler and the mapping is already known .
we draw our ordered trees with the root at the top and proceed downward to the leaves .
the first leaf is the leftmost leaf in the drawing and the remaining leaves are referred to by their positions in a left - to - right order . a preorder ( depth - first ) traversal of the ordered tree provides a well - known correspondence with a lattice path . when an edge of the tree is traversed downward away from the root
we take a @xmath18 ( east ) step in the lattice , otherwise we take a @xmath19 ( north ) step . in this manner ,
an ordered tree with @xmath7 edges corresponds to a unique lattice path from @xmath16 to @xmath17 .
if the path @xmath21 corresponds to the tree @xmath33 , then we need to determine what statistic of the tree corresponds to the area @xmath22 under the path .
we let @xmath47 be this statistic of the tree and claim that it depends only on the vertex levels .
[ l : paths ] if @xmath33 is an ordered tree on @xmath7 edges , then @xmath48 where @xmath49 is the level of vertex @xmath50 .
let @xmath51 be the rightmost leaf of the tree @xmath33 .
our immediate interest is to calculate the area under the east step that arises in the lattice path by traversing the last edge to this leaf downward away from the root .
this area is equal to the height that the east step has in the lattice path and is equal to the number of north steps that have occurred prior to the east step .
there are a total of @xmath7 north steps in the lattice path ( one for each edge of the tree ) and @xmath52 remaining north steps after the east step .
hence , the east step is at height @xmath53 in the lattice .
if we now delete the leaf @xmath51 from the tree @xmath33 , then the resulting tree has a lattice path with area @xmath53 less than that of @xmath33 .
a formal inductive argument on the number of edges provides the result .
we use the lemma to prove the following continued - fraction result .
the result is the same as that given in equation [ e : gen2 ] .
[ t : area ] if @xmath54 enumerates the set of ordered trees on @xmath7 edges by the area under their corresponding lattice paths , then @xmath55 in which the @xmath14th numerator is @xmath27 .
let @xmath33 be an ordered tree on @xmath7 edges and assign to every nonroot vertex at level @xmath56 the value @xmath27 .
the product of all these values is then @xmath57 .
summing over all ordered trees on @xmath7 edges we have by the lemma @xmath58 the sum of these over all @xmath59 then enumerates ordered trees and the generating function is given by the continued fraction of theorem [ t : main ] with @xmath60 .
the previous problem used an existing bijection between the set of ordered trees and the set of lattice paths to get the desired result . we seek a similar approach for ordered trees and permutations .
there are many ways to map a permutation onto a tree ( often an unordered tree ) but none of these serve our needs .
the mapping we introduce appears to be new .
let @xmath33 be an ordered tree on @xmath7 edges .
we use a preorder traversal of @xmath33 to label the nonroot vertices in decreasing order with the integers @xmath61 .
thus , the first vertex visited gets the label @xmath7 and the last receives @xmath62 .
we now construct a permutation written as a word by _ reading _ the labeled tree in _ postorder_. we again traverse the tree from left to right and record the label of a vertex when we last visit it .
in catalan fashion , the five ordered trees and their corresponding permutations are shown in figure [ f : one ] . note that the only permutation missing from those of length three is @xmath63 . the ( 132 ) pattern has been avoided .
also note that there is exactly one permutation with a ( 123 ) pattern ( the first permutation shown ) .
thus , recalling the definition of @xmath11 given in the introduction we have @xmath64 , @xmath65 , and @xmath66 .
we generalize these observations after introducing some useful notation .
if @xmath33 is an ordered tree on @xmath7 edges , then we let @xmath67 be its corresponding permutation written as a word on the numbers @xmath68 .
we let @xmath69 be the same permutation except we use the corresponding numbers @xmath70 . for example , the first tree shown in the figure has @xmath71 and @xmath72 . for emphasis
, we denote the concatenation of two words @xmath6 and @xmath73 as @xmath74 ( instead of the usual @xmath75 ) .
the following lemma describes how the permutation of a tree can be constructed from those of its subtrees .
note that @xmath78 is the total number of vertices to be labeled , that @xmath81 is the total number of vertices to be labeled after those of the subtree @xmath82 , and so on .
since the vertices of @xmath33 are labeled in decreasing order using a preorder traversal , @xmath83 implies that all vertices of @xmath84 receive labels greater than those of @xmath85 .
since @xmath67 is constructed by reading these labels in postorder , @xmath83 implies that the vertex labels in @xmath84 appear in @xmath67 before any of those in @xmath85 .
thus , @xmath67 begins as @xmath86 with each number incremented by @xmath81 , i.e. , as @xmath87 .
the root of this subtree is the first vertex visited in the preorder traversal and receives the label @xmath88 .
it is read last among the vertices of @xmath82 in postorder , however , so that @xmath67 begins as @xmath89 .
the general case follows similarly .
note that the lemma also provides the means to prove that the mapping @xmath90 is injective . that it is onto the set of all @xmath0-pattern avoiding permutations
is proved in the next theorem . before proceeding to this theorem we present another lemma which enables us to count @xmath5 patterns and their generalization .
an _ increasing _ pattern of _ length _
@xmath1 , @xmath32 , in a permutation @xmath6 of length @xmath7 is a @xmath1-tuple @xmath91 of indices for which @xmath92 .
[ l : increase ] let @xmath33 be an ordered tree on @xmath7 edges and @xmath93 a subset of @xmath1 vertices , @xmath94 , none of which are the root , then the labels of these vertices provide an increasing pattern of length @xmath1 in @xmath67 if and only if they lie along a path from the root to some leaf .
we induct on @xmath7 . if @xmath95 , then @xmath96 and the lemma is obvious .
in fact , this is the case for all @xmath7 whenever @xmath96 .
we assume that the lemma is true for any ordered tree on @xmath7 or fewer edges , @xmath97 , and let @xmath33 be a tree on @xmath98 edges .
suppose that @xmath67 contains an increasing pattern of length @xmath99 .
we let @xmath100 be the vertex in @xmath33 that provides the @xmath101 in such a pattern , @xmath102 .
then @xmath31 must receive a larger label than @xmath40 implies that the subtree containing @xmath31 precedes or is the same as that of @xmath40 ( see the proof of lemma [ l : perm ] ) .
however , the label of @xmath31 must be read after that of @xmath40 implies that its subtree must follow or be the same as that of @xmath40 .
the conclusion is that they are in the same subtree .
a similar argument applies to @xmath100 and @xmath40 , @xmath103 , so that all the vertices must be in the same subtree , say @xmath84 .
thus , the increasing pattern of length @xmath1 lies entirely within @xmath104 .
recall that it is the root of the subtree that receives the label @xmath105 .
if the root of the subtree is one of the vertices providing the increasing pattern , then it must be @xmath31 .
we must consider two cases depending on whether the subtree root is @xmath31 or not .
if it is not , then the pattern lies entirely within @xmath106 and corresponds uniquely to an increasing pattern of length @xmath1 in @xmath107 . by the inductive hypothesis , the vertices @xmath108 must lie along some path from the root of the subtree to a leaf . necessarily
, this is also a path from the root of @xmath33 to a leaf as required by this lemma .
if the root of the subtree is @xmath31 , then @xmath109 provide the labels for an increasing pattern of length @xmath110 in @xmath106 .
this pattern corresponds to a unique increasing pattern of length @xmath110 in @xmath107 and again by hypothesis the vertices @xmath109 must lie along a path from the root of the subtree to some leaf of the subtree .
this path together with the root of the subtree provides the needed path of this lemma .
conversely , if @xmath111 is a subset of nonroot vertices of @xmath33 , where we may assume the label of @xmath100 is less than that of @xmath112 whenever @xmath83 , and the vertices lie along a path from the root to a leaf , then @xmath100 lies below @xmath112 on this path whenever @xmath83 .
thus , when the labels are read in postorder , the label of @xmath100 is read prior to that of @xmath112 , @xmath83 .
the vertices then provide an increasing pattern of length @xmath1 in @xmath67 .
the two previous lemmas enable us to prove the following interesting combinatorial theorem .
its corollary establishes a continued fraction as the generating function of @xmath0-avoiding permutations by number of increasing patterns of length @xmath1 .
[ t : perm ] a permutation @xmath6 avoids the @xmath0 pattern if and only if @xmath113 for some tree @xmath33 .
if this is the case , then the number of increasing patterns of length @xmath1 depends only on the levels of the vertices in the tree and is given by @xmath114 .
suppose that @xmath67 contains a @xmath0 pattern and that @xmath33 is among the smallest such trees .
we let @xmath115 be a vertex in @xmath33 that provides the @xmath116 in such a pattern and let @xmath117 be the vertices that provide the corresponding @xmath62 and @xmath118 , respectively .
then @xmath115 must receive a larger label than @xmath40 implies that the subtree containing @xmath115 precedes or is the same as that of @xmath40 ( see the proof of lemma [ l : perm ] ) .
however , the label of @xmath115 must be read after that of @xmath40 implies that its subtree must follow or be the same as that of @xmath40 .
the conclusion is that they are in the same subtree .
a similar argument applies to @xmath119 and @xmath40 so that all three vertices must be in the same subtree , say @xmath84 .
also note that none of them can be the root of the subtree since the root receives the largest label among the vertices of the tree and , hence , can not be @xmath40 or @xmath119 .
its label appears later in @xmath67 than the others in the subtree implies that the root can not be @xmath115 .
thus , the @xmath0 pattern lies entirely within @xmath106 which implies that @xmath107 itself must have a @xmath0 pattern contradicting our choice of @xmath33 . since the number of @xmath0-pattern avoiding permutations of length @xmath7 and the number of ordered trees on @xmath7 edges are the same catalan number , the mapping @xmath120 is a bijection between these sets .
it is an instructive exercise to construct @xmath33 from a @xmath0-pattern avoiding permutation .
it remains to determine the number of increasing patterns of length @xmath1 in @xmath121 . as a result of lemma [ l : increase ]
it is only necessary to count the number of vertex subsets of size @xmath1 , none of which are the root , such that the vertices lie along a path from the root to a leaf .
we claim this number is @xmath122 as stated . to see this ,
let @xmath50 be any vertex of @xmath33 and choose @xmath123 .
there are @xmath124 nonroot vertices other than @xmath50 along the unique path from the root to @xmath50 . from these we select any @xmath110 of them , which together with @xmath50 , form the required subset .
it is clear that every subset with the required properties arises this way and we are done .
we now use the theorem to write a generating function as a continued fraction .
we let @xmath125 denote the number of permutations of length @xmath7 that have no @xmath0 pattern and exactly @xmath12 increasing patterns of length @xmath1 .
the case @xmath2 is that considered by robertson , wilf and zeilberger @xcite .
[ c : perm ] the generating function that enumerates @xmath0-pattern avoiding permutations of length @xmath7 by number of increasing patterns of length @xmath1 is @xmath126 in which the @xmath14th numerator @xmath127 is @xmath128 .
let @xmath33 be an ordered tree on @xmath7 edges and assign to every nonroot vertex at level @xmath56 the value @xmath129 .
the product of all these values is then @xmath130 and the result follows from the previous theorem and theorem [ t : main ] . | we find a generating function expressed as a continued fraction that enumerates ordered trees by the number of vertices at different levels .
several catalan problems are mapped to an ordered - tree problem and their generating functions also expressed as a continued fraction . among these problems is the enumeration of @xmath0-pattern avoiding permutations that have a given number of increasing patterns of length @xmath1 .
this extends and illuminates a result of robertson , wilf and zeilberger for the case @xmath2 . |
many systems of scientific or practical interest are decomposable into subsystems with strong internal and relatively weak external interactions @xcite ; for example , there are groups of friends or collaborators in social networks , sets of topically related documents in hypertexts , or blocs of interlocked countries in international trade .
if systems are modeled as networks , with the system elements as vertices and their interactions as edges , then each subsystem corresponds to a so - called _ community _ , a set of vertices with dense internal connections but sparse connections to the remaining network .
two widely used representations of networks are _ layouts _ , which assign the vertices to positions in a metric space , and _
clusterings _ , which partition the vertex set into disjoint subsets . both representations can group densely connected vertices , by placing them at nearby positions or in the same cluster , and separate sparsely connected vertices , by placing them at distant positions or in different clusters , and can thus naturally reflect the community structure .
requirements like the grouping of densely connected vertices are often formalized as mathematical functions called _ quality measures _ , and the optimization of quality measures is a common strategy for the computation of both layouts @xcite and clusterings @xcite . despite these commonalities , and
although layouts and clusterings are often used together as complementary representations of the same network , there is no coherent understanding of layout quality and clustering quality .
this paper unifies newman and girvan s modularity @xcite , a popular quality measure for clusterings , with energy models of pairwise attraction and repulsion between vertices ( e.g. , @xcite ) , a widely used class of quality measures for layouts . after an introduction of the quality measures in sec .
[ s : def ] , sec .
[ s : density ] shows that layouts with optimal energy and clusterings with optimal modularity represent the community structure similarly , and sec .
[ s : unification ] demonstrates that modularity actually _ is _ an energy model of pairwise attraction and repulsion , if clusterings are considered as restricted layouts .
section [ s : appl ] discusses the application of these results for computing consistent clusterings and layouts .
quality measures for representations of networks formalize what is considered as a _ good _ representation , and allow to compute good representations automatically using optimization algorithms .
mathematically , a quality measure maps network representations to real numbers , such that larger ( or smaller ) numbers are assigned to better representations , and the best representations correspond to maxima ( or minima ) of the measure .
this section introduces two widely used quality measures , namely energy models based on pairwise attraction and repulsion for layouts , and newman and girvan s modularity measure for clusterings . to obtain uniform and general formulations ,
both measures are defined for _ weighted _ networks . in a weighted network ,
each vertex @xmath0 has a nonnegative real _ vertex weight _
@xmath1 , and each unordered vertex pair @xmath2 ( including @xmath3 ) has a nonnegative real _ edge weight _
@xmath4 . intuitively , a vertex ( or edge ) of weight @xmath5 can be thought of as a chunk of @xmath5 vertices ( or edges ) of weight @xmath6 . the commonly studied
un_weighted networks correspond to the special case where the edge weights are either @xmath7 ( no edge ) or @xmath6 , and the vertex weights are @xmath6 .
@xmath9-dimensional layout _
@xmath10 of a network maps each vertex @xmath0 to a position @xmath11 in @xmath12 ; it thereby assigns a _ distance _ to each vertex pair @xmath2 , namely the euclidean distance @xmath13 between the respective vertex positions .
so - called _ energy models _ are an important class of quality measures for layouts . in general , _ smaller _ energy indicates better layouts . because force is the negative gradient of energy , energy models can also be represented as force systems , and energy minima correspond to force equilibria . for introductions to energy - based or force - directed layout ,
see refs .
@xcite .
the most popular energy models for general undirected networks are either similar to stress functions of multidimensional scaling @xcite , or represent force systems of pairwise attraction and repulsion between vertices .
models of the former type ( e.g. , @xcite ) enforce that the distance of each vertex pair in the layout approximates some prespecified distance , most commonly the length of the shortest edge path between the vertices .
they will not be further discussed , because their layouts reflect these path lengths rather than the community structure . in models of the latter type ,
adjacent vertices attract , which tends to group densely connected vertices , and all pairs of vertices repulse , which tends to separate sparsely connected vertices .
the strengths of the forces are often chosen to be proportional to some power of the distance .
formally , for a layout @xmath10 and two vertices @xmath14 with @xmath15 , the attractive force exerted on @xmath16 by @xmath0 is @xmath17 and the repulsive force exerted on @xmath16 by @xmath0 is @xmath18 where @xmath13 is the distance between @xmath16 and @xmath0 , @xmath19 is the unit - length vector pointing from @xmath16 to @xmath0 , and @xmath20 and @xmath21 are real constants with @xmath22 .
the condition @xmath23 ensures that the attractive force between connected vertices grows faster than the repulsive force , and thus prevents infinite distances except between unconnected components .
for most practical force models holds @xmath24 and @xmath25 , i.e. , the attractive force is non - decreasing and the repulsive force is non - increasing with growing distance . in the widely used force model of fruchterman and reingold @xcite , @xmath26 and @xmath27 . by exploiting that force is the negative gradient of energy , the force model can be transformed into an energy model , such that force equilibria correspond to ( local ) energy minima . for a layout @xmath10 and constants
@xmath28 with @xmath22 , the _ @xmath8-energy _ is @xmath29 where @xmath30 must be read as @xmath31 ( because @xmath32 is the derivative of @xmath33 ) .
the @xmath34-energy model has been proposed by davidson and harel @xcite , and the @xmath35-energy model is known as linlog model @xcite . a _ clustering _ @xmath36 of a network partitions the vertex set into disjoint subsets called _ clusters _ , and thereby maps each vertex @xmath0 to a cluster @xmath37 .
proposals of quality measures for clusterings are numerous and scattered over the literature of diverse research fields ; surveys , though non - exhaustive , are provided by refs .
@xcite .
one of the most widely used quality measures was introduced by newman and girvan , and is called modularity .
it was originally defined for the special case where the edge weights are either @xmath7 or @xmath6 and the weight of each vertex is its degree @xcite , and was later extended to networks with arbitrary edge weights @xcite .
( the _ degree _ of a vertex is the total weight of its incident edges , with the edge weight from the vertex to itself counted twice . )
generalized to arbitrary vertex weights , the _ modularity _ of a clustering @xmath36 is @xmath38 where @xmath39 is the set of all vertices in the network , and @xmath40 is the set of clusters ; the weight functions are naturally extended to sets of vertices or edges : @xmath41 is the total edge weight within the cluster @xmath42 , and @xmath43 is the total weight of the vertices in @xmath42 . intuitively , the first term of the modularity measure is the _ actual _ fraction of intra - cluster edge weight . in itself , it is not a good measure of clustering quality , because it takes the maximum value for the trivial clustering where one cluster contains all vertices .
this is corrected by subtracting a second term , which specifies the _ expected _ fraction of intra - cluster edge weight in a network with uniform density .
thus modularity takes positive values for clusterings where the total edge weight within clusters is larger than would be expected if the network had no community structure .
finding a minimum - energy layout or a maximum - modularity clustering of a given network is computationally hard ; in particular , modularity maximization was recently shown to be np - complete @xcite . in practice
, energy and modularity are almost exclusively optimized with heuristic algorithms that do not guarantee to find optimal or near - optimal solutions .
an extensive experimental comparison of energy minimization algorithms for network layout was performed by hachul and jnger @xcite ; however , most of the examined algorithms make fairly restrictive assumptions about the optimized energy model .
more general and reasonably efficient is the force calculation algorithm by barnes and hut @xcite , whose runtime is @xmath44 per iteration for a network with @xmath45 edges ( with nonzero weight ) and @xmath46 vertices ( assuming that the number of dimensions is small and the vertex distances are not extremely nonuniform ) .
the number of iterations required for convergence typically grows sublinearly with @xmath46 .
clustering algorithms for networks are surveyed in refs.@xcite .
a relatively fast yet very effective heuristic for modularity maximization is agglomeration by iteratively merging clusters ( starting from singletons ) , combined with single - level @xcite or multi - level @xcite refinement by iteratively moving vertices ; an efficient implementation requires a runtime of @xmath47 ( assuming @xmath48 hierarchy levels in agglomeration and @xmath48 iterations through all vertices per level in refinement ) .
a set of vertices is called a _ community _ if the density within the set is significantly larger than the density between the set and the remaining network .
density between _ two disjoint sets of vertices @xmath49 and @xmath50 is intuitively the quotient of the actual edge weight and the potential edge weight between @xmath49 and @xmath50 ; formally , it is defined as @xmath51 , where @xmath52 is the total weight of the vertices in @xmath50 , and @xmath53 is the total edge weight between @xmath49 and @xmath50 .
similarly , the _ density within _ a vertex set @xmath50 is @xmath54 .
( this generalizes standard definitions of density from graph theory @xcite to weighted networks with self - edges . )
existing theoretical results , which will be summarized and extended in this section , already show that the community structure of a network is reflected in layouts with optimal @xmath8-energy ( for certain values of @xmath20 and @xmath21 ) and in clusterings with optimal modularity .
what has previously escaped notice is the striking analogy : _ the separation of communities in an optimal layout is inversely proportional to ( some power of ) the density between them , and the separation of communities in an optimal clustering reflects whether the density between them is smaller than a certain threshold . _ as an important limitation , the result for layouts will be derived only for two communities , and can not be expected to hold precisely for more communities .
therefore , the consistency of @xmath8-energy layouts and modularity clusterings will be revisited in sec .
[ s : appl ] , after further evidence has been presented in sec .
[ s : unification ] . in what appears to be the only previous work that formally relates energy - based layout to modularity clustering @xcite
, we did not established similarities between optimal layouts and optimal clusterings , but only noted that the modularity measure is mathematically similar to the density ( called normalized cut in @xcite ) , as both normalize the actual edge weight with a potential or expected edge weight .
this subsection discusses how the distances in a layout with optimal @xmath8-energy can be interpreted in terms of the community structure of the network , and how this interpretation depends on the parameters @xmath20 and @xmath21 . for the simple case of a network with two vertices ,
the minimum - energy layouts can be computed analytically ( theorem 3 in @xcite ) .
if the vertices @xmath16 and @xmath0 have the distance @xmath9 , the @xmath8-energy is @xmath55 the derivative of this function is @xmath7 at its minimum @xmath56 , thus @xmath57 thus the distance of the two vertices in a layout with optimal @xmath8-energy is the @xmath58th power of the density between the vertices . in particular , the distance is the inverse density if @xmath59 , and the distance is almost independent of the density if @xmath60 .
this impact of @xmath61 on the representation of the community structure is illustrated for a larger network in fig .
[ f : random ] .
layouts with small linlog energy ( @xmath59 ) and with small fruchterman - reingold energy ( @xmath62 ) of a pseudo - random network with eight clusters ( intra - cluster density @xmath63 , expected inter - cluster density @xmath64).,title="fig:",width=147 ] layouts with small linlog energy ( @xmath59 ) and with small fruchterman - reingold energy ( @xmath62 ) of a pseudo - random network with eight clusters ( intra - cluster density @xmath63 , expected inter - cluster density @xmath64).,title="fig:",width=147 ] replacing the edge @xmath2 with two edges @xmath65 and @xmath66 , where @xmath67 is a new vertex with weight @xmath7 , increases the optimal distance between @xmath16 and @xmath0 by a factor of @xmath68 . because the @xmath8-energy is only defined for @xmath69 , the factor is @xmath6 if @xmath70 , and greater than @xmath6 if @xmath71 .
this result has a significant implication , given that the addition of @xmath67 increases the path length between @xmath16 and @xmath0 ( from @xmath6 to @xmath72 edges ) without changing the density : the optimal distance of @xmath16 and @xmath0 depends only on the density , and not on the path length , if @xmath70 ( as in the linlog energy model ) , and increases with the path length if @xmath71 .
the results for networks with two or three vertices can be generalized , at least as approximations , to larger networks . in a network with clear communities , for example ,
the density within the communities is ( by definition ) much greater than the density between the communities , and thus the intra - community distances in an optimal layout are much smaller than the inter - community distances ( unless @xmath61 is very large ) .
this can be approximated by assuming that the vertices of each community have the same position , and thus by considering each community as one big vertex . for networks with more than two communities , eq .
( [ e : distdens ] ) can not be expected to hold precisely for all pairs of communities , because this would often imply distances that violate the triangle inequality .
nevertheless , the qualitative reasoning generalizes : distances are less dependent on densities for large @xmath61 , and less dependent on path lengths for small @xmath20 . [ cols="^,^,^ " , ] as motivated in the previous subsection , the parameters of the energy model are set to @xmath70 and @xmath73 , with @xmath74 for networks with very nonuniform density ( @xmath75 ) , and @xmath27 for networks with fairly uniform density ( @xmath76 ) .
the variation of @xmath21 improves the readability by ensuring that vertices are not placed too closely , but otherwise does not affect the grouping of the vertices .
because the exact optimization of @xmath8-energy and modularity is computationally hard , the presented layouts and clusterings are not guaranteed to be optimal ( except for the clustering of the book co - purchase network @xcite ) , but are the best known representations .
the java program used for generating these representations is freely available .
it employs the barnes - hut algorithm for energy minimization , and agglomeration with multi - level refinement for modularity maximization ( see sec .
[ ss : algs ] ) . in the karate club network ( fig .
[ f : karate ] ) , each vertex represents a member of a karate club , and the edge weight of each vertex pair specifies the number of contexts ( like university classes , bars , or karate tournaments ) in which the two members interacted .
the main vertex groups in the @xmath77-energy layout coincide with the four clusters of the modularity clustering , and the layout correctly indicates that joining triangles and circles into a single cluster is almost as good as separating them ( modularity @xmath78 vs. @xmath79 ) .
the clustering and the layout both segregate the members who left the club after the instructor was fired ( gray boxes ) , with the exception of one member who followed the instructor mainly to preserve his chance for the black belt .
@xmath77-energy layout and modularity clustering ( represented by shapes ) of the karate club network .
the modularity of the clustering is 0.445 .
gray boxes represent members who left the club after the instructor was fired.,width=325 ] in the book co - purchase network ( fig .
[ f : polbools ] ) , the vertices represent books on us politics , and edges of weight @xmath6 connect books that were frequently purchased together . the clusters are generally well - separated in the layout ; a few members of the smaller central clusters are placed closely to one of the two large clusters , which correctly indicates that they are densely connected with parts of these large clusters , and their assignment to a smaller cluster is a close decision .
the clustering and the layout , especially their two main groups , conform well to newman s classification @xcite of the books as liberal ( light gray ) , neutral ( dark gray ) , or conservative ( black ) ; the layout is more suitable to represent the liberal - to - conservative ordering of the books .
@xmath80-energy layout and modularity clustering of the book co - purchase network .
the modularity is 0.527 .
shades represent the classification as liberal ( light gray ) , neutral ( dark gray ) , or conservative ( black).,width=325 ] the food classification network ( fig .
[ f : food ] ) represents the categorizations of 45 foods by 38 subjects of a psychological experiment , who were asked to sort the foods into as many categories as they wished based on perceived similarity .
each vertex represents a food , and the edge weight of each vertex pair is the number of subjects who assigned the corresponding foods to the same category .
the clusters correspond well to groups in the layout , but the layout also indicates that the borders between some clusters are rather fuzzy ( e.g. , between snacks and sweets ) , that some clusters could be split into subclusters ( e.g. , fruits and vegetables ) , and that some foods can not be clearly assigned to a single cluster ( e.g. , water , spaghetti ) .
the grouping in both the clustering and the layout largely conforms to common food categories . @xmath77-energy layout and modularity clustering of the food classification network .
the modularity of the clustering is 0.402 .
( the edges are elided to avoid clutter.),width=325 ] the world trade network ( fig .
[ f : trade ] ) models the trade between 66 countries in the year 1999 . the vertices represent countries , and the edge weight of each vertex pair specifies the trade volume between the corresponding countries in us dollar .
the clustering and the layout both group the countries of the three major economic areas ( east asia / australia , america , and europe ) .
the layout also reflects that countries like irn and egy can not be clearly assigned to either the east asian or the european group , and shows many smaller groups of closely interlocked countries like chn and hkg , aus and nzl , gbr and irl , and the nordic countries .
@xmath35-energy layout and modularity clustering of the world trade network .
the modularity of the clustering is 0.275 .
( the edges are elided to avoid clutter.),width=325 ]
as representations for the community structure of networks , layouts subsume clusterings , thus quality measures for layouts subsume quality measures for clusterings , and in fact prominent existing quality measures for layouts
namely , energy models based on the pairwise attraction and repulsion of vertices subsume a prominent existing quality measure for clusterings namely , the modularity measure of newman and girvan .
this result has implications for the entire lifecycle of quality measures : * design : new and existing quality measures for layouts may be applied to clusterings and vice versa .
for example , recent extensions of the modularity measure to directed networks @xcite and bipartite networks @xcite can be directly generalized to energy models for layouts .
* evaluation : the evaluation of quality measures for clusterings and layouts can be partly unified , i.e. , performed without distinguishing between clusterings and layouts .
this has been demonstrated in @xcite with a computation of the expected measurement value for networks with uniform expected density , a particularly important analysis technique @xcite . * optimization : components of clustering algorithms may be reused in layout algorithms and vice versa , for example the agglomeration ( coarsening ) phase of multi - level heuristics .
moreover , energy - based layout algorithms might serve as initial stage of clustering algorithms , similarly to eigenvector - based layout algorithms in existing approaches ( see sec .
[ ss : urel ] ) . *
application : unified quality measures help to ensure the consistency of clusterings and layouts ( see sec . [ s : appl ] ) , which is crucial because both representations are often used together . | two natural and widely used representations for the community structure of networks are clusterings , which partition the vertex set into disjoint subsets , and layouts , which assign the vertices to positions in a metric space .
this paper unifies prominent characterizations of layout quality and clustering quality , by showing that energy models of pairwise attraction and repulsion subsume newman and girvan s modularity measure .
layouts with optimal energy are relaxations of , and are thus consistent with , clusterings with optimal modularity , which is of practical relevance because both representations are complementary and often used together . |
although probabilistic ideas are at the heart of genetics , and have been used since the earliest days of the subject , it was in the study of genetic drift in the 1920s and 1930s that the notion of stochastic processes first played a major part in the theory of genetics . in the simplest case , a diploid population of a fixed size , @xmath0 , was considered , and attention was focused on two alleles at a particular locus . if selective differences between the two alleles , as well as the chances of mutation , are ignored , then the changes in the allele frequencies are purely due to genetic drift . assuming discrete non - overlapping generations
, one can ask : what is the probability that @xmath1 of the @xmath2 genes in the population alive at time @xmath3 are of a given type ?
this formulation of genetic drift is usually referred to as the fisher - wright model , being used implicitly by fisher @xcite , and explicitly by wright @xcite .
although fisher and wright did not use this terminology , the stochastic process they defined through this model is a markov chain , since the probabilities of genetic change from one generation to the next do not depend on the changes made in previous generations .
however the use of markov chains becomes cumbersome when the effects of mutation and selection are introduced , and for this reason there was a move away from the description of the process in terms of a discrete number of alleles @xmath4 , and discrete generations , to a process of diffusion where the fraction of alleles of one type is a real random variable @xmath5 in the interval @xmath6 $ ] and time is continuous .
this type of model actually predates the markov chain description @xcite , and was further studied by wright @xcite and kimura @xcite .
it is this formulation that will concern us here , specifically we will be interested in neutral evolution
that is , the dynamics of a randomly mating population which may , or may not , be subject to genetic mutation .
our motivation for the work presented here lies in the study of an evolutionary model of language change @xcite . as a first step in the mathematical formulation of this model
, we developed a `` drift model '' of language change @xcite , which turns out to be identical to the diffusion models of population genetics discussed above .
particularly relevant in this application to language change is the situation where the number of variants ( the linguistic counterpart to alleles ) may be large .
a survey of the population genetics literature revealed a considerable amount of work on the diffusion equation for two alleles @xcite , some on three alleles @xcite and very little on the general case of an arbitrary number of alleles @xcite .
much of the work originated with kimura in the 1950s , and developments since then appear scattered throughout the literature . in this work
we fill a considerable gap in the literature by giving for the first time an analytic solution of the diffusion equation , with and without mutation , in the general case of @xmath7 alleles .
given the rather scattered nature and occasionally the apparent obscurity of other results relating to genetic diffusion at a single locus , we take the opportunity here to present them in a systematic fashion . in writing this paper
we also have a third aim , namely to present this information in such a way that is accessible to geneticists who may not wish to work through a great deal of mathematical analysis .
we will shortly describe the plan of the paper which we hope will allow us to accomplish this .
after kimura s pioneering work in the 1950s , there was further work by him @xcite and others @xcite on single - locus models , but interest naturally moved to models which involved multiple loci and the interactions between them @xcite . despite the continuing development of population genetics , very little has been achieved by way of exact solutions for single - locus models since kimura @xcite and littler @xcite .
occasional work has appeared , both of an analytic nature @xcite and involving numerical simulations @xcite , although single locus diffusion models continue to receive proportionately less attention .
a number of other factors have been responsible for the relative neglect of this particular area of population genetics .
one of the main ones is that the focus has shifted to experimental investigations and there has been the consequential realisation that reality is very much more complex than what simple models allow . however , it is also the case that the theory is perceived as mathematically difficult .
even kimura @xcite states when studying the extension to three alleles that for the partial differential equation describing diffusion `` the general case of an arbitrary number of alleles can be solved [ by separation of variables ] .
however , additional techniques will be needed to make the mathematical manipulations manageable '' .
we will show here that this is not the case : in fact it is no harder to solve the case of the diffusion of a general number of alleles than the three - allele case .
we believe that , while our motivation for this work originates with models of language change , geneticists will also find the results we obtain useful .
the diffusion theory is described in several standard texts on population genetics @xcite and general consequences of the results previously found in the form of simple rules are widely known and utilised .
the problem , as we have already indicated , is the mathematical nature of the theory .
the diffusion approximation in the case without mutation results in a partial differential equation which is of the type which is used to describe the diffusion of heat in a metal or dye in a fluid @xcite .
the inclusion of mutation corresponds in these physical problems to a deterministic force which biases the diffusion in one direction rather than another .
for example , if the particles diffusing were charged , then this deterministic force could arise from an electric field being imposed on the system .
not surprisingly , diffusion equations of this kind have been extensively studied by physicists who call them fokker - planck equations @xcite . here
we will use the nomenclature used in the genetics literature where they are called kolmogorov equations .
there are other confusing changes in nomenclature across disciplines : for example , the term representing the deterministic motion discussed above in the kolmogorov equation is called the drift in the physics literature , but in the genetics literature it is the diffusive term which is called the drift . in order to avoid confusion we shall now drop the further use of the term `` drift '' in this work . the outline of the paper is as follows . in section [ overview ]
we will give on overview of our results which is devoid of proofs and too much mathematical detail , so that it may be read by those who simply wish to have a reasonably non - technical summary of the results of the paper .
it will also serve as a general introduction to those who wish to go on and absorb the more technical sections of the paper .
these technical sections start with section [ soln ] , where the general solution of the kolmogorov equation for an arbitrary number of alleles , with and without mutations , is derived . in section
[ other ] we give details of the calculation of various quantities of interest that are given briefly at the end of section [ overview ] .
we end in section [ discuss ] with an overview of the paper s main results and point out some remaining unsolved cases .
further technical material has been relegated to three appendices so as not to interrupt the flow of the arguments in the main text .
in this section we will describe the problems we investigate , explaining our methods of analysing them and give the results we have obtained without proof .
the simplest , and most widely studied , version of the problems we will be investigating in this paper , is where two alleles , @xmath8 and @xmath9 , are segregating in a randomly - mating population of @xmath0 monoecious individuals .
we begin with the purely random case in which the only way that allele frequencies may change is through the random sampling of gametes in the reproductive process .
if @xmath10 is the frequency of allele @xmath8 and @xmath11 the frequency of allele @xmath9 , we would expect that the value of @xmath10 would change through time until eventually one or other of the alleles would become fixed , _
i.e. _ @xmath10 would take on the value 0 or 1 . the mathematical quantity which describes the process is the conditional probability distribution function ( pdf ) @xmath12the probability that the frequency of the @xmath8 allele at time @xmath3 is @xmath10 , given the frequency of this allele at @xmath13 was @xmath14 .
it satisfies the kolmogorov equation @xcite @xmath15 \ , , \ \ \
d(x ) = \frac{1}{2 } x(1-x)\ , , \label{kol_2}\ ] ] in the interval @xmath16 .
this is a diffusion equation for the frequency of the @xmath8 allele , but with a diffusion coefficient , @xmath17 , which depends on the frequency .
the form of the diffusion coefficient is such that it gets much weaker as @xmath10 approaches the boundaries at @xmath18 and @xmath19 .
however as we show using a simple argument in appendix a it is still sufficiently strong that fixation can occur , but once this has happened there is no possibility of re - entering the region @xmath16 .
( in feller s classification @xcite , these are exit boundaries ) .
as we shall see , it will be necessary to consider the boundaries separately from the interior region when calculating probabilities .
it should also be noted that in the derivation of eq .
( [ kol_2 ] ) ( see section [ soln ] ) , one unit of time is taken to be @xmath2 generations , so that the time between generations is @xmath20 .
although this choice is useful in derivations involving the diffusion approximation , it is frequently more useful to work with @xmath21 , so that the unit of time is that between generations
. then eq .
( [ kol_2 ] ) takes the form @xmath22 \ , , \label{tau_form}\ ] ] which is the one that generally appears in textbooks @xcite .
when @xmath7 alleles @xmath23 are present , these ideas can be generalised .
if @xmath24 are the frequencies of the alleles , then there are only @xmath25 independent variables the frequency @xmath26 may be replaced by @xmath27 .
the kolmogorov equation is now an @xmath28-dimensional diffusion equation @xcite @xmath29 \ , , \label{kol_m}\ ] ] where the diffusion coefficient is now a matrix : @xmath30 if mutations occurring at constant rates are allowed for , these kolmogorov equations acquire an extra term , which involves only first order derivatives . in the case of two alleles ,
if @xmath9 mutates to @xmath8 at a constant rate given by @xmath31 and @xmath8 mutates to @xmath9 at a constant rate given by @xmath32 , then in the absence of the random process described above , we would expect the frequency of the @xmath8 allele to change according to the _ deterministic _ differential equation @xmath33 , where @xmath34 .
if we include both this deterministic process and the random diffusion process , the kolmogorov equation takes the form @xcite : @xmath35 + \frac{\partial^{2 } } { \partial x^{2 } } \left [ d(x ) p \right]\ , , \label{kol_2_mut}\ ] ] where @xmath36 finally , if @xmath7 alleles are present with mutation , we shall in common with other authors @xcite , only study the situation in which the rate of mutation of the alleles @xmath37 to @xmath38 occurs at a constant rate @xmath39 , independent of @xmath40 .
then the deterministic equation would be @xmath41 the fact that @xmath42 depends only on @xmath43 is a consequence of mutation rates depending only on the end product of the mutation .
it is this feature that allows solution by separation of variables of the resulting kolmogorov equation which reads @xmath44 + \sum^{m-1}_{i=1 } \sum^{m-1}_{j=1 } \frac{\partial^{2 } } { \partial x_{i } \partial x_{j } } \left [ d_{ij}({\underline}{x } ) p \right ] \
, , \label{kol_m_mut}\ ] ] where @xmath45 and @xmath46 are given by eqs .
( [ a_m ] ) and ( [ diff_m ] ) respectively .
it is sometimes useful to use the backward form of the kolmogorov equation , in which the derivatives are with respect to the initial values @xmath14 and @xmath47 : @xmath48 since for the processes we will be interested in here , @xmath49 will be a function only of @xmath50 , the left - hand side may also be written as @xmath51 , and @xmath47 chosen to be equal to zero . a detailed derivation and explanation of the relationship between the forward and backward forms can be found in @xcite .
note that immigration is commonly introduced @xcite by the use of @xmath52 where @xmath53 represents the mean of the population where the immigrants come from .
this situation can be represented by eq .
( [ kol_2_mut ] ) simply by setting @xmath54 and @xmath55 .
when one or more of the mutation parameters is zero , we have irreversible mutation , when the final result is fixation to those alleles which have @xmath39 not equal to zero .
we will not consider this special case here , but see for example @xcite . in purely mathematical terms
, the purpose of this paper is to obtain the conditional pdf @xmath56 obtained by solving eqs .
( [ kol_m ] ) and ( [ kol_m_mut ] ) , subject to the initial conditions that @xmath57 at @xmath13 and the appropriate boundary conditions .
as far as we are aware , only the cases @xmath58 without mutation and @xmath59 with mutation have been solved to date @xcite .
the key to making progress is to find a change of variable which makes these equations separable and also to note that the solution with @xmath1 alleles is nested within the solution with @xmath60 alleles , so many properties follow by induction on @xmath1 . while the @xmath59 kolmogorov equations ( [ kol_2 ] ) and ( [ kol_2_mut ] ) can be solved by separation of variables writing @xmath61their counterparts for general @xmath7 , eqs .
( [ kol_m ] ) and ( [ kol_m_mut ] ) can not .
the standard way to proceed in such a situation is to look for a transformation to a coordinate system in which these partial differential equations are separable .
kimura @xcite gave such a transformation for values of @xmath7 up to 4 without mutation , but not in general .
there is no systematic way to discover such transformations , but we have found one which works for all @xmath7 , namely : @xmath62 with the inverse transformation @xmath63 as discussed in section [ soln ] , this change of variables means that the most general form of the kolmogorov equation we consider , eq .
( [ kol_m_mut ] ) , may be written in diagonal form : @xmath64 + \sum^{m-1}_{i=1}\frac{\partial^{2 } } { \partial u_{i}^{2 } } \left [ \mathcal{d}_{i}({\underline}{u } ) \mathcal{p } \right ] \
, , \label{kol_m_u}\ ] ] that is , there are no mixed second - order derivatives . for completeness
we give the expressions for the functions @xmath65 and @xmath66 : @xmath67 note that this equation has been derived using the new variables just changing variables using ( [ cofv ] ) in the original kolmogorov equation ( [ kol_m_mut ] ) will not give ( [ kol_m_u ] ) , but will instead simply give an equation for @xmath56 with @xmath43 replaced by @xmath68 .
the new conditional pdf in ( [ kol_m_u ] ) , denoted by @xmath69 , is related to @xmath56 through the jacobian of the transformation to the new variables : @xmath70 the jacobian is calculated in appendix b. when @xmath71 it is simply @xmath72 , so that @xmath73 and in the general case one finds that @xmath74 we may summarise this subsection in the following way . we have given a change of variables which allows the kolmogorov equation to be written in a form which is separable .
it is easier to work with the conditional pdf appropriate to these variables , and then to transform back to the original variables of the problem and the original conditional pdf using eqs .
( [ cofv ] ) and ( [ trans_prob_m ] ) respectively .
the details of the solution of the kolmogorov equation ( [ kol_m_u ] ) are given in section 3 and appendix c. here we will present only the final expressions .
for orientation , let us first give the expression in the case @xmath59 and where no mutation is present .
this is well known and appears in the standard texts @xcite . in this case
no change of variable is required : @xmath75 .
then @xmath76 the constant @xmath77 is given by @xmath78/(l+1)$ ] and the @xmath79 are polynomials , described below .
the function ( [ soln_2 ] ) satisfies both the initial and boundary conditions and holds for all @xmath10 in the interval @xmath80 , that is , excluding the boundaries at @xmath18 and @xmath19 . for not too small values of @xmath3 ,
the solution is well approximated by keeping only the first few terms in @xmath81 . in these cases
the polynomials have a simple form , since @xmath82 is a polynomial in @xmath83 of order @xmath81 .
for instance , @xmath84 the natural variable for these polynomials is @xmath85 , so that they are defined on the interval @xmath86 . for general @xmath81
they are the jacobi polynomials @xmath87 @xcite .
we plot the time development of the solution ( [ soln_2 ] ) in fig .
[ neut_2a ] . beginning from a delta function at the initial point ,
the distribution initially spreads out until the boundaries are reached .
the distribution is soon nearly flat , and subsides slowly as probability escapes to the boundaries .
note that the probability deposited on the boundaries is not shown here ( but will be discussed later ) .
time evolution of the pdf for a biallelic system as determined from the analytic solution with no mutation and initial condition @xmath88 . ] for general @xmath7 , the solution may be written in the form @xmath89 as discussed in section [ soln ] . the solution is written in terms of the @xmath90 variables since the problem is separable and so the eigenfunctions are separable . the function @xmath91 is given by @xmath92 and @xmath93 depends on a set of @xmath25 non - negative integers @xmath94 according to @xmath95 this means that the sum over @xmath93 in eq .
( [ soln_m_first ] ) is in fact an @xmath96dimensional sum .
the property , which we have already remarked upon , that the system with @xmath25 alleles is nested in that with @xmath7 , manifests itself here in the fact that the functions @xmath97 are very closely related to the functions which already appear in the two allele solution ( [ soln_2 ] ) .
first of all , since the problem is separable in this coordinate system , they may be written as @xmath98 each of the factors @xmath99 in this product separately satisfies the same differential equation .
we shall derive this equation in section [ soln ] , from which we will learn that @xmath99 depends only on @xmath100 and on the integers @xmath101 and @xmath102 .
specifically , @xmath103^{1/2 } \left ( 1 - u_{i } \right)^{l_{i-1 } } p^{(1,2l_{i-1}+1)}_{l_i } ( 1 - 2u_i)\ , .
\label{soln_m_third}\ ] ] here the @xmath104 are the analogue of the @xmath77 that appeared in eq .
( [ soln_2 ] ) , but in this case we have included them in the function @xmath97 , so that it satisfies the simple orthogonality relation @xmath105 the explicit form of these constants is @xmath106 the @xmath107 , now with @xmath108 , are again jacobi polynomials @xcite .
the general solution for @xmath109 has previously only been found for the case @xmath71 .
we have checked that our solution agrees with the published result @xcite in this case , although the labels on the alleles are permuted between kimura s solution and ours ( @xmath110 ) due to a different change of variable being made .
we plot this solution for a triallelic system in fig .
[ neut_3a ] at several different times . just as in the two allele system ,
the distribution initially spreads out and forms a bell shape , which quickly collapses , and becomes nearly flat , then subsides slowly as probability escapes to the boundaries .
note that the probability deposited on the boundaries is not shown here .
notice also the triangular shape of the region over which the distribution is supported .
time evolution of the pdf for a triallelic system as determined from the analytic solution with no mutation and initial condition @xmath111 . ] when mutation is present , the solution is again well known @xcite for the @xmath59 case .
again , no change of variable is required : @xmath75 . then @xmath112 the constant @xmath77 is given by @xmath113 while @xmath114 , @xmath115 and the parameter @xmath116 .
the @xmath117 are jacobi polynomials @xcite .
unlike the case without mutations , @xmath118 , and they can not be written in terms of simpler polynomials as in case ( i ) .
the function ( [ soln_2_m ] ) satisfies both the initial and boundary conditions and holds for all @xmath10 in the interval @xmath6 $ ] , that is , now including the boundaries at @xmath18 and @xmath19 .
this difference is due to the nature of the boundary conditions in the mutation case , which will be described in detail in section 3 . for not too small values of @xmath3 ,
the solution is well approximated by keeping only the first few terms in @xmath81 . in these cases
the polynomials do have a relatively simple form , since @xmath119 is a polynomial in @xmath83 of order @xmath81 . for instance , @xmath120\ , , \ldots \ , .
\label{low_order_2_m}\ ] ] also note that once again the natural variable for these polynomials is @xmath85 , so that they are defined on the interval @xmath121 . in terms of the allele @xmath8 frequency @xmath10 and
the mutation rates @xmath31 and @xmath32 , these polynomials take the form @xmath122\ , , \ldots \ , .
\label{low_order_2_mx}\ ] ] for concreteness , we plot in figs . [ mut_2a ] and [ mut_2b ] the time evolution pdfs for biallelic ( @xmath59 ) systems from a fixed initial condition of @xmath88 , i.e. , that 70% of the population have one particular allele type . in the former case ,
the mutation rates are low ( @xmath123 ) and one finds a high probability of one allele dominating the population .
conversely , in fig . [ mut_2b ] which has higher mutation rates @xmath124 ,
the most probable outcome is for both alleles to coexist in the population , as indicated by the peak of the distribution occurring for some value of @xmath10 far from either boundary . in each of the figures , we compare the distribution obtained from a monte carlo simulation of the population dynamics against the analytic result ( [ soln_2_m ] ) with good agreement .
time evolution of the pdf for a biallelic system as determined from the analytic solution ( left ) and monte carlo simulation ( right ) with mutation parameters @xmath123 and initial condition @xmath88.,title="fig : " ] time evolution of the pdf for a biallelic system as determined from the analytic solution ( left ) and monte carlo simulation ( right ) with mutation parameters @xmath123 and initial condition @xmath88.,title="fig : " ] time evolution of the pdf for a biallelic systems as determined from the analytic solution ( left ) and monte carlo simulation ( right ) with mutation parameters @xmath124 and initial condition @xmath88.,title="fig : " ] time evolution of the pdf for a biallelic systems as determined from the analytic solution ( left ) and monte carlo simulation ( right ) with mutation parameters @xmath124 and initial condition @xmath88.,title="fig : " ] for general @xmath7 , the solution may be written in the form @xmath125 as discussed in section [ soln ] . here
@xmath126 is the left - eigenfunction of the kolmogorov equation ( @xmath127 being the right - eigenfunction , as usual ) .
again , @xmath93 depends on a set of @xmath25 non - negative integers @xmath94 , with @xmath128 the left - eigenfunction @xmath129 is equal to @xmath130 @xcite , where @xmath131 is the stationary pdf .
this is equivalent to the form ( [ soln_m_first ] ) , appropriate when there are no mutations , since @xmath132 is what is there called the weight function ( up to normalisation ) . since the problem is separable in this coordinate system , @xmath133
may again be written as @xmath134 once again each of the factors @xmath99 is separately a solution of a single differential equation , and depends on @xmath100 , the integers @xmath135 and @xmath136 and the mutation rates @xmath39 .
specifically , @xmath137 where @xmath138 in a similar way , @xmath139 may be written as @xmath140 each @xmath141 depends on @xmath100 and on the integer @xmath136 and the mutation constants @xmath142 and @xmath39 . specifically , @xmath143 where @xmath144 the general solution has previously only been found for the case @xmath59 .
our solution agrees with the published result @xcite in this case . we can calculate the time evolution of various statistics of the population . as previously , we present here a summary of the results for ease of reference , deferring detailed calculations to section [ other ] . in the absence of mutation
, alleles can become extinct since , once they have vanished from the population , there is no mechanism by which they can be replaced .
as we have already mentioned , the solution we have derived ( [ soln_m_first ] ) gives the pdf for all states in the interior .
the pdf on a boundary is also given by this equation , so long as this boundary is not the point of fixation of the final allele , since it corresponds to the solution with @xmath7 reduced by 1 .
so , for example , one can read off the pdf in , say , the two - dimensional subspace where alleles @xmath8 and @xmath9 coexist and all others present in the initial population have become extinct since it is just the @xmath59 solution .
kimura @xcite finds the large - time form for this quantity given an initial condition comprising @xmath71 alleles using an approach involving moments of the distribution which , as we shall show below , can be found directly from the kolmogorov equation without recourse to the full solution we have obtained here .
it remains then to determine the fixation probabilities that are excluded from the pdf we have derived . using a simple argument ( outlined below )
one can find the probability that a particular allele has fixed by time @xmath3 by reduction to an equivalent two - allele description , the properties of which are well - known @xcite .
using the backward kolmogorov equation ( [ bke ] ) , one can find further quantities relating to extinction events .
although some of the formul we quote have appeared in the literature before @xcite they seem not to be widely known and some were stated without proof .
we therefore feel there is some value in reviving them here and giving what is hopefully a clear derivation in this section or in section [ other ] .
define @xmath145 as the probability that allele @xmath8 , which had initial frequency @xmath14 in a two allele system , has become fixed by time @xmath3 .
this is given by @xcite @xmath146 e^{- ( l+1)(l+2 ) t /2 } \,.\ ] ] here @xmath147 is a legendre polynomial which is a particular case of the jacobi polynomials : @xmath148 .
conversely , the probability that @xmath8 has become extinct by time @xmath3 is equal to that of @xmath9 become fixed , viz , @xmath149 e^{-(l+1)(l+2)t/2 } \;.\ ] ] now , consider an initial condition with @xmath109 alleles , divided into two groups @xmath150 and @xmath151 .
let now @xmath10 be the sum of the frequencies of @xmath150 alleles .
we may think of the set of alleles @xmath150 as a single allele in a two allele system ( for instance , @xmath8 in the above discussion ) and the set @xmath151 as the other allele .
then the probability that _ all _ of the @xmath151 alleles ( and possibly some of the @xmath150 alleles ) have become extinct by time @xmath3 is simply @xmath145 where @xmath14 is the initial combined frequency of @xmath150 alleles .
similarly , the probability that all of the @xmath150 alleles are extinct at time @xmath3 , leaving some combination of @xmath151 alleles is @xmath152 .
arguments of this kind , where a set of alleles is identified with a single allele in the @xmath59 problem , can be frequently employed to obtain results for the general case of @xmath7 alleles in terms of those already explicitly calculated for @xmath59 . by combining the above reduction to an equivalent two - allele problem with combinatorial arguments , the probability that exactly
@xmath153 alleles coexist at time @xmath3 can be calculated @xcite .
the result is @xmath154 where @xmath153 can take any value from @xmath155 up to @xmath7 , and the second summation is over all subsets of size @xmath156 drawn from the @xmath7 alleles .
it is a simple matter to use this formula to calculate the mean number of coexisting alleles at time @xmath3 . in fig .
[ r_alleles ] , we compare the time - dependent probabilities for @xmath157 alleles to be present in a system initially comprising @xmath158 alleles given by the exact formula ( [ prob_exactly_r ] ) and as obtained from a monte - carlo simulation with the allele frequencies initially equal .
we also show the evolution of the mean number of coexisting alleles from starts with @xmath71 and @xmath158 alleles . in both cases ,
the exact formul correspond extremely well with the simulations , illustrating the utility of the diffusion approximation to the discrete population dynamics .
alleles are present as a function of time for @xmath159 in a system with 4 alleles initially . expected number of alleles present as a function of time for systems with 3 and 4 alleles initially present ( right ) .
symbols are monte carlo simulation data , solid curves are analytic calculations.,title="fig : " ] alleles are present as a function of time for @xmath159 in a system with 4 alleles initially .
expected number of alleles present as a function of time for systems with 3 and 4 alleles initially present ( right ) .
symbols are monte carlo simulation data , solid curves are analytic calculations.,title="fig : " ] [ num_alleles ] by solving the appropriate backward kolmogorov equation , one can straightforwardly find the mean time to the first ( and only ) extinction event among two alleles to be @xmath160\,.\ ] ] a combinatorial analysis , performed in @xcite and which we will summarise in section [ other2 ] , reveals the mean time to the @xmath161 extinction can be calculated from an @xmath59 result .
one finds @xmath162 \;,\ ] ] where @xmath163 is the initial fraction of allele @xmath164 present in the system .
the second sum is over all possible @xmath156-subsets @xmath165 of the @xmath7 alleles initially present in the population .
not all extinction probabilities can be calculated by reduction to an equivalent a two - allele problem .
a notable example is the probability that allele @xmath166 goes extinct first , followed by @xmath167 leaving only @xmath168 in the final population , since in this case we ask for _ all _ , rather than just _ some _ of a subset to be present in the population at a given time .
this probability is given by @xmath169 in which @xmath170 is the initial frequency of allele @xmath171 in the population .
this result appears in @xcite , albeit without an explicit derivation .
we shall provide one in section [ other2 ] .
an immediate consequence of the previous result is the probability that a particular allele , @xmath171 is the _ first _ to become extinct , that is , at a time when all other alleles have a nonzero frequency .
this is obtained by summing ( [ qp ] ) over all possible sequences of extinctions in which @xmath171 goes extinct first .
for example , when the initial number of alleles @xmath71 , the probability that @xmath8 goes extinct first can be found by summing over the cases @xmath172 in ( [ qp ] ) .
one finds @xmath173 where @xmath174 .
this agrees with the result in @xcite . by a similar method
, one could determine the probability for allele @xmath171 to be the _
second _ to go extinct or any other quantity of this type .
when mutations are absent , we see from ( [ f_1 ] ) that allele @xmath171 fixes with a probability equal to its initial frequency in the population . hence in this case
, the stationary distribution @xmath175 is zero everywhere except at those points that correspond to fixation of a single allele . on the other hand ,
when all the mutation rates @xmath39 at which all alleles @xmath176 mutate to @xmath171 are nonzero , the stationary distribution is nonzero everywhere .
furthermore , this distribution is reached from any initial condition and takes the well - known form @xcite @xmath177 where in keeping with the 2 allele case @xmath178 .
recall that the frequency @xmath26 is implied through the normalisation @xmath179 .
one way to calculate moments of the distribution is to integrate the exact solution we have obtained above .
however , it is also possible to derive differential equations for the moments in terms of lower - order moments from the kolmogorov equation , a procedure that leads more quickly to simple , exact expressions for low - order moments .
we quote some examples for general @xmath7 here , noting that previously results for @xmath59 and @xmath180 without mutations have previously appeared in @xcite and for @xmath59 with mutations in @xcite .
it is also worth remarking that by calculating moments , kimura reconstructed the solution for the pdf for the case @xmath71 and no mutations @xcite .
when mutations are absent , the mean of @xmath43 is conserved by the dynamics , i.e. , @xmath181 meanwhile , the variance changes with time as @xmath182 this result is clear , at least in the case @xmath59 .
the delta - function initial condition has zero variance , and the stationary distribution comprises delta - functions at @xmath18 and @xmath19 with weights @xmath183 and @xmath14 respectively . for the latter one
then has @xmath184 and so @xmath185 .
the covariance , which does not appear in the two allele system is : @xmath186 crow and kimura @xcite give moments only for the biallelic case , and then in terms of hypergeometric functions .
we provide simpler results in explicit form here , again for any @xmath7 .
the mean frequency of any allele behaves as @xmath187 where @xmath188 the mean ( [ meanx_m ] ) exponentially approaches the fixed point of the deterministic motion . recalling that @xmath189
, we see in eq .
( [ kol_2_mut ] ) the function @xmath190 vanishes at the point @xmath191 .
the variance is more complicated : @xmath192 as @xmath193 , this converges to the limit @xmath194 solid lines : time evolution of the mean ( left ) and variance ( right ) for a biallelic system for three different mutation parameter combinations , and with @xmath195 .
dashed lines : evolution of mean and variance when mutation is absent .
, title="fig : " ] solid lines : time evolution of the mean ( left ) and variance ( right ) for a biallelic system for three different mutation parameter combinations , and with @xmath195 . dashed lines : evolution of mean and variance when mutation is absent .
, title="fig : " ] in fig .
[ mean_var ] we show the mean and variance as a function of time in the biallelic case , for three different pairs of mutation parameters .
the mean approaches the stationary distribution value exponentially , at a faster rate the larger the values of the mutation parameters .
when @xmath196 , the variance rises asymptotically to the stationary limit , again faster when the mutation parameters are larger .
notice that for small values of @xmath31 and @xmath32 , the final variance is close to that when no mutations occur , while the larger the mutation parameters become , the narrower the final distribution .
when the parameters are unequal , the variance can reach a maximum value before descending slowly to the final limit .
the time development of the heterozygosity with two alleles , for the case with no mutations ( dashed curve ) and for three different cases of mutation ( solid curves ) , @xmath195 in each case . ]
the covariance ( with @xmath197 ) is @xmath198 as asserted by kimura @xcite , in the absence of mutation the heterozygosity decays as @xmath199 regardless of the initial number of alleles .
the heterozygosity is the probability that two alleles chosen at random from the population will be different .
when there are multiple alleles , this probability can be calculated from the first and second moments using @xmath200\;.\ ] ] when expressions for the various moments are substituted into this expression , it is found that , in the absence of mutations , the constant terms cancel , leaving only terms decaying as @xmath199 . therefore we have @xmath201 where we find @xmath202 .
kimura derived this result from the expected change in @xmath203 with each generation .
the calculation with mutation present , however , does not give such a simple result .
the heterozygosity can also be found for any number of alleles when mutation is present . however , the form becomes increasingly complicated as @xmath7 increases .
therefore we give as an example only the two allele result here : @xmath204 e^{-(2r+1)t } \ , .
\label{h_mut}\end{aligned}\ ] ] this is plotted in fig .
we see that the presence of mutation between the two alleles maintains the heterozygosity at a finite level .
in this section we will derive the solution of the kolmogorov equation ( [ kol_m_u ] ) , which was given in section [ cond_pdf ] .
in particular we will derive the differential equation that is satisfied by every single one of the factors @xmath205 that appear in the eigenfunction @xmath206 in eq .
( [ soln_m_second ] ) which applies when no mutations occur .
we shall also present the more general differential equation that must be solved to determine the corresponding factors in ( [ soln_m_second_m ] ) that give the eigenfunctions when mutations are present .
the more technical points of the discussion are relegated to appendices b and c , in order to avoid interrupting the flow of the arguments .
it is worthwhile to recall briefly the derivation of the kolmogorov equation itself , since it will turn out that the most efficient way of changing from the @xmath207 variables to the @xmath90 variables is go back to an early stage of this derivation .
the starting point for the derivation is the chapman - kolmogorov equation which is one of the defining equations for markov processes : @xmath208 if we take @xmath209 and expand @xmath210 in powers of @xmath211 , we find in the limit @xmath212the kramers - moyal expansion for @xmath56 @xcite : @xmath213 here the @xmath214 are the jump moments defined through the relation @xmath215 the angle brackets are averages over realisations of the process @xmath216 ; @xmath216 is a stochastic process , whereas @xmath217 and @xmath207 are simply the initial and final states .
so far very little has been assumed other than the markov property of the stochastic process .
the markov nature of the process follows from the fact that the sampling of gametes from the gamete pool depends only on its current state , and not on its past history .
if we assume that no deterministic forces are present , only the random mating process , then @xmath218 . if deterministic forces are present , then there will be an @xmath219 contribution to @xmath220 which is equal to @xmath221 . here
@xmath222 is the rate of change of @xmath223 in the deterministic ( @xmath224 ) limit .
the higher jump moments reflect the statistics of the sampling process . in the case of two alleles
this is binomial and so the second jump moment is proportional to @xmath225 , which is the variance of this distribution . within the diffusion approximation , a unit of time
is taken to be @xmath2 generations , so that the time between generations is @xmath226 .
this implies that all jump moments higher than the second are higher order in @xmath227 .
this gives the results ( [ kol_2_mut ] ) and ( [ aandd_2 ] ) .
for the case of general @xmath7 , the multinomial distribution applies , which again means that the @xmath214 for @xmath228 vanish , and the second order jump moment is given by ( [ diff_m ] ) , up to a factor of @xmath229 .
this is the briefest summary of the derivation of eq .
( [ kol_m_mut ] ) .
we refer the reader to @xcite for a fuller account . however , the above discussion is sufficient for our purposes , which is to note that the derivation involving eqs .
( [ c - k_eqn])-([jm_defn ] ) may be carried out for any coordinate system , and specifically in the coordinate system @xmath230 for which the kolmogorov equation is separable . in this case the kramers - moyal expansion for @xmath69 is @xmath231 where @xmath232 are the jump moments in the new coordinate system : @xmath233 the most straightforward way of deriving the kolmogorov equation in the @xmath90 variables is to start from eq .
( [ km_expansion_new ] ) and determine the @xmath232 by relating them to the known @xmath214 .
this is carried out in appendix b and the result is the equations ( [ kol_m_u ] ) and ( [ cala_cald ] ) .
the next step in the solution of the kolmogorov equation is to separate out the time and @xmath100 coordinates .
that is , we write the general solution as a linear combination of such solutions : @xmath234 where the function @xmath127 satisfies @xmath235 \right\ } \phi_{\lambda } ( { \underline}{u } ) = - \lambda \phi_{\lambda } ( { \underline}{u } ) \label{eigen_eqn}\ ] ] and @xmath236
. we may apply initial conditions directly on ( [ lin_comb ] ) : @xmath237 by assuming the orthogonality relation @xmath238 where @xmath91 is an appropriate weight function , we then have that @xmath239 and so @xmath240 to determine @xmath127 we have in principle to solve the partial differential equation ( [ eigen_eqn ] ) in @xmath241 variables .
we will now prove that this equation is separable in the @xmath90 variables .
for clarity we will start where mutation does not take place ( that is , omit the factor @xmath242 in ( [ eigen_eqn]))the argument goes through in an identical fashion when this term is present , as we will discuss later .
when @xmath243 @xmath127 satisfies the equation @xmath244 = - \lambda^{(1 ) } \phi_{\lambda^{(1 ) } } ( u_1)\ , , \label{neutral_1}\ ] ] where we have introduced the superscript @xmath245 on to the eigenvalue @xmath93 to identify it as belonging to the @xmath246 problem .
this ordinary differential equation may be solved in a straightforward fashion as we discuss in the next subsection below . when @xmath247 @xmath127 satisfies the equation @xmath248 & + & \nonumber \\ \frac{1}{2}\,\frac{1}{1-u_1}\,\frac{\partial } { \partial u^2_2 } \left [ u_{2}(1-u_2 ) \phi_{\lambda^{(2 ) } } ( u_1 , u_2 ) \right ] & = & - \lambda^{(2 ) } \phi_{\lambda^{(2 ) } }
( u_1 , u_2)\ , .
\label{neutral_2}\end{aligned}\ ] ] we look for a separable solution of the form @xmath249 where @xmath250 = - \lambda^{(1 ) } \phi_{\lambda^{(1 ) } } ( u_2)\,,\ ] ] that is , @xmath251 satisfies the equation corresponding to the @xmath246 problem .
it is straightforward to show that a separable solution exists if @xmath252 satisfies the equation @xmath253 = \left\ { - \lambda^{(2 ) } + \frac{\lambda^{(1)}}{(1-u_1 ) } \right\ } \psi_{\lambda^{(2)};\lambda^{(1 ) } } ( u_1)\ , . \label{psi_eqn_1}\ ] ] we have introduced the subscript @xmath254 on @xmath252 to show that it depends on both the eigenvalues of the @xmath255 and the @xmath246 problems . as we will see , it is a remarkable fact that the solution of the problem with @xmath256 alleles only involves the function @xmath257 appearing in eq . ( [ psi_eqn_1 ] ) .
since @xmath258 may also be written as @xmath259 we may write the separable form of the solution in the @xmath255 case as @xmath260 we can now prove that the general case of @xmath256 alleles is separable as follows .
we look for a solution of ( [ eigen_eqn ] ) of the form @xmath261 where @xmath262 is the solution of ( [ eigen_eqn ] ) with @xmath263 alleles , but with the replacements @xmath264 . by explicit substitution into ( [ eigen_eqn ] ) it is not difficult to demonstrate this , and to find that @xmath252 satisfies eq .
( [ psi_eqn_1 ] ) , where now @xmath265 and @xmath266 are replaced by @xmath267 and @xmath268 respectively .
therefore we have shown that @xmath269 this shows that if the @xmath263 allele problem is separable , then the @xmath1 allele problem is separable .
but eq . ( [ sep_soln_2 ] ) shows that the @xmath255 problem is separable , and so by induction the problem for general @xmath1 is separable .
the explicit solution is @xmath270 where @xmath271 and where @xmath272 satisfies the equation @xmath273 = \left\ { - \lambda + \frac{\lambda'}{(1-u ) } \right\ } \psi_{\lambda;\lambda ' } ( u)\ , .
\label{psi_eqn}\ ] ] thus we have finally shown that in order to obtain the eigenfunction @xmath274 it is necessary only to solve this _ single _ ordinary differential equation for general @xmath93 and @xmath275 .
the full eigenfunction is then just a product of solutions of this equation .
a similar equation arises from an analogous argument when mutation is included , that is , the factor @xmath242 is restored into eq .
( [ eigen_eqn ] ) : all that is required is to make the substitution @xmath276 \longrightarrow \frac{1}{2 } \frac{\partial } { \partial u_i } \left [ u_{i}(1-u_i ) \right ] + \left ( r_{i } u_{i } - m_{i } \right)\ ] ] in the above argument .
the only difference is that now the function @xmath257 explicitly depends on @xmath277 because of the existence of the terms @xmath142 and @xmath39 .
thus the results corresponding to eqs .
( [ sep_soln_n ] ) and ( [ psi_eqn ] ) are : @xmath278 where @xmath279 satisfies the equation @xmath280 & + & \frac{1}{2 } \frac{{{\rm d}}^2}{{{\rm d}}u^2_i } \left [ u_i(1-u_i ) \psi^{(i)}_{\lambda;\lambda ' } ( u_i ) \right ] \nonumber \\ & = & \left\ { - \lambda + \frac{\lambda'}{(1-u_i ) } \right\ } \psi^{(i)}_{\lambda;\lambda ' } ( u_i)\ , .
\label{psi_eqn_mut}\end{aligned}\ ] ] the rest of this section will be devoted to solving the ordinary differential equations ( [ psi_eqn ] ) and ( [ psi_eqn_mut ] ) subject to the boundary conditions of the problem . to solve the ordinary differential equation ( [ psi_eqn_mut ] ) ( which includes the less general eq .
( [ psi_eqn ] ) as a special case ) it is useful to cast it in a standard form . to this end
, we introduce the function @xmath281 by writing @xmath282 where @xmath283 is a constant which is to be appropriately chosen shortly . substituting eq .
( [ f_defn ] ) into eq .
( [ psi_eqn_mut ] ) one finds ( dropping the subscripts and superscripts on @xmath284 ) @xmath285 f ' \\ + \left [ 2\lambda - ( \gamma_{i}+1)(\gamma_{i}+2 ) + 2r_{i}(\gamma_{i}+1 ) \right ] f \\ = \frac{\left[2\lambda ' - \gamma_{i}(\gamma_{i } + 1 ) - 2\gamma_{i}(m_{i}-r_{i})\right]}{1-u_{i}}\,f\ , .
\label{f_eqn}\end{gathered}\ ] ] we will choose @xmath283 so that the term involving @xmath286 in the denominator vanishes , that is , we ask that @xmath287 this is a quadratic equation for @xmath283 , but it is simple to see that if @xmath283 is one solution , the second one is simply @xmath288 .
it will turn out that either choice leads to the same form for @xmath279 . if we choose @xmath283 to satisfy eq .
( [ gammai_choice ] ) , then eq . ( [ f_eqn ] ) has the form @xmath289 f ' - ab f = 0\ , , \label{hypergeo}\ ] ] where @xmath290 the solution of the equation has now been reduced to a standard form , since eq .
( [ hypergeo ] ) is the hypergeometric equation @xcite , whose solutions are hypergeometric functions @xmath291 .
the details of the analysis of this equation are given in appendix c ; here we only describe the general features and specific form of the solution , without going into their derivation .
that is , we find the constants @xmath283 and @xmath93 . in the following discussion it is convenient to treat the cases where mutations do and do not occur separately .
we begin with the situation where mutations are absent , and so the pdf is found by solving eq .
( [ hypergeo ] ) , but where now @xmath292 the @xmath283 are chosen so that @xmath293 .
we begin the discussion , for orientation , by assuming that only two alleles are present .
this amounts to solving eq .
( [ neutral_1 ] ) or eq .
( [ psi_eqn ] ) with @xmath294 , and the solution is therefore the function @xmath295 .
this function also appears in the solution for arbitrary @xmath7 , given by eq .
( [ sep_soln_n ] ) , and so this result is also required in the general case . since in this simple case of two alleles @xmath296 , it follows that @xmath297 or @xmath298 . as discussed in appendix c ,
either choice gives the same solution , and we therefore take @xmath297 . we have now only to solve the second order ordinary differential equation ( [ hypergeo ] ) subject to boundary conditions at @xmath299 and @xmath300 . as is familiar in such problems @xcite , the general solution is first expressed as a linear combination of two independent solutions .
one of these is ruled out by one of the boundary conditions ( e.g. the one at @xmath299 ) and the application of the other boundary condition constrains a function of @xmath93 to be an integer ; in our case one finds @xmath301 .
details are in appendix c , but we do wish to mention the nature of the boundary conditions here .
technically @xcite these are _ exit _ boundary conditions which means that all the probability which diffuses to the boundary is `` extracted '' from the interval @xmath80 .
navely , one might imagine that it is appropriate to impose absorbing boundary conditions , i.e. , that the pdf vanishes at @xmath302 and @xmath303 .
however , since the diffusion coefficient , defined in eq .
( [ kol_2 ] ) , vanishes on the boundaries the diffusion equation itself imposes some sort of absorption .
a more careful mathematical analysis , such as the informal argument presented in appendix a , reveals that the appropriate constraint on the eigenfunctions is that they should diverge less rapidly than a square - root at both boundaries .
the result @xmath304 imposed by the boundary conditions together with the definitions ( [ abc_defn_neut ] ) imply that @xmath305 and so @xmath306 the solutions of the differential equation will be labelled by the integer @xmath81 and consist of polynomials of order @xmath81 .
these have already been mentioned in section [ overview ] ( eq .
( [ soln_2 ] ) ) , whilst more technical information can be found in appendix c. now we can move on to the solution for general @xmath7 . as explained above it
is sufficient to solve eq .
( [ psi_eqn ] ) where @xmath307 is given by the @xmath93 found in the solution of the @xmath25 allele problem .
we have seen that for @xmath59 , this is given by ( [ lambda_1_quant ] ) , and so this will be the @xmath307 for the @xmath71 problem .
we will see that this will have the general structure @xmath308 , where @xmath309 is an integer , and therefore we take this form for @xmath307 . from the way that the @xmath25 problem is embedded in the @xmath7 allele problem , we always have only to solve the @xmath310 equation and so we need to determine @xmath311 .
this is determined by the equation @xmath312 from which we see that @xmath313 or @xmath314 .
again , both give the same result and so we take the former . applying the boundary conditions again gives @xmath304 , with @xmath315see appendix c for the details .
using eq .
( [ abc_defn_neut ] ) one sees that @xmath316 and so @xmath317 where @xmath318 .
this justifies the choice for @xmath307 . in summary
, we have found that @xmath319 , @xmath320 . if we define @xmath321 then @xmath322 , and we have recovered ( [ lambda_and_l ] ) .
the general solution is given by eqs .
( [ soln_m_first])-([soln_m_fourth ] ) and discussed further in appendix c. when mutations are present we have to solve eq .
( [ hypergeo ] ) subject to the general set of parameter values given by eq .
( [ abc_defn ] ) . while it is true that this slightly complicates the analysis as compared with that of the mutation - free case , the most significant change is the nature of the boundary conditions .
the introduction of mutations in the way we have done here renders the boundaries reflecting , which are defined as having no probability current through them .
the current @xmath323 satisfies the continuity equation @xcite @xmath324 where ( compare with eqs .
( [ kol_m_u ] ) and ( [ cala_cald ] ) ) , @xmath325 - \frac{\partial } { \partial u_{i } } \left [ \mathcal{d}_{i}({\underline}{u } ) \mathcal{p } ( { \underline}{u},t ) \right ] \nonumber \\ & = & \frac{1 } { \prod_{j < i}(1-u_j ) } \left\ { \left ( m_{i } - r_{i } u_{i } \right ) \mathcal{p } - \frac{1}{2 } \frac{\partial } { \partial u_i } \left [ u_i(1-u_i ) \mathcal{p } \right ] \right\}\ , .
\label{current_defn}\end{aligned}\ ] ] using the separable form of the solution ( [ sep_soln_n_mut ] ) , and asking that the current is zero on the boundaries lead to the conditions @xmath326 \right\}_{u_{i}=0,1 } = 0\ , , \label{reflectbc_psi}\ ] ] or in terms of the solutions of the hypergeometric equation @xmath327 \left ( 1 - u_{i } \right)^{\gamma_i } f^{(i)}_{\lambda_{i};\lambda_{i-1 } } ( u_i ) \right . &
+ & \nonumber \\
u_i \left ( 1-u_i \right)^{\gamma_{i}+1 } \frac{{{\rm d}}}{{{\rm d}}u_i } f^{(i)}_{\lambda_{i};\lambda_{i-1}}\right\}_{u_{i}=0,1 } & = & 0\ , . \label{reflectbc_f}\end{aligned}\ ] ] as before , we will discuss the general aspect of the solution here , deferring technical details to appendix c. we begin with the case of two alleles .
the solution to the kolmogorov equation when only two alleles are present has been given by crow and kimura @xcite and in our case corresponds to the solution of eq .
( [ psi_eqn_mut ] ) with @xmath328 .
it then follows from eq .
( [ gammai_choice ] ) that we may take @xmath297 ( note that @xmath277 takes on the value @xmath155 in this case , since there is only one variable in the problem : @xmath329 ) .
it therefore follows from eq .
( [ abc_defn ] ) that @xmath330 and @xmath331 .
we begin by examining the nature of the solutions of eq .
( [ hypergeo ] ) in the vicinity of @xmath299 .
three separate cases have to be examined : ( i ) @xmath332 not an integer , ( ii ) @xmath333 , and ( iii ) @xmath334 .
when the boundary condition ( [ reflectbc_f ] ) is applied , one of the two independent solutions is ruled out .
the application of the other boundary condition again imposes a condition on @xmath335 . in appendix c , we show that @xmath336 where @xmath81 is a non - negative integer and , as @xmath59 , @xmath337 .
the solutions of the differential equations are again labelled by an integer @xmath81 , and once again turn out to be jacobi polynomials which have been given in section [ overview ] . when determining the solution for general @xmath7 , the eigenvalue @xmath307 in eq .
( [ psi_eqn_mut ] ) is given by the eigenvalue @xmath93 found in the @xmath25 allele problem , as occurred in the special case of no mutations .
however , more care is required here , since the iterative nature of the problem manifests itself in such a way that the solution with @xmath7 alleles is determined in terms of a function @xmath252 and the solution with @xmath25 alleles but with @xmath338 and @xmath339 , where @xmath340 . therefore although @xmath266 is given by eq .
( [ lambda_1_quant_mut ] ) , the @xmath307 we use for the @xmath341 problem is actually @xmath342 . for the general case we will see the @xmath307 will have the general structure @xmath343 where @xmath309 is an integer .
so having obtained the solution for @xmath344 from the @xmath25 solution , we need to solve the @xmath345 equation ( [ psi_eqn_mut ] ) where @xmath346 is a solution of eq .
( [ gammai_choice ] ) , that is , @xmath347 this equation has two solutions : @xmath348 and @xmath349 , and since both give the same result we take the former .
the implementation of the boundary conditions is carried out in the same way as in the two allele case .
this is discussed in appendix c where it is shown that this implies that @xmath350 where @xmath351 here the @xmath352 are non - negative integers .
the solutions are given explicitly by equations ( [ soln_m_first_m])-([soln_m_sixth_m ] ) .
in the study of most stochastic systems , the quantities which are of interest and which are most easily calculated are the mean and variance of the distribution and also the stationary pdf which the pdf tends to at large times .
this is still true in the problems we are considering in this paper , where these quantities can , in general , be rather easily obtained .
many are already known and have been given in section [ calculations ] and their derivation will be briefly discussed in this section . we draw particular attention to the added subtleties that exist when no mutation is allowed . in this case
the exact solutions obtained so far only hold in the _ open _ interval @xmath353 and do not include the boundaries @xmath354 or @xmath355 .
it is clear that with increasing time various alleles will become extinct and the stationary pdf will be concentrated entirely on the boundaries where it will accumulate .
this is somewhat different to the usual picture of absorbing states where the probability is removed entirely from the system . in order to calculate moments of allele frequencies , for example
, one must take care to add in the contributions from the boundaries to those obtained by averaging the frequencies over the pdfs we have so far determined .
this procedure will be demonstrated later in this section .
first , however , we calculate a few quantities of interest when the absence of a mutation process admits the fixation and extinction of alleles .
the probability that one allele has fixed by a time @xmath3 was first obtained by kimura @xcite by way of a moment formula for the distribution . here
we derive this quantity directly from the explicit solution when there are only two alleles .
we have described in section [ fixation ] how this can be extended to any number of alleles .
the definition of the current ( [ current_defn ] ) can be written as @xmath356 = - \frac{1}{2 } ( 1 - 2x ) p(x , t ) - \frac{1}{2 } x(1-x ) \frac{\partial p}{\partial x}\ ] ] in which the function @xmath357 is the probability distribution _ excluding _ any boundary contributions .
we find then at the boundary points @xmath18 and @xmath19 , @xmath358 we then find the probability for allele @xmath8 to have fixed by time @xmath3 is the sum of the current at the boundary : @xmath359 which can be evaluated by inserting the explicit expression ( [ soln_2 ] ) .
this procedure yields @xmath360 using the identities ( [ ident1 ] ) and ( [ ident2 ] ) given in appendix c this can be written @xmath361 e^{-(l+1)(l+2)t/2}\ , \label{pi_xequals1_2_alt3}\ ] ] where @xmath147 is a legendre polynomial , in accordance with the result of @xcite .
the probability that @xmath8 has been eliminated by time @xmath3 is the same as the probability that @xmath9 ( with initial proportion @xmath362 ) has become fixed . in other words , @xmath363 e^{- ( l+1)(l+2 ) t/2}\ , , \label{pi_xequals0_2_alt1}\end{aligned}\ ] ]
where we have used the fact that @xmath364 . as noted in section [ calculations ] ,
combinatorial arguments can be used to calculate the probability that exactly @xmath153 alleles coexist at time @xmath3 from the fixation probability just derived . to do this ,
divide the complete set of @xmath7 alleles into two groups @xmath150 and @xmath151 , the first containing a particular subset of @xmath153 alleles . as argued in section [ calculations
] , the probability that all of the @xmath365 alleles in @xmath151 have become extinct by time @xmath3 , leaving _ some _ of those in @xmath150 remaining is just @xmath366 , where @xmath14 is the total initial frequency of all alleles in @xmath150 . to find the probability that _ all _ the alleles in @xmath150 continue to coexist at time @xmath3 , we must subtract from @xmath366 the probability that one or more of them has been eliminated .
for example the probability that only the pair of alleles @xmath171 and @xmath367 coexist at time @xmath3 is @xmath368 as shown by kimura @xcite .
one then finds the probability that exactly two alleles remain at time @xmath3 to be the sum of the previous expression over all distinct pairs of indices @xmath369 .
this gives @xmath370 similarly , the probability that only the triple of alleles @xmath166 , @xmath371 and @xmath372 all coexist at time @xmath3 is the probability @xmath373 that some subset of these alleles remains at time @xmath3 , minus the probability that only any particular pair or single allele from @xmath374 and @xmath372 coexist , that is , @xmath375 \\
-[f_1(x_{i_2,0}+x_{i_3,0},t)-f_1(x_{i_2,0},t)-f_1(x_{i_3,0},t ) ] \\ -[f_1(x_{i_1,0}+x_{i_3,0},t)-f_1(x_{i_1,0},t)-f_1(x_{i_3,0},t ) ] \\
-f_1(x_{i_1,0},t)-f_1(x_{i_2,0},t)-f_1(x_{i_3,0},t)\ , .
\label{3fs}\end{gathered}\ ] ] simplifying , we find that @xmath376 where @xmath377 is a subset of @xmath378 with @xmath156 elements and @xmath379 are the elements of that subset .
proceeding in this way , one finds that the probability that the @xmath153 alleles @xmath380 all coexist at time @xmath3 is given by @xmath381 where now the second summation is over a subset of @xmath382 with @xmath156 elements . the result ( [ specific_comb_r ] )
can be proved by induction , by beginning with the expression for @xmath383 ( compare with eq .
( [ 3fs ] ) which is the case @xmath384 ) : @xmath385 if we now assume the result ( [ specific_comb_r ] ) up to and including @xmath153 alleles , then by substituting these expressions for @xmath386 into the right - hand side of eq .
( [ chi_r+1 ] ) , we obtain the expression ( [ specific_comb_r ] ) , but with @xmath153 replaced by @xmath387 .
since we have explicitly verified the result for low values of @xmath153 , the result is proved . during the course of the proof
the double summation is rearranged using @xmath388 and the fact that a particular set @xmath389 appears @xmath390 times in the sum is used . the analogue of eq .
( [ any_2 ] ) in this case , namely the probability that exactly @xmath153 alleles remain at time @xmath3 is the sum of eq .
( [ specific_comb_r ] ) over all distinct @xmath153-tuplets of indices @xmath391 : @xmath392 substituting the result ( [ specific_comb_r ] ) into this equation , and using exactly the same manipulations that were required in the inductive proof of ( [ specific_comb_r ] ) , gives the general result for @xmath153 alleles stated earlier : eq .
( [ prob_exactly_r ] ) ; see also @xcite .
as is well - known ( see , e.g. , @xcite ) the mean time @xmath393 to reach any boundary from an initial position @xmath394 @xmath395 where @xmath396 is the backward kolmogorov operator appearing in ( [ bke ] ) .
the boundary conditions on @xmath393 are that it vanish for any @xmath394 corresponding to a boundary point . for the case @xmath59 and no mutations ,
we seek the solution of @xmath397 that has @xmath398 .
two successive integration steps yield the required answer ( [ tau_1 ] ) .
a related quantity , which is obtained in a similar fashion , is the mean time to fixation of allele @xmath8 , _ given _ that it does become fixed .
this is given by @xmath399 since @xmath145 is the probability that allele @xmath8 has become fixed by time @xmath3 .
since @xmath400 as @xmath401 , the denominator equals @xmath14 . to find the numerator , we multiply the backward kolmogorov equation by @xmath3 and integrate over all @xmath3 .
use of eqs .
( [ flux_2 ] ) and ( [ f_and_p ] ) then shows that the numerator obeys the equation @xmath402 this yields the result @xmath403 to find the mean time to the @xmath153th extinction event from an initial condition with @xmath7 alleles , note that the probability @xmath404 that @xmath405 or fewer alleles coexist at a time @xmath3 can be found by summing ( [ prob_exactly_r ] ) appropriately .
one finds @xmath406 where we have rearranged the double summation using @xmath407 and also used the identity @xmath408 .
differentiating @xmath404 with respect to time gives the probability that a state comprising @xmath405 alleles is entered during the time interval @xmath409 $ ] .
this corresponds to the @xmath153th extinction event , where @xmath410 .
hence the mean time to this extinction , the first moment of the distribution , is given by @xmath411 the denominator being unity , since @xmath412 . however , from eqs .
( [ fixtime ] ) and ( [ fixtime_result ] ) , @xmath413 this gives an expression for the function which we wish to evaluate in eq .
( [ tau_r(m ) ] ) . changing the summation variable from @xmath156 to @xmath414 , and so identifying @xmath415 with @xmath416 gives @xmath417 \ln \left [ x_{i_1,0 } + \cdots + x_{i_s,0 } \right]\,.\ ] ] let @xmath418 be the probability that in the evolution allele @xmath7 becomes extinct first , followed by allele @xmath25 , @xmath419 and so on , ending with fixation of allele @xmath155 .
littler @xcite found the result given in eq .
( [ qp ] ) , but we give a derivation here .
we define @xmath420 as the probability that this sequence of extinctions has occurred by time @xmath3 .
so for example , in the biallelic system ( @xmath59 ) , @xmath421 .
just as one can show that @xmath422 obeys the @xmath59 backward kolmogorov equation ( by setting @xmath10 equal to its value on the boundary @xmath19 and relating @xmath422 to @xmath423 ) , then in general one can show that @xmath424 satisfies the backward kolmogorov equation ( [ bke ] ) @xmath425 the function @xmath418 is the stationary ( i.e. , @xmath401 ) solution of this equation that satisfies the following boundary conditions .
first , @xmath426 for any @xmath394 that corresponds to any allele other than @xmath7 already being extinct .
that is , for any @xmath394 that has @xmath427 for any @xmath428 . on the hyperplane @xmath429
, we must have that @xmath430 equal to the probability of the subsequent extinctions taking place in the desired order by time @xmath3 . taking the limit @xmath193 , this boundary condition implies @xmath431 we have chosen this particular order of extinctions to demonstrate the result as it corresponds to the ordering implied in our definition of @xmath90 . in the @xmath90 variables ,
the backward kolmogorov operator reads @xmath432 we conjecture a solution @xmath433 clearly , @xmath396 annihilates this product , so it is a stationary solution of ( [ qfp ] ) .
the boundary condition that @xmath434 for @xmath435 , @xmath436 is also obviously satisfied .
furthermore , if @xmath437 , @xmath438 hence the recursion ( [ recur ] ) is satisfied . it is easily established that @xmath439 by finding the stationary solution of the backward fokker - planck equation ( [ qfp ] ) with @xmath59 and imposing the boundary conditions @xmath440 and @xmath441 .
therefore by induction , ( [ qm_u ] ) is the solution required .
rewriting in terms of the @xmath10 variables , @xmath442 the probability for an arbitrary sequence of extinctions : allele @xmath443 to go extinct first , followed by @xmath444 leaving only allele @xmath445 can be determined by permuting the indices in ( [ qm ] ) .
this then gives us ( [ qp ] ) .
we earlier obtained the time - dependent pdf valid when mutation rates are nonzero by imposing reflecting boundary conditions .
these have the effect of preventing currents at the boundaries , which in turn ensure that the limiting @xmath193 solution is nontrivial and hence the complete stationary distribution .
one can check that this is the case from ( [ soln_m_first_m])([soln_m_sixth_m ] ) .
when all the integers @xmath101 are zero , the eigenvalue in ( [ lambda_and_l_m ] ) is also zero , indicating a stationary solution .
the remaining eigenvalues are all positive , which relate to exponentially decaying contributions to the pdf .
retaining then only the term with @xmath446 in ( [ soln_m_first_m ] ) we find immediately that @xmath447 it is easy to check that this distribution is properly normalised over the hypercube @xmath448 , @xmath449 . to change variable back to the allele frequencies @xmath43 we use the transformation ( [ trans_prob_m ] ) , and using the fact that @xmath450 , @xmath451 and @xmath452 , we arrive at the result quoted earlier , ( [ pstarmut ] ) . when the mutation process is suppressed , one finds from ( [ lambda_and_l ] ) that all the eigenvalues are nonzero , indicating that the distribution we have derived is zero everywhere in the limit @xmath401 .
this means that the stationary solution comprises the accumulation of probability at boundary points , as stated in section [ calculations ] .
one way to obtain the moments is to perform averages over the distribution .
the mean and variance when mutation is present can easily be calculated in this way from ( [ soln_2_m ] ) .
calculating the mean directly from the explicit formula for the probability distribution ( [ soln_2 ] ) when fixation can occur is tricky because one must include the full , time - dependent formula for the fixation probability at the right boundary and the integrals one has to evaluate are not particularly convenient .
alternatively , we can exploit the kolmogorov equation written in the form of a continuity equation ( [ continuity ] ) .
this leads to a differential equation for each moment which depends on lower moments .
the first few moments can then be found iteratively in a relatively straightforward way .
we demonstrate the method by giving the derivation of the moments of the distribution when two alleles are initially present .
we then show that this method extends in a straightforward way to the case when @xmath7 alleles are present .
specifically for @xmath59 we have @xmath453 where the current @xmath454 and where @xmath116 .
when mutations in either direction are present ( i.e. , both @xmath455 and @xmath456 ) the mean of @xmath457 is given by the expression @xmath458 and so @xmath459_{x=0}^{1}\end{aligned}\ ] ] where we have used the continuity equation ( [ continuity_m2 ] ) and integrated by parts .
we have already stated that there is no current of probability through the boundaries .
in other words , here @xmath460 and so the last term in ( [ meanxbyparts ] ) is zero .
when fixation is a possibility , one _ does _ have a current at the boundaries and , the probability that allele @xmath8 is fixed at time @xmath3 being @xmath461 . since the function @xmath462 excludes contributions from the boundary in this case , we must add these explicitly into the mean of @xmath457 .
that is , @xmath463 taking the derivative of this expression and carrying out the integration by parts as in ( [ meanxbyparts ] ) , we obtain @xmath464_{x=0}^{1 } + \frac{\partial f_1(x_0 , t)}{\partial t } \;.\ ] ] these last two terms cancel , and so in either case we are left with @xmath465 inserting the expression ( [ current_m2 ] ) for the current we find @xmath466 \;.\ ] ] this reveals that moments of the distribution can be calculated iteratively .
this equation is valid whether or not the @xmath39 are non - zero .
the equation for the mean ( @xmath467 ) can be solved directly , and the results used to find @xmath468 ( @xmath469 ) and so on . when there are more than 2 alleles , a similar derivation gives the equation for the general moment @xmath470 : @xmath471\langle x_i^{k_i-1 } \prod_{j\neq i}x_j^{k_j}\rangle - \right.\\ \left .
[ \frac{1}{2}(\sum_lk_l-1)+r]\langle \prod_i x_i^{k_i}\rangle \right\}\,.\end{gathered}\ ] ] thus , again we can calculate any moment by iteration .
for example @xmath472 , with @xmath473 , obeys the equation @xmath474 and using ( [ meanx_m ] ) we find the result given in ( [ covarx_m ] ) .
in the last decade or two the ideas and concepts of population genetics have been carried over to a number of different fields : optimisation @xcite , economics @xcite , language change @xcite , among others . while the essentials of the subject such as the existence of alleles , and their mutation , selection and drift are usually present in these novel applications , other aspects may not have analogues in population genetics .
furthermore , phenomena such as epistasis , linkage disequilibrium , the relationship between phenotypes and genotypes , which form a large part of the subject of population genetics , may be unimportant or irrelevant in these applications .
our motivation for the work presented here has its roots in the mathematical modelling of a model of language change where it is quite plausible that the number of variants ( alleles ) is much larger than two .
it was this which led us to systematically analyse the diffusion approximation applied to multi - allelic processes , having noted that many of the treatments in the context of biological evolution tend to be couched in terms of a pair of alleles , the `` wild type '' and the `` mutant '' . in this paper we have shown how the kolmogorov equation describing the stochastic process of genetic drift or the dynamics of mutation at a single locus may be solved exactly by separation of variables in which the number of alleles is arbitrary .
the key to finding separable solutions is , of course , to find a transformation to a coordinate system where the equation is separable .
the change of variable ( [ cofv ] ) we used is similar , but slightly different to one suggested by kimura @xcite which he showed achieved separability up to and including @xmath475 alleles .
kimura was of the opinion that novel mathematical techniques would be needed to proceed to the general case of @xmath7 alleles .
what we have shown here is that with our change of variables this generalisation is possible without the need to invoke any more mathematical machinery than was required in the standard textbook case of @xmath229 alleles .
a large part of the reason for this is the way that the problem with @xmath476 alleles can be constructed in a straightforward way from the problem with @xmath7 alleles and that with @xmath229 alleles .
this simple `` nesting '' of the @xmath7-allele problem in the @xmath476-allele one is responsible for many of the simplifications in the analysis and is at the heart of why the solution in the general case can be relatively simply presented .
an illustration of the simple structure of this nesting is the fact that the general solutions , valid for an arbitrary number of alleles , with or without mutation , is made up of products of polynomials more specifically jacobi polynomials . although the higher order versions of these polynomials can be quite complex , even after relatively short times only those characterised by small integers are important .
as mentioned in section [ u_variables ] , we have given the solutions in terms of the transformed variables , but their form in terms of the original variables of the problem can be found using eqs .
( [ cofv ] ) and ( [ trans_prob_m ] ) .
we have also presented the derivations of many other quantities of interest . in the interests of conciseness
we have given only a flavour of some of these : some are new , some have already been derived by other means and yet others are simple generalisations of the two - allele results . yet
other results are more naturally studied in the context of the backward kolmogorov equation , and we also took the opportunity to gather together the most significant , but nevertheless little - known , ones here .
we have thus provided a rather complete description of genetic drift and mutation at a single locus .
nevertheless , there are some outstanding questions . for example , as discussed in appendix b , the transformation ( [ cofv ] ) does not render the kolmogorov equation separable for an arbitrary mutation matrix @xmath477 , only one where that rate of mutation of the @xmath478 alleles to @xmath479 alleles ( @xmath480 ) occurs at a rate independent of @xmath40 .
this is the reason for this simplified choice for the mutation matrix a choice also made in all other research we are aware of .
it might seem possible in principle to find another transformation which would allow a different form of @xmath477 to be studied , but this seems difficult for a number of reasons .
for instance , the transformation must still ensure that the diffusion part of the kolmogorov equation is separable .
furthermore , the matrix has @xmath481 entries , and the transformation only @xmath28 degrees of freedom so for only certain , restricted , forms of @xmath477 will a transformation to a separable equation be possible .
other questions involve the study of selection mechanisms or interactions between loci using the results here as a basis on which to build .
we are currently investigating these questions in the context of language models , but we hope that the work reported in this paper will encourage further investigations along these lines among population geneticists .
gb acknowledges the support of the new zealand tertiary education commission .
rab acknowledges support from the epsrc ( grant gr / r44768 ) under which this work was started , and a royal society of edinburgh personal research fellowship under which it has been continued .
as we have mentioned several times in the main text , although it might be thought that adding the possibility that mutations occur would complicate the problem , in many ways it is the situation of pure random mating that is the richer mathematically .
so , for example , the case with mutations has a conventional stationary pdf given by eq .
( [ pstarmut ] ) , whereas without mutations the stationary pdf is a singular function which exists only on some of the boundaries .
of course , it is clear from the nature of the system being modelled that this has to be the case , but what interests us in this appendix is the nature of the boundary conditions which have to be imposed on the eigenfunctions of the kolmogorov equation ( [ kol_m ] ) to obtain the correct mathematical description .
the nature of the boundary conditions required when fixation can occur are discussed in the literature both from the standpoint of the kolmogorov equation in general @xcite and in the specific context of genetic drift @xcite . here
we will describe a more direct approach , which while not so general as many of these discussions , is easily understood and illustrates the essential points in an explicit way .
it is sufficient to discuss the one - dimensional ( two allele ) case , since all the points we wish to make can be made with reference to this problem .
the question then , can be simply put : what boundary conditions should be imposed on eq .
( [ kol_2 ] ) ? on the one hand
one might feel that that these should be absorbing boundary conditions @xcite , because once a boundary is reached there should be no chance of re - entering the region @xmath482 . on the other hand
, however , the diffusion coefficient `` naturally '' vanishes on the boundaries which automatically guarantees absorption , so there would seem to be no need to further impose a vanishing pdf ( as an absorbing boundary condition would require ) . in that case , what boundary conditions should be imposed ? also is it even clear , given the fact that the diffusion coefficient @xmath483 becomes vanishingly small as the boundaries are approached , that the boundaries can be reached in finite time ? to address these questions let us begin with the kolmogorov equation ( [ kol_2_mut ] ) which includes a deterministic component @xmath484 as well as a random component represented by the diffusion coefficient @xmath17 .
although we are interested in the case when @xmath484 is not present , it will help in the interpretation of the results if we initially include it .
we also make two notational changes we put a subscript @xmath485 on @xmath484 which will be explained below and we will write @xmath486 .
the function @xmath487 is real , since @xmath488 .
therefore our starting equation is @xmath489 + \frac{1}{2 } \frac{\partial^{2}}{\partial x^2 } \left [ g^{2 } ( x ) p \right]\ , .
\label{ito}\ ] ] it turns out that it will more useful for our purposes to write this in the form @xmath490 + \frac{1}{2 } \frac{\partial } { \partial x } \left [ g(x ) \frac{\partial } { \partial x } \left ( g ( x ) p \right ) \right ] \nonumber \\ \nonumber \\ & = & -\frac{\partial } { \partial x } \left [ a_{s}(x ) p \right ] + \frac{1}{2 } \frac{\partial } { \partial x } \left [ g(x ) \frac{\partial } { \partial x } \left ( g ( x ) p \right ) \right]\ , .
\label{strat}\end{aligned}\ ] ] where @xmath491 equations ( [ ito ] ) and ( [ strat ] ) are known as the ito and stratonovich forms of the kolmogorov equation respectively @xcite .
there is no need for us to explore the differences between these two formulations ; the only relevant point here is that the stratonovich form is more convenient for our purposes .
we now transform eq .
( [ strat ] ) by introducing the pdf @xmath492 which is a function of the new variable @xmath493 defined by @xmath494 the transformed equation reads @xmath495 + \frac{1}{2 } \frac{\partial^{2}q}{\partial y^2}\ , , \label{pdf_y}\ ] ] where @xmath496 the system with an @xmath10-dependent diffusion coefficient has now been transformed to one with a state - independent diffusion coefficient , at the cost of adding an additional factor to the deterministic term .
the problem of interest to us has @xmath497 and @xmath498 . from eq .
( [ x_to_y ] ) we find the required transformation to be @xmath499 or @xmath500 with @xmath501 , and from eq .
( [ trans_a_defn ] ) we find that @xmath502 .
one way to understand this process intuitively is through a mechanical analogy : if we set @xmath503 , then the kolmogorov equation can be thought of as describing the motion of an overdamped particle in the one - dimensional potential @xmath504 , subject to white noise with zero mean and unit strength .
an examination of fig .
[ potentials ] shows that the boundaries are reached in a finite time .
more importantly , from the relation @xmath505 we see that imposing the absorbing boundary conditions @xmath506 only implies that the pdf @xmath507 must diverge with a power smaller than @xmath508 at the boundaries @xmath18 and @xmath19 .
these results should be contrasted with the hypothetical situation where @xmath509 , but @xmath510 . in this situation
we have that @xmath511^{-1}$ ] or @xmath512 with @xmath513 , and from eq .
( [ trans_a_defn ] ) we find that @xmath514 .
this corresponds to a process in which an overdamped particle is moving in the potential @xmath515 and subject to white noise . for large @xmath516 , @xmath517 , which indicates that it will take an infinite time to reach the boundaries
this potential is shown on the right - hand side of fig .
[ potentials ] .
considerations such as these complement more mathematically rigorous studies by giving insights into the nature of the processes involved when the diffusion coefficient is dependent on the state of the system . in the actual genetic drift system .
right : potential @xmath518 in the hypothetical system described in the text.,title="fig : " ] in the actual genetic drift system .
right : potential @xmath518 in the hypothetical system described in the text.,title="fig : " ]
the transformation from the coordinate system defined in terms of the set of allele frequencies , @xmath43 , to that denoted by @xmath100 , for which the kolmogorov equation is separable , is given by eq .
( [ cofv ] ) , with the inverse transformation given by eq .
( [ cofv_inverse ] ) . in this appendix
we explore the nature of this transformation , determine the jacobian of the transformation and calculate how the jump moments in the new coordinate system are related to those in the old one .
this enables us to derive the kolmogorov equation in the new variables given by eqs .
( [ kol_m_u ] ) and ( [ cala_cald ] ) in the main text .
we begin by looking at the nature of the transformation for low values of @xmath7 .
@xmath519 : @xmath520 .
@xmath521 .
@xmath522 : @xmath523 . since @xmath524 also satisfies this condition , then @xmath525 .
@xmath526 therefore , @xmath527 , with the three lines @xmath528 and @xmath529 go over to the three lines @xmath530 and @xmath531 respectively .
the point @xmath532 goes over to the line @xmath533 .
this is illustrated in fig .
[ cov_3 ] @xmath534 : @xmath535 with the additional condition @xmath536 .
@xmath537 therefore , @xmath538 .
the planes @xmath539 get mapped on to the planes @xmath354 ( @xmath540 ) , the plane @xmath541 gets mapped on to the plane @xmath542 , the line @xmath543 gets mapped to the plane @xmath544 and the point @xmath545 gets mapped to the plane @xmath546 .
this is illustrated in fig .
[ cov_4 ] .
: @xmath547 with the additional condition @xmath548 . from eqs .
( [ cofv ] ) and ( [ cofv_inverse ] ) we see that @xmath549 .
the pattern from the @xmath71 and @xmath158 cases should now be clear : the hyperplanes @xmath539 get mapped on to the hyperplanes @xmath354 ( @xmath550 ) , the hyperplane @xmath551 , gets mapped on to the hyperplane @xmath552 , the point @xmath553 gets mapped to the hyperplane @xmath554 .
so , in summary , the region @xmath555 lying between the origin and the hyperplane @xmath556 , gets mapped into the unit hypercube @xmath557 .
however , parts of the hyperplane @xmath556 of dimensions @xmath558 get mapped on to the hyperplanes @xmath559 .
the jacobian of the transformation from @xmath90 to @xmath207 is easily found if we note from ( [ cofv_inverse ] ) that @xmath560 if @xmath561 .
this implies that the jacobian has zeros if the column label is greater than the row label , and so the determinant is just the product of the diagonal elements .
since the @xmath277th diagonal element is simply @xmath562 , we have that @xmath563 as explained in section [ u_variables ] , the pdfs in the @xmath90 variables , denoted by @xmath564 , are related to those in the @xmath207 variables , denoted by @xmath507 , by a multiplicative factor which is simply the jacobian : @xmath565 from which ( [ trans_prob ] ) follows . the final point which we wish to discuss in this appendix
is the form of the kolmogorov equation in the new @xmath90 variables .
as discussed in section [ soln ] , the easiest way to determine this is to relate the jump moments in the old coordinate system , defined through eq .
( [ jm_defn ] ) , and those in the new coordinate system , defined through eq .
( [ jm_defn_new ] ) . to do this ,
let us begin with the first jump moment : @xmath566 where we have used ( [ trans_prob ] ) and where @xmath567 is the transformation ( [ cofv ] ) .
use of taylor s theorem then gives @xmath568 since jump moments higher than the second vanish in the limit @xmath212 , we find from eq .
( [ trans_jm_1 ] ) that @xmath569 similarly , @xmath570 the results ( [ trans_jm_1.1 ] ) and ( [ trans_jm_2 ] ) tell us how the jump moments transform from one coordinate system to another . in our case , we have from eq .
( [ cofv ] ) @xmath571 and @xmath572 the key feature of the two results ( [ du_by_dx ] ) and ( [ d2u_by_dxdx ] ) is that they only depend on @xmath277 , and not on @xmath573 or @xmath81 .
this simplifies the summations which appear in ( [ trans_jm_1.1 ] ) and ( [ trans_jm_2 ] ) .
a still fairly tedious calculation shows that the second term on the right - hand side of ( [ trans_jm_1.1 ] ) is zero and that the terms with @xmath197 in ( [ trans_jm_2 ] ) are also zero .
specifically , @xmath574 in this paper the only deterministic mechanism which we will consider will be mutation , described by the differential equation @xmath575 where @xmath477 is the rate of mutation of allele @xmath40 into allele @xmath277 . from eqs .
( [ du_by_dx ] ) and ( [ jm_1 ] ) @xmath576 we will denote the term in the brackets in eq .
( [ explicit_firstjm ] ) by @xmath577 , so that @xmath578 using eqs .
( [ km_expansion_new ] ) and ( [ jm_2 ] ) , we find the kolmogorov equation in the new coordinate system to be @xmath579 + \frac{1}{2 } \frac{\partial^{2 } } { \partial u_i^2 } \left [ u_{i}(1-u_i)\,\mathcal{p } \right ] \right\}\ , .
\label{kol_new_gen}\ ] ] in section [ soln ] we prove that the kolmogorov equation is separable in the @xmath90 coordinate system if @xmath577 depends on @xmath580 only , and not on the @xmath581 .
since @xmath582 involves a sum over the expression ( [ general_mut ] ) , which has itself to be written in terms of the @xmath90 using eq .
( [ cofv_inverse ] ) , it is clear that this will not be the case for general @xmath477 .
however , if it is assumed that the rate of mutation of the alleles @xmath583 to @xmath479 occurs at a constant rate @xmath39 , independent of @xmath40 , then @xmath584 for all @xmath585 and ( [ general_mut ] ) becomes @xmath586 in this case @xmath587 where @xmath588 and where we have used eq .
( [ cofv_inverse ] ) .
we see that @xmath577 only depends on @xmath100 and that @xmath589 where @xmath590 .
this leads to the kolmogorov equation @xmath591 + \frac{1}{2 } \frac{\partial^{2 } } { \partial u_i^2 } \left [ u_{i}(1-u_i)\;\mathcal{p } \right ] \right\}\ , .
\label{kol_new_spec}\ ] ] this is the content of eqs .
( [ kol_m_u ] ) and ( [ cala_cald ] ) in the main text .
in this appendix we will provide further details relating to the solution of the equations ( [ psi_eqn_1 ] ) and ( [ psi_eqn ] ) subject to the appropriate boundary conditions .
some of the details have already been given in section [ soln ] , in particular how the substitution ( [ f_defn ] ) leads to the hypergeometric equation ( [ hypergeo ] ) .
consequently , much of this appendix will be concerned with the hypergeometric functions @xmath592 which are the solutions of this equation .
we will refer to chapter 15 of the standard handbook by abramowitz and stegun @xcite for the formul that we use , but several appear sufficiently often that it is worth stating them explicitly here : * if @xmath593 and @xmath594 , then @xmath595 * @xmath596 where @xmath597 and @xmath598 .
note that @xmath599 . *
if @xmath335 or @xmath600 is a non - positive integer then the power series for @xmath592 terminates : @xmath601 where @xmath602 is a jacobi polynomial of order @xmath1 .
as we will see , the solution of the hypergeometric differential equation typically proceeds as follows .
the general solution to the equation ( [ hypergeo ] ) is written out as @xmath603 , where @xmath604 and @xmath605 are two independent solutions of the equation , and @xmath606 and @xmath607 are two arbitrary constants .
examining the general solution in the vicinity of the boundary at @xmath608 rules out one of the solutions , so that , for instance one must take @xmath609 in order to satisfy the boundary condition there .
the remaining solution is now examined in the vicinity of the other boundary at @xmath610 .
if the solution has the form @xmath611 with @xmath612 then we may use result 1 above .
if @xmath613 , then we may use result 2 followed by result 1 .
if it is found that either @xmath614 or @xmath615 is a non - positive integer , then result 3 may be invoked .
the values of the parameters are given in section [ soln_neutral ] , where it is seen from eq .
( [ abc_defn_neut ] ) that @xmath616 .
one of the solutions of the hypergeometric equation when @xmath616 ( suppose it is @xmath617 ) has the following behaviour for small @xmath618 @xcite : @xmath619 and so this solution would not lead to a finite pdf as @xmath620recall from appendix a that only divergences less strong than @xmath621 will be tolerated .
therefore @xmath609 .
the remaining solution is @xmath622 . from eq .
( [ abc_defn_neut ] ) , @xmath623 , so if we choose @xmath624 , as described in section [ soln_neutral ] , we may use result 2 above to show that @xmath625 now @xmath626 and so we may use result 1 to show that @xmath627 we may once again use the result of appendix a to deduce that this term can not be present since it is diverging too fast as @xmath628 ( the analysis of appendix a can be straightforwardly generalised to cover the equations that are found from separation of the @xmath7 allele kolmogorov equation ) .
we therefore require that either @xmath335 or @xmath600 equals @xmath629 , where @xmath81 is a non - negative integer .
since the hypergeometric function is symmetric in @xmath335 and @xmath600 it does not matter which we choose : let @xmath304 . from result 3
this means that the hypergeometric function is terminating , and more specifically , @xmath630 as shown in section [ soln_neutral ] , @xmath631 , if we take @xmath624 , and so up to a constant @xmath632 we discuss some properties of jacobi polynomials below .
note that if we had made the choice @xmath633 in section [ soln_neutral ] ( _ i.e. _ @xmath634 ) , then the argument given above would still rule out solution @xmath605 using eq .
( [ neut_utozero ] ) , but we would have found the remaining solution to be @xmath635 where @xmath636 and @xmath637 are the parameter solutions relevant for the choice @xmath638 and are related to those for the former @xmath624 choice by @xmath639 and @xmath640 . this can be easily seen from eq .
( [ abc_defn_neut ] ) : if @xmath641 is one solution of @xmath642 , then a second one is @xmath643 .
but then eq . ( [ abc_defn_neut ] ) is unchanged if @xmath644 and @xmath600 are replaced by @xmath645 and @xmath646 .
so for the choice @xmath647 we have shown that @xmath648 where the @xmath644 and @xmath600 are those for the @xmath624 choice .
this is exactly eq .
( [ neut_utoone ] ) and so we again deduce that @xmath304 , where @xmath81 is a non - negative integer . using result 2
we then have that @xmath649 which is eq .
( [ nuet_finalpsi ] ) .
the final step in calculating the solutions is to apply the initial conditions to find the appropriate coefficients .
we use the appropriate orthogonality relationship ( which we shall give below ) and use the method described in ( [ initial_conds])-([post_initial_conds ] ) , and find @xmath650 and @xmath651 as stated in ( [ soln_m_first ] ) and ( [ soln_m_third ] ) .
the values of the parameters are given in section [ soln ] , where it is seen from eq .
( [ abc_defn ] ) that @xmath652 .
the nature of the solutions of eq .
( [ hypergeo ] ) depends on whether @xmath332 is an integer , and if it is , what its value is @xcite .
the simplest case is if @xmath332 is not an integer , then the general solution has the form @xmath653 both of these hypergeometric functions are of the form @xmath654 with @xmath655 and so have power - series expansions near @xmath608 @xcite . substituting the expression ( [ c_noninteger ] ) into the boundary condition ( [ reflectbc_f ] ) gives @xmath656 . the analysis when @xmath333 and @xmath334 has to be done separately .
one of the solutions has a factor of @xmath657 in these cases , but it turns out that , while these solutions have a current which does not diverge at @xmath608 , the current is non - zero , which is forbidden by the reflecting boundary conditions . therefore this solution must be absent .
it is found that in all cases the implementation of the boundary condition at @xmath608 shows that the required solution is @xmath658 .
we now note that the conditions which determine the parameters @xmath335 and @xmath600 from eq .
( [ abc_defn ] ) can be simplified if we introduce a new parameter @xmath659 through @xmath660 .
then @xmath661 , where @xmath662 .
so choosing @xmath663 , this being one of the solutions of eq .
( [ gamma_eqm_mut ] ) , we have that @xmath664 and @xmath665 and therefore @xmath666 . using result 2
we may write this in an alternative form : @xmath667 we may now apply the boundary condition that the current vanishes when @xmath610 , which leads to eq .
( [ reflectbc_f ] ) . after some algebra we find that this condition implies @xmath668 we may now use result 1 to show that for @xmath669 the term in the second set of curly brackets has a finite limit as @xmath628 , specifically this term tends to @xmath670 thus either @xmath671 or @xmath672 must be a non - positive integer if we are to obtain a finite current ( let alone a zero - current ) as @xmath628 .
the case @xmath673 may be treated separately , since the first term in the bracket , to which result 1 does not immediately apply is absent .
the same result is found . to show that these conditions also imply a zero current
, we may substitute @xmath674 or @xmath675 into eq .
( [ alt_form ] ) and find , when @xmath676 for instance that @xmath677 since by result 3 the two hypergeometric functions are polynomials of order @xmath81 , the right - hand side is indeed zero if @xmath669 , and also when @xmath673 , since the first term in the second curly bracket of eq .
( [ bc_uequals1 ] ) is absent .
note that if we had made the choice @xmath675 then the hypergeometric functions which would have appeared would have been of the form @xmath678 , which again are polynomials by result 3 . therefore , returning to eq .
( [ alt_form ] ) , we have found that @xmath679 using result 3 this is , up to a constant , @xmath680 where @xmath681 and @xmath682 .
writing it in terms of the function @xmath257 we have that @xmath683 it is also straightforward to check that if we made the choice @xmath684 , rather than @xmath663 , then the analysis near @xmath608 would be as before , but then use of result 2 would then show that the result we obtain was exactly as before ( but with @xmath335 and @xmath600 interchanged ) . in summary
, we have shown that @xmath674 , where @xmath81 is a non - negative integer .
thus for @xmath59 , when @xmath685 , and so from @xmath686 , we obtain eq .
( [ lambda_1_quant_mut ] ) . for @xmath71 , we use eq .
( [ lambda_prime_mut ] ) , with @xmath687 and so obtain @xmath265 given by @xmath688 where @xmath689 , and @xmath81 is a non - negative integer and where @xmath690 .
continuing in this way we find the result given by eq .
( [ lambda_n_mut ] ) .
all the solutions for the pdfs considered in this paper consist of jacobi polynomial @xmath691 .
all properties of these functions that we require are given in chapter 22 of abramowitz and stegun @xcite , but here we will list the key results that we use .
the first point to note is that the natural variable for these functions is @xmath692 .
thus while @xmath693,\ z \in [ -1,1]$ ] .
they satisfy the following orthogonality relation in the @xmath618 variable : @xmath694 where @xmath695 . in the eigenfunctions
calculated when mutations are present , the functions @xmath99 which make up the right eigenfunction @xmath97 as defined in eq .
( [ soln_m_second_m ] ) , contain the factor @xmath696 which appears in eq .
( [ orthog ] ) , as can be seen in eq .
( [ soln_m_third_m ] ) . on the other hand
the functions @xmath141 which make up the left eigenfunction @xmath139 as defined in eq .
( [ soln_m_fifth_m ] ) , do not contain this factor , as can be seen in eq .
( [ soln_m_sixth_m ] ) .
the result is that the right and left eigenfunctions satisfy the simple orthogonality relation @xmath697 . in the derivation of the fixation probability in section [ fixation ] we make use of the following relationship between jacobi and legendre polynomials which can be found in ref .
@xcite : @xmath700\,,\ ] ] as well as this identity for legendre polynomials , which can be found in ref .
@xcite : @xmath701\;.\ ] ] g. b. golding , c. strobeck , 2-locus , 4th - order gene - frequency moments - implications for the variance of squared linkage disequilibrium and the variance of homozygosity , theoretical population biology 24 ( 1983 ) 173191 .
r. k. p. zia , r. j. astalos , statistics of an aged stuructured population with two competing species : analytic and monte carlo studies , in : d. p. landau , s. p. lewis , h .- b .
schuttler ( eds . ) , computer simulation studies in condensed matter physics xiv , springer - verlag , berlin , 2002 , pp .
235254 .
m. kimura , diffusion models in population genetics with special reference to fixation time of molecular mutants under mutational pressure , in : t. ohta , k. aoki ( eds . ) , population genetics and molecular evolution , japan sci .
soc . press / springer - verlag , 1985 , pp .
1939 .
j. crow , m. kimura , some genetic problems in natural populations , in : j. neyman ( ed . ) , proceedings of the third berkeley symposium on mathematical statistics and probability , volume 4 , university of california press , berkeley , 1956 , pp . 122 . | we give an exact solution to the kolmogorov equation describing genetic drift for an arbitrary number of alleles at a given locus .
this is achieved by finding a change of variable which makes the equation separable , and therefore reduces the problem with an arbitrary number of alleles to the solution of a set of equations that are essentially no more complicated than that found in the two - allele case .
the same change of variable also renders the kolmogorov equation with the effect of mutations added separable , as long as the mutation matrix has equal entries in each row .
thus this case can also be solved exactly for an arbitrary number of alleles .
the general solution , which is in the form of a probability distribution , is in agreement with the previously known results which were for the cases of two and three alleles only .
results are also given for a wide range of other quantities of interest , such as the probabilities of extinction of various numbers of alleles , mean times to these extinctions , and the means and variances of the allele frequencies .
to aid dissemination , these results are presented in two stages : first of all they are given without derivations and too much mathematical detail , and then subsequently derivations and a more technical discussion are provided . , , population genetics , diffusion , kolmogorov equation , genetic drift , single - locus . |
gamma rays are widely used in contemporary technologies for material modification , food sterilization and testing for weak points in welded structures .
medical applications of @xmath4 rays include the imaging technique of positron emission tomography and radiation therapies to treat cancerous tumors as well as detecting brain and cardiovascular abnormalities . since the discovery of recoilless nuclear resonance by mssbauer mossbauer58 , craig59 , studies of the interaction between @xmath0 rays and mssbauer nuclear ensembles have undergone rapid development and have yielded many real and potential applications in , e.g. , mssbauer spectroscopy @xcite and quantum information @xcite . due to their small wavelength ,
@xmath0 rays are naturally suitable for achieving high spatial resolution and for making small quantum photonic circuits @xcite . however ,
control of @xmath0 rays still remains a challenging problem .
coherent effects , such as level mixing induced transparency coussement02 , electromagnetically induced transparency @xcite , @xmath0 echo @xcite , phase modulation @xcite , and the nuclear lighthouse effect @xcite , can be adopted to manipulate @xmath0 radiation .
modulation of mssbauer radiation by pulsed laser excitation was demonstrated in @xcite .
the total reflection of the grazing incidence was used to reflect @xmath0 rays , but application of this technique is limited due to small grazing angle .
development of the @xmath0-ray optics led to the design of the laue lens barriere09 via nuclear bragg diffraction @xcite .
it has also been suggested that @xmath0 rays can be manipulated using delbrck scattering @xcite .
effective control of @xmath0 rays requires further advancements and innovations .
development of a fast switch of @xmath0 rays is important for extending the time resolution of @xmath0-ray sources and for increasing the operating speed of @xmath0-ray quantum information processing .
nanosecond @xmath0-ray switching has been realized by magnetically manipulating nuclear excitation based on the quantum beat in nuclear bragg scattering @xcite .
picosecond x - ray bragg switch utilizing laser - generated phonons was proposed @xcite and later demonstrated experimentally @xcite . in this paper
we investigate a way to control propagation of a @xmath0-ray beam through a crystal by controlling its collective absorption and reemission by many nuclei .
collective spontaneous emission from atomic ensembles has been a subject of long - standing interest since the pioneering work of dicke @xcite .
the collective nature of light interaction yields fascinating effects such as superradiance and radiation trapping even at the single - photon level .
recent studies focus on collective , virtual and nonlocal effects in such systems scully06,eberly06,mazets07,svid08,svid08a , frie08a , frie08,porras08,scully09science , pedersen09,svid09,svid10a , friedberg10,svid10,berman11,svid12,ji07,zhangxf13 .
the josephson effect for photons in two weakly linked microcavities is an example of the collective physics in coupled atom - cavity systems @xcite .
the interaction of light with ordered arrays of nuclei in crystals offers new perspectives .
for example , a photon collectively absorbed by a random medium ( e.g. , gas ) will be reemitted in the same direction as the incident photon @xcite .
however , in the case of a crystal lattice , collective reemission can occur in several directions ( bragg angles ) .
the interaction strength between the @xmath0-ray beam and the crystal depends on the detuning @xmath5 of the photon frequency from the nuclear transition .
here we show that one can redirect a @xmath0-ray beam into a desirable bragg angle by making the crystal lattice coherently vibrate with frequency @xmath3 which lies , e.g. , in the infrared region .
such lattice vibrations are in the combination parametric resonance with the frequency difference between two eigenmodes of the coupled light - nuclear system which results in resonant energy transfer from the incident @xmath0-ray beam to the wave propagating at the bragg angle .
this process is analogous to the parametric frequency mixing in propagating circuits @xcite .
nuclear vibrations can be generated by a driving laser pulse and can be turned on and off on a short time scale .
@xmath0-ray redirection , produced by parametric resonance , occurs on a time scale determined by the collective nuclear frequency @xmath1 which typically lies in the terahertz region .
this mechanism allows us to control propagation of high frequency @xmath0 photons by driving the system , e.g. , with an infrared laser .
we consider a perfect crystal composed of two - level ( @xmath6 and @xmath7 ) nuclei with transition frequency @xmath8 as shown in fig .
[ setup](a ) .
the nuclear transition frequency @xmath8 typically lies in the hard x - ray or @xmath0-ray region .
nuclei are located at positions @xmath9 and form a periodic lattice , where the index @xmath10 labels different nuclei .
typically , the inter nuclei spacing is much larger than the nuclear radiation wave length @xmath11 , where @xmath12 is the speed of light .
we assume that the lattice is coherently excited so that nuclei oscillate along the direction given by a unit vector @xmath13 around their equilibrium positions @xmath14 .
the oscillation frequency @xmath15 lies in the infrared or visible region . in ionic crystals such oscillations
can be produced , e.g. , by a strong linearly polarized driving laser pulse with frequency @xmath15 .
a typical example is potassium iodide crystal , which has a face - centered - cubic unit cell of iodide ions with potassium ions in octahedral holes . by applying an external driving field
one can make ions k@xmath16 and i@xmath17 move in opposite directions such that nuclei of the same species will oscillate in unison .
both k and i have mssbauer isotopes .
namely , @xmath18k has a mssbauer transition with energy @xmath19 kev and spontaneous decay rate @xmath20 s@xmath21 , while @xmath22i has a transition with energy @xmath23 kev and @xmath24 s@xmath21 .
we consider an interaction of high - frequency ( x- or @xmath0-ray ) photons with a coherently vibrating nuclear lattice of a particular mssbauer isotope .
the presence of nuclei of another species in the crystal is irrelevant since they have a very different transition frequency .
we assume that the motion of each nuclei @xmath10 involved in the interaction is given by @xmath25where@xmath26here @xmath27 and @xmath28 is the amplitude of the laser induced nuclei oscillations . in our model a weak , plane ,
linearly polarized @xmath0-ray wave with the wave vector @xmath29 and frequency @xmath30 detuned from the nuclear transition frequency @xmath8 by an amount @xmath31 enters the crystal and collectively interacts with the oscillating recoilless nuclei [ see fig . [ setup](b ) ] . for the sake of simplicity , we consider only the interaction of the wave with the nuclei and disregard interaction with electrons .
processes such as internal conversion , the photoelectric effect @xcite , and electron rayleigh scattering black64,stepanov74 are neglected .
we treat the problem in a semiclassical formalism .
namely , the electromagnetic field @xmath32 of the @xmath0 ray is described by the classical maxwell equation@xmath33 in which @xmath34 is the permeability of free space and the polarization of the medium @xmath35is determined by the off - diagonal elements of the nuclear density matrix @xmath36 . in eq .
( [ p1 ] ) the summation is taken over nuclei that are treated as point particles located at positions @xmath37 .
assuming that the nuclear transition matrix element @xmath38 is real and introducing the rabi frequency of the @xmath0-ray field @xmath39 , we obtain@xmath40equation ( [ m1 ] ) must be supplemented by the evolution equation for the nuclear density matrix@xmath41we assume that nuclear excitation remains weak , so the population of the excited state @xmath42 can be disregarded .
we look for a solution in the form@xmath43@xmath44where @xmath45 and @xmath46 are slowly varying functions of @xmath47 as compared to the fast oscillating exponentials . in the slowly varying amplitude approximation , eqs .
( [ m1 ] ) and ( [ c1 ] ) reduce to @xmath48 \bigg\ } \omega ( t,\mathbf{r } ) \notag \\ & = i\frac{\omega _ { a}^{2}}{n}\sum_{j}\rho ^{j}(t)\delta ( \mathbf{r}-\mathbf{r}% _ { j}(t))\ , , \label{s1}\end{aligned}\ ] ] @xmath49 where @xmath50is the collective nuclei frequency proportional to the square root of the average nuclei density @xmath51 and @xmath52 is the spontaneous decay rate of the nuclear transition .
physically , @xmath1 determines the time scale of the collective resonant absorption of the incident photon by the medium @xcite and typically is of the order of terahertz .
for example , for a @xmath19-kev transition of the @xmath18k mssbauer isotope that spontaneously decays at the rate @xmath20 s@xmath21 if we take the nuclei density to be @xmath53 @xmath54 we obtain @xmath55 s@xmath21 .
a crystal is made up of a periodic arrangement of atoms ( bravais lattice ) that form an infinite array of discrete points given by @xmath56 , where @xmath57 ( @xmath58 ) are any integers and @xmath59 are the primitive lattice vectors .
as a consequence , @xmath60 is a periodic function of @xmath61 with periods @xmath62 and , thus , it can be expanded in the fourier series as@xmath63 } \notag \\ & = n\sum_{m}e^{i\mathbf{k}_{m}\cdot \left [ \mathbf{r-}\hat{n}f(t)\right ] } \ , , \label{s}\end{aligned}\]]where @xmath64 , @xmath65 are the primitive vectors of the reciprocal lattice and @xmath51 is the average nuclear density .
we look for @xmath66 in the form @xmath67multiplying both sides of eq .
( [ s ] ) by @xmath68 we obtain@xmath69 } .
\label{s0}\]]this sum enters the right hand side of eq .
( [ s1 ] ) . in the fourier series ( [ s0 ] )
we are interested in terms that are in resonance with the left hand side of eq .
( [ s1 ] ) . for simplicity
we assume that only two vectors , namely , @xmath29 and @xmath70 have absolute values close to @xmath71 , where @xmath72 is a reciprocal lattice vector , see fig .
[ reciprocal lattice ] .
the other terms in ( [ s0 ] ) are off resonance and thus can be disregarded .
therefore , one can write approximately@xmath73this approximation implies that the incident wave @xmath29 is coupled only with one bragg wave @xmath74 . equation ( [ f1 ] ) suggests that one can look for a solution for @xmath45 in the form of a superposition of these coupled waves@xmath75then eqs .
( [ s1 ] ) and ( [ x0 ] ) yield the following equations for @xmath76 , @xmath77 and @xmath78 ( we take into account that @xmath79 ) @xmath80where @xmath81are detunings of the two coupled waves from the nuclear transition frequency @xmath8 .
taking the time derivative of both sides of eqs .
( [ n1 ] ) and ( [ n2 ] ) and using eq .
( [ n3 ] ) , we obtain evolution equations for two @xmath82-ray waves @xmath83 and @xmath77:@xmath84 = 0\ , , \label{x12 } \\
\bigg ( & \frac{\partial } { \partial t}+i\mathbf{k}_{2}\cdot \hat{n}\dot{f}% \bigg ) \left ( \frac{\partial } { \partial t}+i\delta _ { 2}\right ) \omega _ { 2 } \notag \\ & + \omega _ { a}^{2}\left [ \omega _ { 2}+\omega _ { 1}e^{i(\mathbf{k}_{1}-\mathbf{k% } _ { 2})\cdot \hat{n}f(t)}\right ] = 0\ , .
\label{x13}\end{aligned}\ ] ] equations ( [ x12 ] ) and ( [ x13 ] ) constitute one of our main findings .
these equations describe two coupled harmonic oscillators whose parameters periodically change in time .
the varying of the parameters drives the system .
namely , nuclei vibrations modulate coupling between two oscillators as indicated by the @xmath85 terms and , in addition , they periodically modulate the oscillator s frequency by means of the doppler shift @xmath86 .
it is known that parametric oscillators can have parametric resonances when system s parameters are periodically modulated which can lead to exponentially growing oscillations .
an interesting question appears in this context : can eqs .
( [ x12 ] ) and ( [ x13 ] ) yield exponentially growing solutions which would imply that the high - frequency @xmath0-ray field is being generated at the expense of the energy stored in the low - frequency nuclear vibrations ? in the appendix
we show that the answer to this question is that nuclear vibrations can not excite nuclear transitions in the present model .
specifically , we show that the sum of the energy of the high frequency field @xmath45 and that stored in the nuclear excitation is conserved no matter how nuclei move
. nevertheless , parametric resonance can be useful in the present problem .
namely , it can substantially speed up energy transfer from one coupled oscillator to another , that is , from the incident @xmath0-ray beam to the deflected one .
this mechanism can be used to control propagation of @xmath0 rays on a short time scale , which we discuss next .
first we consider the interaction between the @xmath0-ray field and a static nuclear array . in this case
there is no nuclear motion , so @xmath87 and eqs .
( n1)@xmath88([n3 ] ) can be solved analytically . in particular ,
if @xmath89 the solution satisfying the initial condition @xmath90 , @xmath91 and @xmath92 reads@xmath93 where @xmath94 equations ( [ x14])@xmath88([x16 ] ) yield that on resonance ( @xmath95 ) @xmath96that is , energy is periodically transferred back and forth between two coupled waves on a time scale given by the collective nuclear frequency @xmath97 which is proportional to the square root of the nuclear density .
typically @xmath98 is of the order of picoseconds .
amplitudes of the @xmath0-ray beams undergo collective oscillations svid08a , svid12 with frequency @xmath99 , as shown in fig . omega12onoffa . according to eq .
( [ x17 ] ) , nuclei become excited during the energy transfer between two @xmath0 waves .
namely , the incoming @xmath0 wave is partially absorbed by the nuclear array .
absorption is followed by the superradiant spontaneous emission of photons into the coupled wave .
in contrast , if the wave frequency is off resonance , i.e. @xmath100 , the energy transfer between beams @xmath101 and @xmath102 occurs over a much longer time@xmath103as shown in fig .
[ omega12onoff](b ) .
one should mention that energy oscillations between two @xmath0-ray modes , referred to as the temporal pendellsung effect due to different hyperfine transition frequencies at different nuclear sites , have been discussed in @xcite . in ref .
@xcite the bragg switching of the @xmath0-ray beam was realized using manipulation of nuclear spin states . in our mechanism , oscillations appear due to the collective nature of the interaction between light and the nuclear ensemble .
next we investigate how nuclear motion affects energy transfer between the two coupled @xmath0-ray waves . here
we assume that the nuclear array coherently vibrates with frequency @xmath104 and the consider transformation of the incoming wave @xmath29 into the deflected wave @xmath74 .
we assume that both waves are equally detuned from the nuclear transition , that is @xmath105 .
we also assume that nuclei vibrate with amplitude @xmath106 along the direction @xmath13 perpendicular to @xmath107 , as indicated in fig . [
reciprocal lattice ] . then @xmath108 . introducing the dimensionless modulation amplitude@xmath109eqs .
( [ x12 ] ) and ( [ x13 ] ) reduce to @xmath110 where@xmath111is a function that describes the modulation produced by nuclear motion .
the amplitude of the nuclei vibrations @xmath106 is much smaller than the spacing @xmath6 between nuclei .
however , since the wave length of the nuclear transition is also small compared to @xmath6 the modulation amplitude @xmath112 could be of the order of @xmath113 .
the initial conditions for eqs .
( [ omega1_1 ] ) and ( [ omega2_1 ] ) are @xmath114 and @xmath91 .
we assume that initially there is no nuclear excitation [ @xmath92 ] , which , according to eqs .
( [ n1 ] ) and ( [ n2 ] ) , yield @xmath116 and @xmath117
. equations ( [ omega1_1 ] ) and ( [ omega2_1 ] ) have the integral of motion @xmath118plugging this into eq .
( [ omega2_1 ] ) and introducing @xmath119 according to@xmath120we obtain the following equation for @xmath119 @xmath121which is an equation of the parametric oscillator .
equation ( [ p2 ] ) has a solution in terms of special functions , however , such a solution is not very insightful . instead , we derive an approximate solution that clearly shows the physics behind the parametric speed up of the energy transfer . introducing the function @xmath122@xmath123one can rewrite eq .
( [ p2 ] ) as @xmath124next we expand the exponential factor into the fourier series@xmath125 , \label{p5}\end{aligned}\]]where @xmath126 are the bessel functions .
we assume that @xmath15 is close to @xmath5 while @xmath119 and @xmath127 are slowly varying functions of time on the scale @xmath128 .
then in the fourier expansion ( [ p5 ] ) one can keep only the slowly varying term and approximately write@xmath129as a result , eqs .
( [ p3 ] ) and ( [ p4 ] ) reduce to @xmath130 which can be solved analytically . plugging this solution into eqs .
( [ px1 ] ) and ( [ px2 ] ) we finally obtain@xmath131where @xmath132when @xmath133 we find@xmath134eqs .
( [ pp1 ] ) and ( [ pp2 ] ) show that the rate of energy transfer between two coupled waves depends on the amplitude of the nuclear vibrations .
the optimum value of the modulation amplitude @xmath112 corresponds to maximum of @xmath135 , that is @xmath136 which gives @xmath137 . for larger @xmath112 the transfer rate oscillates following @xmath135 .
for @xmath138 one can use the expansion @xmath139 .
( [ pp2 ] ) yields that energy transfer time between two waves is @xmath140 in fig .
[ omega2_k ] we plot @xmath77 for different values of the modulation amplitude @xmath112 obtained by numerical solution of eqs .
( omega1_1 ) and ( [ omega2_1 ] ) .
our analytical result ( [ pp2 ] ) gives essentially the same curves .
when the incident @xmath0 wave @xmath101 is off resonance with the nuclear transition , the time it takes for the energy to transfer from @xmath141 into the deflected wave @xmath142 can substantially vary with or without nuclear vibrations .
this can be used for fast switching of the wave propagation that can be achieved in the regime @xmath143 .
if the @xmath0-wave detuning @xmath5 is large enough then the wave will pass through the static crystal without deflection .
however , if the nuclear vibrations are suddenly turned on with @xmath133 the incident @xmath0 wave will be deflected on a time scale given by eq .
( pp3 ) , which could be a few picoseconds .
figure [ omega12_compare ] demonstrates the effect for a medium with @xmath144 thz assuming that the incident wave is detuned from the nuclear transition by @xmath145 . for a static crystal the fields @xmath141 and @xmath142
are shown by dashed lines . without nuclear vibrations
it takes @xmath146 ps for the wave @xmath101 to convert into @xmath142 .
if the crystal size is smaller than @xmath147 cm the incident wave passes through . however , if the nuclear array vibrates with modulation amplitude @xmath148 the conversion time becomes @xmath149 ps and , thus , the wave will be deflected at a length of @xmath150 cm ( solid lines in fig .
[ omega12_compare ] ) .
one should note that the @xmath0-ray switch can also operate in the on resonance regime .
namely , when the incident wave is on resonance with the nuclear transition it converts fast into the deflected wave on a time scale @xmath151 . turning on nuclear vibrations would destroy the resonance interaction and make the wave pass through the crystal .
the physics behind the speed up of the energy transfer between two waves can be understood as a parametric resonance in a system of coupled oscillators .
a single oscillator whose frequency is periodically modulated provides a simple example of parametric resonance .
the motion of such an oscillator is described by the mathieu s
equation@xmath152 x=0 , \label{h1}\]]where @xmath153 is the modulation amplitude .
if @xmath154 then the system has two natural frequencies @xmath155 .
if the system s parameters vary with frequency @xmath15 equal to the difference between natural frequencies , that is @xmath156 , the oscillator phase locks to the parametric variation and undergoes a parametric resonance absorbing energy at a rate proportional to the energy it already has .
a similar situation takes place if the system has several natural frequencies ( normal modes ) . to achieve parametric resonance
the modulation frequency @xmath157 must match the difference between two normal mode frequencies hsu63 .
this is known as the difference combination resonance nayf2000 . in the present problem
the natural frequencies of the coupled light - nuclear system are determined from the solution ( [ x14 ] ) and ( x15 ) obtained for the static lattice .
thus , if the frequency of the nuclear vibrations matches the frequency difference , namely , @xmath158 the system undergoes parametric resonance , which speeds up the energy transfer between two @xmath0 waves .
this phenomenon is analogous to parametric frequency mixing in propagating circuits @xcite , in which power can flow back and forth between the two coupled circuits if the coupling reactance variation frequency matches their frequency difference . combination parametric resonance at the frequency difference between two normal modes of the coupled light - atom system is the essence of the qaser @xcite , a device that can generate high - frequency ( e.g. , xuv ) coherent light by driving an atomic medium with a low frequency ( e.g. , infrared ) field @xcite .
contrary to the laser , the qaser does not require any atomic population in the excited state and yields high - frequency light amplification . in the case of the qaser the external field drives the atomic transition which produces modulation of the atom - field coupling strength and yields gain at high frequency . in the present model , unlike the qaser
, modulation is produced by the nuclear motion , which does not yield amplification of the high - frequency ( @xmath4 ) field .
however , parametric resonance and collective effects of the light interaction with a nuclear array enhance the rates of the radiation absorption and reemission . as a result ,
energy transfer between two @xmath0 waves occurs on a much shorter superradiant time scale determined by the collective nuclear frequency @xmath97 which is of the order of terahertz .
the combination of the dicke superradiance , bragg diffraction , and combination parametric resonance introduces interesting features to our problem and allows us to achieve fast manipulation of the @xmath0-ray propagation .
if the incident @xmath0 wave is far detuned from the nuclear transition by the amount @xmath2 then the light - nuclear interaction is weak and the @xmath0 wave passes through the static nuclear array .
we found that if we make the nuclei vibrate coherently with frequency @xmath5 then the combination parametric resonance effectively enhances the light - nuclear interaction strength . as a result , the incident @xmath0 wave undergoes deflection into a wave propagating at the bragg angle on a short superradiant time scale .
the maximum energy transfer rate is achieved for the amplitude of nuclear oscillations @xmath159 , where @xmath160 is the wavelength of the nuclear transition . since @xmath161 is typically much smaller than the spacing between nuclei in crystals the required nuclear vibrations are also small .
such nuclear motion can be realized , e.g. , in ionic crystals illuminated by a strong driving optical laser field .
our findings could be used for manipulation of the propagation direction of @xmath82 rays on a picosecond time scale .
we thank o. kocharovskaya for useful discussions .
we gratefully acknowledge support from the national science foundation through grants no .
phy-1205868 and no .
phy-1241032 and the robert a. welch foundation ( award no .
we start from eqs . ( [ s1 ] ) and ( [ x0 ] ) describing light propagation through a moving crystal lattice@xmath162 \right ) \omega ( t,\mathbf{r } ) \notag \\ & \qquad \qquad \qquad = i\frac{\omega _ { a}^{2}}{n}\sum_{j}\rho ^{j}(t)\delta ( \mathbf{r}-\mathbf{r}_{j}(t ) ) , \label{s1aa } \\ & \frac{\partial \rho ^{j}(t)}{\partial t}=i\delta \omega _ { ab}(t)\rho ^{j}(t)+i\omega ( t,\mathbf{r}_{j}(t))\ , . \label{x0aa}\end{aligned}\]]in eq . ( [ x0aa ] )
we introduced an additional time - dependent frequency shift @xmath163 to include possible external modulation of the nuclear transition frequency .
multiplying both sides of eq .
( [ s1aa ] ) by @xmath164 we obtain@xmath165 \right ) \omega ( t,\mathbf{r } ) \notag \\ = & i\frac{\omega _ { a}^{2}}{n}\sum_{j}\rho ^{j}(t)\omega ^{\ast } ( t,\mathbf{r}% _ { j}(t))\delta \left ( \mathbf{r}-\mathbf{r}_{j}(t)\right ) \ , .
\label{a1}\end{aligned}\]]while eq .
( [ x0aa ] ) yields@xmath166plugging eq .
( [ mid1 ] ) into eq .
( [ a1 ] ) gives@xmath167 \right ) \omega ( t,\mathbf{r } ) \notag \\ + & \frac{\omega _ { a}^{2}}{n}\sum_{j}\rho ^{j}(t)\dot{\rho}^{j\ast } ( t)\delta ( \mathbf{r}-\mathbf{r}_{j}(t ) )
\notag \\ + & i\delta \omega _ { ab}(t)\frac{\omega _ { a}^{2}}{n}\sum_{j}|\rho ^{j}(t)|^{2}\delta ( \mathbf{r}-\mathbf{r}_{j}(t))=0\ , .
\label{mid2}\end{aligned}\]]adding to eq .
( [ mid2 ] ) its complex conjugate we obtain@xmath168 \notag \\ & + \frac{\omega _ { a}^{2}}{n}\sum_{j}\delta \left ( \mathbf{r}-\mathbf{r}% _ { j}(t)\right ) \frac{\partial } { \partial t}|\rho ^{j}(t)|^{2}=0\ ,
. \label{mid3}\end{aligned}\]]integrating eq .
( [ mid3 ] ) over space and taking into account that for weakly excited nuclei @xmath169 , where @xmath170 is the excited state population of the nucleus @xmath10 , we find@xmath171equation ( [ ec ] ) shows that the sum of the energy of the high frequency field @xmath172 and that stored in nuclear excitation is conserved no matter how nuclei move .
this implies that nuclear motion can not amplify high - frequency field .
n. barrire , j. rousselle , p. von ballmoos , n. v. abrosimov , p. courtois , p. bastie , t. camus , m. jentschel , v. n. kurlov , l. natalucci , g. roudil , n. f. brejnholt , and d. serre , j. appl . crystallogr . * 42 * , 834 ( 2009 ) . | the collective nature of light interactions with atomic and nuclear ensembles yields the fascinating phenomena of superradiance and radiation trapping .
we study the interaction of @xmath0 rays with a coherently vibrating periodic array of two - level nuclei .
such nuclear motion can be generated , e.g. , in ionic crystals illuminated by a strong driving optical laser field .
we find that deflection of the incident @xmath0 beam into the bragg angle can be switched on and off by nuclear vibrations on a superradiant time scale determined by the collective nuclear frequency @xmath1 , which is of the order of terahertz .
namely , if the incident @xmath0 wave is detuned from the nuclear transition by frequency @xmath2 it passes through the static nuclear array . however , if the nuclei vibrate with frequency @xmath3 then parametric resonance can yield energy transfer into the bragg deflected beam on the superradiant time scale , which can be used for fast control of @xmath4 rays . |
a physical theory permits correctly involve all interactions in a magnetization process and to reveal relationship between structure and physical properties of a magnetic material .
applicable mathematical model could be derived from such a theory .
this model will give a possibility to investigate real physical and structural properties of magnetic materials from experimental data .
it is essential for synthesis of new materials with desired properties .
nowadays there are several models for describing magnetization of ferromagnetic materials . more detailed description and analyzes of advantages and disadvantages of these models one can find in works @xcite-@xcite . in the present paper , there suggests a new theory of magnetization and an attempt to derive an applicable general mathematical model to describe magnetization curve for soft and stiff magnetic materials .
such generic special points in magnetization curve are also elicited from the model .
a formula for dynamic magnetic susceptibility is derived from a magnetization field dependence .
a new parameter which can numerically characterizes the magnetic stiffness of a material is introduced and its physical interpretation is given .
two competing interactions could be distinguished in ferromagnetic materials : an exchange interaction ( _ exch _ ) which tends to orient magnetic moments in the same direction and by this magnetizes the system and dipole - dipole interaction ( _ dip - dip _ ) which tends to orient magnetic moments antiparallel to each other and by this demagnetize the system .
a relevant difference between these interactions is that _ exch _ acts between nearest atoms and its energy is independent of a magnetic moment of a system . on the contrary , _ dip - dip _ energy rises as magnetic moment increases . increasing a material size a _ dip - dip
_ energy can overcome _ exch _ energy . in this case ,
two and more domains structures become more favorable .
it is schematically shown in figure [ fig1 ] . as _ exch _
energy is much bigger than _ dip - dip _ energy between nearest atoms magnetic moments inside a domain are firmly connected in the same direction .
the magnetic shell where _ dip - dip _ energy becomes equal to _ exch _ energy could be accepted as a border of the domain and by this defining a size of a domain @xcite .
so , after a domain was completed the next shell of magnetic moments will recline in the opposite direction , figure [ fig1 ] .
it should be mentioned that magnetic moments inside a domain are not strictly directed in the same direction .
they recline under an angle to each other from shell to shell until the domain would not be finished .
the transition layers behaviour is detailed described in the work of landau - lifshitz @xcite . because of it , a magnetic moment of a domain is less than a sum of magnetic moments of atoms inside it . on the assumption of foregoing
we could consider domains like solitary magnetic particles .
the separated quasiparticles we would call supermagneton ( _ sm _ ) analogically to r.harrison @xcite-@xcite where domains are also supposed to be a unit magnetic moments in the magnetization processes . we know that _ exch _ is compensated by _ dip - dip _ between _ _ sm__s .
it means that _ _
sm__s magnetic moments are exempt from _
exch_. so , the problem of ferromagnetic materials is brought to a problem of paramagnetic materials where _ _ sm__s play the role of magnetic moments .
seeing that there are axes of easiest magnetization in ferromagnetic materials @xcite _ _ sm__s are distributed in a field of anisotropy according to the boltzmann distribution . for simplicity
, we would consider a case of uniaxial anisotropy .
_ _ sm__s with a positive projection on any selected direction along the anisotropy axis separated from _ _ sm__s with negative projection by anisotropy barrier , figure [ fig2 ] .
by applying magnetic field _
h _ to a ferromagnetic material a magnetic field _
b _ is induced inside . _ _ sm__s with positive projection on the magnetic field direction obtain energy -@xmath0 , and _
_ sm__s with negative projection obtain energy @xmath1 , where @xmath2 and @xmath3 are magnetic moments of _ _ sm__s with positive and negative projection respectively . taking into account that @xmath2 + @xmath3 = 2@xmath4 , _ m _ is magnetic moment of _ _ sm__s in zero field
, we can conclude that @xmath5 .
it means that domains could be effectively replaced by _ _
sm__s magnetic moment of which remains unchanged and equal to a magnetic moment of domain in nonmagnetized state .
so , potential wells shift on a value 2__mb _ _ , as shown on figure [ fig3 ] . as a result of the energetic shift magnetization of the sample
is appeared . to estimate the magnetization one needs to calculate a difference between magnetic moments with positive and negative projections . here and in further positive and negative directions
would be considered relative to the direction in which external magnetic field is applied .
taking into account that at any field value distribution of _ _
sm__s is obey to the boltzmann statistics we can calculate a difference between amount of _ _
sm__swith negatie and positive projections : @xmath6 @xmath7 by the same way we get : @xmath8 dividing [ eq2 ] on [ eq3 ] we get : @xmath9\hfill \\ \end{gathered}\ \label{eq4}\ ] ] multiplying [ eq4 ] on _ m _ and taking into account that by deffinition : @xmath10 finally we obtain : @xmath11\ \label{eq6}\ ] ] where @xmath12 is saturation magnetization and _ m _ is magnetization in field _
b_. to explore magnetization process more thoroughly it is necessary to understand field distribution inside a material during magnetization . as direction of external field
is given we will concentrate on distribution of dipolar field .
the magnetic field which is induced by magnetic moment in any point could be calculated in first approximation as @xcite : @xmath13 where @xmath14 is a unit vector in the direction to the point , and _
r _ is distance between magnetic moment and the point where the field is calculated .
is an angle between magnetization direction and lines _ a _ and _ b_. 1 , 3 are space regions where dipolar field has positive projection on magnetic moment direction and 2 , 4 are space regions where dipolar field has negative projection on magnetic moment direction.,scaledwidth=30.0% ] it is seen from [ eq7 ] that _ h _ has negative or positive projection on the direction of magnetic moment in different points .
it is not difficult to found points where _ h _ changes its projection sign on direction of the magnetization from [ eq7 ] .
@xmath15 @xmath16 @xmath17 @xmath18 these points belong to the lines _ a _ and _ b _ which decline under angles @xmath19 to magnetization direction as shown on figure [ fig4 ] . lines _
a _ and _ b _ demarcate space around magnetic moment on four regions : 1 , 2 , 3 , 4 .
so , _ h _ has positive projection in any point which belongs to regions 1 and 3 with biggest value when or and has negative projection in any pint of region 2 and 4 with biggest value @xmath20 when @xmath21 or @xmath22 . by this ,
dipolar field plays both magnetize ( positive ) and demagnetize ( negative ) roles .
due to positive influence of depolar field it is possible to magnetize bulk samples .
thus magnetic field inside a sample could be represented as : @xmath23 where @xmath24 is a dipolar term , @xmath25 is a distance between nearest domains domains ( linear size of a @xmath26 ) , @xmath27 is a coefficient which represents difference between positive and negative dipolar influences . as an example , we will consider two - domain rod as shown on figure [ fig8 ] .
let s compare processes of magnetization of such a rod when the external magnetic field is directed along longitude ( x axis ) and when the external magnetic field is directed along width ( y or z axes ) .
when the rod is magnetized along x axis the dipolar field of one domain will magnetize the second domain because it belongs to the region of the space where dipolar field has positive projection on magnetization direction .
the diagram of the process is shown in figure [ fig9 ] . by this dipolar field
will help to magnetize the system . in this case
s - shaped hysteresis loop would be observed @xcite-@xcite .
when the rod is magnetized along z axis one domain will direct the second domain in the opposite direction because it belongs to the region of the space where dipolar field has negative projection on magnetization direction as shown in figure [ fig10 ] . in this case @xmath28 .
because of all written above , the magnetic energy term in [ eq6 ] should be replaced by @xmath29 , where @xmath27 is a parameter which depend on difference between demagnetizing and magnetizing parts of _ dip - dip _ influences , by this @xmath27 depends on should depend on surface and surface volume ratio . at each certain values of external magnetic field the certain distribution of field inside a material exists .
this field distribution changes with external field , and by this @xmath27 change with external field as well .
@xmath30\ \label{eq13}\ ] ] as equation [ eq13 ] is trancendent it could be rewritten in more convenient form as done in @xcite-@xcite : @xmath31 it should be noticed that the last term in [ eq14 ] is independent on the value of magnetic moment as both magnetic moment of the domains and anisotropy energy term are both volume dependant .
after the external magnetic field was abolished a _
dip - dip _ tends to demagnetize the sample . because of this _ _ sm__s would turn from positive projection to negative through anisotropy barrier .
but not all _
_ sm__s can overcome anisotropy barrier and the part of them would remain with positive projection .
so , after the external magnetic field was abolished ferromagnetic sample would steel remain in magnetize condition .
this magnetization is called remanent magnetization .
it is represented in figure [ 11 ] . in order to get expression for remanent magnetization
it is necessary to put _ _ h__=0 into [ eq13 ] : @xmath32 \label{eq15}\ ] ] where @xmath33 is remanent magnetization .
as is known a coercive force is a magnetic field which should be applied to the sample to demagnetize it .
so to get an expression of coercive force one need to put _ _ m__=0 into [ eq14 ] : @xmath34 it is seen that coercive force depends on the field direction ( angle between field and anisotropy axis ) .
the next parameter can characterize magnetic stiffness of ferromagnetic materials : @xmath35 dividing [ eq2 ] on [ eq3 ] it is possible to show that @xmath36 and represents the ratio of the amount of _ _
sm__s which overcame anisotropy barrier to the amount of _ _ sm__s which remain with positive projection after the external magnetic field was abolished , figure [ fig11 ] .
values of _ k _ could change in the range from 0 to 1 .
the bigger the value of _ k _ the softer magnetic material is and vice verse .
for example , in case of strong anisotropy no of _ _ sm__s are able to overcome anisotropy barrier , it means that there are no _ _ sm__s with negative projection ( _ k _ = 0 ) and @xmath37 rectangular - like hysteresis loop , figure [ 12 ] a ) . in the case when all _
_ sm__s were able to overcome anisotropy barrier , the amount of _
_ sm__s with negative projection is equal to the amount of positive projection ( _ k _ = 1 ) and consequently @xmath38 , figure [ 12 ] c ) .
b)@xmath39 ; cases a ) and b ) corresponds to a magneto stiff material and magneto soft materials correspondingly.,scaledwidth=30.0% ]
a low which describes magnetization dependence on external magnetic field gives possibility to find out a law for magnetic susceptibility : @xmath40 = \hfill \\
= \frac{{{m_s}m \cdot \left ( { 1 - \frac{{d\eta } } { { dh}}{m_a } - 4\pi \frac{\eta } { { { a^3}}}\chi } \right ) + { m_s}{\raise0.7ex\hbox{${d{e_a}\left ( \gamma \right)}$ } \!\mathord{\left/ { \vphantom { { d{e_a}\left ( \gamma \right ) } { dh}}}\right.\kern-\nulldelimiterspace } \!\lower0.7ex\hbox{${dh}$}}}}{{{k_b}t \cdot { \text{c}}{{\text{h}}^2}\left [ { \frac{{m\left ( { h - \eta { m_a } } \right ) + { e_a}\left ( \gamma \right)}}{{{k_b}t } } } \right ] } } \hfill \\
\end{gathered } \label{eq18}\ ] ] @xmath41}}{{{k_b}t{\text{c}}{{\text{h}}^2}\left [ { \frac{{m\left ( { h - \eta { m_a } } \right ) + { e_a}\left ( \gamma \right)}}{{{k_b}t } } } \right ] + \frac{\eta } { { { a^3}}}m{m_s } } } \label{eq19}\ ] ] it should be noted that @xmath27 decreases with increasing of _ h _ thus @xmath42 .
this term is one of factors which are responsible for high value of magnetic susceptibility of ferromagnetic materials .
it is seen that a magnetic susceptibility depends on the external magnetic field , the dipolar interaction , the temperature of the sample and the anisotropy energy , as it was expected . in case of high temperature or small external magnetic field
@xmath43 from [ eq19 ] one can get : @xmath44
there was shown that a magnetic sample could be considered as a set of magnetic particles , called superparamagnetons ( _ _ sm__s ) .
magnetic moment of _ sm _ is equal to a magnetic moment of domain plus magnetic moment of the domain wall , before external magnetic field was applied .
the stark difference of the _ sm _ is that it s magnetic moment does nt change during magnetisation precess and _ _ sm__s are free from exchange interaction what gives a possibility to apply the boltzmann statistics for them as it done for paramagnetic samples .
it is important that all energies that take place in the process of magnetization are included additively .
it means that additional energies like energies on domain walls pinning could be easily added in the formula in [ eq13 ] .
there were derived expressions for a remanent magnetization [ eq15 ] and a coercive force [ eq16 ] as special points on _ m_(_h _ ) curve .
it is seen that temperature dependence of remanent magnetization bears exponential character and depend on _ dip - dip _ in a sample and on an anisotropy barrier .
it is also seen how coercive force depend on anisotropy barrier and a magnetic moment of a domain . from _ m_(_h _ ) function there was derived an expression for a magnetic susceptibility [ eq19 ] .
it is shown that in extremal cases , like high temperature or low field , magnetization depends on field linearly , and magnetic susceptibility is field independent like in the paramagnetic case .
it is essential to note that expressions [ eq19 ] [ eq20 ] are applicable at all temperature regions .
d.c jiles , x. fang , w. zhang , _ handbook of advanced magnetic materials _ ,
f.liorzou , b. phelps , and d.l .
atherton , _ macroscopic models of magnetization_,ieee transactions on magnetics , vol.36 , no.2 . , ( 2000 ) .
sergey e. zirka , yuriy l. moroz , robert g. harrison , and krzysztof chwastek , _ on the physical aspects of the jiles - atherton hysteresis models _ , j. appl .
112 , ( 2012 ) .
d.c . jiles and d.l .
atherton , `` theory of ferromagnetic hysteresis '' , j. magn .
61 , 48 , ( 1986 ) .
j. frankel and j. dorfman ; `` spontaneous and induced magnetisation in ferromagnetic bodies '' , ( 1930 ) .
l. landau , e. lifshits , `` on the theory of the dispersion of magnetic permeabillity in ferromagnetic bodies '' , phys . zeitsch .
der sow . 8 , pp .
153 - 169 , ( 1935 ) .
harrison , `` physical model of spin ferromagnetism '' , ( 2003 ) .
harrison , `` variable - domain - size theory of spin ferromagnetism '' , ( 2004 ) .
harrison , `` physical theory of ferromagnetic first - order return curves '' , ieee transition on magnetics , vol .
45 no , 4 , ( 2009 ) .
l. d. landau and e. m. lifshitz , course of theoretical physics , vol . 8 , ( 2005 ) .
c. kittel , `` theory of the structure of ferromagnetic domains in films and small particles '' , phys.rev .
vol.70 , no .
11 and 12 , ( 1946 ) .
l. d. landau and e. m.
lifshitz , course of theoretical physics , vol .
2 : the classical theory of fields ( nauka , moscow , 1988 ; pergamon , oxford , 1975 ) .
alberto p. gimaraes , principles of nanomagnetism , springer , 2009 . | the present article concludes that a ferromagnetic sample could be considered as a paramagnetic system where roles of magnetic moments play magnetic domains . based on this conclusion and taking into account presence of an anisotropy field the formula which describes magnetization dependence on the external magnetic field
is derived .
expressions for a remanent magnetization and a coercive force are presented .
the new parameter to characterize a magnetic stiffness of a material is introduced .
a physical expression for a dynamic magnetic susceptibility as a function of material s characteristics , external magnetic field , and temperature is given . |
scalar fields play a fundamental role in the modern cosmology and astrophysics .
having come in them from the theory of elementary particles , the scalar fields are widely used both at creation of models of compact objects and at research of the evolution of the universe as a whole . in particular , the models of quasi - star objects from scalar fields , the so - called boson stars , are well - known @xcite . in turn in cosmology
scalar fields are the basis for creation of models of the early inflationary universe @xcite .
another field of application of scalar fields is connected with one of the most exciting event in astrophysics and cosmology of the last decades - discovery of the acceleration of the present universe @xcite .
this discovery has stimulated an appearance of a large amount of various models trying to explain this phenomenon .
a basis of all the models consists in violation of different energy conditions .
for cosmology , most essential are two energy conditions ( see , e.g. , ref .
@xcite ) : the first one , the so - called strong energy condition ( sec ) , states that @xmath0 , where @xmath1 and @xmath2 are effective energy density and pressure of matter which determine the evolution of the universe . in hydrodynamical language
this means that the parameter of the equation of state @xmath3 .
this case corresponds to the decelerated expansion of the universe ( the friedmann models ) .
violation of the sec leads to the accelerated expansion of the universe ( the inflationary models ) .
the second condition , the so - called weak energy condition ( wec ) , states that @xmath4 , whence it appears that @xmath5 .
failure satisfying the wec results in exponential or more fast accelerated expansion of the universe .
the substance providing such acceleration was called dark energy .
this term usually means that the equation of state lies in the range @xmath6 ( the cosmological constant , quintessence @xcite , chaplygin gas @xcite , the theories of gravitation with high derivatives @xcite , etc . ) .
all the models describe the present accelerated expansion of the universe within the accuracy of observations .
the mentioned above violation of the sec is normally arisen at some choice of a potential energy of usual scalar fields ( see examples of such potentials in refs .
@xcite ) . however , as it was shown in ref .
@xcite by a model - independent way on the basis of study of data sets containing 172 snia , it is possible in the present epoch that @xmath7 , i.e. the wec can be violated .
( for more recent papers with the latest observational data see refs .
@xcite . ) such an unusual state of matter is also known as phantom dark energy @xcite .
there are models of phantom energy within the framework of higher - order theories of gravity @xcite , braneworld cosmology @xcite , etc .
another popular direction consists in consideration of ghost scalar fields with negative sign before kinetic term @xcite . in ref .
@xcite we considered the cosmological model of the early universe with two interacting ghost scalar fields with special form of the potential energy : @xmath8 here @xmath9 are two scalar fields with the masses @xmath10 and @xmath11 , @xmath12 are the self - coupling constants and @xmath13 - some constant which should be chosen on the assumption of a problem statement .
the essential feature of this potential is presence of two local minima at @xmath14 .
it allows existence of localized solutions with finite energy in problems with such potential . as is known @xcite , for a case of one scalar field
, localized solutions could exist only for a case with two or more minima of potential energy when solutions start in one minimum and tend to another one . in a case of two scalar fields , a situation is possible when a solution starts and finishes in the same minimum . in ref .
@xcite such a case was carried out : the solution started at @xmath15 and returned to the same local minimum at @xmath16
. such type of soliton - like solutions , localized on a spacelike hypersurface , known as spacelike branes ( s - branes ) @xcite . the mentioned in ref .
@xcite possibility of existence of the localized in _ time _
solutions for phantom fields allows us to hope for presence of similar solutions and for a _
static _ case . in this paper
we consider three models created by two interacting phantom and ghost scalar fields with the potential : 1 ) a traversable wormhole ; 2 ) a brane - like solution analog of a domain wall solution but with asymptotically non - flat spacetime ( anti - de sitter spacetime ) ; 3 ) a spherically symmetric particle - like solution . the model of so - called traversable wormhole
was suggested in ref .
@xcite ( for a recent review , see @xcite ) .
this wormhole is created by some special matter with the violated wec .
as phantom scalar fields also violate the wec , they could be used at modelling of traversable wormholes .
the researches of such models were already carried out ( see , e.g. , refs .
@xcite ) . in these works
some effective hydrodynamical energy - momentum tensor with the equation of state @xmath7 was chosen as a source of matter . but a distribution of this matter was added by hand and , correspondingly , the _ nonself - consistent _ models of the traversable wormholes were considered . in this paper
we consider a _ self - consistent _ model of a traversable wormhole created by two interacting phantom and ghost scalar fields with the potential .
the specified above features of this potential allow us to find regular solutions with localized energy density .
domain walls are topological defects and they arise in different aspects both in particle physics and cosmology ( see , e.g. , @xcite and references therein ) .
they separate a spacetime into several domains along a single coordinate . in a case of scalar
fields domain wall solutions exist when the scalar filed potentials have isolated minima .
the domain walls are surfaces separating minima of the potentials with different vacuum expectation values .
the region of fast change of the scalar field corresponds to the domain wall .
the domain wall refers to as _ thin _ wall if the energy density of the scalar field is localized at the domain wall surface and could be replaced by the delta function .
the different variants of the domain wall solutions were found in refs .
@xcite with asymptotically flat , de sitter and schwarzschild spacetimes .
there are also known the so - called _ thick _ domain wall solutions @xcite which could exist at late - time phase transitions in the evolution of the universe . in this paper
the consideration of the thick domain wall model with the potential is presented .
search of spherically symmetric solutions with various matter sources always was an important problem in special and general relativity .
such solutions are using both at investigation of different particle - like models of elementary particles and creation of models of star - like objects and another large - scale configurations .
the source of matter is fields with various spins both interacting between each other and with gravitational field ( see , e.g. , ref .
there are well - known regular solutions for the scalar fields both for noninteracting and self - interacting fields @xcite . for a case of usual ( non - phantom ) scalar fields , the model of boson star with potential
was considered in ref .
it was shown there that there are regular solutions in the case under consideration .
recently the spherically symmetric model with self - gravitating matter with the hydrodynamic equation of state @xmath7 was considered in ref .
below we show that regular solutions for spherically symmetric case exist for phantom and ghost scalar fields also .
the paper is organized as follows : in section [ ge ] the general gravitational and field equations for all three above mentioned models are presented . in sections [ tw ] ,
[ dw ] and [ sss ] the models of traversable wormhole , brane - like and spherically symmetric solution are considered . in section [ concl ]
we present comments and conclusions .
we chose the lagrangian as follows : @xmath17~,\ ] ] where @xmath18 is the scalar curvature , @xmath19 is the newtonian gravity constant and the constant @xmath20 . in the case @xmath21 one
has the theory of usual scalar field plus gravitation .
the case @xmath22 corresponds to the theory of ghost scalar field .
the corresponding energy - momentum tensor will then be : @xmath23 \right\}~,\ ] ] and variation of the lagrangian gives the gravitational and field equations in the form : @xmath24 @xmath25=-\frac{\partial v}{\partial ( \varphi,\chi)}.\ ] ] in our case equations - are the system of ordinary nonlinear differential equations with the potential energy from .
as it follows from the experience of previous researches of similar systems @xcite , finding of solutions of the system - is reduced to search of eigenvalues of the parameters @xmath26 .
procedure of search of solutions and its application for investigation of the models mentioned in introduction will be considered in next three sections .
we will search for static solutions of equations - for the following metric @xcite : @xmath27 where @xmath28 are the even functions depending only on the coordinate @xmath29 which covers the entire range @xmath30 . using this metric , one can obtain from and the following equations ( at @xmath31 ) : @xmath32~ , \\ \label{einstein_c } \frac{1}{4}\left(\frac{a^{\prime}}{a}\right)^2-\frac{1}{a}+\frac{1}{2}\frac{a^{\prime}}{a}\frac{b^{\prime}}{b}&= & -\frac{1}{2}(\varphi^{\prime 2}+\chi^{\prime 2})+v~,\end{aligned}\ ] ] where a prime denotes differentiation with respect to @xmath29 .
eq . was obtained by subtracting @xmath33 component from @xmath34 component of the equations , and the equations and are @xmath35 and @xmath33 components of eqs . .
the corresponding field equations from will be : @xmath36~ , \\
\label{field_b } \chi^{\prime \prime}+\left(\frac{a^\prime}{a}+\frac{1}{2}\frac{b^\prime}{b}\right)\chi^\prime= \chi \left[2\varphi^2+\lambda_2(\chi^2-m_2 ^ 2)\right]~.\end{aligned}\ ] ] in the equations - the following rescaling are used : @xmath37 , @xmath38 , @xmath39 , @xmath40 . as it was shown in previous researches of problems with the potential ( see , e.g. ,
@xcite ) , regular solutions of the system of nonlinear differential equations - could exist only for some values of the self - coupling constants @xmath12 and the masses of the scalar fields @xmath26 , and also depend on boundary conditions which set under the problem statement .
particulary , specifying some values of the parameters @xmath12 , one has already effect on the shape of the potential that , in turn , determine a possibility of existence of regular solutions of the system - .
the further task consists in a search of such parameters @xmath26 which give regular solutions . in this sense
the problem reduces to a search of _ eigenvalues _ of the parameters @xmath26 and corresponding _ eigenfunctions _
@xmath41 and @xmath42 for the nonlinear system of differential equations - .
the technique of solution of systems similar to - is described in ref .
@xcite in details .
the essence of this procedure is the following : on the first step one solve the equation with some arbitrary selected function @xmath42 looking for a regular solution existing only at some value of the parameter @xmath10 . on this step
the influence of gravitation is not taken into account .
then this solution for the function @xmath43 insert into the equation and one search for a value of the parameter @xmath11 yielding a regular solution .
this procedure reiterate several times ( three usually enough ) for obtaining of acceptable convergence of values of the parameters @xmath26 .
the obtained functions @xmath44 are inserting into the gravitational equations - .
. , which is the constraint equation , is using for specifying of boundary conditions ( see below ) .
the obtained solutions for the metric functions @xmath45 are inserting into the complete equations for the scalar fields - and they are solving again for a search of eigenvalues of the parameters @xmath26 with account of gravitation .
this procedure reiterate so many times as it is necessary for obtaining of acceptable convergence of values of the parameters @xmath26 .
the described procedure of a search of solutions of the system - , also known as the shooting method , allows to find rather fast values of the parameters @xmath26 at which regular solutions exist .
we have checked the obtained solutions using the ndsolve routine from _
mathematica _ substituting the eigenvalues @xmath26 and solving , , , directly .
the boundary conditions are choosing with account of @xmath46 symmetry in the following form : @xmath47 where the condition for @xmath48 is choosing to satisfy the constraint at @xmath49 , @xmath50 is the value of the potential at @xmath49 and the self - coupling constants @xmath51 and @xmath52 .
then , using the above procedure for obtaining of solutions of the system - , we have the results presented in fig . - .
these results are obtained for the masses @xmath53 and @xmath54 .
as one can see from fig . , @xmath55 and @xmath56 as @xmath57 .
it corresponds to asymptotic transition of the solutions to the local minimum of the potential ( see introduction ) .
the arbitrary constant @xmath13 was chosen in such a way that the value of the potential in the local minimum was equal to zero , viz @xmath58 .
such a choice of @xmath13 ensures zero value of the energy density as @xmath57 ( fig . ) . for the wormhole model.,width=340 ] for the wormhole model.,width=340 ] in the wormhole model.,width=340 ] in the wormhole model.,width=340 ] let us estimate an asymptotic behaviour of the solutions .
for this purpose we will seek for solutions of the equations - in the form : @xmath59 where @xmath60 as @xmath57 .
then the right hand side of eqs .
- goes to zero and their particular solutions are : @xmath61 where @xmath62 and @xmath63 are constants .
practically @xmath64 defines a total mass of a wormhole and @xmath63 the run of time at the infinity . by corresponding redefinition of the time @xmath65
, these solutions could be reduced to a flat form in spherical coordinates , i.e. one has asymptotically flat minkowski spacetime ( see figs .
. then , taken into account - , the corresponding asymptotic equations for the scalar fields - will be rewritten as : @xmath66 with the exponentially fast damping solutions : @xmath67 where @xmath68 are integration constants .
thus the asymptotic solutions go to vacuum ones with the zero energy density ( fig . ) .
in this section we consider a brane - like solution in 4d .
let us chose the metric in the form : @xmath69 where the metric function @xmath70 depends only on the coordinate @xmath71 .
this metric describes ( 2 + 1)-dimensional spacetime embedded in a ( 3 + 1)-dimensional spacetime . using , and
, one can obtain the @xmath72 and @xmath73 components of einstein equations : @xmath74 and the scalar field equations : @xmath75,\\ \chi^{\prime \prime}+3 \frac{a^\prime}{a}\chi^\prime&=&\chi\left[2\varphi^2+\lambda_2(\chi^2-m_2 ^ 2)\right ] , \label{field_wall_2}\end{aligned}\ ] ] where a prime denotes differentiation with respect to @xmath71 and the arbitrary constant @xmath13 is choosing as follows @xmath76 for the purpose of zeroing of @xmath77 at @xmath78 ( see eqs . and below ) .
( here and further we use the same rescaling for all variables and parameters as in the previous section . )
we will solve the system of equations - with the following boundary conditions at @xmath78 : @xmath79 the procedure of finding of solutions is the same as in the previous section .
the obtained solutions with masses @xmath80 and @xmath81 for the scalar fields are presented in fig .
, for the metric function @xmath70 in fig . and for the energy density @xmath82 in fig .
we can easily estimate an asymptotic behavior of the solutions .
one can see from eq .
that asymptotically the right hand side tends to zero and the solution of this equation is : @xmath83 where @xmath84 and @xmath85 are integration constants .
this solution corresponds to the de sitter - like solution for the space variable @xmath71 .
then , using and seeking for asymptotic solutions of the equations - in the form : @xmath86 where @xmath60 as @xmath87 , we will have the following equations for @xmath88 and @xmath89 from - : @xmath90 with the damping solutions : @xmath91 } , \label{sol1_wall } \\ \delta \chi & \approx & c_{\chi } \exp{\left[-\frac{x}{2}\left(3\alpha+ \sqrt{9\alpha^2 + 4(2 m_1 ^ 2-\lambda_2 m_2 ^ 2)}\right)\right ] } , \label{sol2_wall}\end{aligned}\ ] ] where @xmath68 are integration constants .
so we have the solutions that tend asymptotically to the local minimum of the potential at @xmath92 and @xmath93 .
.,width=340 ] .,width=359 ] near @xmath78 with asymptotic anti - de sitter behavior.,width=340,height=226 ]
for consideration of the spherically symmetric problem we take the metric in schwarzschild coordinates : @xmath94 then the @xmath73 and @xmath95 components of the einstein equations will be : @xmath96 , \label{einst3_sph}\end{aligned}\ ] ] and the scalar field equations are : @xmath97~ ,
\\ \chi^{\prime \prime}+\left(\frac{2}{r}+\frac{b^\prime}{2b}-\frac{a^\prime}{2a}\right)\chi^\prime= a\chi\left[2\varphi^2+\lambda_2(\chi^2-m_2 ^ 2)\right]~ , \label{sfe2_sph}\end{aligned}\ ] ] where a prime denotes differentiation with respect to @xmath29 and in potential @xmath98 for the purpose of zeroing of the energy density as @xmath99 .
equation is consequence of the preceding equations - .
choosing the boundary conditions at @xmath49 in the form : @xmath100 and following the above procedure of obtaining of solutions we find the masses @xmath101 at @xmath81 .
the results of numerical calculations for the scalar fields are presented in fig .
, for the metric functions @xmath28 in fig . and for the energy density @xmath82 in fig . .
one can see from - that asymptotic behavior of the metric functions @xmath102 and @xmath103 is : @xmath104 where @xmath62 and @xmath63 are constants .
practically @xmath62 defines a total mass and @xmath63 the run of time at the infinity . redefining the time variable @xmath65 we can chose @xmath105 as @xmath99 .
i.e. we have asymptotically flat minkowski spacetime .
the corresponding asymptotic scalar field equations - with account of @xmath106 will be : @xmath107 with the exponentially fast damping solutions : @xmath108 where @xmath68 are integration constants .
thus the asymptotic solutions go to vacuum ones with the zero energy density ( fig . ) .
finally , let us show evolution of the effective equation of state @xmath109 where @xmath110 and @xmath2 are the effective energy density and pressure of the scalar fields . in the case under consideration
we have from and : @xmath111,\\ t^1_1&=&-p(r)=-\left[-\frac{1}{2a}\left(\varphi^{\prime 2}+\chi^{\prime 2}\right)+v(\varphi,\chi)\right].\end{aligned}\ ] ] then the corresponding effective equation of state will be : @xmath112 using the numerical solution obtained earlier , we have the following graph for the equation of state ( see fig . ) .
as one can see from this figure , there is some point @xmath113 in which the denominator in tends to zero ( in the case under consideration @xmath114 ) . in the range
@xmath115 we have @xmath116 ( @xmath117 at @xmath49 ) , and in the range @xmath118 we have @xmath119 . at @xmath120 from the left @xmath121 and at @xmath120 from the right @xmath122 .
asymptotically @xmath123 tends to zero , i.e. we have the dust - like equation of state .
it follows from this that in the entire range @xmath115 the wec is violated , i.e. efficiently the scalar fields describe phantom matter . on the other hand in the range @xmath124 the wec
is not violated and the scalar fields are non - phantom .
the scalar fields with the equation of state varying in time are used in the theories describing the present acceleration of the universe ( see , e.g. , @xcite ) . in such theories
there is some point of time in which the violation of the wec and transition of usual scalar fields into phantom ones take place . in our case
we deal with inhomogeneous distribution of the scalar fields with the varying in space equation of state and transition of usual scalar fields into phantom ones in some point @xmath113 . for the spherically symmetric case.,width=340 ] for the spherically symmetric case.,width=340 ] for the spherically symmetric case.,width=340 ] for the spherically symmetric case.,width=340 ]
we have considered three static solutions for the model of two interacting phantom and ghost scalar fields : the model of a traversable wormhole , the brane - like model and the spherically symmetric problem .
the self - consistent problems with account of the back reaction of the scalar fields on gravitation were investigated .
the choice of a potential in the form ensures existence of two local minima that allow to find regular solutions which start and finish in one of these minima . in this case the nonlinear problems on evaluation of eigenvalues of the parameters @xmath26 ensuring the mentioned regular solutions were solved .
note that existence of such solutions depends on values of the self - coupling constants @xmath12 and the boundary conditions .
there is some range of these parameters appropriate for existence of regular solutions .
particularly , this range is defining by conditions on existence of the local and global minima of the potential : @xmath125 ; @xmath126 ; @xmath127 .
for the wormhole model , it was shown that there exist regular solutions in the entire range @xmath30 with asymptotically flat minkowski spacetime ( see eqs .
i.e. the obtained wormhole solution connects two flat spacetimes .
the radius of the wormhole throat is defined by a minimal value of the function @xmath102 , i.e. @xmath128 at @xmath49 . as one can see in fig .
, for a remote observer the phantom energy looks like a compact object localized near the throat of the wormhole with some negative energy density . for the brane - like case ,
the obtained solutions describe ( 2 + 1)-dimensional spacetime embedded in a ( 3 + 1)-dimensional spacetime .
the asymptotic value of the potential , @xmath129 plays the role of a negative cosmological constant , and so the metric is asymptotically anti - de sitter with corresponding anti - de sitter horizon . in this sense
such a solution differs from domain wall solution which has flat asymptotics . for the spherically symmetric case we have found particle - like solutions with asymptotically flat minkowski spacetime .
it was shown that the effective equation of state @xmath123 for the scalar fields changes essentially along the radius @xmath29 ( see fig . ) .
there exists some point @xmath113 dividing the whole space into two regions : in the entire range @xmath115 , @xmath7 that corresponds to phantom - like behavior of the scalar fields . on the other hand , in the range @xmath130 the equation of state @xmath119 that corresponds to non - phantom ( ghost ) scalar fields .
similar behavior of the equation of state could be obtained both for the wormhole and brane - like models as well .
let us note that for all three models , the obtained solutions correspond to soliton - like solutions starting and finishing in the same minimum ( in the cases under consideration , in the local minimum @xmath131 of the potential ) . in models with one scalar field
soliton - like solutions exist only in a case of presence of two or more minima and they start from one minimum of a potential and tend asymptotically to another one . by the terminology of @xcite , such solutions refer to topological solutions , and our solutions - to non - topological ones . note here that the known constraint on possibility of existence of regular static solutions in spaces with dimensionality @xmath132
does not work in our case . as it was shown in ref .
@xcite ( see also ref .
@xcite ) , the mentioned solutions for usual ( nonphantom ) scalar fields do not exist if the potential @xmath133 in the whole space . on the other hand , for phantom fields the condition @xmath134 should be satisfied
always ( see in this connection ref .
@xcite ) . in our case
the potential changes its sign that allow to avoid the constraint of the theorem from ref .
@xcite and obtain static regular solutions .
our attempts to find regular solutions with @xmath135 were not succeeded . however
, this does not exclude a possibility of their existence at some parameters @xmath12 in the potential and some special boundary conditions .
the authors would like to thank the referee for helpful comments and for some suggested references .
99 r. ruffini and s. bonazolla , _ phys . rev . _ * 187 * , 1767 ( 1969 ) .
m. colpi , s. l. shapiro , i. wasserman , _ phys .
lett . _ * 57 * , 2485 ( 1986 ) .
e. mielke and f. schunck , _ gen .
grav . _ * 33 * , 805 , ( 2001 ) , gr - qc/9801063 .
a. linde , `` particle physics and inflationary cosmology '' , hep - th/0503203 .
a. riess et al .
, _ astron .
j. _ * 116 * , 1009 ( 1998 ) , astro - ph/9805201 ; + s. j. perlmutter et al . , _ astroph .
j. _ * 517 * , 565 ( 1999 ) , astro - ph/9812133 .
v. sahni and a.a .
starobinsky , _ ijmp _ * d 9 * , 373 ( 2000 ) , astro - ph/9904398 ; + v. sahni , _ lect .
notes phys . _
* 653 * , 141 ( 2004 ) , astro - ph/0403324 ; + v. sahni , dark matter and dark energy `` , astro - ph/0403324 ; + a.d . dolgov , ' ' problems of vacuum energy and dark energy , hep - ph/0405089 .
starobinsky , phys .
* b 91 * , 99 ( 1980 ) ; + s. capozziello , int . j. mod .
* d 11 * , 483 ( 2002 ) , gr - qc/0201033 ; + s. capozziello , v.f.cardone , s. carloni , a. troisi , _ int . j. mod .
phys . _ * d12 * , 1969 ( 2003 ) ; + s. nojiri and s. d. odintsov , `` introduction to modified gravity and gravitational alternative for dark energy '' , hep - th/0601213 ; + v. folomeev , v. gurovich and h. kleinert , geometric model of dark energy , astro - ph/0501209 ; + v. folomeev , v. gurovich and i. tokareva , _ grav . cosmol . _
* 12 * , 163 ( 2006 ) , astro - ph/0701375 .
a. gong and a. wang , phys .
d 75 * , 043520 ( 2007 ) , astro - ph/0612196 ; + u. alam , v. sahni and a.a .
starobinsky , jcap * 0702 * , 011 ( 2007 ) , astro - ph/0612381 ; + s. nesseris and l. perivolaropoulos , jcap * 0702 * , 025 ( 2007 ) , astro - ph/0612653 ; + t.m .
davis et al . , astro - ph/0701510 ; + y. wang and p. mukherjee , phys . rev .
* d 76 * , 103533 ( 2007 ) , astro - ph/0703780 ; + r.a .
daly et al .
, arxiv:0710.5345 [ astro - ph ] .
l. amendola , _ phys .
rev . _ * d 60 * , 043501 ( 1999 ) .
e. elizalde , s. nojiri and s. d. odintsov , phys .
d * 70 * ( 2004 ) 043539 .
caldwell , _ phys .
* b 545 * , 23 ( 2002 ) , astro - ph/9908168 ; + r.r .
caldwell , m. kamionkowski and n.n .
weinberg , _ phys .
lett . _ * 91 * 071301 ( 2003 ) , astro - ph/0302506 ; + s.m .
carroll , m. hoffman and m. trodden , _ phys .
rev . _ * d 68 * , 023509 ( 2003 ) , astro - ph/0301273 .
m. szydlowski , w. czaja and a. krawiec , _ phys .
* e 72 * , 036221 ( 2005 ) , astro - ph/0401293 ; + e. elizalde , s. nojiri and s.d .
odintsov , _ phys .
* d 70 * , 043539(2004 ) , hep - th/0405034 ; + s. capozziello , s. nojiri , and s.d .
odintsov , unified phantom cosmology : inflation , dark energy and dark matter under the same standard , hep - th/0507182 .
v. dzhunushaliev , v. folomeev , k. myrzakulov , r. myrzakulov , phantom fields : bounce solutions in the early universe and s - branes `` , gr - qc/0608025 .
r. rajaraman , _ an introduction to solitons and instantons in quantum field theory _ ( north - holland publishing company , amsterdam , new york , oxford , 1982 ) .
a. sen , ' ' non - bps states and branes in string theory
, hep - th/9904207 .
s. sushkov , wormhole ssupported by a phantom energy `` , gr - qc/0502084 .
f.s.n . lobo , ' ' stable phantom energy traversable wormhole models , gr - qc/0603091 .
j. magueijo and r. h. brandenberger , `` cosmic defects and cosmology '' , astro - ph/0002030 .
m. cveticand , h. h. soleng , _ phys.rept . _
* 282 * , 159 ( 1997 ) , hep - th/9604090 .
berezin , v.a .
kuzmin , i.i .
tkachev , _ phys.lett . _
* 120b * , 91 ( 1983 ) .
berezin , v.a .
kuzmin , i.i .
tkachev , _ phys.rev . _ * d36 * , 2919 ( 1987 ) .
d. garfinkle , c. vuille , _
class.quant.grav . _
* 6 * , 1819 ( 1989 ) .
ipser , p. sikivie , _ phys.rev . _ * d30 * , 712 ( 1984 ) .
p. laguna - castillo , r.a .
matzner , _
phys.rev . _ * d34 * , 2913 ( 1986 ) .
m. visser , _ lorentzian wormholes : from einstein to hawking _
( woodbury , new york , 1996 ) v. dzhunushaliev , _ hadronic j. suppl . _ * 19 * , 185 ( 2004 ) .
a. a. andrianov , f. cannata , a. y. kamenshchik , _ phys.rev . _
* d72 * , 043531 ( 2005 ) , gr - qc/0505087 . | three static models with two interacting phantom and ghost scalar fields were considered : a model of a traversable wormhole , a brane - like model and a spherically symmetric problem .
it was shown numerically that regular solutions exist for all three cases . |
since the seminal work by vaiana et al .
( 1981 ) it has become clear that all late type stars share the same basic coronal characteristics : the presence of thermal plasma at several million degrees , its magnetic confinement , the presence of flares etc . also the basic plasma physics governing coronae is probably very similar in all these stars . on the other hand ,
coronae cover a large range of x - ray luminosity ( even among stars of the same spectral type and luminosity class ) and a few scientists think that the coronae of the most active late - type stars are different from the solar corona , albeit the crucial differences have rarely been stated clearly .
two otherwise identical stars may have rather different levels of stellar activity , i.e. x - ray luminosity , spottedness , uv luminosity etc . , because of very different rotation speed ( pallavicini et al . , 1981 ) and age ( for a review on the subject micela , 2002 )
; the speed probably determines how effectively the stellar magnetic dynamo generates magnetic fields .
one may then expect that many differences among late - type stars coronae are just due to rather different regimes of the dynamo at work inside stars , while the basic plasma phenomena occurring in the outer magnetized coronal plasma are probably the same and only the amount and scale of such phenomena are different . along this line of thought
, this work explores to which extent most of the differences found among the various coronae can be attributed to their different composition in terms of various kinds of coronal structures ( ranging from the relatively faint and cool structures of the background corona to the very bright and hot flaring regions ) and to the amount of _ x - ray emitting _ coronal structures present ; on the other hand we explore to which extent the plasma phenomena and the structures involved in the coronae are similar even in stars of very different activity levels . according to this paradigm , the solar corona offers the opportunity to study and understand better the building blocks of any late - type - star corona . to this end we consider the average surface flux ( @xmath0 ) and the spectral hardness ratio ( hr ) in the rosat / pspc band , of late - type stars with those of solar coronal structures ; we also study the strong @xmath0hr correlation ( schmitt , 1997 ; henceforth s97 ) . in the following
we do not claim that all the aspects of solar corona are the same in the other stars but that the fundamental features are the same ; for instance there is evidence of polar active regions and of large polar flares ( e.g. schmitt and favata , 1999 ; brickhouse , dupree and young , 2001 ; _ see also schrijver and title , 2001 ) .
we will concentrate on the x - ray emission from stellar coronae and we will compare the solar and the various stellar cases and through analogies and differences we will infer ( or just constraint ) features of x - ray emitting structures present in stellar coronae . _
a few works along this line of thought have been published , either studying the temperature vs. emission measure per unit area on active stars ( schrijver et al .
1984 ; jordan and montesinos , 1991 ) or correlating activity parameters ( e.g. luminosity in given bands or emission measure ) with temperature ( schmitt et al . , 1990 ; guedel et al . , 1997 ; preibish , 1997 ) .
more in detail , schrijver et al .
( 1984 ) analyzed a sample of 34 late - type stars observed with the einstein telescope , related the relevant coronal temperature and emission measure per unit area and rotation rate , determining scaling laws among them and relating the first two through loop models .
jordan and montesinos ( 1991 ) studied various scalings among emission measure , coronal temperature , surface gravity and rossby number .
schmitt et al . (
1990 ) performed an extensive survey of late - type stars observed with the einstein telescope , studying and relating the coronal features and showing some differences among m stars , f and g stars , giants and rs cvn stars .
guedel et al .
( 1997 ) studied the `` x - ray sun in time '' with the help of various x - ray observations of solar - like stars of different ages ; they found a power law dependence of the total x - ray luminosity with the temperature of the hottest component of a two - temperature spectral fit .
preibish ( 1997 ) studied a sample of young late - type stars and found a power law dependence of total x - ray luminosity vs. the temperature of the hottest component .
this work is based on methods and results developed in our program to study the solar - stellar connection in x - rays , mostly based on rosat / pspc and yohkoh / sxt data . _
our approach allows us to tie directly , with just little elaboration and interpretation , the vast amount of data collected with the two aforementioned observatories . also our method and scope is novel with respect to previous works , since we use solar data ( made homogeneous to stellar data ) as a template and a guide .
our findings are eventually interpreted in terms of coronal loops scaling laws .
_ rosat has provided the largest _ homogeneous _ data set of stellar coronal observations so far collected ; this data set is an important reference for studies of stellar coronae because : i ) the relevant data have been collected with just one instrument , ii ) instrumental and observing conditions were relatively stable , iii ) a large number of stars of different spectral types and luminosity classes have been observed and iv ) the data set has an extended time coverage . for analogous reasons ,
yohkoh / sxt data are _ equally _ important to study the solar corona .
it is , therefore , natural to refer both to rosat and to yohkoh data in order to model a scenario of coronae on late - type stars and to connect it to the solar corona . in the following we also use results of previous works on the solar - stellar connection in x - ray band plus some stellar data already published in different contexts ; our scope is to draw a global perspective on coronal structure .
the paper is organized as follows : section 2 describes our method to use the solar corona as a template to study stellar coronae , section 3 presents and discusses some of the related results along with related works on stellar x - ray emission , in section 4 we draw our conclusions .
the scope of this program has been to study the x - ray emission of the whole sun as an x - ray star , and the contribution of the various coronal structures to its emission , using yohkoh / sxt data ; we have translated the solar observations into a format and into a context homogeneous to those of stellar observations .
thanks to this approach , we can now easily perform comparative studies among solar and stellar observations .
yohkoh / sxt data have relatively high time resolution , down to a few seconds , and cover uninterruptedly approximately ten years , from the maximum of the solar cycle in 91 to the following one . all sorts of phenomena
have been observed , ranging from extremely large flares to moments of very low emission , during the cycle minimum . the spatial resolution ( typically @xmath2 )
allows us to identify and study , with high spatial detail , the various regions origin of the x - ray emission and to discriminate , as well as to isolate , the contribution of different coronal structures to the coronal x - ray emission . in brief , from the yohkoh / sxt images of the sun at a given time
, we derive a @xmath3 pixels map of coronal plasma temperature and the corresponding map of emission measure ; from them we derive the distributions of coronal emission measure vs. temperature at each observation time ; then , summing the optically thin spectrum emitted at each temperature ( obtained with the corresponding emission measure value and a spectral synthesis code ) we obtain the spectrum of the whole corona ; folding it with the known instrumental characteristics ( effective area vs. photon energy and photon redistribution matrix ) , we derive the stellar - like focal plane spectrum of the sun , as it would be observed with the x - ray telescope of interest ( orlando et al .
, 2000 - henceforth paper i ; peres et al . , 2000 - henceforth paper ii ) .
in this work we will focus on the rosat / pspc data set and will use the rosat standard data analysis also to analyze solar data translated into stellar format and conditions . the capability to isolate the various kinds of coronal structures has allowed us to study their conditions and their contributions to the coronal spectrum ( for the flares , see reale et al . , 2001 - henceforth paper iii ;
for the background corona , the active regions and the core of active regions , see orlando et al .
2001 - henceforth paper iv ) .
the good time coverage of yohkoh / sxt observations has been exploited to study how the emission measure distribution and the x - ray spectrum of the whole corona changes along the solar cycle , with a regular time sampling and taking care not to include flare events ( paper ii and paper iv ) .
s97 presented the results of a large set of rosat / pspc observations of late - type stars , within 13 pc from the sun , in the form of a diagram ( fig .
[ fig1 ] ) of f@xmath4 ( the average x - ray surface flux of the star ) vs. hr ( the x - ray spectral hardness ratio ) .
the hardness ratio is defined as @xmath5 - where h is the total flux in the rosat / pspc spectral channels between 0.55 and 1.95 kev , and s the total flux in the channels between 0.13 and 0.40 kev .
it is a simple characterization of the relative weight of the hard and soft part of the spectrum .
, i.e. @xmath6 k since the hr becomes inadequate ( see also fig . [ fig2 ] ) ; also small hr errors around that temperature yield large temperature errors . ]
both f@xmath4 and @xmath7 are useful to characterize the strength of coronal activity , albeit averaged over the whole corona .
indeed , f@xmath4 gives the coronal radiative losses of the star _ per unit area _ and is related to the amount of energy delivered into the corona , i.e. the coronal heating , also _ per unit area _ ; it can discriminate between two stars of the same x - ray luminosity but of very different radius and surface area : the smaller star will have a much higher flux than the other , hinting at a much higher level of plasma activity ; f@xmath4 is an activity marker independent of the stellar surface value .
hr is a simple tracer , albeit a non - linear one , of the coronal average temperature .
it is another index of plasma activity , independent from f@xmath4 .
[ fig2 ] shows the relationship between the plasma temperature and the hr of the spectrum derived with the me
l ( mewe , lemen , and van den oord 1986 ; kaastra , j.s .
1992 ; mewe , kaastra and liedahl , 1995 and references therein ) spectral synthesis model .
note that hr is little sensitive to temperatures below @xmath8 k ; it also saturates and is multi - valued in the range 0.0 - 0.4 ; therefore hr is also relatively insensitive to very hot plasma .
moreover hr should not go beyond 0.4 at any plasma temperature .
the s97 results show that late type stars cover a diagonal strip , with hr ranging from -1.0 to 0.0 and f@xmath4 from @xmath9 to @xmath10 ( fig .
[ fig1 ] ) .
the more active a star , the higher are both its f@xmath4 and its hr ; the highest hr ( temperature ) and f@xmath4 values belong to uncommonly active single stars .
the overall trend suggests that the confined plasma in the corona of a late type star , on the average , gets brighter as it gets hotter .
analogously , marino et al .
( 2001 ) reported the data of 40 late type stars of spectral type between f7 and k2 , within 25 pc from the sun , in an analogous f@xmath4 vs. hr diagram , shown in fig .
the stars of this sample are , on the average , more luminous and active than those in s97 ; also , at variance from s97 , m and late k stars were not included .
there were 70 pointed rosat / pspc observations and some of these stars were observed several times ; also some observations , of sufficient signal to noise ratio , were split to allow time - resolved analysis during the observation .
some of these stars moved , in time , along the same diagonal track where most of the stars are located ; however some stars evolved along a path in this graph , less steep ; we claim , in the following discussion , that this different evolution may be due to the contribution of flares .
also steltzer ( 2001 ) generated preliminary f@xmath4 vs. hr diagrams for stars in clusters . in paper
ii we studied the solar corona at the maximum , at an intermediate phase and at the minimum of the solar cycle , taking care not to include flare events ; also the whole solar corona during a large flare event was studied .
paper iii was entirely dedicated to study , with the same method , the evolution of eight representative flare events ranging from small to large ones and covering most of the conditions encountered in flares ; each flaring region was studied separately from the rest of the corona _ so the surface flux and the hr were derived just for the flaring region and its effect was not added to , and diluted within , the whole corona .
_ paper iv studied the solar corona at several moments of the solar cycle , sampling the cycle better than in paper ii and studying separately the various classes of coronal structures , classified as background corona ( very quiet regions ) , active regions and cores of active regions , according to their surface brightness in yohkoh / sxt bands .
the set of whole the data mentioned above yields a good sampling of the physical conditions in the corona from the background corona to the strongest flares and span the whole solar cycle .
_ for the present work we have taken all the flares in the sample of paper iii because they are well observed so they can be studied and characterized accurately ; also , since they span a large range within the flares classification and the various phases of their evolution are well observed , they provide a good sampling of the various conditions encountered in flares , from the largest down to very small events .
most of the attention , in the following will be on the physical conditions of various flare phases .
we did not consider micro - flares because they can hardly be discriminated from small ar variability and their contribution in the context of average stellar @xmath0 and hr is , in practice , the same as bright ars .
[ fig4 ] shows a @xmath0 - hr graph of the whole solar corona during the cycle plus the evolution of the whole solar corona , near maximum of the cycle , in the course of a x-9 flare ; the graph collects some results of paper ii , of paper iv and of orlando et al .
the flux of the whole solar corona was derived dividing the total corona luminosity by the whole solar surface area .
the evolution of the whole solar corona along the solar cycle , apart from * moderate to large * flares , spans the lower part of the diagram , the same covered by low activity stars , i.e. the ones at low f@xmath4 and low hr , well below the levels of most active stars .
k or lower , as well as the relevant emission measure , while the same plasma would have significant emission in the rosat / pspc band ( see judge , solomon , ayres , 2003 ) .
the solar data points relative to background corona and/or the very quiet sun at cycle minimum may have to be shifted at higher flux ; this effect is not crucial , however , in the context of this paper . ]
* solar flares span a quite large range , from very faint , barely detectable , to very large ones ; the flares included in our work are those that at least triggered the yohkoh flare mode ; in this respect they can be considered to be representative of flares from moderate to large . *
if a * moderate to large * flare happens at maximum of the cycle , it makes the luminosity ( and the flux ) of the solar corona change at most by a factor @xmath11 , during large flares .
however , the corresponding increase of hr can be larger than the changes occurring along the solar cycle , for the same increase of luminosity .
analogous flares occurring during the solar minimum carry a similar change in hr but a much larger change of luminosity and flux .
thus , under the effect of * significant * flares , the whole sun moves rightward in the graph , on a path less steep than that covered by stars of various activity levels .
the occurrence of large flares may explain why the evolution of some stars studied by marino et al .
( 2001 ) follows paths , in this diagram , less steep than the track of all the other stars : albeit they show no evidence of large changes of flux , they may undergo a moderate flaring episode which shows up through the hr change and the different path of the star in the diagram ( see also orlando et al , 2002 , 2004 ) .
[ fig5 ] shows f@xmath4 vs. hr for four specific classes of structures ( i.e. background corona , active regions , cores of active regions and flares ) .
it is worth noting that the flux has been derived by dividing the luminosity of the structure ( or of the structures of that class ) by the surface covered in yohkoh / sxt images .
the interpretation of these results is useful for stellar physics : a star entirely covered with solar - like active regions would be in the same part of the f@xmath4-hr diagram where solar active regions are , and analogously for all the other kind of structures , since they would yield the same hr ( average temperature ) and surface flux values .
therefore we can foresee how differently a stellar corona would behave if covered with the various kinds of solar coronal structures .
_ more in general , however , _
a mixture of two different classes of structures would yield values intermediate between the two , the effective value depending on the relative fractions of stellar surface covered .
interestingly enough , solar structures of different kinds ( with the exception of solar flares ) span the same region of the graph covered by stellar coronae of different activity levels .
_ quite likely the stellar surface is not entirely covered with x - ray emitting structures and different classes of structures ( according to our classification ) do coexist on the star , covering different fractions of its surface . while places at the extremes of the graph can most likely be explained with only one kind of structure , a locus between two classes of structures may be explained with an appropriate ( non - linear ) combination of the two .
_ * it is worth noting that , according to the analogy we are drawing here , a star equivalent to a structure , e.g. an active region , would be located in the same place in the diagram if it were fully covered by this type of structure ; similarly a star intermediate between two kinds of structure should have its surface close to 100% covered with the two kinds of structures . in the solar case we know from yohkoh / sxt observations that active regions , their cores and flares ( when they occur ) take a small fraction of the surface of the sun , the rest taken by the very faint background corona ( as shown by very long exposure images ) .
s97 has shown that very faint coronae have a surface flux comparable to that of very faint solar background corona .
we can , therefore , imagine that also on active stars a faint background corona covers the surface not covered by active structures . * on the basis of the above interpretation , then , we see that the coronae of stars with extremely low surface flux and hardness ratio could be just a large solar coronal hole as , on the other hand , s97 suggested .
continuing this exercise , by increasing the fraction of stellar surface covered with active regions , or with the cores of active regions,_or a combination thereof , _
the corona spans the strip covered by coronae of late - type - stars from low to intermediate activity .
therefore we can draw a _
hypothesis according to _ which the levels of stellar coronal x - ray emission are due to different fractions of the stellar surface covered with active regions or even cores of young active regions .
the fraction of more active components increases going towards coronae of higher @xmath0 and higher hr .
with one notable exception : the few , very - high-@xmath0 ( @xmath12 ) and very - high - hr ( @xmath13 ) , late type stars .
the _ average _ characteristics of these few stars are beyond the _ peak _ values of the solar coronal active regions cores .
yet they are well below * conditions of solar flares of interest here*. the most active late type stars can , therefore , be explained as being partly covered with active regions , their cores and with flares or flare - like ( i.e. very hot and dense ) structures _ at all times_. very active stars may then be continuously subject to several uncorrelated flares whose light curves overlap randomly , canceling most of the evidence of variability . some authors have already proposed the presence of continuous flaring activity in stellar coronae ( wood et al .
1997 ; drake et al .
2000 ; guedel et al .
2002 , and references therein ) .
flares can contribute to coronae of low activity on the strip as long as the flares are small ; indeed large flares in low activity stars are very evident and make their @xmath0 - hr point move along a path much less inclined than the strip , bringing them out of the strip itself .
there appears to be no match in the solar corona for those super - hot and super - luminous flares detected on very active stars and on rs cvns with the beppo - sax and asca satellites ( franciosini et al .
, 2001 ; maggio et al . 2000 ; favata et al . 2000 ; favata and schmitt , 1999 ; schmitt and favata , 1999 ) .
the @xmath0 - hr diagram provides information on the average features of the coronal loops forming the structures .
the amount of energy emitted from the coronal part of any loop is approximately one half of the entire heating of the loops per unit cross - sectional area ( approximately equal to the part of the solar surface covered by the loops footpoints ) , the other half being emitted from the transition region ( vesecky et al . , 1979 ) .
therefore , considering that @xmath0 is proportional to the heating @xmath14 and , using the scaling laws by rosner , tucker and vaiana ( 1978 ) , in the following rtv78 , i.e. @xmath15 and @xmath16 we find @xmath17 where @xmath18 , @xmath19 and @xmath20 are proportionality factors , @xmath21 is the total radiated flux , @xmath22 is the heating per unit volume averaged along field lines , p is the coronal plasma pressure , t is the coronal loop * maximum * temperature and l is the half - length of the magnetic loop .
then , given the very weak dependence on t , @xmath0 is very close to being linearly related to the coronal pressure . a closer look to the above derivation shows that , on one hand , the linear relation between @xmath0 and @xmath21 is appropriate only for a large part ( but not all ) of the temperature range , since one expects that for very high t ( above a few @xmath23 k ) or very low t ( below @xmath8 k ) different fractions of @xmath21 fall outside the rosat band .
nonetheless the main point of the above exercise is to show that @xmath0 variations are largely dominated by pressure variations . of course
, hr changes are related to temperature changes . using the simple relation derived above and the hr vs. t relation in fig .
[ fig2 ] , it is straightforward to translate the @xmath0 vs. hr diagram into an `` average pressure '' vs. `` average temperature '' diagram , deriving the temperature from the single - valued part ( @xmath24 k ) of the hardness ratio and then the pressure as @xmath25 .
[ fig6 ] shows the pressure - temperature graphs for solar structures , for the whole solar corona , for the f and g stars in s97 , and for the g stars ( from f7 to k2 ) in marino et al .
( 2001 ) the data set by marino et al .
( 2001 ) is the most homogeneous one of all those reported here .
we have fitted the relevant data , finding the relation @xmath26 shown as a dashed line in all the graphs in fig .
[ fig6 ] ( @xmath27 is the temperature in units of @xmath8 k ) ; more precisely the power index found from the fitting is @xmath28 .
it is quite interesting that all the stars , the solar corona during the cycle and the various coronal structures follow this law , which shows a rather steep dependence of the confined plasma pressure on the plasma temperature , probably governed by the heating mechanism(s ) due to the magnetic field dissipation .
_ flares appear to depart , to some extent , from such a law , more specifically there appear to be a flattening of the pressure - temperature relation with slightly lower pressure than that predicted by the above scaling law . _ the diagonal strip covered by the solar corona at various phases , and by solar coronal structures is a coarse but firm relation between average pressure and temperature of the loops at different evolutionary stages of the solar structures they form , from emergence to decay .
there is an analogous , striking , relation for stellar coronae at different activity levels .
steady loop models can not be applied to the rise phase of flares .
however they can be applied , with caveats , to flares during the slow evolution of the decay phase .
because of this , the pressure values corresponding to solar flares are less precise than all the others and have to be taken with care .
indeed the large spread of solar - flares - related points suggests the limits of using rtv78 scaling laws on flares .
we also note that the pressure values of the solar structures appear slightly lower than _ the values predicted according to eq .
( [ eq : pt ] _ ) ; we should consider , however , that the surface considered , in the yohkoh / sxt image , is invariably not uniformly filled with loops . therefore the local x - ray flux ( and the average pressure ) is higher than what we measure , even for the solar structures .
determining in detail the area really filled with loops , from each yohkoh / sxt image , is a formidable task .
we note in passing that this choice simply skips on the problem of the surface filling factor by coronal structures , probably present also below the angular resolution ( see however appendix a ) .
an equivalent , albeit slightly different , form of the same relation is shown in fig .
[ fig7 ] where we show the average density vs. temperature , derived with the ideal gas law for a fully ionized hydrogen plasma @xmath29 where @xmath30 is the hydrogen number density and @xmath31 is the boltzmann constant . in this figure
we have put together all the solar and the stellar data ; obviously we find @xmath32 using the rtv78 scaling law [ eq : rtvt ] and substituting there the p - t relation found above , it is straightforward to translate these graphs and laws into the analogous ones relating the characteristic structure length ( l ) vs. temperature ( t ) .
[ fig8 ] shows a graph for the same data as above ; the l - t scaling is @xmath33 analogously , substituting the relations [ eq : pt ] and [ eq : lt ] into the scaling law [ eq : rtveh ] we obtain the relation @xmath34 shown as a dashed line in fig .
[ fig9 ] which also shows the same solar and stellar data in a @xmath22 vs. t graph .
_ flares appear to depart from all the above scaling laws and , in particular , to yield a lower volumetric heating than predicted by the above scaling law , as an implication of the departure from the pressure - temperature scaling law .
this may just be an implication of the non - stationarity of the relevant x - ray emitting plasma .
however , since many of such data are related to the decay phase , i.e. when the loops are in quasi - steady conditions and rtv scaling laws should be applicable , we should leave open the possibility that there is indeed a difference in the dissipation mechanisms involved in coronal steady loops and flares . _
since dissipation mechanisms tend to connect volumetric heating with the characteristic size , we are interested to derive a volumetric - heating vs. structure - length law . combining the above scaling laws we find @xmath35
also probably dictated by the dispersal and dissipation mechanisms of the magnetic field .
all the above derivation does not take into account the fraction of the stellar surface covered by the x - ray emitting region , i.e. the surface filling factor , for clarity s sake .
appendix a contains the generalization of the derivation , considering the surface filling factor .
we have described the coronae of late - type stars of different activity levels as covered by correspondingly different fractions of structures similar to those present in the solar corona .
we have classified and grouped these structures as background corona , active regions , active regions cores and flares , in order of increasing surface x - ray flux and hardness ratio .
indeed , we find the same correlation of @xmath0 vs. hr for coronae of late - type stars and for coronal structures present on the sun .
the solar coronal structures , from background corona to large flares , fall in the same region of the @xmath0 and hr diagram where the late - type stars coronae , from very inactive to extremely active , are located .
the occurrence of a large flare on a star of medium _ or low _ activity level should make the star s position move , in this diagram , toward the right ( i.e. with a significant increase of hr ) with relatively moderate increase of @xmath0 ( paper ii ; orlando et al . ,
_ thus significant flares in these stars should be very prominent for their variability but also because they should make the star move outside the strip covered by coronae .
_ the most active late type stars appear to have both @xmath0 and hr significantly higher , even than the active region cores , but one can still explain them assuming a significant presence , at any time , of flares or of flare - like structures . although one can not exclude that even moderate activity stars may have a contribution due to flares ,
most of their behavior can be easily explained with active regions .
flares , in analogy with the solar case , should lead to relatively fast evolution along more horizontal paths slightly departing from the main region of active stars .
many flares may go undetected for a variety of reasons , most likely for the very moderate increase of stellar flux or for the low signal to noise of the observation ; undetected small flares may however lead to a spread along the hr direction of the locus of the stars in the @xmath0-hr diagram .
the diagram in marino et al .
( 2001 ) , containing m stars , indeed shows that the m stars are more spread along the hr direction than the g stars ; it is quite fitting that m stars are more variable than g ones on short time scales ( marino et al .
2002 ) . only the very late evolutionary phases of active region cores , going toward rather low hr but high surface flux appear to deviate a little from this scenario .
however the fraction of surface covered by these very old cores appear to be rather small and so , their emission may be irrelevant with respect to that of the whole stellar corona , as it happens in the solar corona .
the x - ray observations made with skylab , thanks to their several months coverage , allowed for the first time to study the long term evolution of the x - ray corona ( for a review , see vaiana and rosner , 1978 ) .
it became clear that the magnetically confined coronal plasma evolves from the compact , hot , bright and high pressure loops of emerging active regions to progressively longer , cooler , dimmer and relatively low - pressure loops , first of active regions and then of the background corona . in this context , therefore , one can also consider the @xmath0 vs. hr diagram for solar coronal structures as an evolutionary path .
the structures move along the region covered by the various structures starting as very high @xmath0 , pressure and temperature and small l and , as they evolve , become of progressively lower @xmath0 , pressure , and temperature and get longer . the various structures of confined plasma , when reported in the pressure - temperature diagram of fig .
[ fig6 ] or , equivalently , in the others , in figs .
[ fig7 ] , [ fig8 ] and [ fig9 ] , are all located along a particular track , probably resulting from the physical effects governing the dispersion and dissipation of the magnetic field as it evolves from emergence to total dispersion into very background corona .
interestingly enough , this track characterizes both individual structures in the solar corona and the solar corona as a whole at different phases of the solar cycle .
the global characteristics are the results of a sort of averaging the features of all the structures present in the corona , with a weight due to the surface coverage and to the relative brightness .
as a consequence , the average conditions of the solar corona at minimum of the cycle are closer to those of the background corona , even if some active regions may be present , and those of the corona at maximum are closer to active region conditions .
the track also describes the global features of the coronae of solar - type stars . while in the solar case we are able to resolve the entire set of structures at the various evolutionary stages , each star instead just yields one data point in such a diagram , which yields an average over all the structures present in corona .
the coronal configuration of the solar - type stars , including the sun , results from the competing effects of emergence of magnetic field ( generated by the dynamo mechanisms inside the star ) and the magnetic field dispersal and dissipation in the outer stellar layers ; as a consequence , the average values corresponding to each point in the diagram are determined by these competing physical effects .
the higher the emergence rate , the more the point characterizing the stellar corona moves towards the high - pressure - high - temperature part of the graph , as we observe on the sun ( papers ii and iv ) . in general , stellar coronae should have a complex coronal composition , according to this scenario , with different fractions of their surface covered with compact , hot and bright coronal loops and possibly with interactions among magnetic fields of different ar ( drake et al .
a more active star has a higher number of continuously emerging ar with very effective magnetic field amplification and dissipation mechanisms .
we also claim that each star should have a distribution of structures of confined plasma obeying this sort of universal pressure - temperature relationship with loop plasma pressure higher for higher average loop temperature .
we note that flares appear to behave differently , as for the pressure - temperature relation and the related heating - temperature scaling law .
as mentioned , this may depend on the non - stationarity of the relevant x - ray emitting structures but since some of the observations pertain to the decay flares phase when the departure from stationarity is small , we should leave open the possibility that heating mechanisms in flares are indeed different from those of steady structures . _ but , if each of the structures evolve similarly and are equivalent to the solar ones ( as we assume ) and the various coronae are just the sum of the same structures , one should detect all the coronae just as a time - average of identical coronal structures , thus all with the same characteristics , independently of the number of structures present .
so why a higher emergence rate should lead to higher average surface flux and temperature ?
_ on one hand , a higher emergence rate of magnetic structures brings more ars and so more x - ray emitting regions to the surface per unit time , leading to higher luminosity and average x - ray flux of the star . as for the temperature and hr , the emission measure of an ar and of its core
are significantly larger around their emergence ( orlando et al . , 2004 ) ; also flares in an ar are more frequent after emergence .
the unresolved corona approximate the time - average of the structures only if the number of structures of various age visible at any time is so large to sample all the evolutionary phases , from the ar emergence to the `` final decay '' , and including flares .
we doubt that a low - activity star has enough ars to yield a reasonable `` average behavior of structures '' .
flares make the `` poor sampling '' effect even worse .
typically a `` low activity '' star is caught in a non - flaring phase and , if flares occur , they are easily identified and typically analyzed separately from the rest of the observation . instead ,
when we observe a very active star , probably one ( or more ) of its ars is in some phase of a flare . indeed , in order to explain the most active stars we need something brighter and hotter than just active regions , most likely flares , at any time ( drake et al .
, 2000 ; paper iii ) . even on the most active stars ,
however , we may not get a large enough number of ars to sample well the various flare phases , despite we may observe enough ars to sample ( and average ) their non - flaring evolution . the stars in the highest part of the @xmath0 - hr diagram appear to vary somehow from one observation to the other ( micela , private communication ) .
the locus identifying each very active star in the diagram may wander around a bit , probably because of flares ; the centroid of the stars loci at any time may yield the average star behavior and their spread relative to the centroid may yield the stellar variations , apart from experimental uncertainties and stellar peculiarities .
the invariable presence of flares probably makes a further big difference between less active and very active stars .
for completeness we should consider at least two additional effects : 1 ) a large number of ars may subtract substantial amount of surface to the faint , diffuse , structures ; so the corona is slightly biased toward higher temperatures and brightness with respect to the average of the structures evolution we would see on the sun ; 2 ) as mentioned above if the ars crowding is substantial , ars may interact triggering even more flares than expected from just a sum of solar structures .
we have shown that we are able to explain the correlation of @xmath0 vs. hr of late - type - stars coronae present in the data of s97 , and all the scaling laws we have derived , as due to the plasma structures making the coronae . now
the focus moves to explaining why _ x - ray emitting _ solar structures follow that very relationship , a task which probably should be accomplished within solar physics and should address the mechanisms governing magnetic field dispersal and dissipation .
we acknowledge useful suggestions by e. franciosini , a. maggio , a. marino , g. micela , j. sanz - forcada , s. sciortino and b. steltzer _ and by an anonimous referee who helped to improve the paper_. this work was partially supported by agenzia spaziale italiana and by ministero della istruzione , universit e ricerca .
while in the text we have derived the stellar x - ray surface flux by dividing the stellar x - ray luminosity @xmath36 by the stellar surface @xmath37 , a more correct derivation of the x - ray surface flux at the star should consider that only the area @xmath38 on the star may be covered by coronal structures .
therefore the whole x - ray luminosity should be @xmath39 and therefore the surface flux we have derived should include a surface filling factor : argiroffi , c. , maggio , a. , peres , g. 2003 , `` on coronal structures and their variability in active stars : the case of capella observed with chandra / letg '' , a&a , submitted brickhouse , n.s .
, 1996 in iau colloq . 152 : astrophysics in the extreme ultraviolet , p. 105 , brickhouse , n.s . , 2001 , `` chandra and xmm - newton : x - ray spectral synthesis at high resolution '' in `` stellar coronae in the chandra and xmm - newton era '' , proc . of the 35-th eslab symposium , estec , 25 - 29 june 2001 , favata , f. , and drake , j. editors , in press brickhouse , n.s . ,
dupree , a.k .
, young , p.r . , 2001 , , 562 , 75 drake , j.j . , 1996 in proc .
of 9th cambridge workshop , ed . by r. pallavicini and a.k .
dupree , asp conf .
109 , p. 203 drake , j.j . ,
peres , g. , orlando , s. , laming , j.m . , maggio , m. 2000 , , 545 , 1074 favata , f. , reale , f. , micela , g. , sciortino , s. , maggio , a. , matsumoto , h. 2000 , a&a , 353 , 987 favata , f. , schmitt , j. h. m. m. 1999 , a&a , 350 , 900 favata , f. & micela , g. 2003 , space science reviews , in press franciosini , e. , pallavicini , r. , tagliaferri , g. 2001 , a&a , 375 , 196 guedel , m. , audard , m. , skinner , s.l . , horvath , m.i .
2002 , , 580 , 73 guedel , m. , guinan , e.f . , & skinner , s.l . , 1997 , , 483 , 947 huenemoerder , d.p . , canizares , c.r . , and schulz , n.s . , 2001 559 , 1135 huenemoerder , d.p . , canizares , c.r . ,
drake , j.j . , and sanz - forcada , j. 2003 `` the coronae of ar lac '' , submitted jeffries , r. d. 1999 , in `` solar and stellar activity : similarities and differences '' , asp conference series v. 158 , ed .
c. j. butler & j. g. doyle . , p.75 jordan , c. , & montesinos , b. , 1991 , mnras , 252 , 21p judge , p.g . ,
solomon , s.c .
ayres , t.r . , 2003 , , 593 , 534 kaastra , j.s . 1992 , an x - ray spectral code for optically thin plasmas ( internal sron - leiden report , updated version 2.0 ) marino , a. , micela , g. , peres , g. , sciortino , s. 2002 , a&a , 383 , 210 marino , a. , micela , g. , peres , g. , sciortino , s. 2003 , `` x - ray variability in rosat - pspc observations of f7-k2 stars '' , proceedings of the `` x - ray 2000 '' meeting , 4 - 9 september 2000 , giacconi , r. , serio , s. , and stella , l. , editors , asp conf .
series , vol .
234 ( on the cd - rom ) maggio , a. , drake , j.j . ,
kashyap , v. , micela , g. , peres , g. , sciortino , s. , harnden , jr .
murray , s.s .
2003 , `` ad leo observed with chandra letg : emission measure distribution , density , and element abundances of the coronal plasma '' , in `` stellar coronae in the chandra and xmm - newton era '' , proc . of the 35-th eslab symposium , estec , 25 - 29 june 2001 , favata , f. , and drake , j. editors , in press maggio , a. , pallavicini , r. , reale , f. , tagliaferri , g. 2000 , a&a , 356 , 627 mewe , r. , kaastra , j.s . ,
liedahl , d.a .
1995 , legacy , 6 , 16 mewe , r. , lemen , j.r . , and van den oord , g.h.j .
1986 , a&as , 65 , 511 micela , g. 2002 , `` evolution of stellar coronal activity on the main sequence '' in `` the evolving sun and its influence on the planetary environments '' , b. montesinos a. gimenez e.f .
guinan , eds .
asp conference series , vol .
269 , p.107 micela , g. 2002 , in `` stellar coronae in the chandra and xmm - newton era '' .
asp conference series , edited by f. favata and j.j .
drake , ( 2003 ) , v. 277 , p. 263
orlando , s. , peres , g. , reale , f. 2000 , , 528 , 524 orlando , s. , peres , g. , reale , f. 2001 , , 560 , 499 orlando , s. , peres , g. , reale , f. 2002 , `` observing the sun as an x - ray star : active region evolution and rotational modulation '' , in `` stellar coronae in the chandra and xmm - newton era '' , proc . of the 35-th eslab symposium , estec , 25 - 29 june 2001 , favata , f. , and drake , j.j .
editors , a.s.p .
series , v. 277 , p. 341
orlando , s. , peres , g. , reale , f. 2004 , `` the sun as an x - ray star : active region evolution , rotational modulation , and implications for x - ray stellar variability '' , a&a , in preparation pallavicini , r. , golub , l.,rosner , r. , vaiana , g. s.,ayres , t. , linsky , j. l. 1981 , , 247 , 692 peres , g. , orlando , s. , reale , f. , rosner , r. ; hudson , h. 2000 , , 528 , 537 randich , s. 2000 , in `` stellar clusters and associations : convection , rotation , and dynamos '' .
asp conference series , vol .
198 . edited by r. pallavicini , g. micela , and s. sciortino ( 2000 ) , p.40 reale , f. , peres , g. , orlando , s. 2001 , , 557 , 906 rosner , r. , tucker , w.h . , & vaiana , g.s .
1978 , , 220 , 643 sanz - forcada , j. , brickhouse , n.s . , and dupress , a.k . , 2002 , 570 , 799 schmitt , j.h.m.m . , collura , a. , sciortino , s. , vaiana , g.s . ,
harnden , f.r.jr . , & rosner , r. , 1990 , , 365 , 704 schmitt , j. h. m. m. 1997 , a&a , 318 , 215 schmitt , j. h. m. m. , favata , f. 1999 , nature , 401 , 44 schrijver , c.j . ,
mewe , r. & walter , f.m .
1984 , a&a , 138 , 258 schrijver , c.j .
, & title , a.m. 2001 , , 551 , 1099 steltzer , b. , 2001 , `` x - ray variability of young stars in taurus - auriga - perseus '' poster presented at jenam 2001 , held in munich , sept .
10 - 14 , 2001 vaiana , g. s. , and 15 coauthors 1981 , , 245 , 163 vaiana , g. s. , rosner , r. 1978 , ara&a , 16 , 393 vesecky , j. , antiochos , s. , & underwood , j. 1973 , , 233 , 987 wood , b.e . ,
linsky , j.l . , ayres , t.r .
1997 , , 478 , 745 | this work is dedicated to the solar - stellar connection , i.e. the close similarity of the sun and the late - type stars ; in particular this work shows that stellar coronae can be composed of _ x - ray emitting _ structures similar to those present in the solar corona . to this end
we use a large set of rosat / pspc observations of late - type - stars of all spectral types and activity levels , and a large set of solar x - ray data collected with yohkoh / sxt .
solar data have been analyzed and formatted to study the sun as an x - ray star ; they include observations of the solar corona at various phases of the cycle and data on various kinds of _ x - ray _
coronal structures , from flares to the background corona , i.e. the most quiet regions .
we use the x - ray surface flux ( @xmath0 ) vs. spectral hardness ratio ( hr ) diagram as a fundamental tool for our study .
we find that @xmath0 is strongly correlated to hr in stellar coronae , in the solar corona at all phases of the cycle , and in the individual solar coronal structures ; all the above follow the same law .
schmitt found the same correlation in stellar coronae .
we therefore claim that coronae of late type stars are formed with _
x - ray _ structures very similar to the solar ones , since their behavior is identical to that of the solar coronal structures and of the whole solar corona .
the spatial location of the _ x - ray _
structures on the star , however , can be very different than on the sun .
in this scenario , the fraction of the stellar surface covered with active regions and with their bright cores increases with activity ; the most active stars are brighter and hotter than if they were entirely covered with active regions so they can be explained only with the additional presence of several flares ( or flare - like structures ) at any time . on the basis of the @xmath0 vs. hr correlation ,
_ corresponding to @xmath1 _ , we then derive a set of new laws relating the temperature , pressure , volumetric heating and characteristic loop length of the coronal plasma , on all the late type stars .
also , individual solar coronal structures and the whole solar corona , follow the same laws
. these laws also agree with recent findings of higher plasma density at higher temperatures in stellar coronae .
we claim that the strong correlation between surface flux and temperature and the laws mentioned above are just the effect of more fundamental physical mechanisms driving the coronal structures of all the late - type stars from the emergence of new magnetic structures to their dispersal and dissipation . |
a time projection chamber ( tpc ) is foreseen as the main tracker of the detector for a linear collider such as tesla @xcite .
currently the possibility to use gas electron multipliers ( gem ) @xcite for the charge amplifying system is studied extensively . when using gems for the tpc end plate , the pads directly detect the amplified electron cloud which results in a fast and narrow charge signal .
the fine structures of the gem of the order of 100 @xmath1 m size allow the detection of the charge cloud with high spatial resolution .
in contrast to wires , a gem shows no preferred direction , thus any @xmath2 effects will be isotropic .
finally , using different electric fields on both gem sides , the drift of ions produced inside the gem holes into the tpc drift volume ( ion feedback ) can be suppressed and the need of a gating grid can be potentially avoided . to demonstrate the advantages of a tpc with gem readout , a prototype chamber will be built within the r&d activities of the linear collider tpc group @xcite resembling a sector of the tpc as in the tdr design
. it will be operated in test beams and in magnetic fields up to 5 t. the end plates of this prototype will contain up to three planes of charge amplifying gem foils .
the operation conditions of this multi - gem - structure have to be studied beforehand , because in the case of three gem planes six electric fields have to be set . for these studies two small test chambers
were set up at rwth aachen and installed in a 5 t magnet at desy hamburg .
the first chamber was dedicated to the measurement of charge transfer in high magnetic field and the second one to the determination of the transverse charge spread as a function of the magnetic field .
the measurements described below were accomplished in a 5 t superconducting magnet at desy ( see figure [ fig : magnet ] ) . after formerly serving as magnet compensating the main field of the zeus experiment
, the solenoid has been reinstalled in the cryogenic plant of the hera accelerator at desy to provide a high magnetic field facility for detector r&d projects for a future linear collider such as tesla .
the magnet has an aperture of 28 cm and a coil length of 120 cm . including its cryostat , the total length amounts to 186 cm .
the coil is cooled by 4.5 k liquid helium to ensure superconductivity .
the solenoid is operated at currents up to 1000 a corresponding to a central magnetic field of up to 5.3 t as shown in figure [ fig : magnet ] . with the ramp rate of 0.5 @xmath3
, it takes about half an hour to ramp up the magnet from 0 t to 5.3 t. more detailed information on the magnet facility is presented in @xcite .
the mechanical and electric setup of our test chambers has been described in detail in reference @xcite .
the gas volume consists of a composite frame enclosing a stack of three standard @xmath4 gem foils . using thin absorbers the radiation from an @xmath0fe source of 1 gbq activity
is diminished such that about @xmath5 photons per second reach the chamber . a small window on top of the chamber covered by a 24 @xmath1 m thick mylar foil allows the photons to penetrate into the chamber .
the chambers were operated with a gas mixture as it is proposed in the tesla technical design report @xcite which consists of ar(93%),ch@xmath6(5%),co@xmath7(2% ) .
a regulation system was used to keep the absolute pressure of the chamber at a constant value within 1 mbar .
in addition the chamber s temperature was continuously monitored and recorded . each electrode ( gem surfaces and cathode )
is connected to an individual channel of a caen sy127 hv power supply . for the current measurements
the anode plane consists of a solid copper electrode of the same size as the gem structures @xcite .
current monitors with a resolution of about 0.1 na are inserted into the supply line of each hv channel . to measure the anode current ,
the anode plane is connected to ground via an additional current monitor .
the high voltage control and current readout are handled by a custom application running on a linux pc .
for the measurement of individual pulses an anode with a finely segmented area with 8 strips of 0.8 mm pitch is used @xcite .
the signal pulse of each strip is read out via a preamplifier and digitised using a 100 mhz flash adc .
the eight fadcs digitise their input signal continuously and store it into a ring - buffer memory .
after arrival of a trigger all fadcs are stopped and the memory is read out by a linux computer . a trigger is generated if one of the strip signals exceeds the level of an analog discriminator . in figure
[ fig : setup ] an overview of the electric setup and read - out system of the two test chambers is given .
figure [ fig : chamber ] shows one of the chambers just before insertion into the aperture of the superconducting magnet .
our primary motivation for making tests in a magnet was to investigate whether there is a significant drop in collection efficiency for gems in high b - fields parallel to the electric field as suggested by the langevin formula @xmath8 where @xmath9 and @xmath10 are unit vectors of the fields , @xmath11 is the mean time between two collisions of the drifting electron , and @xmath12 is the cyclotron frequency . the last term proportional to @xmath13 , which gives the contribution along the magnetic field lines , could cause a drop in collection efficiency for high magnetic fields .
when this term dominates , most electrons will no longer be collected into a gem hole , but stay on drift lines perpendicular to the gem surface and eventually reach the gem s copper coating .
those charges would be lost for the signal and consequently decrease the chamber s @xmath14 capabilities due to the loss in primary ionisation statistics .
figure [ fig : collection ] shows a numerical simulation using the programs maxwell @xcite , magboltz @xcite and garfield @xcite . in total 20 drift lines of electrons starting in front of a gem hole
are shown for a 4 t magnetic field perpendicular to the gem surface .
none of the drift lines end up on top of the gem , but all electrons are collected into the gem hole along curling tracks . from these simulations
no significant drop in collection efficiency is expected in high magnetic fields . on the other hand
the numerical simulations predict an increase in extraction efficiency .
the situation is shown in figure [ fig : extraction ] , where the drift lines of electrons out of a gem hole at 0 t and at 4 t magnetic field are compared . in both simulations
the same number of electrons start within the gem hole . from the number of extracted drift lines
one observes that more of the electrons are extracted at 4 t. this is explained by the third term in equation [ langevin ] which is bending the drift lines along the b field direction . in figure
[ fig : juelich ] the measured anode current as function of the magnetic field is shown .
measurements done in a 2 t magnet at the forschungszentrum jlich @xcite are nicely reproduced by our new measurements at the 5 t magnet at desy .
it rises approximately by a factor of 2 between 0 t and 5 t. simultanously the extraction efficiency of the electrons out of the last gem in front of the anode plane was measured . because the chamber was operated with a symmetric setup ( all gem voltages and electric fields at the same value ) the increase of the anode current can be calculated from the product of collection efficiency , gain and extraction efficiency of a single gem to the power of three .
the independently measured extraction efficiency can explain the rise in the anode current .
this shows that the product of collection efficiency and gain is only slightly affected by the change of the magnetic field .
we see no indication for severe losses of primary ionisation charge .
one important charge transfer quantity is the ion feedback , which describes how much ion charge is transferred into the drift volume per electron charge collected on the anode plane .
ions reaching the tpc drift volume would represent a positive space charge and deteriorate the electric drift field .
the ion feedback can be derived from the measurements by dividing the cathode current by the anode current .
highly asymmetric electric fields on both sides of a gem lead to ion suppression .
the small electric field in the drift volume for example causes many of the drift lines from the amplification region inside the gem hole to end on the copper plane facing the tpc drift volume .
therefore , the relatively small drift field typical of a tpc automatically leads to ion feedback suppression .
moreover , the ion feedback can be minimised by the variation of the electrical fields within the gem structure @xcite .
we searched for a setting with optimum ion feedback suppression with the drift field constrained at 200 v / cm and requiring an effective gain of @xmath15 .
the first two gems are set at voltages of 310 v , the last gem at 350 v. the first transfer field and induction field are set very high ( 6000 v / cm and 8000 v / cm respectively ) whereas the second transfer field is set rather low ( 60 v / cm ) . using this optimised
setting allows a long term and stable operation of a triple - gem structure with ion feedback well below 1% .
the ion feedback is then measured as a function of the magnetic field .
the result is presented in figure [ fig : ionfeedback ] and shows that the ion feedback is further reduced with increasing magnetic field strength .
this is due to the improved electron extraction already mentioned before . in a magnetic field of 4 t
an ion feedback of only 2.5 permille has been achieved .
the information on the charge spread within a gem structure is important to estimate the optimum pad size for the gem tpc .
the spatial resolution of a tpc improves with a magnetic field parallel to the electric field , because the transverse diffusion of the electron cloud is suppressed . a very narrow primary charge cloud , for example 300 @xmath1 m in ar(95%):cf@xmath6(5% ) at 4 t after a drift distance of 2.5 m ,
will reach the gem readout structure .
then the charge cloud is further broadened by effects within and between the gem foils .
possible effects are the gas amplification , highly divergent electrical field lines and the diffusion within high electric fields . the part of the charge broadening which is caused by diffusion effects can be reduced in high magnetic fields .
we present new measurements of the charge spread by gem structures in high magnetic fields .
the cluster width distribution for individual events has been analysed using a segmented pad anode .
only events of the photo peak , where the total photon energy has been deposited inside the chamber , are used for this analysis . in these events the 6 kev photon of the @xmath0fe source is converted into a 3 kev photo - electron , which is predominantly emitted perpendicular to the incoming photon .
additionally , a 3 kev auger electron is emitted from the excited argon ion .
the two electrons ionise the gas along their tracks and a primary charge cloud is produced .
the primary charge cloud reaches the gem structure after a maximum drift distance of 4 mm .
then the width of the charge cloud is further broadened inside the gem structure . in this experiment
the width of the charge cloud is determined from the charge sharing between adjacent strips .
first the centre of the charge cloud is calculated from the centre of gravity of the measured signals on each of the eight strips . the charge sum measured on the left half of the anode ( strips 03 )
is compared to the charge sum on the right side ( strips 47 ) .
the charge ratio , @xmath16 is calculated for each event .
this ratio is then plotted versus the charge centre in figure [ fig : eta ] . as result one
observes the typical shape of an error function as expected for a charge cloud of gaussian shape . from the slope of the data
points one determines the width of the charge distribution .
the charge width of the primary ionisation cloud is estimated to be in the range of 3070 @xmath1 m , depending on the b field , and can therefore be neglected .
the result of the cluster width measurement is shown in figure [ fig : spread ] .
the rms width of the charge cloud is reduced from 0.5 mm without magnetic field to 0.3 mm at 4 t. from a numerical simulation using the program magboltz we would expect a charge width between 400 @xmath1 m at 0 t and 240 @xmath1 m at 4 t due to the diffusion in the high transfer fields between the gem foils . if we add a b - field independent charge broadening effect of the gem structure of 215 @xmath1 m size quadratically , measurement and simulation agree .
the charge transfer and the charge broadening of triple - gem structures have been studied in magnetic fields up to 5 t perpendicular to the gem surface . with increasing magnetic field ,
an improved electron extraction efficiency has been measured , leading to a reduction of ion feedback by a factor of 2 .
no indication for a large drop in electron collection efficiency has been observed .
the charge broadening of gem structures has been measured and we conclude that the charge broadening nature of a triple gem structure is not severely diminished by magnetic fields up to 4 t.
we thank the cern printed - circuits workshop and r. oliveira for providing the gem foils .
those of us from rwth aachen thank desy for its hospitality .
we acknowledge the financial support of this r&d project by the german bundesministerium fr bildung und forschung .
e. gadwinkel _ et .
_ , _ cryogenics for a 5 tesla superconducting solenoid with large aperture at desy _ ,
proceedings of cryogenic engineering conference cec 2003 , anchorage , alaska , usa , to be published in _ advances in cryogenic engineering_. | we report on measurements of charge transfer in gem structures in high magnetic fields .
these were performed in the framework of the r&d work for a time projection chamber at a future linear collider .
a small test chamber has been installed into the aperture of a superconducting magnet with the gem structures mounted perpendicular to the b field direction .
the charge transfer is derived from the electrical currents monitored during irradiation with an @xmath0fe source .
no severe loss of primary ionisation charge is observed , but an improved ion feedback suppression is achieved for high magnetic fields .
additionally , the width of the charge cloud released by individual @xmath0fe photons is measured using a finely segmented strip readout after the triple gem structure .
charge widths between 0.3 and 0.5 mm rms are observed , which originate from the charge broadening inside the gem readout .
this charge broadening is only partly suppressed at high magnetic fields .
pitha 04/01 + , , , , , , , , time projection chamber , tpc ; gas electron multiplier , gem ; ion feedback ; electron transparency ; spatial resolution |
cluster algebras were introduced by fomin - zelevinsky @xcite , and have been influential in many settings including algebraic combinatorics , lie theory , poisson geometry , teichmller theory , total positivity and quiver representations . the first link to quiver representations and tilting theory for algebras was given in @xcite .
cluster categories were defined in @xcite , giving a categorical model for cluster combinatorics . for type @xmath1 ,
an independent approach was given in @xcite .
a cluster algebra is in essence a commutative ring with a distinguished countable family of generators , called _
cluster variables_. the cluster variables can be grouped into overlapping sets of equal finite size @xmath2 , called _ clusters_. any given cluster variable can be uniquely expressed as a rational function in the elements of a fixed initial cluster . by the laurent phenomenon , @xcite , these rational functions are actually laurent polynomials with integer coefficients .
corresponding to each cluster there is a quiver @xmath3 , with @xmath2 vertices .
the cluster algebra is said to be acyclic if it admits a cluster for which the corresponding quiver @xmath3 has no oriented cycles . in this case
there is a corresponding finite dimensional hereditary algebra , the path algebra of @xmath4 ( for some field @xmath5 ) giving rise to a cluster category . in @xcite
a surjective map @xmath6 from the cluster variables of an acyclic cluster algebra to the exceptional indecomposable objects in the corresponding cluster category was given .
in @xcite a map was defined in the opposite direction , and it was shown that this map is a bijection with @xmath6 as inverse ( see also @xcite ) .
our exposition is made in such a way that it is easy to see where the fact that @xmath6 is a bijection is used .
an interesting property of the map @xmath6 from @xcite is the following .
choose a cluster whose corresponding quiver is acyclic . if a cluster variable ( not from this cluster ) is expressed as a laurent polynomial in the cluster variables from this cluster there is a natural interpretation of the monomial in the denominator in terms of the composition factors of its image under @xmath6 ( regarded as a @xmath4-module ) .
this generalized earlier results in this direction @xcite .
it is interesting to ask whether for an arbitrary choice of initial cluster it is possible to interpret the monomial in the denominator of a cluster variable in terms of the composition factors of the module ( over an appropriate cluster - tilted algebra ) corresponding to the image of the cluster variable under @xmath6 .
the cluster - tilted algebra involved should be the endomorphism algebra of the cluster - tilting object @xmath7 obtained by forming the direct sum of the images of the fixed initial cluster under @xmath6 .
we remark that such an interpretation has been found in the dynkin case in @xcite . in this paper
we give a precise answer to this question for the general case by giving necessary and sufficient conditions on the cluster - tilting object @xmath0 for this to hold .
for a given acyclic cluster algebra , these conditions hold for all cluster - tilting objects @xmath0 in the corresponding cluster category if and only if the algebra is of finite cluster type or of rank 2 . in the tame case
we show that a cluster - tilting object @xmath0 satisfies the conditions if and only if no indecomposable regular summand of @xmath0 has quasilength exactly one less than the rank of the tube in which it lies , and that this is equivalent to @xmath8 for each indecomposable summand of @xmath0 .
in fact this last condition is necessary for any path algebra @xmath4 , and we conjecture that it is also sufficient in general .
we would like to thank otto kerner for answering several questions on wild algebras , and especially for assisting us with the proof of proposition [ p : endnotk ] .
let @xmath3 be a finite connected acyclic quiver and @xmath5 an algebraically closed field .
let @xmath9 be the corresponding ( finite dimensional ) path algebra and @xmath10 the bounded derived category of finite dimensional @xmath4-modules .
it has autoequivalences @xmath11 ( the auslander - reiten translate ) and @xmath12 $ ] ( the shift ) .
let @xmath13 $ ] be the corresponding cluster category .
we shall regard modules over @xmath4 as objects of @xmath14 ( via the natural embedding of the module category over @xmath4 in @xmath10 ) . for @xmath15 a vertex of @xmath3 let @xmath16 denote the corresponding indecomposable projective @xmath4-module .
let @xmath17 be the ( acyclic , coefficient - free ) cluster algebra defined using the initial seed @xmath18 , where @xmath19 is the transcendence basis @xmath20 of @xmath21 . for an object @xmath22 of @xmath23 ,
let @xmath24 . for a polynomial @xmath25
, we say that @xmath26 satisfies the _ positivity condition _ if @xmath27 where @xmath28 ( with a @xmath29 in the @xmath15th position ) for @xmath30 .
this condition was crucial in the investigations in @xcite , and also plays an essential role here .
we first recall the following results .
@xcite [ bmrtfirst ] every cluster variable of @xmath31 is either of the form @xmath32 ( for some exceptional @xmath4-module @xmath22 ) or of the form @xmath33 for some @xmath15 , where in either case @xmath34 satisfies the positivity condition .
every indecomposable exceptional @xmath4-module arises in this way , and if @xmath22 and @xmath35 are two such modules satisfying @xmath36 then @xmath37 .
a map @xmath6 is then defined from cluster variables of @xmath31 to ( isomorphism classes of ) exceptional indecomposable objects of @xmath14 in the following way . if @xmath32 is a cluster variable for some exceptional indecomposable @xmath4-module @xmath22 then @xmath38 .
if @xmath33 is a cluster variable then @xmath39 $ ] .
recall that a _ tilting seed _ for @xmath31 is a pair @xmath40 where @xmath0 is a cluster - tilting object of @xmath14 and @xmath41 is the quiver of @xmath42 .
@xcite [ bmrtsecond ] the map @xmath43 is surjective .
it induces a surjective map @xmath44 , and a surjective map @xmath45 , preserving quivers .
we also have the following result of @xcite , see also @xcite .
[ calderokellerresult ] there is a bijection @xmath46 from indecomposable exceptional objects of @xmath14 to cluster variables of @xmath31 such that for any indecomposable exceptional @xmath4-module @xmath22 , we have @xmath47 as an irreducible fraction of integral polynomials in the @xmath48 . in particular @xmath6 is an inverse of @xmath49 and for the cluster variables @xmath33 considered in theorem [ bmrtfirst ] , we can only have @xmath50 .
let @xmath51 be a quiver mutation - equivalent to @xmath3 .
then there is a seed @xmath52 of @xmath31 , where @xmath53 is a transcendence basis of @xmath21 . for @xmath30 let @xmath54 so that we have @xmath55 .
it follows from theorem [ bmrtsecond ] that @xmath56 is a cluster - tilting object in @xmath14 and that the quiver of its endomorphism algebra in @xmath14 is @xmath51 .
hence the same holds for @xmath57 .
we can ask whether results analogous to the above hold for cluster variables when they are expressed in terms of the initial cluster @xmath58 . for @xmath22
an object in @xmath14 , let @xmath59 .
an interesting open question is whether @xmath60 ( or , equivalently , the dimensions of the spaces @xmath61 for @xmath30 ) determines @xmath22 uniquely .
this is true if @xmath62 is the direct sum of the non - isomorphic indecomposable projective modules but is not known in general .
in particular , this means that an approach following @xcite exactly in this more general context is not possible .
[ typestar ] let @xmath63 be a cluster variable of @xmath31 .
we say that @xmath63 expressed in terms of the cluster @xmath64 has a _
@xmath0-denominator _ if either : + ( i ) we have that @xmath65 for some exceptional indecomposable object @xmath22 of @xmath14 not isomorphic to @xmath66 for any @xmath15 , and @xmath67 , or + ( ii ) we have that @xmath68 for some @xmath15 and @xmath69 . + in either case , @xmath26 is a polynomial in the @xmath70 satisfying the positivity condition . our first main result is the following : [ main1 ] let @xmath3 be a finite quiver with no oriented cycles , and let @xmath5 be an algebraically closed field .
let @xmath14 be the cluster category associated to @xmath4 , and let @xmath57 be a cluster - tilting object in @xmath14 .
let @xmath71 be the cluster algebra associated to @xmath3 . then
: * if no indecomposable direct summand of @xmath0 is regular then every cluster variable of @xmath31 has a @xmath0-denominator . * if every cluster variable of @xmath31 has a @xmath0-denominator , then @xmath8 for all @xmath15 .
we note that , as in theorem [ bmrtfirst ] , the bijectivity of @xmath6 can be used to show that in case ( ii ) of definition [ typestar ] , only @xmath72 occurs .
the proof of theorem [ main1 ] will be completed at the end of section [ s : thepreprojectivecase ] . in the special case where @xmath4 is a tame algebra
, we have the following : [ main2 ] suppose that we are in the situation of theorem [ main1 ] and that , in addition , @xmath4 is a tame algebra .
then the following are equivalent : * every cluster variable of @xmath31 has a @xmath0-denominator . *
no regular summand @xmath73 of quasi - length @xmath74 lies in a tube of rank @xmath75 .
* for all @xmath15 , @xmath8 .
the proof of theorem [ main2 ] will be completed at the end of section [ s : thetamecase ] . as a consequence of our main results ,
we have in the general case : [ main3 ] let @xmath3 be a finite quiver with no oriented cycles .
let @xmath14 be the cluster category associated to @xmath4 .
let @xmath71 be the cluster algebra associated to @xmath3 .
then every cluster variable of @xmath31 has a @xmath0-denominator for every cluster - tilting object @xmath0 if and only if @xmath3 is dynkin or has exactly two vertices .
the proof of corollary [ main3 ] will be completed at the end of section [ s : thepreprojectivecase ] .
in this section we consider certain conditions on a cluster - tilting object which are equivalent to all cluster variables in @xmath31 having a @xmath0-denominator .
fix an almost complete ( basic ) cluster - tilting object @xmath76 in @xmath23 .
let @xmath77 be the two complements of @xmath76 , so that @xmath78 and @xmath79 are cluster - tilting objects ( see @xcite ) .
let @xmath80 @xmath81 be the exchange triangles corresponding to @xmath22 and @xmath82 ( see @xcite ) , so that @xmath83 is a minimal right @xmath84-approximation of @xmath22 in @xmath23 and @xmath85 is a minimal right @xmath84-approximation of @xmath82 in @xmath23 .
the following definition appears to be important for understanding the link between representation theory and denominators of cluster variables : [ d : excomp ] let @xmath86 be an exceptional indecomposable object of @xmath14 .
we say that @xmath86 is _ compatible with an exchange pair @xmath87 _ if the following holds whenever @xmath88 : @xmath89 if @xmath86 is compatible with every exchange pair @xmath87 in @xmath14 we call @xmath86 _ exchange compatible_. we will investigate when this condition is satisfied . note that the case when @xmath90 or @xmath91 is covered by the following .
[ l : oldc3 ] let @xmath86 be an exceptional indecomposable object of @xmath23 and suppose that @xmath90 or @xmath91 .
then we have that @xmath92 assume that @xmath93 ( the other case is similar ) .
then there are exchange triangles @xmath94 and @xmath95 .
we have that @xmath96 .
this space is one - dimensional , since @xmath97 is an exchange pair ( by @xcite ) .
since @xmath86 is exceptional , @xmath98 , and we have @xmath99 , since @xmath100 can be completed to a cluster - tilting object .
[ p : c1toc2 ] suppose that neither @xmath22 nor @xmath82 is isomorphic to @xmath101 .
then the following are equivalent : * we have that @xmath102 * either the sequence @xmath103 is exact , or the sequence @xmath104 is exact .
consider the ( not necessarily exact ) sequence : @xmath105 by exactness at the second term we have that @xmath106 .
let @xmath107 , @xmath108 , @xmath109 and @xmath110 .
then we have that @xmath111 .
similarly , by exactness at the fourth term , we have that @xmath112 .
then the sequence is exact if and only if @xmath113 is surjective and @xmath114 is injective , if and only if @xmath115 and @xmath116 . by the above argument that @xmath117 , we see that this is true if and only if @xmath118 .
similarly , the sequence is exact if and only if @xmath119 .
hence ( b ) holds for @xmath86 if and only if @xmath118 or @xmath119 .
since in any case @xmath117 and @xmath120 , we see that ( b ) holds if and only @xmath121 as required .
let @xmath122 be indecomposable objects of @xmath14 .
we can always represent such objects by indecomposable @xmath4-modules or objects of the form @xmath123 $ ] for some @xmath15 . call these also @xmath124 and @xmath125 , by abuse of notation . by @xcite , @xmath126 .
we call an element of @xmath127 an _ @xmath128-map _ from @xmath124 to @xmath125 , and an element of @xmath129 an _ @xmath130-map _ from @xmath124 to @xmath125 .
we note that the composition of two @xmath128-maps is an @xmath128-map and the composition of two @xmath130-maps is zero .
the composition of an @xmath128-map and an @xmath130-map is an @xmath130-map .
if @xmath86 is an indecomposable exceptional object with @xmath131 , then there is an indecomposable exceptional object @xmath132 such that @xmath133 is an exchange pair and @xmath86 is not compatible with this exchange pair .
let @xmath134 be the bongartz complement of @xmath86 , as defined in @xcite .
then @xmath135 is a tilting module such that @xmath136 ( see @xcite ) .
now consider @xmath135 as a cluster - tilting object in @xmath14 , and consider the exchange triangles @xmath137 and @xmath138 , where @xmath139 and @xmath140 are in @xmath141 .
then the map @xmath142 is a sum of @xmath130-maps . since by assumption @xmath143
, we have a non - zero @xmath130-map @xmath144 .
the composition @xmath145 is zero , as the composition of @xmath130-maps is zero .
so @xmath146 applied to @xmath137 does not give a short exact sequence .
it is clear that @xmath146 applied to @xmath147 does not give a short exact sequence , since the identity map @xmath148 does not factor through @xmath149 .
we claim that @xmath86 is not compatible with @xmath150 .
we have that @xmath151 , so @xmath152 is not an exchange pair by @xcite .
it follows that @xmath153 .
we also have that @xmath154 , since @xmath86 is exceptional .
so we are done by combining the above with proposition [ p : c1toc2 ] .
we have the following immediate consequence .
[ c : endpropc ] if @xmath86 is an exchange compatible object of @xmath14 , then @xmath155 .
there is the following description of the indecomposable exceptional modules @xmath86 with @xmath131 .
we know that such @xmath86 must be regular . in the tame case the indecomposable exceptional modules in a tube of rank @xmath156 are those of quasilength at most @xmath157 . in the wild case , for a component of the ar - quiver
there is some @xmath75 with @xmath158 , where @xmath2 is the number of simple modules , such that @xmath86 is exceptional if and only if @xmath86 has quasilength at most @xmath75 .
let @xmath124 be an indecomposable regular module .
then we denote by @xmath159 the wing of @xmath124 , as defined in @xcite .
the proof of the next result , which is based on the proof of @xcite , @xcite , was pointed out to us by kerner .
[ p : endnotk ] assume that @xmath128 is hereditary of infinite type , and that @xmath86 is an indecomposable exceptional regular @xmath128-module .
then @xmath131 if and only if @xmath86 has maximal quasilength amongst the indecomposable exceptional modules in its component .
let @xmath86 be indecomposable exceptional regular of quasilength @xmath156 , and let @xmath160 be indecomposable with an irreducible epimorphism @xmath161 .
then @xmath131 if and only if @xmath162 , and @xmath160 is exceptional if and only if @xmath163 , if and only if @xmath86 does not have maximal quasilength amongst the indecomposable exceptional modules in its component ( note that all indecomposable modules in a @xmath11-orbit of an ar - component of a hereditary algebra are always either all exceptional or all not exceptional ) . consider the exact sequences @xmath164 and @xmath165 . since @xmath166 is an epimorphism and
@xmath167 is a monomorphism , it follows that @xmath162 implies @xmath168 .
assume conversely that there is a nonzero map @xmath169 .
since the wing @xmath170 determined by @xmath171 is standard @xcite , we have that @xmath172 ( see figure [ f : endnotk ] ) .
hence the composition of the map @xmath173 with @xmath174 is zero .
it follows that the map @xmath175 factors through @xmath86 and therefore that there is a non - zero map @xmath176 .
similarly , considering the wing @xmath177 , we see that @xmath178 .
hence the composition of @xmath179 with the map @xmath180 is zero , and @xmath179 factors through @xmath181 and therefore there is a non - zero map @xmath182 .
hence we see that @xmath131 if and only if @xmath86 is exceptional of maximal quasilength .
the proof in the tame case can be made even shorter .
@xmath183 & & n ' \ar[dr ] \\ & & \tau^2 n \ar[ur ] \ar[dr ] & & \tau n \ar[ur ] \ar[dr ] & & n \\ & & & \ar[ur ] & & \ar[ur ] \\ \\ \\ a \ar@{--}[uuurrr ] & & & & & & & & b \ar@{--}[uuulll ] } \endxy\ ] ] we note the following consequence of lemma [ c : endpropc ] and proposition [ p : endnotk ] : [ thm3.12 ] assume that @xmath128 is of infinite type .
if @xmath86 is indecomposable regular and has maximal quasilength amongst the indecomposable exceptional modules in its component , then @xmath86 is not exchange compatible .
in this section we show that for a given cluster - tilting object @xmath0 in the cluster category of @xmath4 , all indecomposable direct summands @xmath184 of @xmath0 are exchange compatible if and only if all cluster variables in @xmath185 have @xmath0-denominators
. let @xmath186 be a seed , with @xmath187 , and assume that each @xmath188 has a @xmath0-denominator .
let @xmath189 for @xmath30 and set @xmath190 . by theorem [ bmrtsecond ] , we have that @xmath191 is a cluster - tilting object in @xmath14 and that @xmath41 is the quiver of @xmath192 . mutating @xmath193 at @xmath194 we obtain a new cluster variable @xmath195 .
let @xmath196 .
let @xmath197 be the unique indecomposable object in @xmath14 with @xmath198 such that @xmath199 is a cluster - tilting object .
[ p : inductionstep ] with the above notation and assumptions , the cluster variable @xmath195 has a @xmath0-denominator if and only if each summand @xmath73 of @xmath0 is compatible with the exchange pair @xmath200 .
we consider the exchange triangles corresponding to the almost complete cluster - tilting object @xmath76 in @xmath14 ( see triangles and ) .
for an indecomposable direct summand @xmath1 of @xmath191 , we denote the corresponding cluster variable in @xmath201 by @xmath202 .
note that all summands of @xmath139 and @xmath140 are also summands of @xmath76 .
we therefore have , by assumption , that for any summand @xmath1 of @xmath139 or @xmath140 ( or @xmath203 ) , either @xmath204 for any @xmath15 , and @xmath205 , or @xmath206 for some @xmath15 and @xmath207 , where in either case @xmath208 is a polynomial satisfying the positivity condition .
we also define @xmath209 and @xmath210 .
we note that by @xcite , @xmath211 . if @xmath212 are summands of @xmath191 , we write @xmath213 and @xmath214 we also write @xmath215 for any @xmath216 we have @xcite that @xmath217 the argument now falls into different cases .
* case ( i ) : * + suppose first that neither @xmath22 nor @xmath82 is isomorphic to @xmath66 for any @xmath15 .
write @xmath218 , where no summand of @xmath219 is of the form @xmath66 for any @xmath15 .
similarly , write @xmath220 , where no summand of @xmath221 is of the form @xmath66 for any @xmath15 . then by assumption @xmath222 .
similarly @xmath223 .
let @xmath224 and @xmath225 . we have : @xmath226 noting that @xmath227 since @xmath228 for all @xmath229 , and similarly @xmath230 .
assume now that each of the @xmath73 is compatible with @xmath87 , so that we have @xmath231 hence @xmath232 by definition of least common multiple , @xmath233 and @xmath234 are coprime . since @xmath139 and @xmath140 have no common factors @xcite , @xmath235 and @xmath236 are coprime .
suppose that @xmath233 and @xmath236 had a common factor @xmath70 .
then there would be a summand @xmath124 of @xmath221 such that @xmath237 , and @xmath66 was a summand of @xmath140 .
but then @xmath238 contradicting the fact that @xmath139 is the direct sum of summands of a cluster - tilting object . therefore @xmath233 and @xmath236 are coprime , and similarly @xmath234 and @xmath235 are coprime .
it follows that @xmath239 and @xmath240 are coprime .
let @xmath241 .
it follows from our assumptions that @xmath242 and @xmath243 .
it is clear that @xmath244 and @xmath245 .
since @xmath239 and @xmath240 are coprime , these two numbers can not be simultaneously zero , so @xmath246 .
it follows that @xmath247 satisfies the positivity condition . by assumption
, @xmath248 also satisfies the positivity condition .
since @xmath249 is a laurent polynomial in @xmath250 by the laurent phenomenon @xcite , @xmath251 is also a laurent polynomial .
since it is defined at @xmath252 for all @xmath15 , it must be a polynomial . by the above
, it satisfies the positivity condition .
it follows that @xmath249 has a @xmath0-denominator .
now assume that at least one summand of @xmath0 is not compatible with @xmath87 .
then @xmath253 for some nontrivial laurent monomial @xmath254 .
we then obtain that @xmath255 where , as before , @xmath251 satisfies the positivity condition .
it follows that the numerator @xmath256 is either not a polynomial or does not satisfy the positivity condition and thus that @xmath257 does not have a @xmath0-denominator .
* case ( ii ) : * + next , suppose that @xmath258 for some @xmath15 .
since @xmath259 for all @xmath260 , we can not have @xmath261 for any @xmath262 ( as @xmath263 ) .
assume that the @xmath184 are all compatible with the exchange pair @xmath87 .
then we have @xmath264 , noting that @xmath265 . here
we use , in particular , that @xmath266 holds for @xmath267 , by lemma [ l : oldc3 ] . as in the above case
, we have @xmath268 . the same analysis , together with the fact that @xmath269 for some polynomial @xmath26 in the @xmath270 satisfying the positivity condition , provides : @xmath271 as above
, we see that the numerator is a polynomial which satisfies the positivity condition , and it follows that @xmath249 has a @xmath0-denominator .
now assume that at least one summand of @xmath0 is not compatible with @xmath87 .
summands @xmath272 which are not compatible with @xmath87 , will not satisfy @xmath273 for @xmath274 .
it follows that @xmath275 for some nontrivial laurent monomial @xmath254 . arguing as above we obtain that @xmath276 since @xmath277 satisfies the positivity condition , it follows that the numerator @xmath278 is either not a polynomial or does not satisfy the positivity condition and therefore that @xmath279 does not have a @xmath0-denominator .
* case ( iii ) : * + next , suppose that @xmath280 for some @xmath15 . since @xmath259 for all @xmath260 , we can not have @xmath281 for any @xmath262 ( as @xmath263 ) .
assume that the @xmath184 are all exchange compatible with @xmath282 .
we have , using proposition [ p : c1toc2 ] , that @xmath283 , noting that @xmath284 . as in the above case
, we have @xmath268 .
the same analysis provides : @xmath285 as above , we see that @xmath251 is a polynomial which satisfies the positivity condition and it follows that @xmath249 has a @xmath0-denominator .
now assume that at least one summand of @xmath0 is not compatible with @xmath87 .
then , as in case ( ii ) , it follows that @xmath286 for some nontrivial laurent monomial @xmath254 .
arguing as above we obtain that @xmath287 since @xmath277 satisfies the positivity condition , it follows that @xmath278 is either not a polynomial or does not satisfy the positivity condition and therefore that @xmath279 does not have a @xmath0-denominator .
[ t : all ] all indecomposable direct summands @xmath184 of @xmath0 are exchange compatible if and only if all cluster variables in @xmath288 have a @xmath0-denominator .
we note that @xmath289 for @xmath30 and that the quiver @xmath51 of the seed @xmath52 is the quiver of @xmath290 ( see the start of section [ s : mainresults ] ) .
suppose first that all indecomposable direct summands @xmath184 of @xmath0 are exchange compatible .
let @xmath63 be a cluster variable of @xmath288 .
it follows from proposition [ p : inductionstep ] by induction on the smallest number of exchanges needed to get from @xmath64 to a cluster containing @xmath63 that @xmath63 has a @xmath0-denominator .
suppose that there is an indecomposable direct summand of @xmath0 which is not exchange compatible .
consider the cluster @xmath291 of @xmath288 which is a minimal distance in the exchange graph from @xmath64 such that some summand of @xmath0 is not compatible with an exchange pair @xmath292 of modules coming from an exchange pair @xmath293 of cluster variables with @xmath294 .
arguing as above , we see that the cluster variables in @xmath291 all have a @xmath0-denominator . by proposition [ p : inductionstep ]
we see that @xmath295 does not have a @xmath0-denominator .
the fact that @xmath6 is a bijection gives us more information , since it follows that @xmath68 if and only if @xmath296 .
for the rest of this article , we freely use this fact .
we therefore have : [ c : correctform ] all indecomposable direct summands @xmath184 of @xmath0 are exchange compatible if and only if for every cluster variable @xmath63 of @xmath288 we have : + ( a ) the object @xmath297 is not isomorphic to @xmath66 for any @xmath15 , and @xmath298 for some polynomial @xmath26 in the @xmath70 satisfying the positivity condition , or + ( b ) we have that @xmath299 for some @xmath15 and @xmath296 .
recall that the exponent of @xmath70 in @xmath60 is @xmath300 .
it follows from ( * ? ? ?
* theorem a ) that @xmath301 is the multiplicity of the simple top of the projective @xmath302-module @xmath303 as a composition factor in @xmath304 : we have that @xmath305 using the fact that @xmath306 ( as @xmath0 is a cluster - tilting object in @xmath14 ) .
thus @xmath307 is the dimension vector of the @xmath308-module @xmath304 .
in this section we adopt the notation from the previous section , but assume in addition that @xmath86 is a preprojective module , a preinjective module , or the shift of a projective module .
our aim is to show that such modules are exchange compatible .
[ p : oneexact ] suppose that neither @xmath22 nor @xmath82 is isomorphic to @xmath101
. then at least one of the following holds ( where the maps are induced from triangle or triangle respectively ) : + ( a ) the sequence @xmath309 is exact , or + ( b ) the sequence @xmath310 is exact . assume first that @xmath86 is projective .
* case ( i ) : * + suppose that @xmath22 and @xmath82 are both modules . by @xcite , we know that either triangle or triangle arises from a short exact sequence of modules .
suppose first that triangle arises from a short exact sequence @xmath311 of modules .
applying the exact functor @xmath312 to this , we obtain the exact sequence : @xmath313 noting that @xmath314 as @xmath86 is projective . since @xmath315 for all modules @xmath124 ( cf .
@xcite ) ) , we obtain ( a ) above . similarly ,
if triangle arises from a short exact sequence of modules , we obtain ( b ) above .
* case ( ii ) : * + suppose that @xmath316 $ ] is the shift of a projective indecomposable module @xmath317 .
since @xmath318,p'[1])=0 $ ] for any projective module @xmath319 , we must have that @xmath82 is a module .
let @xmath320 be the indecomposable injective module corresponding to @xmath317 . since @xmath128 is the path algebra of an acyclic quiver , either @xmath321 or @xmath322 . if @xmath323 , then , applying @xmath146 to triangle , we obtain the exact sequence @xmath324 noting that @xmath325)=0 $ ] as @xmath86 is projective .
we see that ( a ) follows . if @xmath326 , we note that @xmath327=p[2]\simeq \tau p[1]\simeq i$ ] , so applying @xmath146 to triangle , we obtain the exact sequence @xmath328 noting that @xmath325)=0 $ ] , and ( b ) follows .
the argument in case @xmath82 is not a module is similar .
now assume that @xmath86 is a preprojective or preinjective module , or @xmath86 is a shift of a projective .
then @xmath329 is projective for some @xmath330 . since @xmath331 , then @xmath332 , and similarly for @xmath82 . by the above , either ( a ) or ( b ) holds for @xmath333 , so one of @xmath334 or @xmath335 is exact .
hence one of @xmath336 or @xmath337 is exact ( since @xmath11 is an autoequivalence of @xmath14 ) and we are done .
[ c : preprojectivec ] each indecomposable preprojective module , preinjective module , or shift of an indecomposable projective module is exchange compatible .
this follows from proposition [ p : oneexact ] . 0.2 in * proof of theorem [ main1 ] :
* we observe that theorem [ main1](a ) follows from corollary [ c : preprojectivec ] and theorem [ main1](b ) follows from corollary [ c : endpropc ] .
@xmath338 0.2 in * proof of corollary [ main3 ] : * if @xmath3 is dynkin , then @xmath4 has no regular modules , and if @xmath3 has only two vertices , there are no exceptional regular modules .
it follows from theorem [ main1](a ) that in these cases , every cluster variable has a @xmath0-denominator for every cluster - tilting object @xmath0 .
conversely , suppose that every cluster variable has a @xmath0-denominator for every cluster - tilting object @xmath0 .
we note that any exceptional indecomposable @xmath4-module can be completed to a tilting module and therefore to a cluster - tilting object by @xcite .
if @xmath3 was not dynkin and had more than two vertices , @xmath4 would have an indecomposable regular exceptional module @xmath86 of maximal quasilength amongst the indecomposable exceptional modules in its ar - quiver component ( see @xcite for the wild case ) . by proposition
[ p : endnotk ] , we would have @xmath339 , a contradiction to theorem [ main1](b ) .
in this section we shall interpret exchange compatibility further .
the results here hold for any finite dimensional path algebra .
they will be used in the tame case in the next section .
to ease notation we let @xmath340 and @xmath341 ( note that these coincide with the corresponding @xmath342 and @xmath343-spaces over @xmath128 if both objects are @xmath128-modules ) .
we also let @xmath344 and @xmath345 .
[ lem3.3 ] let @xmath346 be an exact exchange sequence of @xmath128-modules , and let @xmath86 be an exceptional indecomposable object in @xmath14 .
* @xmath347 gives a right exact sequence if and only if any @xmath128-map @xmath348 factors through @xmath349 .
* @xmath347 gives a left exact sequence if and only if any @xmath128-map @xmath350 factors through @xmath349 .
applying @xmath347 gives the long exact sequence @xmath351 & ( n , m^{\ast } ) \ar[r ] & ( n , b ) \ar[r ] & ( n , m ) \ar[r ] & \\ & & & & \amalg & \amalg & \amalg & \\ 0 \ar[r ] & ( \tau n , m^{\ast } ) \ar[r ] & ( \tau n , b ) \ar[r ] & ( \tau n , m ) \ar[r ] & { } ^1(\tau n , m^{\ast } ) \ar[r ] & { } ^1(\tau n , b ) \ar[r ] & { } ^1(\tau n , m ) \ar[r ] & 0 } \endxy\ ] ] the claim follows directly from this .
[ assumepreproj ] given a finite number of indecomposable objects in a cluster category , where each of them is preprojective , preinjective or the the shift of a projective , then by changing the hereditary algebra @xmath128 , if necessary ( i.e. via a tilt ) , we can assume that in fact all objects are preprojective ( or preinjective ) .
we make use of this in the sequel .
[ lem3.4 ] let @xmath352 be an exchange triangle , where @xmath22 is regular and @xmath82 is preprojective .
let @xmath86 be an indecomposable regular exceptional module .
* @xmath347 gives a left exact sequence if and only if for any nonzero @xmath130-map @xmath353 the composition @xmath354 is nonzero .
* @xmath347 gives a right exact sequence if and only if for any nonzero @xmath130-map @xmath355 the composition @xmath356 is nonzero .
\(a ) one direction is obvious .
assume that for any nonzero @xmath130-map @xmath353 the composition @xmath354 is nonzero , then we need to see that the map @xmath357 is injective .
any non - zero map @xmath358 in @xmath14 must have the form @xmath359 , where @xmath360 is an @xmath128-map and @xmath361 is an @xmath130-map , by @xcite .
suppose , for a contradiction , that its composition with the map from @xmath22 to @xmath140 is zero .
then there is a map @xmath362 such that @xmath363 , where @xmath364 . since @xmath365 is preprojective and @xmath22 is regular , we have that @xmath366)\simeq ( m^{\ast},m[1])\simeq { } ^1(m^{\ast},m)\simeq d(m,\tau m^{\ast})=0 $ ] since @xmath367 is preprojective .
it follows that @xmath156 is an @xmath128-map .
similarly , @xmath368 since @xmath86 is regular and @xmath365 is preprojective , so @xmath369 is an @xmath130-map .
comparing @xmath128-maps and @xmath130-maps , we see that @xmath370 and @xmath371 .
hence the @xmath130-map @xmath361 is non - zero while its composition with the map from @xmath22 to @xmath140 is zero , contradicting the assumption in ( a ) .
@xmath372 \ar@{--}_{s}[dl ] & & \\
\tau^{-1}m^{\ast } \ar^{t}[r ] & m \ar[r ] & b'\ \ar[r ] & m^{\ast } } \endxy\ ] ] \(b ) there is a long exact sequence @xmath373 hence @xmath374 is an epimorphism if and only if @xmath375 is a monomorphism .
this clearly holds if and only if @xmath376 is a monomorphism . by ( a ) , this holds if and only if for any non - zero @xmath130-map @xmath355 , the composition @xmath377 is nonzero .
we have the following direct consequence of lemmas [ lem3.3 ] and [ lem3.4 ] . [ prop3.7 ]
let @xmath282 be an exchange pair with @xmath22 regular and @xmath82 preprojective .
let @xmath378 and @xmath352 be the corresponding exchange sequence and triangle , where @xmath379 and @xmath380 with @xmath381 preprojective and @xmath382 regular .
let @xmath86 be an indecomposable regular exceptional module .
* applying @xmath347 to @xmath383 gives a short exact sequence if and only if any @xmath128-map @xmath384 factors through @xmath385 .
* applying @xmath347 to @xmath386 gives a short exact sequence if and only if for any nonzero @xmath130-map @xmath387 the composition @xmath388 is nonzero .
\(a ) if @xmath384 is an @xmath128-map , it can not factor through @xmath389 .
we apply lemma [ lem3.3 ] .
\(b ) since @xmath390 is preprojective , the map @xmath391 is an @xmath130-map .
hence if @xmath387 is an @xmath130-map , then the composition @xmath392 , must be zero .
we apply lemma [ lem3.4 ] .
in this section , we investigate the case in which @xmath128 is tame in more detail . in particular ,
we determine which exceptional regular modules are exchange compatible in this case .
[ lem3.6 ] let @xmath128 be a tame hereditary algebra and let @xmath282 be an exchange pair with @xmath22 regular and @xmath82 preprojective .
let @xmath378 and @xmath352 be the corresponding exchange sequence and triangle , where @xmath379 and @xmath380 with @xmath381 preprojective and @xmath382 regular
. then the maps @xmath385 and @xmath393 are @xmath128-maps .
the first claim holds obviously , since the exchange sequence is in @xmath394 . for the second claim , by remark [ assumepreproj ] , by replacing @xmath128 with @xmath395 via a tilt
, we can assume that @xmath82 is preinjective , and similarly that @xmath390 is preinjective .
then the triangle @xmath352 is the image of an exact sequence of @xmath395-modules @xmath396 .
then @xmath397 is an @xmath395-map and consequently also an @xmath128-map .
[ ansubcategory ] let @xmath398 denote a category which is either the module category of the path algebra @xmath399 of a quiver of type @xmath400 with linear orientation , or the abelian category associated with a tube , i.e. a connected component of regular modules over some tame hereditary algebra @xmath401 . for an indecomposable exceptional object @xmath22 in @xmath398 ,
we let @xmath402 denote the additive subcategory generated by the subfactors of @xmath22 inside @xmath398 , or equivalently the extension closure of @xmath403 .
the indecomposable objects in @xmath402 form a full subquiver of the ar - quiver of @xmath401 , which is a triangle with @xmath22 sitting on the top ( see figure [ f : matthetop ] ) .
@xmath404 & & & & \\ & & & & \circ \ar[ur ] \ar[dr ] \ar@{.}[rr ] & & \circ \ar[dr ] & & & \\ & & & \circ \ar@{--}[ddll ] \ar[ur ] \ar@{.}[rr ] & & \circ \ar[ur ] \ar@{.}[rr ] & & \circ \ar@{--}[ddrr ] & & \\ & & & & & & & & & \\ & \circ \ar[dr ] & & & \ar@{--}[rr ] & & & & & \circ \ar[dr ] & \\
\circ \ar[ur ] \ar@{.}[rr ] & & \circ & & & & & & \circ \ar[ur ] \ar@{.}[rr ] & & \circ } \endxy\ ] ] [ lem3.8 ] consider the quiver of type @xmath405 with linear orientation and let @xmath401 be its path algebra .
let @xmath139 be the indecomposable projective and injective @xmath401-module .
assume @xmath406 is a sequence of irreducible maps , where @xmath407 .
if @xmath0 is a tilting module over @xmath401 , it must contain one of the @xmath408 ( for @xmath409 ) or @xmath410 ( for @xmath411 ) as an indecomposable direct summand .
assume to the contrary that @xmath0 is a tilting module having none of the @xmath412 as indecomposable direct summands .
the object @xmath139 is necessarily a summand of @xmath0 , since it is projective - injective .
since @xmath413 only has extensions with the the proper factors of @xmath139 , at least one such factor must be a summand of @xmath0 .
let @xmath414 , with length @xmath262 maximal .
then , by assumption @xmath415 .
let @xmath16 be the indecomposable projective of length @xmath15 , and let @xmath416 .
then @xmath417 is in @xmath418 , since , for @xmath419 , @xmath420 and all factors of @xmath420 which are not in @xmath421 have non - trivial extensions with @xmath422 .
note that @xmath417 has @xmath157 nonisomorphic indecomposable summands .
the subcategory @xmath421 has clearly at most @xmath262 summands in @xmath0 , so at least @xmath423 summands @xmath424of @xmath0 are in @xmath425 .
note that @xmath425 is clearly equivalent to @xmath426 , so @xmath427 forms a tilting module in this subcategory , and hence @xmath428 is a summand in @xmath191 .
but this is a contradiction , since @xmath428 is equal to some @xmath408 .
we illustrate the situation in figure [ f : typea ] . .
] [ ancover ] let @xmath22 be an indecomposable regular module in a tube @xmath429 of rank @xmath2 for a tame hereditary algebra @xmath128 .
assume @xmath430 ( where @xmath431 denotes quasilength ) .
let @xmath432 be the path algebra of a quiver of type @xmath405 with linear orientation .
then there is a functor @xmath433 , which is clearly fully faithful ( from the structure of the tube ) , exact and dense and preserves indecomposability .
the following is due to strauss ( * ? ? ?
[ lem3.9 ] let @xmath22 be an indecomposable exceptional @xmath128-module for @xmath128 tame hereditary , lying in a tube @xmath429 of rank @xmath2 .
let @xmath434 ( note that necessarily @xmath435 ) .
let @xmath436 be a tilting @xmath128-module .
then @xmath437 , where @xmath438 , for a tilting module @xmath439 in @xmath440 .
@xmath441 & & \circ \ar[dr ] & & \circ \ar[dr ] & & \circ \\ & & & \circ \ar[ur ] \ar[dr ] & & \circ \ar[ur ] \ar[dr ] & & \circ \ar[ur ] \\ & & & & \circ \ar[ur ] \ar[dr ] & & \circ \ar[ur ] \\ & & & & & m \ar[ur ] \ar[dr ] \\ & & & & \circ \ar[dr ] \ar[ur ] & & \circ \ar[dr ] \\ & & & \circ \ar[dr ] \ar[ur ] & & \circ \ar[dr ] \ar[ur ] & & \circ \ar[dr ] \\ & & \ar@{--}[ll ] \ar@{--}[rr ] \circ \ar[ur ] & & \circ \ar@{--}[rr ] \ar[ur ] & & \circ \ar@{--}[rr ] \ar[ur ] & & \circ \ar@{--}[rr ] & & } \endxy\ ] ] [ lem3.10 ] let @xmath429 be a tube of rank @xmath442 .
let @xmath443 be indecomposable objects in @xmath429 with @xmath444 , where @xmath445 , and @xmath446 , where @xmath447 .
suppose that @xmath448 is a sequence of irreducible maps with @xmath449 and @xmath450 and that @xmath451 is a sequence of irreducible maps with @xmath452 and @xmath453 .
\(a ) let @xmath454 and @xmath455 be non - zero maps .
then the number of modules @xmath456 , @xmath457 , that @xmath26 factors through plus the number of modules @xmath458 , @xmath457 , that @xmath349 factors through is at least @xmath156 .
+ ( b ) statement ( a ) holds if we assume instead that @xmath459 or @xmath460 or both .
\(a ) suppose first that @xmath461 .
let @xmath462 be a sequence of irreducible maps with @xmath463 and @xmath464 for all @xmath15 .
then @xmath465 for some @xmath114 , where @xmath466 , so @xmath467 . from the structure of the tube
it follows that @xmath26 factors through @xmath468 , a total of @xmath114 modules .
let @xmath469 be a sequence of irreducible maps with @xmath470 and @xmath471 for all @xmath15 .
then @xmath472 for some @xmath113 , where @xmath473 , so @xmath474 . from the structure of the tube
it follows that @xmath349 factors through @xmath475 , a total of @xmath113 modules .
since @xmath476 and @xmath444 , we have that @xmath477 , so @xmath478 since we have assumed @xmath479 .
see figure [ f : tubefactoring ] .
the argument in case @xmath480 is similar , so we are done .
for ( b ) we just need to remark that a non - zero map @xmath481 factors through more of the @xmath456 than @xmath26 does , and that a non - zero map @xmath482 factors through more of the @xmath458 than @xmath349 does .
therefore ( a ) holds if @xmath86 is replaced by @xmath101 , or @xmath181 is replaced by @xmath101 , or both . statement ( b ) follows .
@xmath483 & & & & & & & & & & \tau^2 n \ar@{-}[ddll ] \\ & & & & & & & & & & \\ & & u_t \ar@{-}[dr ] \ar@{.}[rr ] & & m \ar@{-}[dl ] \ar@{-}[ddrr ] \ar@{.}[rrrr ] & & & & v_t \ar@{-}[ddll ] \\ & & & u_{t - a } \ar@{-}[ddll ] \\ & & & & & & v_{t - b } \ar@{-}[dr ] \\
\ar@{--}[rrrrrrrrrr ] & & & & & & & & & & } \endxy\ ] ] [ lem3.11 ] let @xmath128 be tame and @xmath86 indecomposable exceptional in a tube of rank @xmath484 , so that @xmath485 .
then @xmath486 if and only if @xmath487 . in this case , there is a non - zero @xmath130-map from @xmath86 to @xmath86 and @xmath488 .
it is easy to see that @xmath162 ( and is one - dimensional ) if and only if @xmath486 , using @xmath489 .
furthermore , @xmath490 .
we have that @xmath491 , and @xmath492 . [ helpthm3.13 ]
let @xmath128 be tame and @xmath86 indecomposable exceptional in a tube of rank @xmath493 , with @xmath494 .
let @xmath87 be an exchange pair , where @xmath22 and @xmath82 are complements of some almost complete cluster - tilting object @xmath76 .
we assume that @xmath22 is regular .
then we have an exchange sequence @xmath378 and exchange triangle @xmath495 .
write @xmath496 and @xmath497 , where @xmath498 are regular and @xmath499 are preprojective .
then either : * any @xmath128-map @xmath384 factors through @xmath500 , or * for any nonzero @xmath130-map @xmath501 , the composition @xmath502 is nonzero .
we can clearly assume that there is some nonzero @xmath128-map @xmath384 and some nonzero @xmath130-map @xmath501 , and hence there is some nonzero @xmath128-map @xmath503 .
in particular , @xmath86 and @xmath22 are in the same tube . by lemma [ lem3.9 ]
, @xmath76 is the direct sum of the image under @xmath504 of a tilting module @xmath439 in @xmath440 and another summand ( using the notation of remark [ ancover ] ) .
let @xmath505 . * case ( i ) : * + we first assume that there is a non - zero @xmath128-map from @xmath86 to @xmath22 and a non - zero @xmath128-map from @xmath22 to @xmath181 .
by lemma [ lem3.8 ] , combined with lemma [ lem3.10](a ) , we see that either some non - zero @xmath128-map from @xmath86 to @xmath22 factors through some indecomposable direct summand of @xmath76 or that some non - zero @xmath128-map @xmath506 factors through some indecomposable direct summand of @xmath76 . in the first case , we note that , since @xmath507 , any non - zero @xmath128-map from @xmath86 to @xmath22 factors through @xmath76 . since any non - zero @xmath128-map from @xmath101
to @xmath22 must factor through an @xmath128-map from @xmath86 to @xmath22 , it follows that any non - zero @xmath128-map from @xmath508 to @xmath22 must factor through @xmath76 , and therefore through @xmath500 , and we see that ( i ) holds . in the second case , arguing similarly , we see that any non - zero @xmath128-map from @xmath22 to @xmath509 factors through the @xmath128-map component of @xmath510 ( noting that an @xmath128-map between two modules can not factor through an @xmath130-map ) .
it follows that the map from @xmath511 to @xmath512 given by composition with the @xmath128-map component of @xmath510 is surjective , so that the induced map from @xmath513 to @xmath514 is injective .
hence the induced map from @xmath515 to @xmath516 is injective .
so the map from @xmath517 to @xmath518 is injective , and therefore the map from @xmath519)$ ] to @xmath520)$ ] is injective .
hence the map from @xmath521 to @xmath522 is injective , and we see that the composition of any non - zero @xmath130-map from @xmath523 to @xmath22 with the @xmath128-map component of the map @xmath510 is non - zero . note that composition of any such @xmath130-map with the @xmath130-map component is always zero .
it follows that the composition of any non - zero @xmath130-map from @xmath523 to @xmath22 with the map @xmath510 is non - zero .
we see that ( ii ) holds and we are done .
* case ( ii ) : * + we now assume that @xmath524 , that there is a non - zero @xmath128-map from @xmath101 to @xmath22 and a non - zero @xmath128-map from @xmath22 to @xmath181 . by lemma [ lem3.8 ] , combined with lemma [ lem3.10](b ) , we see that either some non - zero @xmath128-map from @xmath101 to @xmath22 factors through some summand of @xmath76 or that some non - zero @xmath128-map @xmath506 factors through some summand of @xmath76 .
in the first case , we note that , since @xmath525 , any non - zero @xmath128-map from @xmath101 to @xmath22 factors through @xmath76 . since @xmath524 , we see that any non - zero @xmath128-map from @xmath508 to @xmath22 must factor through @xmath76 , and therefore through @xmath500 , and we see that ( i ) holds .
the argument in the second case is as in case ( i ) .
* case ( iii ) : * + we now assume that there is a non - zero @xmath128-map from @xmath86 to @xmath22 , that @xmath526 and that there is a non - zero @xmath128-map from @xmath22 to @xmath101 . by lemma [ lem3.8 ] , combined with lemma [ lem3.10](b ) ,
we see that either some non - zero @xmath128-map from @xmath86 to @xmath22 factors through some summand of @xmath76 or that some non - zero @xmath128-map @xmath482 factors through some summand of @xmath76 .
the argument in the first case is as in case ( i ) . in the second case , since @xmath526
, we see that any non - zero @xmath128-map from @xmath22 to @xmath527 factors through the @xmath128-map component of @xmath510 ( noting that an @xmath128-map between two modules can not factor through an @xmath130-map ) . as in case ( i )
, we see that ( ii ) holds .
* case ( iv ) : * + finally , we assume that @xmath524 , @xmath526 , that there is a non - zero @xmath128-map from @xmath101 to @xmath22 and that there is a non - zero @xmath128-map from @xmath22 to @xmath101 . by lemma [ lem3.8 ] , combined with lemma [ lem3.10](b ) , we see that either some non - zero @xmath128-map from @xmath101 to @xmath22 factors through some summand of @xmath76 or that some non - zero @xmath128-map @xmath482 factors through some summand of @xmath76 . in the first case we argue as in case ( ii ) , and in the second case we argue as in case ( iii ) , and we are done .
[ thm3.13 ] let @xmath128 be tame and @xmath86 indecomposable exceptional in a tube of rank @xmath493 , with @xmath494 .
then @xmath86 is exchange compatible .
let @xmath87 be an exchange pair , with @xmath22 and @xmath82 complements of some almost complete cluster - tilting object @xmath76 .
* case ( i ) : * + we assume first that @xmath22 is regular and @xmath82 preprojective
. then we have an exchange sequence @xmath378 and exchange triangle @xmath495 .
write @xmath496 and @xmath497 , where @xmath498 are regular and @xmath499 are preprojective . by lemma [ helpthm3.13 ] , either * any @xmath128-map @xmath384 factors through @xmath500 , or * for any nonzero @xmath130-map @xmath501 , the composition @xmath502 is nonzero .
if ( i ) holds , we see by lemma [ lem3.3 ] that the sequence is exact .
if ( ii ) holds , we see by lemma [ lem3.4 ] that the sequence is exact . * case ( ii ) : * + next , we assume that @xmath22 and @xmath82 are both preprojective
. we can still assume that we have an exchange sequence @xmath528 of @xmath128-modules , so that @xmath139 is also preprojective .
applying @xmath347 , arguing as in lemma [ lem3.3 ] and using that @xmath529 for any preprojective @xmath124 we see that the long exact sequence @xmath530 reduces to @xmath531 * case ( iii ) : * + finally , we consider the case where both @xmath22 and @xmath82 are regular modules . by lemma [ lem3.3 ]
( and arguing as in lemma [ lem3.4 ] ) we see that it is enough to show that either + ( a ) any @xmath128-map @xmath532 factors through @xmath349 in the sequence @xmath533 or + ( b ) for any non - zero map @xmath534 , the composition with @xmath360 is non - zero , where @xmath360 is the map in the triangle : @xmath535 note that statement ( a ) is the same as statement [ helpthm3.13](i ) .
in order to do this , it is clear that we can assume that there is a non - zero @xmath128-map from @xmath536 to @xmath22 and that there is a non - zero map ( in @xmath14 ) from @xmath523 to @xmath22 .
from the former , it follows that @xmath86 and @xmath22 must lie in the same tube .
we first of all show the following claim : * claim : * if there is a non - zero h - map from @xmath537 to @xmath22 whose composition with @xmath360 is zero then any non - zero @xmath128-map from @xmath508 to @xmath22 factors through @xmath538 * proof of claim : * if @xmath539 is a non - zero @xmath128-map and the composition @xmath540 is zero , then @xmath541 lifts to @xmath365 .
it follows that @xmath365 lies in the rectangle spanned by @xmath86 and @xmath22 in the tube ( see figure [ f : regreg ] ) .
hence @xmath82 lies in the rectangle spanned by @xmath542 and @xmath101 in the tube ( the dotted rectangle in figure [ f : regreg ] ) . the middle term in the short exact sequence is the direct sum of two indecomposable modules , @xmath543 and @xmath544 , which are at the other two corners of the rectangle spanned by @xmath22 and @xmath82 .
we see from the structure of the tube that @xmath541 factors through at least one of @xmath543 and @xmath544 ( using the fact that @xmath545 ) .
it follows that @xmath541 factors through the map @xmath546 . since @xmath507
, we see that every @xmath128-map from @xmath86 to @xmath22 factors through @xmath546 .
since a non - zero @xmath128-map from @xmath101 to @xmath22 factors through any @xmath128-map from @xmath86 to @xmath22 ( note that such maps , if they exist , are unique up to a scalar multiple ) , it follows that any non - zero @xmath128-map from @xmath101 to @xmath22 factors through @xmath547 also .
thus any non - zero @xmath128-map from @xmath508 to @xmath22 factors through @xmath546 .
if @xmath548 is a non - zero @xmath128-map and its composition with @xmath360 is zero , then , as above , @xmath541 must factor through @xmath547 .
then any non - zero @xmath128-map from @xmath86 to @xmath22 must factor through @xmath541 and therefore , by the above , through @xmath547 . again arguing as above , we see that any non - zero @xmath128-map from @xmath101 to @xmath22 factors through @xmath547 and therefore that any non - zero @xmath128-map from @xmath508 to @xmath22 factors through @xmath547 .
the claim now follows .
since @xmath22 is regular , lemma [ helpthm3.13 ] applies , and we see that either [ helpthm3.13](i ) or ( ii ) holds .
if ( i ) holds , we are done , as ( a ) above is the same as [ helpthm3.13](i ) .
if [ helpthm3.13](ii ) holds , it follows that for any non - zero @xmath130-map @xmath549 , the composition @xmath550 is nonzero . by the claim above , we know that if there is a non - zero @xmath128-map @xmath549 whose composition with @xmath360 is zero then ( a ) holds , and we are done .
so we are left with the case in which for any non - zero @xmath128-map or @xmath130-map from @xmath537 to @xmath22 the composition with @xmath360 is non - zero .
we claim that this means that for any non - zero map from @xmath537 to @xmath22 , the composition @xmath551 is non - zero .
we write maps in @xmath14 between indecomposable objects as pairs @xmath552 where @xmath26 is an @xmath128-map and @xmath179 is an @xmath130-map .
let @xmath553 be a non - zero map such that the composition with @xmath510 is zero .
then there are maps @xmath554 and @xmath555 such that @xmath556 .
then @xmath557 , so that @xmath558 composed with @xmath510 is zero ; hence @xmath370 ( as @xmath360 is a @xmath128-map ) .
then @xmath361 composed with @xmath510 is zero , so @xmath559 ( as @xmath361 is an @xmath130-map ) .
it follows that @xmath560 , a contradiction .
see figure [ f : hfargument ] .
we see that the map from @xmath561 to @xmath562 is injective , and thus that ( b ) above holds .
@xmath563 \ar@{--}[dddddddddrrrrrrrrr ] & & \circ \ar@{-}[ddddllll ] \ar@{-}[ddddrrrr ] & & & & & & & & \\ & & & & & & & & & & & & & & \\ & & & & & & & & & & & & & & \\ & & & & & & & & & & & & & & \\
\tau n \ar@{--}[dddddddddrrrrrrrrr ] & & n \ar@{-}[dddddddddrrrrrrrrr ] & & & & & & & & b_1 \ar@{-}[dddll ] \ar@{-}[dddddrrrrr ] & & & & \\ & & & & & & & & & & & & & & & \\ & & & & & & & & & & & & & & \\ & & & & & & & & m^{\ast } \ar@{-}[dddddrrrr ] & \tau^{-1}m^{\ast } & & & & & \\ & & & & & & & & & & & & & & \\ & & & & & & & & & & & & & \circ \ar@{--}[ddddllll ] & & m \ar@{-}[dddlll ] \\ & & & & & & & & & & & & & & \\ & & & & & & & & & & & & & & & \\ & & & & & & & & & & & & b_2 \ar@{-}[dl ] & & & \\ & & & & & & & & & \circ & & \circ & & & \\ & & & & & & & & & & & & & & } \endxy\ ] ] @xmath564_{(t , t ' ) } \ar[d]^{(h , h ' ) } \\ \tau^{-1}m^{\ast } \ar[r]_{(s , s ' ) } & m \ar[r ] & b ' } \endxy\ ] ] we can now conclude that if @xmath88 , then @xmath565 thus @xmath86 is exchange compatible , and we are done .
let @xmath128 be a tame hereditary algebra , and let @xmath86 be an indecomposable exceptional object in @xmath14 .
then @xmath86 is exchange compatible if and only if @xmath566 .
this follows immediately from theorem [ thm3.13 ] and corollary [ thm3.12 ] together with proposition [ p : endnotk ] .
* proof of theorem [ main2 ] : * by theorem [ t : all ] , theorem [ main2](a ) is equivalent to the statement : + ( a ) all indecomposable direct summands @xmath73 of @xmath0 are exchange compatible .
+ by lemma [ lem3.11 ] , [ main2](c ) implies [ main2](b ) ; by theorem [ thm3.13 ] , [ main2](b ) implies ( a ) , and by corollary [ c : endpropc ] , ( a ) implies [ main2](c ) .
the proof is complete .
in this final section we give two examples to illustrate our main result .
if we choose a cluster - tilting object @xmath0 where the quiver of @xmath42 is @xmath41 and compute all cluster variables with respect to the initial cluster @xmath568 , we get the picture shown in figure [ f : a3example ] , where in the ar - quiver we have represented the indecomposable objects by their corresponding cluster variables .
@xmath570 & & \frac{y_1 + y_2 + y_3}{y_1 y_2 } \ar[dr ] & & \\ & \frac{y_1 + y_2}{y_3 } \ar[ur ] \ar[dr ] & & \frac{y_2 + y_3}{y_1 } \ar[ur ] \ar[dr ] & & \frac{y_1 + y_3}{y_2 } \ar[dr ] & \\ y_1 \ar[ur ] & & \frac{y_1 + y_2 + y_3}{y_1 y_3 } \ar[ur ] & & y_3 \ar[ur ] & & \frac{y_1 + y_2 + y_3}{y_2 y_3 } } \endxy\ ] ] [ e : a2tilde ] in the next example we consider the quiver @xmath3 shown in figure [ f : a2tilde](a ) .
this is mutation equivalent to the quiver @xmath51 shown in figure [ f : a2tilde](b ) .
then the ar - quiver of @xmath4 has a tube of rank two with quasisimple modules given by the simple module @xmath571 and @xmath572 , a module with composition factors @xmath573 and @xmath574 .
we choose a cluster - tilting object @xmath0 such that the quiver of the cluster - tilted algebra @xmath575 is @xmath51 , for example @xmath576 . then @xmath577 .
let @xmath70 be the cluster variable corresponding to @xmath66 for @xmath580 .
then there is a seed @xmath581 of @xmath288 .
the cluster variables corresponding to some of the objects in the preprojective / preinjective component of the ar - quiver of @xmath582 are shown in figure [ f : a2tildevariables ] .
@xmath583 \ar[ddr ] & & \frac{y_1 + y_3}{y_2 } \ar[rr ] \ar[ddr ] & & \frac{y_1 ^ 2 + y_2 + y_1 y_3}{y_2 y_3 } \ar[rr ] \ar[ddr ] & & \frac{f}{y_1 y_2 y_3 ^ 2}\ar[ddr ] & & \\ \cdots & & & & & & & & & \cdots \\ & & y_3 \ar[rr ] \ar[uur ] & & y_1 \ar[rr ] \ar[uur ] & & \frac{y_1 ^ 2 + y_2}{y_3 } \ar[rr ] \ar[uur ] & & \cdot & } \endxy\ ] ] the cluster variables corresponding to the two quasisimple modules in the tube of rank two are @xmath584 , corresponding to @xmath585 , and @xmath586 corresponding to @xmath22 .
however @xmath587 is easily seen to be two dimensional , while @xmath584 appears with multiplicity one in the denominator of the cluster variable corresponding to @xmath22 . c. m. ringel , _ the regular components of the auslander - reiten quiver of a tilted algebra _ , a chinese summary appears in chinese ann .
ser . a 9 ( 1988 ) , no . 1 , 102 .
chinese ann .
b 9 ( 1988 ) , no . | associated to any acyclic cluster algebra is a corresponding triangulated category known as the cluster category .
it is known that there is a one - to - one correspondence between cluster variables in the cluster algebra and exceptional indecomposable objects in the cluster category inducing a correspondence between clusters and cluster - tilting objects .
fix a cluster - tilting object @xmath0 and a corresponding initial cluster . by the laurent phenomenon
, every cluster variable can be written as a laurent polynomial in the initial cluster .
we give conditions on @xmath0 equivalent to the fact that the denominator in the reduced form for every cluster variable in the cluster algebra has exponents given by the dimension vector of the corresponding module over the endomorphism algebra of @xmath0 .
[ section ] [ lem]proposition [ lem]corollary [ lem]theorem [ lem]remark [ lem]definition [ lem]example |
recently there has been much attention for a certain multi - variable version of linear system theory which presents evolution equations of the form @xmath0 this is called a noncommutative fornasini - marchesini system in @xcite . here
@xmath1 where @xmath2 is the free semigroup with @xmath3 generators ( which we denote @xmath4 and @xmath5 or @xmath6 )
. the elements in @xmath2 are words in the letters @xmath4 , including the empty word @xmath7 .
composition is defined by concatenation of words , for example @xmath8 in the formula above is the concatenation of a word @xmath9 and a generator @xmath10 . now @xmath11 are functions on @xmath2 taking values in vector spaces @xmath12 and @xmath13 are linear operators . for @xmath14
we have @xmath15 and this is the classical setting of linear system theory ( state space models ) .
the functions @xmath11 are interpreted as input , internal state , output of the system .
( 15,3.5 ) ( 1.0,1)(0,1)2 ( 1.0,1)(1,0)2.9 ( 3.9,1)(0,1)2 ( 1.0,3)(1,0)2.9 ( 5.4,1)(0,1)2 ( 5.4,1)(1,0)3 ( 8.4,1)(0,1)2 ( 5.4,3)(1,0)3 ( 9.9,1)(0,1)2 ( 9.9,1)(1,0)2.8 ( 12.7,1)(0,1)2 ( 9.9,3)(1,0)2.8 ( 5.4,1.7)(-1,0)1.5 ( 7.4,1.7)(-1,0)1.0 ( 9.9,1.7)(-1,0)1.5 ( 1.3,2.0)output space @xmath16 ( 10.2,2.0)input space @xmath17 ( 5.7,2.0)internal space @xmath18 ( 4.5,1.2)@xmath19 ( 6.7,1.2)@xmath20 ( 9.0,1.2)@xmath21 ( 2.5,0.5)(1,0)8.8 ( 2.5,0.5)(0,1)0.5 ( 11.3,0.5)(0,1)0.5 ( 6.8,0.2)@xmath22 it has been established that many important mathematical concepts and results in linear system theory generalize nicely for all @xmath3 , see @xcite .
we mention two such concepts which will later be studied in this paper .
the first is the _ observability map _
@xmath23 which is a linear map from @xmath18 into the @xmath16-valued functions on @xmath2 .
( in this paper we use the convention that for any variables @xmath24 and a word @xmath25 we have @xmath26 , @xmath27 , if @xmath28 we interpret it as an identity . ) by studying the observability map we can find out what we are able to know about the internal space @xmath18 by observing the output .
the second concept we want to mention here is the _ transfer function _ which is a description of how inputs are transferred into outputs . in the multi - variable context above the transfer function
can be defined as a formal power series @xmath29 where @xmath30 with indeterminates @xmath31 freely noncommuting among each other but commuting with the linear maps , @xmath32 , and @xmath33 denoting a transpose of the row vectors so @xmath34 and @xmath35 are column vectors of linear maps . in the classical case
@xmath14 we call the single variable @xmath36 and then this reduces to a familiar formula which gives the transfer function as an analytic function . explicitly @xmath37
we come back to this at the end of section [ section : structure ] .
see also @xcite for further discussions of similar formulas .
mathematical system theory is an abstraction from the physical dynamics .
but of course its relevance to the real world depends on the fact that such a physical dynamics exists in the background . the basic idea behind our work is the proposal that the physical dynamics which implements a multi - variable system theory ( such as the one sketched above ) is described by processes describing quantum physics . in @xcite we started a project to investigate such connections , compare further @xcite for a recent generalization of the model in @xcite and also @xcite for some complementary work . in these papers
the quantum processes are based on infinite tensor products which are natural from the point of view of approximating the fock spaces of continuous time processes but which sometimes obscure the more fundamental level where the connections with operator and system theory originate . in the present paper
we go directly to this level by investigating the proposal that it is the concept of ` weak markov processes ' , worked out by bhat and parthasarathy in @xcite to catch the most basic features of quantum markov processes , which provides a systematic and very general point of departure for our project .
the connections to operator theory are evident here because the concept of a weak markov process may be seen as operator theoretic dilation theory studied from a probabilistic point of view .
dilation theory will be mentioned only in side remarks in this paper however , the emphasis lies on the structure of the full process .
the benefits of such studies go in both directions : access to operator and system theory tools for concrete quantum models and guidance for the development of the general theory from the questions arising in the models .
to achieve such an interaction between the disciplines a more systematic development of the theory is needed .
the following sketch of the contents of this paper should be read with this motivation in mind . in section [ section : weak ] we start with a self - contained but concise description of the basic theory of weak markov processes in discrete time : the dynamics is described by a row isometry and then from a co - invariant subspace certain structures emerge which can be given a quantum probabilistic interpretation , such as transition operators and weak filtrations .
most of this is well known but note how in the end we give an operational meaning to the elements of the free semigroup @xmath2 by interpreting them as certain measurement protocols . in section [ section : structure ]
we define the notion of a representation of structure maps @xmath38 ( as described above ) by a weak markov process and in this way we get an explicit systematic procedure to identify multi - variable linear systems ( as described above ) within quantum physical models . at this point
it remains quite abstract . as a preparation for seeing representations of structure maps in use
we first introduce in section [ section : cat ] the theory of subprocesses and quotient processes of ( discrete weak markov ) processes and then show that within a suitably defined category of processes this can be reformulated as a short exact sequence .
the main result here is a classification of extensions appearing in such short exact sequences by a construction which we call a @xmath39-cascade of processes .
there is a set of contractions from which @xmath39 can be chosen which gives a parametrization of all the ways in which two processes can be put together as subprocess and quotient process , with @xmath40 yielding the direct sum .
cascades are a well known concept for systems and their structure maps .
the terminology of a @xmath39-cascade for processes is justified in section [ section : cascades ] where we show that within a @xmath39-cascade of processes we have a representation of a cascade of the original systems and that in fact this construction is a natural tool to represent cascades of systems and of structure maps .
let us remark here that the notion of cascades and of more general networks of processes has been around for some time in a continuous time setting and this theory has been investigated intensely because of promising applications in quantum filtering and quantum control @xcite .
there is no explicit comparison with continuous time systems in this paper , however we provide the basis for such a comparison by describing weak markov processes in terms of product systems which suggests how to build the theory starting from continuous product systems .
our justification for concentrating on the discrete time setting here is the same as in helton s seminal paper @xcite in which he connects classical system theory ( @xmath14 ) with operator models and scattering theory , saying : ` we concentrate on discrete time systems because it is for these that the relationship is most clear ' ( @xcite , p.15 ) .
in fact , very similarly to the path followed by helton in @xcite it is quite natural in our setting to connect observability of represented multi - variable linear systems with scattering in quantum physical models .
we investigate observability in section [ section : observable ] and show that in a quantum model it can be given the same interpretation as in classical system theory , namely that by observing the outputs we have an indirect way to measure and to investigate those parts of the internal space which have the character of a black box .
we show that we get a particularly interesting situation if we have a @xmath39-cascade and if we use the input space of the subprocess as the output space of the cascade , so that looking for observability amounts to the question how much we can find out about the cascade by observations which only involve the subprocess .
the extreme case when we can find out everything about the cascade by such special observations we call asymptotically complete .
we give a convenient criterion for asymptotic completeness in terms of asymptotic properties of the transition operator of the process which is a completely positive map .
this criterion is a generalization of a criterion for asymptotic completeness @xcite in a scattering theory for operator algebraic stationary quantum markov processes introduced by kmmerer and maassen in @xcite .
it turns out that our definition of asymptotic completeness for short exact sequences of weak processes can be seen as a generalization of the definition of kmmerer and maassen .
such a connection has been hinted at in @xcite but the theory developed in this paper allows us to explain the relevant facts in a much more conceptual and elegant way .
the details are in section [ section : scattering ] .
we include the general fact that normal invariant states lead to weak subprocesses and we work out how for a stationary markov chain in an operator algebraic setting we can construct an associated weak process together with a subprocess ( with one - dimensional internal space ) and hence we can establish the connection to our previous results .
this improved conceptual understanding of asymptotic completeness is an excellent example of the cross - fertilization between operator and system theory on the one hand and quantum probability on the other hand which we have in mind .
it uses only a very special case of our general theory and suggests further work .
unitary dynamics on a hilbert space is the most basic way of describing quantum mechanical evolution .
if causality is taken into account and one restricts the attention to observables belonging only to the future ( or only to the past ) then it becomes natural to study @xmath41-endomorphisms of @xmath42 , the algebra of bounded linear operators on a hilbert space @xmath43 . a convincing argument in this direction
is presented in @xcite , section 1.2 . from now on let @xmath44 be a @xmath41-endomorphism and @xmath43 a separable hilbert space .
then it follows from the representation theory of @xmath42 that there exists a separable hilbert space @xmath45 and an isometry @xmath46 such that for all @xmath47 @xmath48 we assume @xmath49 , then @xmath50 is automatically injective .
see @xcite for more details .
if @xmath51 ( with @xmath52 or @xmath6 ) is an orthonormal basis of @xmath45 then with @xmath53 we can also write @xmath54 ( limits to be understood in the strong operator topology if @xmath6 ) which is called a _ kraus decomposition_. the @xmath55 are isometries with orthogonal ranges , so alternatively ( and with the same notation ) we also think of @xmath56 as a _ row isometry _ @xmath57 @xmath56 is called a _ row unitary _ if @xmath58 and this is equivalent to @xmath50 being unital , i.e. , @xmath59 .
if this additional assumption is used in the following then it will always be explicitly stated .
+ + using induction we define row isometries @xmath60 ( where @xmath61 ) by @xmath62 and , for @xmath63 and @xmath64 @xmath65 later we need the following equivalent description for the adjoints : if @xmath66 and @xmath67 then @xmath68 it is not difficult to check that this implements the @xmath69-th power of @xmath50 : @xmath70 here the notation is @xmath71 if @xmath72 and @xmath73 is the length of the word .
this kraus decomposition where a sum over all words of length @xmath69 occurs gives the first connection to the multi - variable formalism sketched in section [ section : intro ] .
+ remark : the tensor products @xmath74 appearing in these formulas represent the ( discrete ) product system associated to the endomorphism @xmath50 and though we do not go into continuous time systems in this paper it is worth noting that the natural starting point to translate our results to continuous time would be to consider continuous product systems and in this way make the connection with the theory exposed in @xcite . to introduce processes which resemble markov processes from probability theory we need to specify a subspace @xmath75 ( by which we always mean a closed subspace if not otherwise stated ) .
let us denote the orthogonal projection onto @xmath76 by @xmath77 .
( in this paper we use consistently the notation @xmath78 for the orthogonal projection onto a subspace @xmath79 . )
given @xmath75 we have ( with @xmath80 ) a family of normal @xmath41-homomorphisms @xmath81 and the compressions @xmath82 defined by @xmath83\ ] ] which are contractive completely positive maps . for the processes to be defined below the @xmath84 play the role of _ noncommutative random variables _ and the @xmath85 are _ transition operators_. the subspace @xmath76 of @xmath43 is called _ invariant _ if @xmath86 for all @xmath87 and _ co - invariant _ if @xmath88 for all @xmath87 .
the importance of co - invariant subspaces in this context has been observed by many , see for example @xcite for various related topics .
we note some useful properties equivalent to co - invariance .
[ lem : co ] the following are equivalent : * @xmath89 * @xmath90 * @xmath76 is co - invariant let us write @xmath91 for @xmath92 and state the following modification of the previous lemma : [ lem : co+ ] the following are equivalent : * @xmath93 * @xmath94 * @xmath95 and @xmath76 is co - invariant * @xmath96 below we prove lemma [ lem : co ] .
it is then easy to get lemma [ lem : co+ ] by checking that ( @xmath97 ) is nothing but ( @xmath98 ) together with @xmath95 ( for @xmath99 ) .
the equivalence of ( 1 ) and ( 4 ) is immediate because , by definition , @xmath100 .
if @xmath56 is a row unitary then all the properties in lemmas [ lem : co ] and [ lem : co+ ] are equivalent ; for example @xmath101 and @xmath102 are equivalent for a row unitary because in this case we have @xmath59 .
this means that for a row unitary we can always use the simpler statements ( 1)-(4 ) when dealing with co - invariant subspaces .
consider an orthogonal projection @xmath103 onto @xmath104 .
then @xmath105 is the orthogonal projection onto @xmath106 .
hence @xmath107 projects onto @xmath108 .
this gives @xmath109 .
applying @xmath110 to ( 2 ) we obtain @xmath111 , hence @xmath112 and @xmath88 for all @xmath87 which is ( 3 ) .
conversely , from ( 3 ) we get @xmath113 which is ( 1 ) .
the following definition is consistent with the terminology used by bhat and parthasarathy in @xcite where also continuous time and versions with @xmath114-subalgebras of @xmath115 are considered which allows the inclusion of classical markov processes by restricting to commutative subalgebras .
we only consider discrete time steps and focus on the algebra @xmath115 of all bounded operators .
this allows a lean formulation from which further ( probabilistic ) features of the processes can be pulled out afterwards .
[ def : weak ] a _ discrete weak markov process _
is a triple @xmath116 where @xmath56 is a row isometry on a hilbert space @xmath43 which contains the hilbert space @xmath76 as a co - invariant subspace with respect to @xmath56 .
further we assume that @xmath117 if @xmath56 is a row unitary then we call the process _
unital_. for simplicity we refer to discrete weak markov processes as processes .
note that if @xmath56 is originally defined on a larger hilbert space which contains the hilbert space @xmath76 as a co - invariant subspace with respect to @xmath56 then we can always restrict to a space @xmath43 satisfying the additional assumption which ensures that @xmath43 is the smallest invariant subspace containing @xmath76 . because @xmath76 is co - invariant it follows that @xmath43 is even reducing in the original larger hilbert space .
remark : from the point of view of dilation theory @xmath56 is the minimal isometric dilation of its compression to @xmath76 and as such it is determined by it up to unitary equivalence , see @xcite .
let us now investigate the following increasing sequence of subspaces which is called the _ weak filtration _ associated to the process : @xmath118 ( for @xmath119 ) .
then @xmath120 and @xmath121 note that if the process is unital then @xmath122 for all @xmath123 and the arguments simplify . for unital processes we have @xmath124 and
the inclusion @xmath125 is mapped by @xmath126 to an inclusion @xmath127 . hence in this case
we have an identification of @xmath43 with the inductive limit of the sequence @xmath128 . [
prop : weak ] let @xmath116 be a process .
then * @xmath129 ( for all @xmath130 ) * @xmath131 ( for all @xmath132 ) * if @xmath133 then @xmath134 . *
if @xmath133 then @xmath135 \(a ) states an adaptedness property for the process and the filtration while ( b),(c ) and ( d ) resemble properties of markov processes in classical probability and hence motivate the terminology ` weak markov process ' . versions of ( d ) appear in bhat s papers as the ` weak markov property ' .
@xmath136 is the projection onto @xmath137 .
hence ( a ) is nothing but a reformulation of the obvious @xmath138 . we prove ( b ) by induction .
@xmath139 is the definition of @xmath91 .
now suppose that for @xmath140 @xmath141 then we can use ( 2 ) of lemma [ lem : co ] in the form @xmath142 to get @xmath143 v^*|_{\ensuremath{\mathfrak{h}}\xspace}\\ & = & p v ( p \otimes { \ensuremath{{\rm 1\kern-.25em l}}\xspace } ) \big [ \ldots \big ] ( p \otimes { \ensuremath{{\rm 1\kern-.25em l}}\xspace } ) v^*|_{{\ensuremath{\mathfrak{h}}\xspace } } \\ & = & p v ( z^{n-1}(x ) \otimes { \ensuremath{{\rm 1\kern-.25em l}}\xspace } ) v^*|_{{\ensuremath{\mathfrak{h}}\xspace } } = z^{n}(x ) \\\end{aligned}\ ] ] to prove ( c ) we note that ( 2 ) of lemma [ lem : co ] in the form @xmath144 can be iterated to yield @xmath145 for all @xmath146 .
hence for all @xmath147 we find , by applying @xmath148 , @xmath149 which implies @xmath150 .
together with @xmath151 we obtain @xmath152 the iterative definition of the @xmath153 shows that their ranges do not increase if @xmath154 increases .
hence also @xmath134 whenever @xmath155 . to get ( d ) we start from the definition of @xmath156 , @xmath157 and apply @xmath158 to get @xmath159 now because of ( b ) and ( c )
we can replace @xmath156 by @xmath160 and @xmath161 by @xmath162 .
the following observation , also noted in @xcite , is crucial for our approach .
bhat remarks in @xcite , p.562 , in this context : ` roughly speaking there is also an additive structure when we deal with general quantum dynamical semigroups a detailed study of such systems is yet to be undertaken . '
the work presented in this paper goes into this direction .
[ prop : wandering ] let @xmath116 be a process .
then @xmath163 is a wandering subspace , i.e. , @xmath164 if @xmath165 , and @xmath166 to prove that @xmath167 is wandering it is enough to show that @xmath168 for all @xmath169 .
suppose @xmath169 . because @xmath170 we conclude by ( 2 ) of lemma [ lem : co ] that @xmath171 for all @xmath172 . in particular @xmath173 .
writing @xmath174 with @xmath10 and @xmath172 we find @xmath175 .
finally if @xmath176 but @xmath177 then @xmath178 because @xmath55 and @xmath179 have orthogonal ranges .
putting it all together we have @xmath168 .
the other assertions are now immediate .
the following results indicate how the @xmath180 are related to the quantum physical behaviour of the process .
for this we apply the standard interpretations of quantum physics to the mathematical objects .
suppose @xmath181 and @xmath182 are linear operators . then for @xmath80 and @xmath183 we have the following linear operators on @xmath43 @xmath184 ( empty tensor products to be omitted for @xmath185 and @xmath186 ) .
we refer to these operators as observables .
[ prop : operational ] suppose that the process @xmath116 is unital
. then the observables @xmath187 commute with each other and @xmath187 commutes with @xmath188 whenever @xmath155 .
suppose further that @xmath189 for the orthonormal basis @xmath190 of @xmath45 which we use to define the isometries @xmath191 .
if the process is prepared in a vector state given by a unit vector @xmath192 in the range of @xmath180 so that the @xmath154-th letter in @xmath9 is equal to @xmath123 then the measurement of @xmath187 yields the outcome @xmath123 with certainty . for @xmath193
we define @xmath194 ( empty tensor products omitted in the cases @xmath186 and @xmath195 ) . because the process is unital , @xmath56 is a row unitary and in particular @xmath196 .
it follows that @xmath197 whenever @xmath155 .
writing @xmath187 in this way it becomes obvious that it commutes with @xmath188 and @xmath198 .
let @xmath199 be the orthogonal projection onto the one - dimensional subspace @xmath200 of @xmath45 .
then @xmath201 is the probability that a measurement of @xmath187 yields the result @xmath123 , according to the standard rules of quantum mechanics .
now suppose that the process is prepared in a vector state with a unit vector @xmath192 in the range of @xmath180 , i.e. , @xmath202 for some @xmath66 , so that the @xmath154-th letter in @xmath9 is equal to @xmath123 .
then from @xmath203 ( with @xmath204 for @xmath205 is short for @xmath206 ) we see that only the terms in the sum with @xmath207 can be non - zero and hence the application of @xmath199 at the @xmath154-th copy of @xmath45 always acts identical on all the non - zero terms .
so the probability above is the squared length of a unit vector , i.e. , it is equal to @xmath208 .
this result gives an operational meaning to the words @xmath1 by identifying them with measurement protocols for certain observables .
we can think of measuring @xmath209 as performing a certain type @xmath210 of measurement at the consecutive times @xmath211 and the commutation properties proved above ensure that these measurements can be performed without perturbing the system ( non - demolition measurements ) .
for example with @xmath212 it makes sense to consider @xmath213 for @xmath214 conditioned on @xmath209 ( conditioning in the sense of classical probability theory ) .
such schemes and their application to quantum filtering and quantum control have been pioneered by belavkin @xcite and this is a rapidly developing interdisciplinary field of study .
we mention the recent introductory surveys @xcite which give many references , the latter focusing on discretized models and containing constructions similar to the one above ( @xcite , section 2.5 ) .
it is of course interesting to prepare the system in other states where the outcomes are not deterministic but only statistical information can be obtained , see for example @xcite which uses discretization of continuous processes and thus can be directly connected with the approach here . for a unital process the hilbert space @xmath43 is always the orthogonal sum of the ranges of the @xmath180 for all @xmath9 with a given length , so in principle the problem can be dealt with for an arbitrary vector state by decomposing the state vector with respect to such an orthogonal sum and then using proposition [ prop : operational ] . this may not always be the most practical path to follow for data given in a different way but it should be of theoretical significance . in this place
we only want to indicate the relevance of the @xmath180 for applications and so we do nt go into further details here .
if for the multi - variable systems introduced in section [ section : intro ] the spaces are hilbert spaces and the linear maps are contractions between these hilberts apaces then we want to think of them as appearing inside the weak markov processes introduced in section [ section : weak ] .
the following definitions give a precise meaning to that .
then we justify the definitions by discussing how the represented structure maps can help us to understand the properties of the process .
[ def : structure ] a _ representation of an input pair _ @xmath215 ( with column contractions @xmath216 and @xmath217 ) is given by a process @xmath116 such that @xmath218 and @xmath219 together with an isometry @xmath220 such that , with @xmath221 the represented input space , we have @xmath222 . a _ representation of an output pair _ @xmath223 ( with column contractions @xmath216 and @xmath224 ) is given by a process @xmath116 such that @xmath218 and @xmath219 together with an isometry @xmath225 such that the represented output space @xmath226 is wandering for @xmath56 and we have @xmath227 . a _ representation of the structure maps _
@xmath228 , where @xmath229 , is given by representations of @xmath215 as an input pair and @xmath223 as an output pair as above with the same process @xmath116 such that @xmath230 . in the following
we shall often suppress the isometries @xmath231 and @xmath232 and treat them as identifications whenever this simplifies the notation .
note that a represented input space @xmath233 is always wandering because by definition it is a subspace of the wandering subspace @xmath167 , see proposition [ prop : wandering ] .
an interesting example for a represented output space is @xmath234 which is the wandering subspace arising from the wold decomposition of the row isometry @xmath56 , see @xcite .
let us write @xmath235 for @xmath236 and @xmath237 for @xmath238 , further we denote by @xmath239 the closed linear span of all @xmath240 , by @xmath241 the closed linear span of all @xmath237 ( with @xmath1 ) . with suitable identifications we can think of @xmath239 as an orthogonal direct sum of copies of the input space @xmath17 , similarly @xmath241 is an orthogonal direct sum of copies of the output space @xmath16 .
we sometimes refer to these copies as @xmath9-translated input resp . output spaces .
representations of structure maps as above are always _ causal _ in the sense that @xmath242 for all @xmath243 with @xmath244 .
more general it follows further that @xmath245 whenever @xmath246 which we also refer to as causality .
note that from a representation with @xmath247 we get a kraus decomposition for the transition operator of the process , namely @xmath248 ( limit in the strong operator topology if @xmath6 ) . given a process @xmath116 any subspace of @xmath167 interpreted as
a represented input space gives rise to a representation of an input pair @xmath215 and any wandering subspace of @xmath249 interpreted as a represented output space gives rise to a representation of an output pair @xmath223 and together we have a representation of structure maps @xmath228 .
[ prop : system ] given a representation of structure maps @xmath228 by a process @xmath116 let @xmath250 be an element of @xmath43 , with @xmath251 and @xmath252 . if ( for all words @xmath9 ) @xmath253 then we have ( for all words @xmath9 and generators @xmath254 ) @xmath255 if further @xmath256 then we have @xmath257 hence we have a noncommutative fornasini - marchesini system ( compare section [ section : intro ] ) .
the first assertion follows from @xmath258 using causality we find that @xmath259 = c x(\alpha ) + d u(\alpha).\end{aligned}\ ] ] note that the proof uses causality but does not use the fact that @xmath260 is wandering .
hence the result is valid for generalized ( non - wandering but causal ) output spaces . with the maximal choice @xmath261 we have for any vector @xmath262 that @xmath263 which is relevant information with respect to proposition [ prop : operational ] above .
let us now work out the observability map @xmath264 and the transfer function @xmath265 ( already introduced in section [ section : intro ] ) for represented structure maps .
if we identify @xmath266 then @xmath267 in other words , we obtain the @xmath9-entry @xmath268 of the observability map @xmath264 by projecting the represention @xmath76 of the internal space @xmath18 to the @xmath9-translated output space @xmath237 . similarly with @xmath266 and @xmath269
we obtain @xmath270 in other words , we obtain the @xmath9-coefficient @xmath271 in the formal power series expansion of the transfer function @xmath265 ( already introduced in section [ section : intro ] ) by projecting the represention @xmath233 of the input space @xmath17 to the @xmath9-translated output space @xmath237 .
causality implies that further , with identifications @xmath266 and @xmath272 , @xmath273 this pattern in the operator - valued kernel @xmath274 describes what is called a multi - analytic kernel .
it is an operator - valued matrix describing the contraction @xmath275 for the orthogonal decompositions with respect to the translated output resp .
input spaces .
it follows from this description that this contraction intertwines the row shifts on @xmath276 and @xmath277 which are obtained by restricting the row isometry @xmath56 .
this is the defining property of a multi - analytic operator and we have verified it for the contraction @xmath275 .
multi - analytic operators have been introduced and studied by popescu @xcite , the situation above involving transfer functions for pairs of wandering subspaces is worked out in more detail in @xcite .
we can think of it as a kind of multiplication by the transfer function @xmath265 .
in fact for @xmath14 it is literally multiplication by an ordinary analytic function .
but for all @xmath278 , for a noncommutative fornasini - marchesini system , we can still verify the following analogy : if we introduce formal power series for the entries @xmath279 and @xmath280 ( see proposition [ prop : system ] ) @xmath281 then , with the initial condition @xmath282 , we have the input - output relation @xmath283 recall the convention @xmath284 for @xmath285 which gives the multiplication rule @xmath286 .
we introduce additional concepts with the intention to describe substructures of processes .
we say that @xmath287 is a _
subprocess _ of the process @xmath116 if @xmath288 is a closed subspace of @xmath76 which is co - invariant for @xmath56 and @xmath289 where @xmath290 .
note that @xmath288 is also co - invariant for @xmath291 and @xmath287 is a process in its own right . given a subprocess @xmath287 of a process
@xmath116 we can form the _ quotient process _
@xmath292 where @xmath293 .
let us check that @xmath294 is a process .
indeed , because @xmath288 is co - invariant for @xmath56 we see that @xmath295 is contained in @xmath296 .
hence @xmath297 is contained in @xmath298 which is a @xmath56-invariant subspace orthogonal to @xmath299 .
hence @xmath299 is co - invariant for @xmath300 which proves our claim .
note that in general @xmath299 is not co - invariant for @xmath56 .
considering adjoints we see that @xmath301 but only @xmath302 .
so we need to distinguish carefully between subprocesses and quotient processes .
it is convenient to reformulate these concepts within a _ category of processes _ which we define now .
the objects of the category are the processes with a common multiplicity space @xmath303 .
a _ morphism _ from @xmath304 to @xmath305 is a contraction @xmath306 which intertwines the adjoints of the row isometries , i.e. @xmath307 ( or written differently : @xmath308 for @xmath254 and @xmath309 ) .
composition of morphisms is given by composition of operators , the identity morphism is given by the identity operator .
we add some immediate observations justifying this definition .
first note that a morphism @xmath33 is an isomorphism if and only if @xmath33 is unitary and we get a reasonable meaning for two processes to be isomorphic . in the following we often identify isomorphic processes .
also note that if we have a morphism given by a contraction @xmath306 then @xmath33 can always be extended to a contraction @xmath310 such that @xmath311 and @xmath312 and @xmath313 this is nothing but the commutant lifting theorem in the version of popescu @xcite .
we call @xmath314 an _ extended morphism _ associated to @xmath33
. we can also think of a morphism as the class of all extended morphisms associated to @xmath33 , this attaches a global interpretation to it .
finally it follows from @xmath315 that @xmath316 is co - invariant for @xmath317 , and hence from a morphism @xmath33 from @xmath304 to @xmath305 , by defining @xmath288 to be the closure of @xmath316 and @xmath318 , we always obtain a subprocess @xmath287 of @xmath305 .
in fact , we can reformulate subprocesses and quotient processes in terms of morphisms as follows .
if @xmath287 is a subprocess of a process @xmath116 then @xmath319 , interpreted as an embedding of @xmath288 into @xmath76 , is an isometric morphism : @xmath320 conversely , given an isometric morphism @xmath33 from a process @xmath287 to a process @xmath116 we can interpret @xmath33 as an embedding @xmath319 and the properties of a morphism ensure that @xmath288 is co - invariant for @xmath56 . with an extended morphism @xmath321 , interpreted as an embedding of @xmath322 into @xmath43 , we can also arrange that @xmath323 ( by uniqueness of a minimal isometric dilation up to unitary equivalence ) .
so we get a subprocess . for the corresponding quotient process
@xmath294 we can check that the orthogonal projection @xmath324 is a coisometric morphism : @xmath325 conversely , if we have a coisometric morphism @xmath33 from @xmath116 to a process @xmath294 then we can interpret @xmath33 as an orthogonal projection @xmath324 , then use the properties of a morphism to check that @xmath326 is co - invariant for @xmath56 , giving rise to a subprocess so that the corresponding quotient process is the process @xmath294 we started from .
note that @xmath327 and @xmath328 are not morphisms , in general .
the image of @xmath319 is equal to the kernel of @xmath329 and we can proceed to give a concise description of the situation as a short exact sequence of processes : @xmath330 & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^{{\ensuremath{{\rm 1\kern-.25em l}}\xspace}_{\ensuremath{\mathfrak{g}}\xspace } } & ( { \ensuremath{{\mathcal h}}\xspace } , v , { \ensuremath{\mathfrak{h}}\xspace } ) \ar[r]^{p_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] & 0 } \ ] ] here @xmath7 stands for the process on the @xmath7-dimensional space which takes the role of a zero object in our category . in the following
we suppress the embeddings in the notation whenever this is convenient .
[ lem : gamma ] given a short exact sequence of processes @xmath330 & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^{{\ensuremath{{\rm 1\kern-.25em l}}\xspace}_{\ensuremath{\mathfrak{g}}\xspace } } & ( { \ensuremath{{\mathcal h}}\xspace } , v , { \ensuremath{\mathfrak{h}}\xspace } ) \ar[r]^{p_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] & 0 } \ ] ] then for the relative position of the wandering subspaces @xmath331
@xmath332 we obtain @xmath333 more general , for all @xmath1 @xmath334 we have @xmath335 where @xmath336 . because @xmath337 we conclude that @xmath338 .
but by the definition of @xmath339 we also have @xmath340 and it follows that @xmath341 .
hence if @xmath342 and @xmath343 then @xmath344 and @xmath345 and it follows that @xmath346 for all @xmath343 .
obviously also @xmath347 and so @xmath348 which is the first formula we intended to prove .
it is clear that @xmath349 . to get the equality @xmath350 we argue as follows .
consider @xmath351 . because @xmath352 it follows that @xmath353 and
because @xmath167 is wandering for @xmath56 we conclude that @xmath354 if @xmath343 and hence @xmath355 now , because @xmath356 , we can compute @xmath357 for the last equality above note that @xmath358 but @xmath359 . + + because @xmath360 and @xmath339 are wandering it is now immediate that @xmath361 to get the first line of the general formulas we only have to note additionally that always @xmath362 . to get the second line we have to prove that additionally @xmath363 indeed , if @xmath364 this follows from @xmath350 shown above . then for other @xmath9 reduce it to the previous case by using the fact that @xmath365 is a row unitary and hence @xmath366 .
these observations suggest the following construction which allows us to classify the processes which can be obtained by short exact sequences , given the subprocess and the quotient process .
[ def : cascade ] given processes @xmath287 and @xmath294 ( with a common multiplicity space @xmath303 ) and any contraction @xmath367 we define the @xmath39-cascade @xmath368 where @xmath369 we add the following explanations for this definition .
@xmath370 is the defect operator for @xmath371 and @xmath372 , the closure of its range , is the defect space .
then @xmath373 is the orthogonal sum of a family of copies of @xmath372 indexed by the free semigroup @xmath374 , so what we mean by the canonical row shift on this space is just moving elements between these copies .
further , to explain the action of @xmath56 on @xmath288 , note that @xmath291 maps @xmath288 into @xmath375 and now @xmath376 acts as identity on @xmath288 , the contraction @xmath371 maps @xmath360 into @xmath377 and @xmath378 maps @xmath360 into @xmath372 which we interpret as @xmath379 .
note that for @xmath40 the @xmath39-cascade is nothing but the direct sum of the two processes @xmath287 and @xmath294 .
[ thm : cascade - p ] given processes @xmath287 and @xmath294 ( with a common multiplicity space @xmath303 ) and a contraction @xmath367 . then the @xmath39-cascade @xmath380 is a process , @xmath381 and we have a short exact sequence @xmath382 in other words , @xmath287 is a subprocess and @xmath294 is a quotient process of the @xmath39-cascade .
+ conversely , in a short exact sequence @xmath330 & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^{{\ensuremath{{\rm 1\kern-.25em l}}\xspace}'_{\ensuremath{\mathfrak{g}}\xspace } } & ( { \ensuremath{{\mathcal h}}\xspace } ' , v ' , { \ensuremath{\mathfrak{h}}\xspace } ' ) \ar[r]^{p'_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] & 0 } \,,\ ] ] where @xmath383 resp .
@xmath384 are an isometric resp . a coisometric morphism , the process @xmath385 is isomorphic to a @xmath39-cascade for a contraction @xmath367 such that moreover the corresponding extensions are equivalent , i.e. the following diagram commutes : @xmath330 & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^{{\ensuremath{{\rm 1\kern-.25em l}}\xspace}'_{\ensuremath{\mathfrak{g}}\xspace } } & ( { \ensuremath{{\mathcal h}}\xspace } ' , v ' , { \ensuremath{\mathfrak{h}}\xspace } ' ) \ar[r]^{p'_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] & 0 \\ 0 \ar[r ] & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^(0.35){{\ensuremath{{\rm 1\kern-.25em l}}\xspace}_{\ensuremath{\mathfrak{g}}\xspace } } \ar[u]_{= } & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \lhd_\gamma ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[u]^{j}_{isom . } \ar[r]^(0.65){p_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] \ar[u]_{= } & 0 } \ ] ] two cascades with @xmath386 are equivalent as extensions if and only if @xmath387 and @xmath388 are equal .
we can summarize the theorem by saying that there is a one - to - one correspondence between equivalence classes of extensions of the process @xmath294 by the process @xmath287 and contractions from @xmath339 to @xmath360 and this correspondence is given by the @xmath39-cascade construction .
working within the @xmath39-cascade as defined above we have @xmath360 identified via the isometry @xmath389 with a subspace of @xmath390 .
note that @xmath339 is the wandering subspace arising from the wold decomposition of @xmath300 , see @xcite , and @xmath391 is another wandering subspace with all translates orthogonal to the translates of @xmath339 .
hence the embedded @xmath360 is wandering for @xmath56 . with this
it is now easy to check that @xmath56 is a row isometry , that @xmath116 is a process with subprocess @xmath287 and quotient process @xmath294 and that @xmath39 and @xmath371 satisfy the formulas stated .
now suppose that the process @xmath385 is given by a short exact sequence , i.e. as an extension of @xmath294 by @xmath287 .
we define @xmath392 and then form the corresponding @xmath39-cascade @xmath393 .
we verify that this yields an equivalent extension by constructing a unitary @xmath394 which intertwines the row isometries @xmath395 and @xmath56 and which maps @xmath396 to @xmath397 if @xmath398 and @xmath399 to @xmath400 if @xmath401 .
in fact , then the adjoint @xmath402 is an extended morphism and its restriction @xmath403 is the isomorphism we look for . to see that @xmath404 exists it
is enough to check that the remaining parts can be matched correctly .
we invoke the following lemma which is a standard tool in operator theory .
[ lem : defect ] let @xmath405 and @xmath79 be hilbert spaces and @xmath406 a closed subspace of @xmath79 . if @xmath407 is an isometry such that @xmath408 then @xmath409 is unitarily equivalent to @xmath410 the unitary from @xmath79 to @xmath411 needed in the lemma is the identity on @xmath406 and it is @xmath412 for @xmath413 .
we apply lemma [ lem : defect ] with @xmath414 , with @xmath409 being the isometric embedding of @xmath360 into @xmath415 and with @xmath416 .
lemma [ lem : gamma ] shows that @xmath417 in this case and moreover that the embeddings of translates @xmath418 follow exactly the pattern exposed by the @xmath39-cascade .
hence we can put all the pieces together and get @xmath404 .
finally , if @xmath419 and @xmath420 are equivalent extensions then the unitary intertwiner @xmath404 constructed above maps for each element of @xmath360 its first embedding to the second .
the same happens to elements of @xmath339 .
hence if we suppress the embeddings we find that @xmath421 is the same operator in both cases , i.e. , @xmath422 .
remark : it is instructive to look at the situation described in the previous theorem from the point of view of dilation theory .
then we start with a row contraction on @xmath423 of the form @xmath424 which is called a lifting ( in our application this is @xmath425 ) .
it is well known that in such a situation @xmath210 must have the form @xmath426 with a contraction @xmath427 , see @xcite .
chapter iv , lemma 2.1 for @xmath14 and @xcite , prop .
3.1 for the general case .
hence a @xmath39-cascade can also be thought of as a description of the structure of the minimal isometric dilation of such a row contraction .
liftings and their dilations are studied in @xcite and the results can be interpreted in the language of processes which we use in this paper .
one of the things one can do with linear systems is to stick them together in various ways . the most basic way to do that is to take the output of one system @xmath428 and to use it as the input of another system @xmath429 . the combined system
is then called a _ simple cascade _ , see for example @xcite for the classical theory ( @xmath14 ) .
it also works for the noncommutative fornasini - marchesini systems we have been considering here .
we need a slight generalization where the input of system @xmath429 is obtained from the output of system @xmath428 by applying a transformation @xmath430 to it .
we call this a @xmath430-cascade of systems .
( 15,2.5 ) ( 4.4,0.2)(0,1)2 ( 4.4,0.2)(1,0)6.4 ( 10.8,0.2)(0,1)2 ( 4.4,2.2)(1,0)6.4 ( 4.4,0.7)(1,0)1.8 ( 4.4,1.7)(1,0)1.8 ( 6.2,0.7)(0,1)1 ( 5.1,1.1)@xmath431 ( 9.0,0.7)(1,0)1.8 ( 9.0,1.7)(1,0)1.8 ( 9.0,0.7)(0,1)1 ( 9.8,1.1)@xmath432 ( 4.4,1.2)(-1,0)1.4 ( 9.0,1.2)(-1,0)1.1 ( 7.3,1.2)(-1,0)1.1 ( 12.2,1.2)(-1,0)1.4 ( 7.6,1.2 ) ( 3.6,0.9)@xmath433 ( 6.6,0.9)@xmath434 ( 8.3,0.9)@xmath435 ( 11.4,0.9)@xmath436 ( 7.5,1.1)@xmath430 we assume here that the two noncommutative fornasini - marchesini systems both have the same multiplicity @xmath3 .
then the internal space of the combined system is defined to be the direct sum of the internal spaces of systems @xmath428 and @xmath429 and , with @xmath437 ( for all @xmath1 ) , it follows , by eliminating variables , that the structure maps @xmath228 of the combined system are obtained from the structure maps @xmath438 of system @xmath428 and @xmath439 of system @xmath429 by @xmath440 @xmath441 in this case we also speak of a @xmath430-cascade of structure maps . the transfer function @xmath265 of such a @xmath430-cascade of systems ( or of structure maps ) factorizes .
if @xmath314 is a power series with coefficients @xmath442 and @xmath430 maps @xmath443 to @xmath444 then we denote by @xmath445 the power series with coefficients @xmath446 . with this convention
it is not difficult to check that the transfer function @xmath265 of the combined system is obtained from the transfer functions @xmath447 and @xmath448 of systems @xmath429 and @xmath428 by @xmath449 ( with @xmath450 ) .
+ now we prove that if @xmath116 is a @xmath39-cascade @xmath451 of processes then we can think of its fornasini - marchesini system , from proposition [ prop : system ] , as a cascade of the systems associated to @xmath287 and @xmath294 .
this is not so obvious if you arrive at the notion of a @xmath39-cascade of processes from a dilation point of view but it nicely clarifies the system theory involved and of course it motivates our choice of terminology . + we assume that we have structure maps @xmath452 represented by the process @xmath287 and @xmath453 represented by the process @xmath294 , according to definition [ def : structure ]
now we form a @xmath39-cascade of processes @xmath454 and we introduce output and input spaces represented by @xmath116 : @xmath455 we have the inclusions @xmath456 and @xmath457 and @xmath458 is wandering for @xmath291 and hence for @xmath56 , so indeed this yields a representation of structure maps @xmath459 by the process @xmath116 .
further we define @xmath460 where the latter equality follows from @xmath461 together with the geometry of the @xmath39-cascade ( see definition [ def : cascade ] and lemma [ lem : gamma ] ) .
a subspace @xmath79 of @xmath43 is called a _ left support _ of @xmath39 if @xmath462 and it is called a _
right support _ of @xmath39 if @xmath463 . with these preparations
we can now find a @xmath430-cascade of systems in the @xmath39-cascade of processes .
[ thm : cascade - s ] suppose that @xmath464 is a left support and @xmath465 is a right support of @xmath39 .
then the structure maps @xmath459 are a @xmath430-cascade of @xmath453 and @xmath452 . explicitly ( for @xmath466 ) : @xmath467 @xmath468 roughly speaking , @xmath464 left and @xmath465 right supporting @xmath39 means that they are chosen big enough to transport the information contained in @xmath39 .
an example is given by the choice @xmath469 and @xmath470 and in this case @xmath471 .
we verify the explicit formulas step by step .
the arguments are based on the geometry of a @xmath39-cascade as given in definition [ def : cascade ] , in particular : @xmath472 maps @xmath288 into @xmath375 and @xmath473 is orthogonal to the range of @xmath291 , further @xmath474 .
@xmath475 it is possible to iterate the construction shown in theorem [ thm : cascade - s ] . because @xmath476 we have @xmath477 .
hence if @xmath478 is another process ( with the same multiplicity ) and @xmath479 is a contraction then not only can we form @xmath480 but also @xmath481 $ ] with @xmath482 given by @xmath483 theorem [ thm : cascade - s ] applies iteratively .
this may be seen as another justification for calling such a structure a cascade .
to make use of the system theory now available to us for the study of processes we discuss the control theory concept of observability in the multi - variable setting .
see for example @xcite for a recent treatment in a purely operator theoretic spirit .
[ def : observable ] given an output pair @xmath223 for an internal space @xmath18 and an output space @xmath16 , a subset @xmath484 is called observable if @xmath485 , the observability map restricted to @xmath486 , is injective ( as a map from @xmath486 to the @xmath16-valued functions on @xmath2 ) .
if @xmath223 is represented by the process @xmath116 then we also say in this case that @xmath487 is observable in @xmath116 by the output space @xmath260 .
the interpretation of observability is that every @xmath488 can be reconstructed from the outputs @xmath489 .
[ prop : observable ] the subset @xmath490 is observable in @xmath116 by the output space @xmath260 if and only if @xmath491 is injective
. for @xmath488 we have @xmath492 and we conclude that @xmath486 is observable if and only if @xmath493 is injective . because the projection @xmath494 is the supremum of the projections @xmath495 we can replace the family @xmath493 by the single contraction @xmath491 .
if we have a @xmath39-cascade and we use the maximal input space @xmath360 of the subprocess as output space for the cascade , roughly speaking if we confine our observations to the subprocess , then questions of observability become an interesting issue of theoretical and practical importance . [
thm : observable ] consider the @xmath39-cascade @xmath454 and the output space @xmath496 .
the following assertions are equivalent : * @xmath299 is observable in @xmath116 by @xmath260 .
* @xmath497 is injective .
* @xmath498 * @xmath499 has dense range .
* @xmath500 * there is an extended morphism associated to the morphism @xmath501 + which is a unitary from @xmath43 to @xmath43 .
* @xmath300 is a row shift and @xmath502 is injective .
* @xmath300 is a row shift and @xmath502 is isometric .
if the transition operator @xmath91 of @xmath116 is unital then we also have the following equivalent condition : * @xmath503 ( in the strong operator topology ) the equivalence of ( 1a ) and ( 1b ) is proposition [ prop : observable ] .
further @xmath504 which shows ( 1b ) @xmath505 ( 2b ) . because @xmath506 assertion ( 2b ) is equivalent to @xmath507 which is ( 2a ) . on the other hand @xmath508 because @xmath43 is the closed linear span of the @xmath509 with @xmath1 and we have @xmath510 , hence @xmath511 which is dense in @xmath509 .
but @xmath322 is a closed subspace of @xmath43 and we conclude that ( 2b ) implies ( 3a ) .
the converse , ( 3a ) implies ( 2b ) , is obvious .
it is easy to check that in a @xmath39-cascade the morphism @xmath319 , the embedding of @xmath288 into @xmath76 , always has @xmath321 , the embedding of @xmath322 into @xmath43 , as an associated extended morphism . if we have ( 3a ) , i.e. @xmath500 , then @xmath321 is nothing but the identity operator @xmath512 on @xmath43 , hence ( 3a ) implies ( 3b ) .
conversely ( 3a ) is implicit in the statement of ( 3b ) . from definition [ def : cascade ] of the @xmath39-cascade we have @xmath513 with @xmath377 and @xmath514 , from which we get @xmath515 . if we have ( 3a ) , i.e. @xmath500 , then , because @xmath516 and @xmath517 , we find @xmath518 and hence @xmath519 which means that @xmath300 is a row shift . from ( 3a )
we have @xmath520 and together with @xmath347 and @xmath521 for all @xmath343 ( by lemma [ lem : gamma ] ) we conclude that @xmath339 is actually a subspace of @xmath360 which means that @xmath522 is an isometric embedding .
hence ( 3a ) implies ( 4b ) .
obviously ( 4b ) implies ( 4a ) and from ( 4a ) , together with lemma [ lem : gamma ] , we get an injective map @xmath523 and ( 1b ) follows . finally if @xmath91 is unital then the projections @xmath524 increase with @xmath69 and converge to their supremum @xmath525 ( see section [ section : weak ] ) , hence @xmath526 converges to @xmath527 ( in the strong operator topology ) . now ( 3a ) implies that @xmath528 and this implies ( 2b ) , hence all these assertions , including ( 5 ) , are equivalent .
remark : of course ( 3b ) does not mean that @xmath287 and @xmath116 are isomorphic as processes .
in fact , if @xmath529 then the morphism @xmath319 , the embedding of @xmath288 into @xmath76 , is not unitary .
[ def : ac ] the short exact sequence @xmath330 & ( { \ensuremath{{\mathcal g}}\xspace } , v^{\ensuremath{{\mathcal g}}\xspace } , { \ensuremath{\mathfrak{g}}\xspace } ) \ar[r]^{{\ensuremath{{\rm 1\kern-.25em l}}\xspace}_{\ensuremath{\mathfrak{g}}\xspace } } & ( { \ensuremath{{\mathcal h}}\xspace } , v , { \ensuremath{\mathfrak{h}}\xspace } ) \ar[r]^{p_{\ensuremath{\mathfrak{k}}\xspace } } & ( { \ensuremath{{\mathcal k}}\xspace } , v^{\ensuremath{{\mathcal k}}\xspace } , { \ensuremath{\mathfrak{k}}\xspace } ) \ar[r ] & 0 } \ ] ] is called _ asymptotically complete _ if one ( and hence all ) of the assertions in theorem [ thm : observable ] are satisfied .
this terminology is borrowed from scattering theory
. it will be explained in the final section of this paper how the setting above is related to already existing scattering theories for markov chains .
[ cor : inner ] let an asymptotically complete sequence as in definition [ def : ac ] be given together with an input space @xmath233 and the output space @xmath530 , both represented by @xmath116
. then @xmath531 , the restriction of the observability map to @xmath299 , is isometric and the transfer function @xmath265 is inner , i.e. , the corresponding multi - analytic operator is isometric .
note that @xmath532 indeed gives us @xmath533 for all @xmath9 .
we have @xmath534 .
the multi - analytic operator corresponding to the transfer function @xmath265 is @xmath275 . from asymptotic completeness
it follows that @xmath535 , see theorem [ thm : observable ] ( 3a ) .
hence , because @xmath299 and @xmath239 are orthogonal to @xmath288 , they are both subspaces of @xmath241 and it follows that @xmath497 and @xmath275 are isometric .
we describe a few alternative ways to interpret asymptotic completeness .
first , if we are in an asymptotically complete situation and we choose @xmath233 maximal , i.e. @xmath261 , then the sum of the identity @xmath319 ( on @xmath288 ) , of the observability map @xmath536 ( on @xmath299 ) and of the multi - analytic operator associated to @xmath265 ( on @xmath239 ) is equal to the identity @xmath512 .
this is a direct consequence of corollary [ cor : inner ] .
but it is the identity @xmath512 presented with a change of basis that describes the relative position of the weak filtrations of the process @xmath537 and of the process @xmath538 .
second , from the point of view of dilation theory , in the case of asymptotic completeness we deal with socalled subisometric dilations .
we do nt go into this here , see @xcite for details .
third , asymptotic completeness and its observability aspect can also be given an explicit quantum physical interpretation by reconsidering the observables discussed in proposition [ prop : operational ] .
indeed , in the case of asymptotic completeness the hilbert space @xmath43 is the closed span of the spaces @xmath539 ( with @xmath80 , the case @xmath185 to be interpreted as @xmath288 ) .
it follows that the algebras @xmath540 generate @xmath42 ( as a von neumann algebra ) .
on the other hand , the observables @xmath541 if we consider all @xmath542 and all @xmath543 , generate @xmath544 ( as a von neumann algebra ) .
conclusion : we can approximate arbitrary observables in @xmath42 in the weak ( or strong ) operator topology by observables generated by these @xmath188 and @xmath545 .
in particular , in physics language , we can answer all questions about observables on the part @xmath299 ( and hence also find the state ) from measuring observables on the orthogonal part @xmath288 of the internal space ( the @xmath188 ) together with field observables ( the @xmath545 ) .
we have the following quantitative statement about these approximations which stresses the role of the observability operator @xmath264 : if @xmath546 then the norm distance squared of @xmath547 from the space @xmath548 is given by @xmath549 .
if we have asymptotic completeness then @xmath536 is isometric and this distance tends to @xmath7 for @xmath550 .
the following way of finding subprocesses gives a connection to a topic which is of natural interest for quantum markov processes and more general for quantum probability : invariant states .
recall that if @xmath551 is a normal state of a von neumann algebra then in this von neumann algebra there exists a smallest orthogonal projection @xmath552 such that @xmath553 , called the support projection @xmath554 .
for all elements @xmath555 we have @xmath556 essentially the following is lemma 6.1 from @xcite , for convenience we include a proof which uses our now familiar terminology and notation .
[ prop : state ] given a process @xmath116 , suppose that @xmath551 is a normal state of @xmath115 which is invariant for the transition operator @xmath91 , i.e. , @xmath557 then with @xmath558 the subspace @xmath288 is co - invariant for @xmath56 .
let @xmath559 .
then @xmath560 hence @xmath561 ( because @xmath551 restricted to the subalgebra @xmath562 , obtained by compression with its support , is a faithful state ) . but @xmath563 with @xmath564 .
hence @xmath565 which shows that @xmath288 is co - invariant for @xmath56 .
using proposition [ prop : state ] we can always find a subprocess from a normal invariant state and this subprocess is nontrivial , in the sense that @xmath529 , if and only if the state is not faithful .
definition [ def : ac ] for asymptotic completeness applies and we can use the criteria in theorem [ thm : observable ] . instead of further analyzing the general case we concentrate for the rest of this section on an application to a class of noncommutative markov processes which are not originally weak markov processes but which nevertheless can be studied successfully by our methods .
the following construction represents the most basic way of introducing stationary markov chains in an operator algebraic context . in this form
it is taken from @xcite , see more details there .
a more leisurely introduction to the topic is @xcite .
let @xmath566 and @xmath567 be @xmath114-algebras and let @xmath568 be a non - zero @xmath41-homomorphism .
this can be iterated to yield the @xmath41-homomorphisms @xmath569 where @xmath570 and then inductively @xmath571 if @xmath551 resp .
@xmath572 are states on @xmath566 respectively @xmath567 and we impose the stationarity condition @xmath573 then we can think of the sequence @xmath574 as noncommutative random variables which form a noncommutative stationary markov chain . we can obtain an associated weak process in the sense of this paper by applying the gns - construction which from @xmath575 produces @xmath576 and from @xmath577 produces @xmath578 , the gns - hilbert space and a cyclic vector representing the state ( in each case )
. then the stationarity condition translates into the fact that @xmath579 ( with @xmath580 ) is an isometry . in chapter 1 of @xcite @xmath581
is called the _ associated isometry _ and it plays a central role there in the analysis of the markov chain .
we are now in a position to deepen this analysis and to make the conceptual framework more elegant by putting it into the context provided in this paper .
we start with the adjoint @xmath582 which is a ( row ) coisometry .
hence its minimal isometric dilation @xmath583 is a row unitary and @xmath116 is a unital process .
( it is a well known fact in dilation theory which can be checked directly that the minimal isometric dilation @xmath56 on @xmath43 of a row contraction on @xmath76 is a row unitary if and only if the row contraction is coisometric . )
we call @xmath116 the ( weak ) process associated to the original stationary markov chain
. + remark : if we think of @xmath110 as a kind of ( schrdinger ) dynamics of vector states for a weak process then we see that for an associated weak process this is provided on @xmath76 by the associated isometry @xmath581
. note however that if we consider the original noncommutative random variables @xmath584 as steps of a ( heisenberg ) dynamics of observables then there is a time reversal involved if we go to the noncommutative random variables @xmath84 ( as in section [ section : weak ] ) as steps of a ( heisenberg ) dynamics of observables for the associated weak process . +
but there is more structure available from a stationary markov chain .
one can check that the vector state represented by @xmath585 is invariant for the transition operator @xmath91 of the ( weak ) process and hence we can invoke proposition [ prop : state ] to see that its support , the one - dimensional subspace @xmath586 , is co - invariant for @xmath56 .
in fact it is easy to check this directly : consider the one - dimensional subspace @xmath587 . because @xmath588 and @xmath589 we find @xmath590 which implies that @xmath288 is co - invariant for @xmath56 and hence we have a subprocess @xmath287 .
combined with the analysis in section [ section : cat ] we obtain the surprising result that the gns - hilbert space of a stationary markov chain always carries automatically the structure of a @xmath39-cascade @xmath591 and we can think of the whole short exact sequence @xmath382 as being _ associated _ to the stationary markov chain .
we have @xmath586 and @xmath592 .
let us investigate what asymptotic completeness means in this case .
we can define the associated isometry @xmath593 for the @xmath69-th noncommutative random variable @xmath584 which , because it arises from the same iterative procedure , can be expressed by the iteration @xmath594 of @xmath56 . with @xmath580 and @xmath595 we have @xmath596 and from that @xmath597 which yields the norm equality @xmath598 because @xmath599 increases to the projection @xmath525 ( which is the limit for @xmath550 in the strong operator topology ) it follows that the property @xmath500 , equivalent to asymptotic completeness by theorem [ thm : observable ] , is also equivalent to @xmath600 for all @xmath251 .
if the gns - representation is faithful then we can interpret the hilbert space norm as a norm @xmath601 on the @xmath114-algebra and we can write @xmath602 for all @xmath580 , where @xmath603 denotes the conditional expectation obtained by evaluating the state @xmath551 on @xmath566
. however the latter condition is well known to be equivalent to the asymptotic completeness of the stationary markov chain in a scattering theory meaning introduced by kmmerer and maassen in @xcite .
as mentioned earlier here we followed a variant developed in @xcite .
the equivalence of the convergence in the norm @xmath604 above with the property of asymptotic completeness in the scattering theory meaning is stated in lemma 1.5 of @xcite .
hence , proceeding as above , all these meanings of asymptotic completeness coincide .
let us summarize this insight as follows : [ prop : complete ] the short exact sequence of weak processes associated to a stationary markov chain is asymptotically complete in the sense of definition [ def : ac ] if and only if the stationary markov chain is asymptotically complete in the scattering theory meaning of @xcite or @xcite
. a necessary and sufficient criterion is given by @xmath605 in fact , the last statement is nothing but criterion ( 5 ) for asymptotic completeness from theorem [ thm : observable ] applied to this special situation .
this reproduces a criterion for asymptotic completeness in terms of the completely positive map @xmath91 ( which in @xcite is called the dual extended transition operator @xmath606 , see @xcite , section 4 , in particular theorem 4.3 ) .
because in this case @xmath586 is one - dimensional the criterion is also equivalent to the ergodicity of @xmath91 , i.e. the fixed point set of @xmath91 being equal to @xmath607 , see @xcite , section 3 , for more details . the construction of the scattering theory itself with its intertwiners ( mller operators ) between a free and a perturbed dynamics requires a two - sided process ( i.e. with time variable in @xmath608 for discrete time ) as in @xcite or the construction of a two - sided extension as in @xcite and it would take us too far to include it here .
the reader who follows the given sources will recognize that the change of basis studied in section [ section : observable ] relating the weak filtration of @xmath609 to the weak filtration of @xmath538 and given by the identity on the one - dimensional @xmath586 plus the observability map from @xmath299 to @xmath241 plus the multi - analytic operator associated to the transfer function ( on @xmath239 ) , is a gns - version of the mller operator in the scattering theory , in the same way as the associated weak markov process is a gns - version of the stationary markov chain .
if we take into account the identification of @xmath322 with the inductive limit of the sequence @xmath610 mentioned in section [ section : weak ] and note that because @xmath586 is one - dimensional we can identify it further with an infinite tensor product @xmath611 then we arrive essentially at the version of the scattering theory worked out in @xcite , chapter 2 .
but the setting of this paper is more general . for example we could also consider stationary markov chains which are not originally constructed by tensor products of algebras and nevertheless associate weak processes via gns - construction and apply proposition [ prop : state ] to find subprocesses ( with @xmath586 ) . or we can study higher - dimensional co - invariant subspaces @xmath288
the systematic use of weak processes adds conceptual clarity and cohesion to such investigations .
+ + * acknowledgement * : this research has been supported by the epsrc grant ep / g039275/1 .
1 w. arveson , _ noncommutative dynamics and e - semigroups . _
springer monographs in mathematics .
springer , 2003 .
j.a.ball , v. bolotnikov , q. fang , _ multivariable backward - shift - invariant subspaces and observability operators . _ multidimens .
syst . signal process . 18 ( 2007 ) , no . 4 , 191 - 248 .
ball , g. groenewald and t. malakorn , _ structured noncommutative multidimensional linear systems _ , siam j. control optim . 44 ( 2005 ) , no . 4 , 1474 - 1528 .
ball , g. groenewald and t. malakorn , _ conservative structured noncommutative multidimensional linear systems . _
the state space method : generalizations and applications ( ed .
d. alpay and i. gohberg ) , ot 161 , birkhuser ( 2006 ) , 179 - 223 .
j.a . ball , v. vinnikov , _ lax - phillips scattering and conservative linear systems : a cuntz - algebra multidimensional setting .
_ memoirs of the ams , vol . 178 ( 2005 ) , no . 837 .
v. p. belavkin , _ theory of the control of observable quantum systems .
control , 44 ( 1983 ) , 178 - 188 . v. p. belavkin ,
_ nondemolition stochastic calculus in fock space and nonlinear filtering and control in quantum systems .
_ proceedings xxiv karpacz winter school , r. guelerak and w. karwowski , eds .
, stochastic methods in mathematics and physics , world scientific , singapore ( 1988 ) , 310 - 324 . b.v.r .
bhat , k.r .
parthasarathy , _ kolmogorov s existence theorem for markov processes in c * algebras .
_ k. g. ramanathan memorial issue .
indian acad .
104 ( 1994 ) , no . 1 , 253 - 262 . b.v.r .
bhat , k.r .
parthasarathy , _ markov dilations of nonconservative dynamical semigroups and a quantum boundary theory .
h. poincar probab .
31 ( 1995 ) , no . 4 , 601 - 651 .
bhat , _ an index theory for quantum dynamical semigroups .
( 1996 ) , no . 2 , 561 - 583 .
bhat , _ cocycles of ccr flows . _ mem .
( 2001 ) , no .
l. bouten , r. van handel , m. james , _ an introduction to quantum filtering .
siam j. control and optimization 46 , no .
6 ( 2007 ) , 2199 - 2241 .
l. bouten , r. van handel , m. james , _ a discrete invitation to quantum filtering and feedback control .
_ siam review 51 , ( 2009 ) , 239 - 316 .
o. bratteli , p. jorgensen , a. kishimoto , r.werner , _ pure states on @xmath612 .
_ j.operator theory 43 ( 2000 ) , 97 - 143 .
s. dey , r. gohm , _ characteristic functions for ergodic tuples .
_ integral equations and operator theory 58 ( 2007 ) , 43 - 63 .
s. dey , r. gohm , _ characteristic functions of liftings . _
journal of operator theory * 65 * ( 2011 ) , 17 - 45 . s. dey , k. haria , _ generalized repeated interaction model and transfer functions _ , preprint arxiv:1112.4590 . c. foias , a.e .
frazho , _ the commutant lifting approach to interpolation problems . _ operator theory , advances and applications , vol .
44 , birkhuser , 1990 .
i. gohberg , p. lancaster , l. rodman , _ invariant subspaces of matrices with applications .
_ classics in applied mathematics , vol .
51 , siam , 2006 .
r. gohm , b. kmmerer , t. lang , _ noncommutative symbolic coding . _
ergodic theory and dynamical systems 26 ( 2006 ) , 1521 - 1548 .
r. gohm , _ noncommutative stationary processes . _ lecture notes in mathematics 1839 , springer , 2004 .
r. gohm , _ kmmerer - maassen scattering theory and entanglement .
_ infinite dimensional analysis , quantum probability and related topics , vol.7(2 ) , world scientific ( 2004 ) , 271 - 280 .
r. gohm , _ decompositions of beurling type for @xmath613-semigroups . _ banach center publications , vol . 73 , institute of mathematics , polish academy of sciences , warzawa ( 2006 ) , 167 - 176 .
r. gohm , _ noncommutative markov chains and multi - analytic operators . _ j. math .
364(1 ) ( 2009 ) , 275 - 288 .
r. gohm , _ transfer functions for pairs of wandering subspaces . _ preprint arxiv:1102.5519 , to appear in : iwota 2010 proceedings , operator theory : advances and applications , vol .
* 221 * ( 2012 ) , 389 - 402 .
j. gough , r. gohm , m.yanagisawa , _ linear quantum feedback networks .
a 78 , 062104 ( 2008 ) j. gough , m.r .
james , _ quantum feedback networks : hamiltonian formulation .
, volume 287 ( 2009 ) , no . 3 , 1109 - 1132 . j. gough , a. sobolev , _ stochastic schrdinger equations as limit of discrete filtering .
_ open systems & information dyn . 11 ( 2004 ) , 1 - 21 . j.w .
helton , _ discrete time systems , operator models and scattering theory .
_ j. functional analysis 16 ( 1974 ) , 15 - 38 . b. kmmerer , h. maassen , _ a scattering theory for markov chains . _ inf . dim .
analysis , quantum prob . and related topics , vol.*3 * ( 2000 ) , 161 - 176 . b. kmmerer , _ quantum markov processes .
_ a. buchleitner , k. hornberger ( eds . ) , coherent evolution in noisy environments , springer lnp 611 ( 2002 ) , 139 - 198
. m. laca , _ endomorphisms of @xmath42 and cuntz algebras . _ j. operator theory 30 , no.1 ( 1993 ) , 85 - 108 .
g. popescu , _ isometric dilations for infinite sequences of noncommuting operators . _
( 1989 ) , 523 - 536 .
g. popescu , _ characteristic functions for infinite sequences of noncommuting operators .
_ j. operator theory 22 ( 1989 ) , 51 - 71 .
g. popescu , _ on intertwining dilations for sequences of noncommuting operators . _ j. math .
appl . 167 (
1992 ) , no . 2 , 382 - 402 . g. popescu , _ multi - analytic operators on fock spaces .
( 1995 ) , no . 1 , 31 - 46 . | a noncommutative fornasini - marchesini system ( a multi - variable version of a linear system ) can be realized within a weak markov process ( a model for quantum evolution ) . for a discrete time
parameter this is worked out systematically as a theory of representations of structure maps of a system by a weak process .
we introduce subprocesses and quotient processes which can be described naturally by a suitable category of weak processes . a corresponding notion of cascade for processes
induces a represented cascade of systems .
we study the control theoretic notion of observability which turns out to be particularly interesting in connection with a cascade structure . as an application we gain new insights into stationary markov chains where observability for the system is closely related to asymptotic completeness in the scattering theory of the chain .
this motivates a general definition of asymptotic completeness in the category of weak processes . |
the first experimental observation of bose - einstein condensate ( bec ) @xcite in bose atom vapors have initiated an exciting field of research , both theoretically and experimentally .
one of the most interesting developments in this field is the formation of multi - component condensates .
multi - component becs have been observed experimentally by myatt _
_ @xcite and hall _
et al._@xcite for two different hyperfine spin sates of @xmath2 , by modugno _
@xcite for different atoms @xmath3 and @xmath4 , and papp _ et al . _
@xcite for different isotopes of the same atom @xmath5 and @xmath4 .
a rich variety of various interesting effects exhibited by these two - component becs have inspired a number of theoretical studies covering various aspects of a these systems @xcite .
the common feature of the bose systems in these experiments is that the intra - component and the inter - component boson - boson interactions are all repulsive .
it rises the curiosity about the properties of a multi - component condensate in which one kind of atoms have repulsive interactions while another kind of atoms have attractive interactions .
recently , fukuhara _
et al._@xcite observed bec of spin - zero @xmath6 isotopes by implementing an all - optical cooling protocol .
the bose - bose mixture in these experiments contain @xmath1 atoms having a positive s - wave scattering length and @xmath0 atoms having a negative s - wave scattering length@xcite .
the s - wave scattering length between @xmath1 and @xmath0 is also positive@xcite .
such a two - component condensate can be expected to show dynamical properties far more complex than a one - component condensate of attractively interacting bosons , which becomes unstable when the number of atoms exceed a critical value@xcite .
for a two - component bec with attractive interactions between bosons in one component and repulsive interactions in the second component , two most basic questions are of that of its stability and that of the collapse dynamics . in this paper
, we theoretically study the ground state as well as dynamic properties of such a two - component condensate in a three dimensional ( 3d ) anisotropic harmonic potential .
we specifically consider the @xmath1-@xmath0 bose - bose mixture in the anisotropic harmonic confining potential as in the experiment of fukuhara _
et al._@xcite .
the paper is arranged as follows . in sec .
ii , we describe the theoretical model for the study of a two - component bec . in sec .
iii , we discuss the ground state profile of a stable two - component bec . in sec .
iv , we present the collapse dynamics of the system and compare it with the experimental results@xcite .
the conclusions are given in sec .
the ground state and dynamic properties of a two component bose - einstein condensate ( bec)is well described by a set of coupled gross - pitaeveskii equations @xcite ( gpe ) given by @xmath7 where @xmath8 are indices for the two components @xmath9@xmath10 for @xmath1 and @xmath11 for @xmath0@xmath12 , @xmath13 , @xmath14 is the spatial coordinate vector , @xmath15 ) is the trapping potential , @xmath16 is the intra - species interaction and @xmath17 the inter - species interaction strength between atoms in the condensed state , @xmath18 is intra - species and @xmath19 is inter - species s - wave scattering length , @xmath20 is the reduced mass in which @xmath21 and @xmath22 are atomic masses .
the normalization condition for each component is @xmath23 where @xmath24 is number of atoms in each component .
we non - dimensionalize eq.@xmath25([nls_eqn ] ) through a set of linear transformations : @xmath26 . after dropping the wiggles on the symbols ,
we obtain @xmath27 where @xmath28 @xmath29 since the masses of the two isotopes are nearly equal , we have taken @xmath30 . in order to find a stationary solution of eq.@xmath25([nls_non_dim_eqn ] ) , we do a separation of variables @xmath31 $ ] , where @xmath32 is the chemical potential of the @xmath33th component . starting from eq.@xmath25([nls_non_dim_eqn ] ) , we obtain @xmath34
in this section , we discuss the ground state properties of a two - component bec of yb isotopes by numerically solving the coupled gpe ( eq.@xmath25[stationary_state_eqn ] ) .
the ground state solution of the gpe is found by the imaginary time propagation method . in this method , the time dependent gpe is evolved in imaginary time starting from an initial guess using a finite difference crank - nicholson ( fdcn ) scheme@xcite . in imaginary time propagation we have taken the space step as @xmath35 and the time step as @xmath36 .
we have used a set of parameters corresponding to the @xmath1-@xmath0 system in the experiment@xcite : @xmath37 , @xmath38 , @xmath39 , @xmath40 , @xmath41 , @xmath42 , @xmath43 . due to the attractive interaction between @xmath0 atoms , the condensate of @xmath0 undergo a collapse if the particle number exceeds a critical value @xmath44 @xcite . for a single component @xmath0 this value is given by @xmath45 , where @xmath46 and @xmath47 . for the parameters given above , @xmath48 .
but , the critical number is modified due to the presence of @xmath1 atoms with a positive scattering length .
the physical origin of this modification is in the effective potential felt by the bosons in the attractive component .
the interaction with the repulsive component changes the effective potential from @xmath49 to @xmath50 .
this reduces the effect of the attractive interaction and leads to a reduction of the critical number .
alternatively , one may say that the mean - field contribution from the repulsive component leads to a flattening of the effective potential felt by the attractive component .
if @xmath51 , then for the given parameters , @xmath44 is calculated to be @xmath52 by our numerical simulation . in order to prepare a stable condensate ,
we have taken @xmath51 and @xmath53 .
the ground sate profile of the two - component bec is presented in fig .
we observe that the @xmath0 atoms are at the center of trap and are surrounded by a large cloud of @xmath1 atoms .
in this section , we study the collapse dynamics of a two component bec composed of @xmath0 and @xmath1 atoms using the coupled time - dependent gpe s . due to the negative s - wave scattering length ,
the @xmath0 condensate becomes unstable if the number of atoms becomes greater than a critical value @xmath44 .
this condensate collapses by emitting atoms out of it .
due to high density of atoms in the attractive condensate , the loss of atoms from the condensate occurs through three - body collisions . to model this collapse we add a imaginary three body quintic loss term @xcite to the rhs of gpe eq.@xmath25([nls_eqn ] ) given by @xmath54 where @xmath55 is the three - body loss coefficient for each component .
we have neglected the two - body dipolar loss term as they make negligible contribution in this case@xcite .
we have also left out the loss terms for @xmath56 , @xmath57 , @xmath58 , and @xmath59 collisions , since the losses due to these are comparatively small @xcite .
so , eq.@xmath25([nls_non_dim_eqn ] ) becomes @xmath60 where @xmath61 we time evolve the coupled time - dependent gp equations eq.@xmath25([quintic gpe ] ) using the finite difference crank - nicholson ( fdcn ) scheme with a known initial condition . in real time propagation , we have taken the space step as @xmath35 and the time step as @xmath62 .
the time evolution of the number of @xmath0 and @xmath1 is shown in fig . 2 along with the experimental data of fukuhara _ et al._@xcite . considering the complex dynamics of the the two - component system with mixed interactions
, the theoretical results may be said to be in reasonable agreement with the experimental results .
we observe that there is a significant loss of @xmath0 atoms .
we also see that the decay of @xmath0 is very rapid .
it is due to the collapse of the @xmath0 condensate .
the number of @xmath1 atoms does show a very small decrease , which is not visible on the scale of this figure . to understand the details of the decay process , we study the condensate profiles of each component at different times . the results are shown in the fig . 3 for @xmath0 and in fig . 4 for @xmath1 .
at @xmath63 , the @xmath0 are at the center of the trap ( top left panel of fig .
3 ) surrounded by @xmath1 ( top left panel of fig .
4 ) . since the number of atoms in the attractive component is higher than the critical number for stability , the system is unstable .
when the system evolves in time , the attractive component explodes as is evident from the spreading of this component with time , in real space , as shown in fig .
the spiky structures in these figures represent the inhomogeneities produced due to the on - going explosion process .
as mentioned earlier , the explosion also leads to a spread of the ground state profile . due to the coupling between the attractive and repulsive components ,
the condensate of @xmath1 is also redistributed in real space during the time evolution , as shown in fig .
4 . the numbers of remaining atoms in each condensate component during the time evolution , for a longer period of time ,
is shown in fig .
we note that the best agreement with the experimental results ( fig .
2 ) is obtained for the measured values of the @xmath64 and @xmath65 .
the disagreement at later times is likely to be originating from the neglect of atomic loss due to collisions involving @xmath0 and @xmath1 . during the initial stages of the time evolution
, the bosons distribution gets heavily mixed due to the explosion and due to the coupling between the two components .
then , at later times , the inter - component collisions is likely to affect the atom loss .
we are unable to include these loss terms since their values are not known at present .
in this paper , we presented a study of some static and dynamic properties of a two - component bose condensate consisting of repulsively interacting @xmath1 atoms and attractively interacting @xmath0 atoms in an anisotropic harmonic confinement . in the stable state
, the ground state has @xmath0 atoms in the center of the trap surrounded by @xmath0 atoms . when the number of atoms in the attractive component exceed a critical value ,
the the system undergo a collapse .
we analyzed the time evolution of this collapse process for the specific system parameters of the @xmath66-@xmath1 system studied in the experiment of fukuhara _
et al._. the details of the collapse dynamics are found to be in reasonable agreement with experimental results .
the critical number for stability of the attractive condensate is reduced by it s interaction with the repulsive condensate .
gopesh kumar chaudhary thanks ugc , government of india , for financial support during this work .
gkc thanks prof .
g. v. shlyapnikov for a discussion during the icts international school on cold atoms and ions held recently in kolkata .
we thank prof .
j. k. bhattacharjee for a discussion during an serc school on nonlinear dynamics held recently in delhi .
99 m. h. anderson , j. r. ensher , m. r. matthews , c. e. weiman , and e. a. cornell , science * 269 * , 198 ( 1995 ) .
k. b. davis , m. -o .
mewes , m. r. andrews , n. j. van druten , d. s. durfee , d. m. kurn , and w. ketterle , phys .
lett . * 75 * , 3969 ( 1995 ) . c. c. bradley , c. a. sackett , j. j. tollett , and r. g. hulet , phys .
lett . * 75 * , 1687 ( 1995 ) . c. j. myatt , e. a. burt , r. w. ghrist , e. a. cornell , and c. e. wieman , phys .
* 78 , * 586 ( 1997 ) .
d. s. hall , m. r. matthews , j. r. ensher , c. e. wieman , and e. a. cornell , phys .
lett . * 81 * , 1539 ( 1998 ) .
g. modugno , m. modugno , f. riboli , g. roati , and m. inguscio , phys .
* 89 * , 190404 ( 2002 ) .
s. b. papp , j. m. pino , and c. e. wieman , phys .
* 101 * , 040402 ( 2008 ) .
ho and v. b. shenoy , phys . rev .
* 77 * , 3276 ( 1996 ) .
h. pu and n. p. bigelow , phys .
lett . * 80 * , 1130 ( 1998 ) .
m. trippenbach , k. goral , k. rzazewski , b. malomed , and y. b. band , j. phys .
b * 33 * , 4017 ( 2000 ) .
s. k. adhikari , phys .
a * 63 * , 043611 ( 2001 ) .
k. kasamatsu and m. tsubota , phys .
a * 74 * , 013617 ( 2006 ) .
k. kasamatsu and m. tsubota , j. low temp .
phys . * 150 * , 599 ( 2008 ) .
t. fukuhara , s. sugawa , y. takasu , and y. takahashi , phys .
a * 79 * , 021601(r ) ( 2009 ) .
m. kitagawa , k. enomoto , k. kasa , y. takahashi , r. ciurylo , p. naidon , and p. s. julienne , phys .
a * 77 * , 012719 ( 2008 ) .
p. a. ruprecht , m. j. holland , k. burnett , and m. edwards , phys .
a * 51 * , 4704 ( 1995 ) .
e. a. donley , n. r. claussen , s. l. cornish , j. l. roberts , e. a. cornell , and c. e. wieman , nature * 412 * , 295 ( 2001 ) .
e. p. gross , nuovo cimento * 20 * , 454 ( 1961 ) .
l. p. pitaevskii , zh .
40 * , 646 ( 1961 ) [ sov .
jetp * 13 * , 451 ( 1961 ) ] .
a. sinatra , p. o. fedichev , y. castin , j. dalibard , and g. v. shlyapnikov , phys .
lett . * 82 * , 251 ( 1999 ) . for a review , see : p. muruganandam and s. k. adhikari , comp .
. comm . * 180 * , 1888 ( 2009 ) .
kagan , b. v. svistunov , and g. v. shlyapnikov , pisma zh .
. fiz . * 42 * , 169 ( 1985 ) [ jetp lett . * 42 * , 209 ( 1985 ) ] .
kagan , a. e. muryshev , and g. v. shlyapnikov , phys .
lett . * 81 * , 933 ( 1998 ) .
h. saito and m. ueda , phys .
lett . * 86 * , 1406 ( 2001 ) .
r. a. duine and h. t. c. stoof , phys .
lett . * 86 * , 2204 ( 2001 ) . | in this paper , we present a theoretical study of a two - component bose - einstein condensate composed of ytterbium ( yb ) isotopes in a three dimensional anisotropic harmonic potential .
the condensate consists of a mixture of @xmath0 atoms which have a negative s - wave scattering length and @xmath1 atoms having a positive s - wave scattering length .
we study the ground state as well as dynamic properties of this two - component condensate . due to the attractive interactions between @xmath0 atoms ,
the condensate of @xmath0 undergo a collapse when the particle number exceed a critical value .
the critical number and the collapse dynamics are modified due to the presence of @xmath1 atoms .
we use coupled two - component gross - pitaevskii equations to study the collapse dynamics .
the theoretical results obtained are in reasonable agreement with the experimental results of fukuhara _ et al . _
[ pra*79 * , 021601(r ) ( 2009 ) ] . |
we ( @xcite , @xcite , @xcite , @xcite , @xcite ) model binaries in which the primary star ( bh progenitor ) evolves as a single star until its core begins he burning . when he burning begins in the core the stellar radius expands a further @xmath1 to @xmath2 relative to the red giant radius , it fills its roche lobe and begins to transfer mass to the secondary star , i.e. case c mass transfer . as a result , the orbit shrinks and the binary goes into a common envelope .
the h envelope of the primary is removed at the expense of orbital angular momentum .
eventually the orbit , originally @xmath3 , decays to a few @xmath4 .
the ratio of the stellar radius to orbital separation leads the primary to spin up by tidal synchronization .
hence , having a good estimate of the orbital period and , assuming angular momentum conservation during the collapse , one may obtain the natal kerr parameter of the bh ( see column 6 in tab .
[ tab - results ] and the left ( l ) plot in fig .
[ fig:1 ] ) .
the natal kerr parameter may be modified by two post - formation mechanisms .
the first one is by using the rotational energy to produce the grb / hn , e.g. through the bz mechanism @xcite , a grb / hn explosion ( see column 9 in [ tab - results ] ) .
a drastic decrease of the natal kerr parameter is not expected given that a small percentage of the available energy is enough to power the grb / hn .
further constraints on the energy of the explosion can be placed through the blaauw - boersma kick ( @xcite , @xcite ) and the fact that the binary remains bound ( and , usually , has small eccentricity ) .
the second mechanism is by accreting mass into the bh ( see the right ( r ) plot in fig .
[ fig:1 ] ) once the secondary , or its wind , fills its roche lobe .
measurements by @xcite coincide with the predictions of @xcite for the kerr parameters of gro j1655@xmath540 and 4u 1543@xmath547 .
our estimates are consistent with the observed kerr parameter in lmc x@xmath53 ( fig.3 in @xcite ) , and the measurements for the galactic grs 1915@xmath6105 @xcite , as well as the extragalactic lmc x@xmath51 @xcite and m33 x@xmath57 @xcite can be explained through a phase of mass transfer ( with wind mass transfer and hypercritical accretion for the last 2 , see @xcite and @xcite ) . | the collapsar model , in which a massive star ( @xmath0 ) fails to produce a sn and forms a bh , provides the main framework for understanding long gamma - ray bursts ( grb ) and the accompanying hypernovae ( hn ) . however , single massive - star models that explain the population of pulsars , predict cores that rotate too slowly to produce grbs / hne .
we present a model of binary evolution that allows the formation of kerr black holes ( bh ) where the spin of the bh can be estimated from the pre - collapse orbit , and use the blandford - znajek ( bz ) mechanism to estimate the available energy for a grb / hn .
a population synthesis study shows that this model can account for both , the long grb and the subluminous grb populations .
address = e - mail : [email protected] argelander - institut fr astronomie , bonn university , auf dem hgel 71 , 53121 bonn , germany |
a quantum system undergoing an adiabatic evolution accumulates a geometrical phase , which adds to the standard dynamical phase .
this phase , introduced by berry in his seminal paper @xcite , has become a central concept in our understanding of quantum mechanics .
the signature of this berry phase has been sought in a broad range of domains in physics , like in neutron beams @xcite , nmr measurements in molecules@xcite and solid - state superconducting qubits @xcite . in condensed matter
, the berry phase is intimately related to band theory .
it provides a deep understanding of , say , the topological nature of the quantum hall effect or some peculiar features of graphene @xcite . despite its fundamental nature ,
no direct experimental signatures have been reported , except through reinterpretation of anomalous hall - related effects @xcite .
one of the cornerstones of condensed - matter physics is the bloch theorem and the associated concept of band structure .
bloch s theorem states that , inside a crystal , the electronic wave function with momentum @xmath0 in a band @xmath1 takes the form @xmath2 where the plane wave part @xmath3 describes the long distance physics of the system while @xmath4 captures the short ( atomic ) distance physics .
when @xmath0 varies ( due to elastic collisions with defects , in the present study ) , the wave function picks up the so called _ band structure berry phase _ @xcite , @xmath5 .
this berry phase is therefore ubiquitous in condensed matter .
however , up to now , it has been revealed only indirectly .
two ingredients are indeed required to observe it : a quantum interferometer to probe any phase changes and an adjustable parameter to control the berry phase accumulated by the carrier wave function upon traversing closed paths on the fermi surface .
universal conductance fluctuations ( ucfs ) in disordered materials are a well - known manifestation of quantum interference effects .
these aperiodic but reproducible ucfs appear as an external magnetic field is varied @xcite and have been extensively used as a probe of the phase coherence in mesoscopic samples since the mid-80s .
the fluctuation pattern is a signature of quantum interference between different electronic trajectories in real space , reflecting the singular microscopic configuration of disorder for a given sample , yielding the so - called magneto - fingerprint " of that individual sample .
the mean value of the amplitude of these fluctuations is universal , and of the order of the quantum of conductance , @xmath6 .
hence , ucf experiments can be viewed as a poor man s quantum interferometer " picking up the long distance particular physics of the system .
we need now a tunable parameter to modify the berry phase .
as mentioned above , this phase is intrinsically related to the band structure , so the accumulated berry phase is related to the band configuration at the fermi surface . in ( ga , mn)as compounds the valence band is split due to the effective field arising from the exchange interaction between the hole spin and the polarized mn moments.@xcite because of the spin - orbit coupling , the band splitting is highly anisotropic with respect to the direction of magnetization .
therefore the magnetization , and more specifically its orientation , appears to be a relevant parameter to produce changes in the band structure , and hence in the associated berry phase acquired by the quasi - particle wave function .
phase - coherent transport has been investigated in ( ga , mn)as devices over the past few years : ucf@xcite and non - locality effects@xcite were reported .
also experimental evidence for weak localization@xcite , as well as theoretical studies@xcite , has been presented .
it was found that low energy spin wave modes hardly affect the phase coherence . as shown in ref .
[ ] , neither magnetization rotation nor magnetic texture destroy coherence in those systems .
more strikingly , very unusual patterns , consisting of fast fluctuations of the conductance , were observed at low fields in addition to the conventional ucfs@xcite .
this unexpected regime appeared as the magnetic field was swept perpendicular to the easy axis of magnetization , below the anisotropy field .
such behavior was conjectured to be related to the presence of magnetic domain walls ( dws ) , yielding to an additional phase term connected to the magnetic disorder .
however , in those experiments , the knowledge of the sample magnetic configuration was not accessible during the transport measurements , and no conclusive argument could be provided .
more recently , hals and co - workers@xcite proposed a new physical mechanism accounting for the existence of this regime
. the fast oscillations would be the berry phase fingerprint of effective magnetic monopoles in momentum space , which arise from energy - band crossings@xcite .
when a quasi - particle wave function traverses a closed loop in momentum space , it accumulates a geometrical phase from the monopole field . in ( ga , mn)as the position of these energy - crossings , or monopoles , depends on the magnetization orientation , due to the strong spin - orbit coupling
hence , magnetization reorientation leads to a relocation of the monopoles in @xmath7-space , which in turn yields a berry phase change .
these conflicting interpretations called for further investigations .
to that end , we designed a new sample geometry so that the magnetic state could be well established .
ucf experiments were performed for different magnetic field orientations and temperatures . as we will show later
, our results indicate that the dw contribution can be clearly discarded as being the origin of these fast fluctuations , and point toward a fingerprint of the band structure berry phase .
our experiments were performed on a ferromagnetic semiconductor thin film .
a 50 nm thick ( ga@xmath8mn@xmath9)(as@xmath10p@xmath11 ) epilayer was grown at 250 @xmath12c by molecular beam epitaxy on a ( 001 ) gaas substrate .
doping ( gamn)as with phosphorus produces a variation of the cell parameter of the compound and , in turn , the substrate induced strains are modified .
this allows us to tune the magnetic anisotropy of ( gamn)(asp ) epitaxial layers deposited directly on gaas substrates by changing only the phosphorus concentration@xcite .
the film used in this work presents an out - of - plane easy axis of magnetization .
after conventional annealing treatment , the curie temperature was @xmath13 113 k and the sheet resistance @xmath14 k@xmath15 .
this film exhibits a low density of dw pinning centers.@xcite submicrometric hall bars were defined on the wafers using electron beam lithography and ion beam etching , with the geometry shown in fig .
[ fig.1 ] . , @xmath16 .
* b * denotes the applied magnetic field , * m * denotes magnetization and @xmath17 the magnetization angle from the easy axis [ 001 ] . in the present work , @xmath16
is fixed to @xmath18 .
( b ) scanning electron microscope image of a hall bar indicating the numbering of voltage contacts and the direction of electric current @xmath19 .
( c)-(e ) the voltage at each contact is indicated for different magnetization configurations .
@xmath20 is the hall voltage , @xmath21 is the longitudinal voltage drop and @xmath22 is an offset voltage referred to ground .
( c ) the magnetization is homogeneous throughout the sample .
( d)-(e ) there is a dw in the constriction , with the magnetization presenting opposite sign at both sides . ]
the distance between the voltage contacts for longitudinal conductance measurements is @xmath23 440 nm , comparable to the phase coherence length measured at low temperature for similar compounds in previous works ( @xmath24 nm at @xmath25 mk)@xcite .
an @xmath26100 nm wide constriction in the middle of the bar acts as an efficient pinning center for dws ; this constriction was designed to trap domain walls at will and to probe their influence on the dynamical phase .
finally , ti / au ( 20/200 nm ) ohmic contacts were thermally evaporated .
the double hall bar geometry ( fig . [ fig.1 ] ) allows for detection of a domain wall ( dw ) trapped in the constriction by comparing the hall voltages on both sides of the constriction .
indeed , in ( ga , mn)(as , p ) , the hall voltage ( @xmath20 ) is completely dominated by the anomalous hall effect , which is proportional to the local magnetization component perpendicular to the sample plane .
therefore , @xmath20(@xmath27 ) reproduces the magnetization loops .
the presence of a dw at or close to the constriction would result in hall voltages with opposite sign on both sides of the constriction . if there are no domain walls , meaning that the magnetization is homogeneous throughout all the sample , the hall voltage on the left should be the same as that on the right , that is , @xmath28 ( see fig . [
fig.1](b ) for the labeling of electrical contacts ) . in this case , the longitudinal voltage will be @xmath29 , with @xmath30 being the resistance between contacts @xmath31 and @xmath32 ( see fig . [
fig.1](c ) ) .
on the contrary , if there is a dw trapped in the constriction , the magnetization will have opposite sign at both sides . in this case ,
@xmath33 , and the longitudinal voltage will be @xmath34 , with the plus or minus sign depending on the magnetization configuration corresponding to fig . [
fig.1](d ) or ( e ) . figure [ fig.2 ] shows the effect of a dw pinned in the constriction , on both the hall ( fig . [ fig.2](a ) and [ fig.2](b ) ) and on the longitudinal ( fig . [ fig.2](c ) and [ fig.2](d ) ) measured voltages . when the magnetic field is swept from the positive to the negative saturation value ,
simultaneous magnetization reversal occurs on both sides of the constriction , yielding to the reversal of both @xmath20 curves for the same negative coercive field .
nucleation and propagation of domain walls are the well - established reversal mechanism in these materials@xcite , so dws are present in this process . however , since the nucleation energy is higher than the depinning one , no dws are trapped in the constriction .
thus , once dws are created from a saturated state , they have enough energy to travel across the constriction .
if the magnetic field is swept to high negative values , the magnetization will be saturated and the same effect will be observed when sweeping back to positive fields , with the reversal on both sides of the constriction occurring at the same positive field value .
however , a different effect occurs if the field is not swept beyond @xmath35 mt , as shown in fig .
[ fig.2 ] . in this case
, the magnetization is not saturated and dws must still be present in the sample , although far away from the voltage contacts .
then , when sweeping the field back to positive values , the first coercive field is observed when one of the @xmath20 signals changes sign ( @xmath36 and @xmath37 in figs .
[ fig.2](a ) and [ fig.2](b ) respectively ) , meaning that a dw has propagated and gotten trapped in the constriction . by sweeping further the magnetic field
, there is a field range where the dw remains trapped , and then the second @xmath20 curve reverses .
the left and right panels in fig .
[ fig.2 ] correspond to the opposite magnetization configurations depicted in figs .
[ fig.1](d ) and ( e ) . furthermore , not only does @xmath20 indicates the presence of a dw , but @xmath38 is also sensitive to this effect , as shown in the bottom panels of fig .
[ fig.2 ] . comparing the top and bottom panels ,
the relationship @xmath34 can be deduced , where the minus sign corresponds to figs .
[ fig.2](a ) and ( c ) , and the plus sign corresponds to figs.[fig.2](b ) and ( d ) .
we have proved here that , despite the lack of control on the magnetic configuration ( that is , we can not choose which configuration to stabilize ) , we can pin and detect a dw , and also distinguish between the two possible configurations .
similar experiments were performed in the mesoscopic regime , i.e. , at temperatures lower than 1 k. in this case , the magnetization reversal is clearly observed in @xmath20 , although conductance fluctuations may veil the associated jumps in @xmath38 . after validating the detection procedure of dws ,
this method was used to study the magnetization configuration with the magnetic field applied at different angles @xmath39 from the easy axis .
a thorough analysis of the magnetization reversal processes in this hall bar was carried out prior to the ucf experiments : we observed that after saturating the magnetization at b = 1 t , the field can be swept back and forth without any dw being created , as long as the field is always kept positive .
this allowed us to work in a single magnetic domain configuration while applying the magnetic field at any angle @xmath40 , which was important in order to get rid of any domain wall contribution to the magnetotransport experiments .
electronic transport experiments were performed in a dilution refrigerator with a base temperature of 40 mk using a standard four - probe lock - in technique .
variable magnetic field was provided by a 3-axis superconducting magnet system , with a high field up to 7 t for the principal axis @xmath41 and a 1 t vector field using any combination of @xmath42- , @xmath43- and @xmath41-axis coils .
since mesoscopic transport had never been measured in this phosphated compound , we first verified the appearance of ucfs . in fig .
[ fig.3](a ) , magneto - conductance curves measured at different temperatures are presented.@xcite the magnetic field was applied perpendicular to the sample s plane , i.e. , along the easy axis of magnetization . in this configuration , with the magnetic field applied along the easy axis , the magnetization remains collinear to the field direction , even at low field .
hence , the band structure remains mostly unchanged over the full field excursion , except for a weak contribution of the band zeeman effect .
the amplitude of the fluctuations decreases with increasing temperature .
normalized to the distance between contacts , _
l_. red symbols are data obtained in this work for gamnasp , and black symbols are data presented in ref .
[ ] , different symbol shapes corresponding to samples of different sizes . ]
the behavior of @xmath44 vs. @xmath45 is shown in fig . [
fig.3](b ) .
here , @xmath46 is quantified as the rms value @xmath47 ^ 2\right>}$ ] and @xmath48 is the distance between contacts .
the obtained values and the @xmath49 dependence , as well as the values of the correlation fields , are in coincidence with those obtained previously in similar mesoscopic devices but made on gamnas layers@xcite ( shown in fig .
[ fig.3](b ) ) , meaning that the intrinsic transport properties are roughly the same , despite the higher concentration of impurities in the present sample .
to probe the effect of magnetization reorientation on the conductance fluctuations , the magnetic field has to be oriented away from the easy axis .
figure [ fig.4 ] shows @xmath50 measured at constant temperature , with the magnetic field applied at different angles @xmath51 .
the field was swept between 0 and 1 t. mk with the magnetic field applied at different angles @xmath39 from the easy axis of magnetization .
the curves for @xmath52 are vertically shifted for clarity . ] as mentioned above , in these experiments the magnetization remains in a single domain configuration . hence ,
during a field sweep with @xmath53 , the magnetization direction ( @xmath17 ) changes with the magnetic field strength : at low field , the magnetization rotates coherently toward the field direction . after reaching the saturation field ,
the magnetization remains collinear to the field direction . above 0.2 t , where the magnetization is aligned with the field ,
the conductance exhibits slow fluctuations similar to those in fig .
[ fig.1 ] , ascribed to the usual ucf . below 0.2 t , a new regime develops as the field is rotated away from the easy axis : fast fluctuations are visible . as @xmath39 increases , the fast fluctuations develop within a wider field range .
this broadening is consistent with the increasing saturation field , as the magnetic field is tilted away from the easy axis .
this first series of experiments clearly confirms that the reorientation of the magnetization produces an additional term in the phase accumulated by the carriers over their trajectories .
_ et al.__@xcite attributed these fast fluctuations to the signature of a berry phase , and we will give further evidence supporting this interpretation , by studying the temperature dependence of fluctuations in both regimes .
let us briefly summarize what is observed in the case of conventional ucfs . as stated before , a ucf fingerprint can be viewed as a poor man s interferometer " , where the randomness of the fingerprint comes from the lack of control on the different paths that are actually interfering .
the relevant length scale is given by the phase coherence length of the charge carriers , @xmath54 , which sets the maximum size of the loops that can contribute to the interference pattern .
when @xmath54 is larger than the system size , the ucf amplitude is of the order of @xmath6 .
but as the temperature increases , @xmath54 gets smaller due to decoherence effects , leading to a reduction of the ucf amplitude as shown in fig .
[ fig.3 ] .
more interesting for us is the typical quasi - period of the @xmath50 oscillations , which is related to the dynamical phase accumulated in a loop of size @xmath54 , which is given by @xmath55 . as the temperature increases ,
the fluctuations become slower ( i.e. , the quasi - period increases ) since a larger field is needed to put one quantum of flux inside a smaller loop . on the contrary , the band structure berry phase @xmath56 depends only on the direction of the magnetization and is independent of @xmath54 .
this is because the relevant trajectory is described in @xmath7 space where the size of the loop , given by @xmath57 , remains nearly unchanged .
a simple corollary is that the quasi - period of the oscillations due to a change of the berry phase is independent of temperature . .
the curves for @xmath58 mk are vertically shifted for clarity .
the vertical gray lines are a guide to the eye , indicating the position of consecutive maxima of @xmath59 . ] to test whether or not these fast fluctuations are a fingerprint of a berry phase , we performed a series of experiments at different temperatures .
figure [ fig.5 ] shows @xmath50 curves measured at different temperatures and fixed magnetic field angle @xmath60 , for a narrow field window ( @xmath61 t ) .
again , both regimes can be clearly identified , for all the temperatures up to 1 k. the fast fluctuations pattern in the low field region , persists as the temperature increases , except for a damping of the amplitude .
moreover , the peaks occur at the same magnetic field values , whatever the temperature . in this regime ,
the oscillation pseudo - period is therefore temperature independent , contrary to what is observed in the high field ucf regime .
a better insight into the temperature dependence of both regimes is given by the comparison of two relevant parameters : the fluctuation pseudo - period @xmath62 and the rms amplitude of the fluctuations @xmath46 .
in the high - field regime , the correlation field @xmath62 was estimated by calculating the autocorrelation function of @xmath50 from measurements at @xmath63 between 0 and 7 t ( like those of fig . [ fig.3](a ) ) . in the low - field regime , the analysis of the curves in fig .
[ fig.5 ] ( @xmath60 ) was restricted to the field range 0 - 200 mt .
in such a limited range , the measured curve displays only a few peaks and valleys and the statistics is not good enough for the autocorrelation function to be calculated ; the pseudo - period was thus estimated as the mean distance between consecutive minima in each curve .
the latter method was also applied in the high - field regime , in order to check its validity , giving roughly the same temperature dependence as that of the calculated autocorrelation function .
the rms amplitude of the conductance fluctuations was calculated as @xmath64 for both regimes .
the results are summarized in figure [ fig.6 ] . and ( b ) the amplitude @xmath46 of the fluctuations in the high - field ( @xmath65 ) and low - field ( @xmath66 ) regimes .
@xmath62 of the low - field regime is multiplied by 10 . ]
the evolution of @xmath62 with temperature ( fig . [ fig.6](a ) ) shows clear evidence that the two oscillating regimes have different origins . indeed , at high fields , the characteristic pseudo - period increases with increasing @xmath45 , as expected for conventional ucfs .
in this case , the accumulated phase is proportional to the magnetic field flux enclosed by the loops contributing to the interferences . as mentioned above
, increasing temperature reduces the loop characteristic size and , in turn , the magnetic field flux ( i.e. , the accumulated phase ) for a given applied field . a larger field variation is therefore needed to accumulate a @xmath67 phase and complete a quasi - period , causing @xmath62 to increase with temperature . on the contrary , as captured in fig .
[ fig.6](a ) , in the fast fluctuations regime @xmath62 is one order of magnitude smaller and is insensitive to the temperature .
we argue that this temperature independence is consistent with a geometrical berry phase .
this phase is accumulated by the carriers as they travel around a closed loop in @xmath7-space enclosing an effective magnetic monopole .
the loop size is roughly defined by @xmath57 and is therefore temperature independent .
this is in sharp contrast to the dynamical phase accumulated during the transport over a loop in _ real _ space , whose size is given by the temperature dependent coherence length @xmath54 .
figure [ fig.6](b ) shows the temperature dependence of @xmath46 for the high- and low - field regimes .
the variation is similar for both regimes , with @xmath46 decreasing with temperature in either case .
this behavior is expected in the high - field regime , where only conventional ucf are visible .
@xmath46 depends on the size of the loops in real space producing interference . again
, these loops have a typical length given by @xmath54 , which decreases with increasing temperature.@xcite this leads to vanishing interference , and hence vanishing oscillations , above 1 k.@xcite in the low - field regime , the fast fluctuations also disappear at high temperature , even though these fluctuations are associated with a geometrical phase .
this is related to the sensibility of our ucf interferometer . in our experiments
, the berry phase term is probed through the interference patterns produced by the wave functions of the quasi - particles describing different _ real _ space paths .
the visibility of the interferences is thus given by the coherence length @xmath54 .
hence , the fast - fluctuation regime is also lost at high temperature , above 1 k.
to conclude , we found that gamnas is a unique material for addressing berry phase physics in condensed matter : the combination of ferromagnetism , strong spin - orbit coupling and semi - conducting like nature provides a direct way to control its band structure ( hence its berry phase ) by controlling the magnetization of the device .
our measurements give clear indications that the fast - fluctuation regime observed at low magnetic fields is the fingerprint of the change in the band - structure berry phase , yielding to a direct experimental evidence of this fundamental mechanism in solid state matter .
the next step will be to measure directly the phase itself .
such an experiment will require having good control of the interfering trajectories by using a real interferometer such as an aharonov - bohm ring .
using this approach in such a class of materials will provide a direct way to explore one of the most fundamental and elusive concepts of condensed - matter physics .
a. bohm , a. mostafazadeh , h. koizumi , q. niu , and j. zwanziger , in _ the geometric phase in quantum systems : foundations , mathematical concepts , and applications in molecular and condensed matter physics _
( springer verlag , berlin , 2003 ) .
the cryostat base temperature was kept constant and the experiments were carried out with different applied currents , ranging between 1 na and 500 na and giving rise to different effective temperatures , as discussed in ref . [ 16 ] . the relationship between the electronic temperature and the applied current was accurately checked by comparing the conductance of the sample with calibration experiments at low current . | we report on experimental evidence of the berry phase accumulated by the charge carrier wave function in single - domain nanowires made from a ( ga , mn)(as , p ) diluted ferromagnetic semiconductor layer .
its signature on the mesoscopic transport measurements is revealed as unusual patterns in the magnetoconductance , that are clearly distinguished from the universal conductance fluctuations .
we show that these patterns appear in a magnetic field region where the magnetization rotates coherently and are related to a change in the band - structure berry phase as the magnetization direction changes .
they should be thus considered as a band structure berry phase fingerprint of the effective magnetic monopoles in the momentum space .
we argue that this is an efficient method to vary the band structure in a controlled way and to probe it directly .
hence , ( ga , mn)as appears to be a very interesting test bench for new concepts based on this geometrical phase . |
one of the major unsolved problems in astrophysics today is the nature of the connection between the formation and evolution of galaxies and supermassive black holes ( smbhs ) .
current models invoke a close interplay ( feedback " ) between smbhs and their host galaxies in order to arrest the growth of the most massive galaxies , which are otherwise over - predicted by models of structure formation ( e.g. , * ? ? ?
* ; * ? ? ?
observations of high - redshift galaxies paint a somewhat confusing picture of the relationship between star formation and black hole growth ( e.g. , * ? ? ?
the best - studied population of high - redshift star - forming galaxies are the lyman - break galaxies ( lbgs ) and closely related types of uv - selected galaxies .
they represent a major phase in the early stages of galaxy formation and evolution @xcite .
they are a primary tracer of high redshift star formation , as they are easily found at @xmath9 in deep pencil beam optical surveys ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the lbgs are believed to be the precursors of present day galaxies undergoing a phase of intense star formation and merging , and a large fraction of lbgs may merge to form early - type galaxies that are situated in present - day groups and clusters ( e.g. , * ? ? ?
most of the lbgs do not harbor an active galactic nucleus ( agn ) luminous enough to be individually detected in the deepest x - ray surveys .
however , a subset ( @xmath23% ) of lbgs with actively accreting smbhs has been identified @xcite .
these agns typically have x - ray luminosities of @xmath10 erg s@xmath1 in the 2 - 10 kev energy range , and often show signs of significant obscuration .
detailed studies of what might trigger both the agns and the starbursts in these forming galaxies are difficult since their distances render them small and faint .
the galaxy evolution explorer ( _ galex _ ) mission is conducting surveys in the far- and near - uv ( @xmath11 1600 and 2300 respectively ) and has been used to identify a rare population of low - redshift ( @xmath120.1 - 0.3 ) galaxies that are remarkably similar to high - redshift lbgs in most physical properties @xcite .
this population was first identified by @xcite and @xcite , based on cross - matching the _ galex _ and sloan digital sky survey data .
these objects were defined by two criteria designed to match the typical lbgs : 1 ) a minimum luminosity in the far - uv ( fuv ) band @xmath13 ; 2 ) an fuv effective surface brightness @xmath14 , where @xmath15 is the mean fuv surface brightness interior to the sdss @xmath16 half - light radius .
these local `` lyman break analogs '' ( lba ) are similar to distant lbgs in size , surface brightness , metallicity , dust , star formation rate ( sfr ) , stellar mass , and gas velocity dispersion .
there are other local star forming galaxies , e.g. , ultra - luminous infrared galaxies ( ulirgs ) and blue compact dwarfs ( bcds ) , which have some similarities with lbgs .
however , ulirgs are much more heavily obscured by dust , and bcds are much smaller in mass and luminosity ( star - formation rate ) compared with lbgs .
these differences make them poorer local analogs to lbgs .
however , it is intriguing that an actively accreting smbh has recently been reported in a nearby dwarf starburst galaxy @xcite .
interestingly , about 20% of the lbas have optical emission - line spectra that are intermediate between those of pure starbursts and those of type 2 ( obscured ) agn . as discussed by @xcite , the optical spectra alone
do not unambiguously establish the presence of a type 2 agn
. they may instead be extreme starburst systems in which the physical conditions in the ionized gas are different from typical local starbursts .
@xcite also presented imaging with the _ hubble space telescope _ that showed that most of these possible starburst - agn composites contain extremely massive ( one to few - billion m@xmath8 ) and compact ( radius @xmath17 pc ) starbursts . these dominant compact objects ( dcos )
would appear to be ideal sites for the formation and/or rapid growth of a smbh .
it has been suggested that the seeds of present - day smbhs may be present at the centers of the massive , compact stellar clumps seen in clumpy galaxies at high redshift ( e.g. * ? ? ?
if these lbas indeed host an agn , they would offer the opportunity to study the relationship between the formation of galaxies and supermassive black holes in detail in objects that appear to be excellent local analogs to a major population of high - redshift galaxies .
unfortunately , the optical spectra alone do not unambiguously establish that these lbas are indeed composite starburst / obscured - agn systems ( as opposed to starbursts with extreme properties ) @xcite .
observations in the hard x - ray band can potentially offer us important new diagnostics of the presence of an obscured agn .
in this paper we report the results of such an investigation .
we describe our data in [ data ] and present the spectral analysis in [ results ] .
the results are discussed in
[ discussion ] , and we come to some brief conclusions in [ conclusion ] .
to select our sample for x - ray observations we begin with the 31 lbas with _
hst _ imaging @xcite and _ spitzer _ ( irac+mips ) photometry @xcite . in fig .
[ f : bpt ] we show the standard
bpt " diagram @xcite used to classify the optical emission - line spectra of galaxies , where the emission line ratios are from table 2 in @xcite .
we plot all the galaxies in the _ galex_-sdss cross - matched sample and then highlight the 31 lbas in our sample ( large symbols ) .
we note that seven ) in our lba - composite " sample , although it lies close to the lower left of the solid line in the bpt diagram and does not harbor a dco . ] of the lbas lie in the composite region in the bpt " diagram , which is known to harbor composite objects " , which are typically ( but not necessarily ) objects having both star - formation and a type 2 agn ( see * ? ? ?
* ; * ? ? ?
hereafter , we refer to this sample as lba - composites " . in this paper , a flat @xmath18 cold dark matter ( @xmath18cdm ) cosmology with @xmath19 , @xmath20 is assumed . _
_ x__mm - newton observations were performed for 6 of the 7 lba - composites .
sdss j0054 + 1554 was given a c " priority and not observed ( marked as green asterisk in fig .
[ f : bpt ] ) .
a summary of observations is shown in table [ t : summary ] .
the reduction of the _ _ x__mm - newton data was done by ` xassist ( ver .
0.97 ) ` , which is a software for automatic analysis of x - ray data .
` xassist ` runs the _
_ x__mm - newton science analysis system ( ` sas ` ) packages automatically to filter the data , generate the light curves and extract the spectra of sources in the field , as well as compute the associated redistribution matrix files ( rmf ) and ancillary region files ( arf ) .
the lba - composites are weak in x - ray emission , and their optical positions were used to define the source regions in x - ray data reduction . ` xassist ` determines the source extraction sizes for source by fitting an elliptical gaussian model to the source image , which in the case of on - axis _ _ x__mm observations typically results in regions of size 20 - 25 .
background regions were chosen as annuli centered on the sources to extract the background spectra .
table [ t : cstat ] lists the pn and mos counts for each source .
the optical monitor ( om ) aboard on _ _ x__mm - newton allows us to make simultaneous observations in optical / uv and x - ray bands .
we processed the om data with the ` omichain ` package in ` sas ` to obtain uv count rates and the corresponding magnitudes of each object , as well as their associated uncertainties .
we obtained exposures in the u ( 344 nm ) , uvw1 ( 291 nm ) , uvm2(231 nm ) and uvw2 ( 212 nm ) band filters , but only the uvw1 was used for all six lbas .
we list the uv detections with each filter and the luminosity measured in uvw1 in table [ t : uv ] . the conversions from uv count rates to specific fluxes ( in units of erg / s/@xmath21/ )
were calculated based on the white dwarf standard stars .
table [ t : loiii ] lists the observed and extinction - corrected [ o iii]@xmath35007 line luminosity ( @xmath22,obs}}$ ] and @xmath22,corr}}$ ] , respectively ) for each lba - composite .
we obtain @xmath22,obs}}$ ] from the sdss data release 7 mpa / jhu catalog .
we derive @xmath22,corr}}$ ] from @xmath22,obs}}$ ] using the observed h@xmath23/h@xmath24 ratio , an intrinsic ( h@xmath23/h@xmath24)@xmath25 ratio of 2.9 , and the standard extinction curve for galactic dust @xcite . in table
[ t : loiii ] , we also list the mid - infrared ( mir ) and far - infrared ( fir ) luminosity , @xmath26 and @xmath27 , respectively @xcite .
x - ray spectral fitting was performed with ` xspec ` ( ver .
x - ray photons are collected by three detectors on _ _ x__mm - newton , i.e. pn , mos1 and mos2 .
the data of three detectors can be fitted simultaneously in ` xspec ` by tying all parameters together except a multiplicative constant factor .
we froze this factor at 1 for pn , and allowed it to vary between 0.8 to 1.2 for mos1 and mos2 . since our targets have limited counting statistics ( see table [ t : cstat ] ) , we first fit their spectra using the @xmath28-statistic @xcite , which is often used in this low counts regime .
the spectra were grouped to 1 count per bin , to improve the performance of ` xspec ` @xcite .
the simplest spectral model often fit to agn is an absorbed power - law model .
the x - ray photons suffer from two sources of absorption .
the first is due to the neutral hydrogen in our own galaxy , and its column density ( @xmath29 ) is listed in table [ t : summary ] . the other is the intrinsic obscuration from the gas at the redshift of the source ( @xmath30 ) .
@xmath30 is a free parameter in the spectral fitting as well as the photon index @xmath31 of the power law .
however , due to the low signal - to - noise ratio ( the number of photons detected by the pn and mos detectors are listed in table [ t : cstat ] ) , we were not able to constrain both @xmath30 and @xmath31 simultaneously in several cases . therefore for sources with low s / n , we alternately froze @xmath30 and @xmath31 and let the other parameter vary .
the same methodology was applied when adding an additional thermal component into the power - law continuum .
as mentioned above we fitted the spectrum in two ways to derive the power - law photon index and the intrinsic absorption : first , we freed the power - law photon index and assumed no absorption at the redshift of the source ( @xmath32 @xmath33 ) and that the continuum power - law spectra is only absorbed due to the galactic column density ( listed in table [ t : summary ] ) . again in many cases there was not sufficient signal for the absorption and slope to both be free parameters .
we list the photon indices derived from this simple absorbed power - law model in column 3 in table [ t : cstat ] , as well as the corresponding @xmath28-statistic in column 4 .
second , the spectra were fitted with the power - law model this time with the intrinsic absorption as a free parameter while the photon index was fixed as @xmath34 , which is a mean value for x - ray detected lbgs @xcite and is also a typical spectrum for an unobscured agn .
the values of @xmath30 are shown in column 5 in table [ t : cstat ] .
the errors of photon indices in column 3 and the upper limits of intrinsic absorptions in column 5 were calculated at a confidence level of 90% ( i.e. , single interesting parameter @xmath35=2.7 ) .
we also list in table [ t : cstat ] their derived luminosities in soft x - ray ( 0.5 - 2.0 kev ) and hard x - ray ( 2 - 10 kev ) bands ( column 7 and 8 , respectively ) , and the ratio of hard x - ray and [ o iii ] luminosities ( column 9 ) .
the associated uncertainties for luminosity correspond to the 90% confidence range of normalization of the continuum power - law component in spectral fit .
[ f : cstat ] for the spectral fits of each source with a fixed photon index @xmath34 .
we also fitted the x - ray spectra of our targets using the @xmath36-statistic to see how different the fits are compared with those by @xmath28-statistic .
the fits were performed in the same way as those using the @xmath28-statistic above .
the spectra were grouped in 10 counts per bin for both pn and mos detectors , and the spectra of mos1 and mos2 were combined in the fits .
table [ t : chi ] lists the photon index and intrinsic column density of each source derived while the other one is frozen in the spectral fitting .
sdss j0802 + 3915 is excluded due to insufficient bins for fitting after grouping into 10 counts per bin . also , spectral plots are shown in fig .
[ f : chi ] , which are fits with photon indices fixed at 1.9 .
fitting the spectra using @xmath36 does not make any improvements or significant difference in estimating the absorption and photon index compared to the @xmath28-statistic method .
thus , we use the x - ray luminosity derived from the @xmath28-statistic fit in the following discussion . additionally , as sdss j0808 + 3948 , sdss j1434 + 0207 and sdss j0922 + 4509 have relatively more photons detected , we also fitted their spectra by thawing both the parameters of photon index and intrinsic absorption ( using both the @xmath28 and @xmath36 statistics ) .
table [ t : both ] shows the values of these parameters and their associated errors or upper limits . since we have two interesting parameters in the fits ,
the errors at the 90 per cent confidence level are calculated according to @xmath37 or @xmath35=4.6 .
the lbas are the sites of strong starbursts , so we would expect to see thermal x - ray emission from the hot gas that is a characteristic features of such systems ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we therefore fit models in which we added a thermal component to the absorbed power - law model aforementioned .
we used the thermal plasma model ` apec ` in ` xspec ` instead of the older model,`vmekal ` , used in @xcite .
the ` apec ` model has three free parameters : the plasma temperature , the metal abundance ( in solar units ) and the normalization . due to the low number of photon counts , we fixed the abundance at the solar value .
the parameter of temperature and normalization were free in fitting .
however , due to insufficient photon counts , we fixed the temperature at 0.5 kev for sdss j0808 + 3948 , sdss j0922 + 4509 and sdss j0802 + 3915 which is a typical value for starburst galaxies ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
here we kept the photon index fixed at 1.9 and allowed the intrinsic absorption at the redshift of the lba to vary .
the results are shown in table [ t : apec ] , which lists the temperature of plasma , the column density , and the derived luminosities of both the thermal and power - law component , as well as the @xmath28-stat for comparison with the single power - law fitting .
the errors for parameters were calculated according to @xmath38 , and the uncertainty of luminosity was proportional to the corresponding error of normalization of each component .
for the three lbas where the temperature was permitted to vary the temperature ranged between 0.3 and 0.6 kev .
the thermal component luminosities were on the order of @xmath39 erg s@xmath1 , with @xmath40 erg s@xmath1 for sdss j0808 + 3948 and about @xmath41 erg s@xmath1 for sdss j0802 + 3915 .
the power - law component luminosity in the hard x - ray range ( 2 - 10 kev ) is within a factor of 2 of those derived from the single power - law model spectral fit mentioned above .
this power - law plus thermal model fit makes no statistically significant difference in the derived column density compared to that of the single power - law model . since the signal - to - noise of the _
_ x__mm - newton spectra is low , we attempted to constrain any thermal component by simultaneously fitting the spectra of all lbas together assuming that they have approximately the same hot gas temperature , i.e. , we tied the temperature parameters together between the fits but let the ` apec ` normalization vary for each lba .
we also fixed the power - law index at 1.9 , and tied the normalization of power - law component in our fit ( although note that overall normalization for each lba was allowed to be free ) .
the temperature given by this joint fit is 0.56 kev , and its 90% confidence level is [email protected] kev .
we list the intrinsic column density , the thermal luminosity and the soft and hard x - ray luminosity contributed by the power - law component in table [ t : apecjoint ] .
we see that the column densities and thermal luminosities given by this joint fit are consistent with those in the individual fits , but with improved constraints . in fig .
[ f : lxlapec ] , we compared the two compositions of soft x - ray luminosity , i.e. , the thermal and power - law luminosities in 0.5 - 2.0 kev , where the dashed line indicates where both values equate .
this comparison was done separately for both the cases in which the spectra were fit individually and those that were fit simultaneously .
these are shown in the left and right panel of fig .
[ f : lxlapec ] respectively .
the results derived from the individual spectra fits imply that these two mechanisms play a nearly equal role in soft x - ray emission , except in j0808 + 3948 where the power - law soft x - ray luminosity is about one dex higher than that of thermal emission . for the joint spectral fit ,
the luminosities from the two emission mechanisms are comparable , within the range of 1 dex .
the obscuration of sdss j0214 + 1259 is compton - thick in the joint fit , and the power - law luminosity at 0.5 - 2 kev is negligible ( not shown ) .
we list the uv luminosity for each lba measured from the om observation in table [ t : uv ] .
we derive the sfr from our uv luminosities using the formula from @xcite , which is based on evolutionary synthesis models of starbursts by assuming a salpeter initial mass function ( imf ) . for consistency with @xcite , we divide the rosa - gonzlez et al .
relation by a factor of 1.5 to reflect the use of a kroupa rather than a salpeter imf ( as described in @xcite ) .
the resulting formula is as follows : @xmath43 in table [ t : sfr ] we compare our sfr derived from the uv luminosity with that derived from a combination of the 24@xmath44 m and h@xmath23 luminosities ( see * ? ? ?
* for details ) . due to the uncertain infrared seds
, no @xmath45-correction was applied to the 24@xmath44 m luminosity in @xcite , and the actual sfrs might be somewhat higher than their derived values .
the uv - derived sfr is lower than that derived from the ir continuum plus optical emission lines .
this is consistent with the fact that the uv photons suffer significant dust attenuation and are re - emitted in the ir ( see also * ? ? ?
we note that any agn contribution to the uv continuum is negligible in these objects .
this is demonstrated most directly by _ hst _ cos fuv spectra of two of our targets @xcite .
as we described in the introduction , the optical emission - line spectra of our sample of lbas suggest that these may be composite objects consisting of an intense starburst and an obscured agn .
however , this evidence is ambiguous . @xcite
have discussed this in detail and suggest as an alternative that the optical spectra of these galaxies could reflect extreme conditions in the ionized gas related to the presence of a starburst with extreme properties .
the x - ray data we have presented provide important new insight into the nature of these objects .
we begin by comparing the amount of detected hard and soft x - ray emission in the lbas compared to pure starbursts with the same star - formation rates ( as traced by the far - ir luminosity ) .
we will then compare the lbas to typical type 1 and type 2 agn in terms of the ratios of the x - ray to [ o iii ] and mid - ir to [ o iii ] luminosities .
the far - infrared ( fir ) luminosity is considered to be a reliable indicator of the global sfr of star forming galaxies . in particular
, @xcite found that the 2 - 10 kev x - ray luminosity has a tight linear correlation with the fir luminosity by investigating a sample of local star forming galaxies .
we show our lbas in the @xmath46 diagram ( see fig .
[ f : sfrlx ] ) with the ranalli relation , which is indicated by the solid line with two dashed lines indicating a factor of 2 above and below .
the hard x - ray luminosities of the lbas are taken from table [ t : cstat ] , which are derived from the absorbed power - law model fit with the photon index fixed at 1.9 .
for comparison , we also show the sample of luminous infrared galaxies ( lirg ) from @xcite and @xcite , where iwasawa et al s sample was divided into two groups of agn and hard x - ray quiet " galaxies ( hxq ) indicated by diamond and asterisk symbols respectively , and lehmer et al s sample was indicated by triangle symbols .
we can see that our lbas are about 0.5 to 1 dex above the ranalli et al .
relation , and even more displaced from the non - agn lirgs .
this implies that a normal star forming process would only account for a minority of the hard x - ray emission of these lbas .
we also fit a @xmath47 correlation from our lba sample by applying the survival analysis ( ` asurv ` rev 1.2 ; * ? ? ?
* ) , which gives @xmath48 the dash - dotted line shows the best fit for the correlation .
we find similar results in the soft x - ray band .
we fit the correlation between the soft x - ray ( 0.5 - 2 kev ) and fir luminosity of our lba sample , and compared it with the result of @xcite .
the survival analysis gave a free slope fit @xmath49 besides the linear slope fit , @xcite gave a free - slope fit of @xmath50 for star forming galaxies . in fig .
[ f : softlfir ] , the dash - dotted line indicates the best - fit of lbas , and the ranalli curve is shown as the dashed line below it . as in the case of the hard x - ray emission ,
only part of the soft x - ray flux is likely to be contributed by normal processes associated with star formation . given that the lbas suffer from less dust extinction than typical local starbursts with similar star - formation rates @xcite , one might worry that the high ratio of x - ray to far - ir luminosity seen in figures 7 and 8 is due to the incomplete conversion of far - uv into far - ir light .
this is not the case . as discussed in @xcite ,
the star - formation rates derived from the observed far - uv are much smaller than those derived from the far - ir data , implying that the far - ir is a good proxy for the star - formation rate . our results in section 3.2 and figure 6 further support this .
the results from [ 4.1 ] are consistent with interpreting the optical spectra as indicating the presence of an agn .
the best tracer of both type 1 and type 2 agn in the optical spectra is the [ o iii]@xmath35007 emission - line . in this case
, we would expect that the amount of hard x - ray emission from the lbas would be consistent with the values found for agn with similar [ o iii ] luminosities . in fig .
[ f : lxlo ] , we plot a histogram of the ratio of the hard x - ray ( 2 - 10 kev ) luminosity obtained from the power - law model spectral fit of our lbas to its observed [ o iii ] luminosity .
we also show the empirical distribution of the luminosity ratios of type 1 ( dashed blue line ) and type 2 agn ( dot - dashed red line ) in this plot @xcite .
the ratios of hard x - ray to [ o iii ] luminosities in the lbas are too low ( by about an - order - of - magnitude ) to be consistent with the values seen in type 1 agn .
however , they do lie within the relatively broad range of values seen in type 2 agn .
the relatively weak hard x - ray luminosity of type 2 agn is due to obscuration of the hard x - rays by the high column densities of gas associated with the agn torus .
in contrast , the [ o iii ] line is formed in the 100 to 1000 pc - scale narrow line region and is not affected by the torus surrounding the black hole . at first sight ,
the presence of significant obscuration of the hard x - ray emission in the lbas would not be consistent with the relatively low absorbing column densities listed in table [ t : cstat ] .
here we see that the column density derived from the spectral fitting of an absorbed power - law gives the 90% confidence level upper limit at the order of @xmath51 @xmath33 ( @xmath52 @xmath33 for sdss j0802 + 3915 ) .
these columns are much too low to attenuate the hard x - ray emission by @xmath2 an order - of - magnitude .
such a seeming paradox is often seen in type 2 agn and is most likely due to fitting an overly idealized x - ray spectral model .
more realistic models ( e.g. partial covering models and/or models with reprocessing ) @xcite can reconcile the observed x - ray spectrum with heavy obscuration .
alternatively , the actual absorbing column densities derived from the fits to the x - ray spectra may be correct . in this case
, the low ratio of hard x - ray to [ o iii ] luminosities relative to unobscured agn would require that the [ o iii ] line is predominantly produced by the starburst rather than the agn .
as discussed by @xcite , this possibility can not be excluded .
assuming for the moment that the relatively high ratios of x - ray to far - ir luminosity indicate that an agn is present in these lbas , our results still do not establish whether this agn is more luminous and heavily obscured in hard x - rays or less luminous and relatively unobscured .
an additional diagnostic of an obscured agn is a strong iron k@xmath23 emission - line with a large equivalent width of about one kev ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in fig .
[ f : cstat ] , we show the comparison between the data and the spectral fit from an absorbed power - law .
we find that there is a large data - to - model ratio at the fe k@xmath23 energy in 1434 + 0207 , 0214 + 1259 and 0802 + 3915 .
we used a power - law plus gaussian model to fit the pn spectra of 1434 + 0207 and 0214 + 1259 ( insufficient photon counts for 0802 + 3915 ) .
we removed the 5 - 8 kev photons first to determine the power - law index , then froze this parameter and added the 5 - 8 kev photons with a gaussian component .
the line energy is 6.6 kev for 1434 + 0207 and 6.2 kev for 0214 + 1259 , and the spectral plots are shown in fig .
[ f : iron ] .
the plots indicate the possibility of an iron emission line in these lbas .
however , the photon counts are too few to give good estimates of the equivalent width of iron lines . in order to further investigate the presence of a fe line in our sample
, we loaded the spectra of all 6 lbas in ` xspec ` to construct an average spectrum due to the small number of photons of each lba ( effectively stacking the spectra in model space ) .
this was done in the same way by which we fitted the thermal component simultaneously in [ thermal ] .
this is not a fit to a stacked spectrum , but rather a joint fit to the set of observed spectra without shifting them to their rest-frame.the averaged spectrum was modeled as a continuum power - law in the range of 3 - 8 kev plus a gaussian emission line .
the intrinsic line width ( @xmath4 ) in the gaussian component was fixed at 0.01 kev , and the photon indices of the continuum power - law were set to 1.9 . by tying the gaussian normalization of all data sets , we obtain the emission line energy at 6.40 kev , with a 90% confidence region ( @xmath35-stat=4.6 for two model components ) of 6.09 - 6.68 kev . the equivalent width ( ew ) is 1.0 kev from this joint fit , and the associated 90% confidence upper limit of ew is about 4.4 kev , where the upper limit is obtained by scaling the confidence region of the line intensity by the best - fit value of the ew , which is likely to be overestimated @xcite .
however , the data has low signal - to - noise ratio , and the emission line profile is not significant .
when we removed the gaussian component in the spectral fitting , the @xmath28-stat only increased by 5.5 compared to the value of 109.6/102(dof ) for the model with gaussian .
we then performed the markov chain monte carlo ( mcmc ) simulation to test the significance of the iron emission line from the joint fit .
we generated 6 chains , and the length of each chain is @xmath53 .
the 90% confidence level for the line energy from the mcmc is 6.10 - 6.93 kev , and the estimated upper limit of ew is 2.5 kev .
the histogram distribution of the line energy and normalization parameters is shown in fig .
[ f : mcmc ] .
the significance of the fe k@xmath23 line is calculated as the fraction of non - zero line normalization in the chains , which is found to be 95.3% , i.e. , at the @xmath54 level .
one other test for the presence of a type 2 agn in the lbas is to compare the amount of mid - ir emission expected from a such an agn with the observed [ o iii]@xmath35007 luminosity . @xcite
have shown that both of these are good proxies for the bolometric luminosity of an obscured agn ( unlike the 2 - 10 kev x - ray luminosity ) . in figure 12
we plot histograms of the ratio of mid - ir/[o iii ] luminosities for the sample of type 2 agn in @xcite and for the lbas in this paper .
the distribution for the type 2 agn has a mean of 1.89 dex and a dispersion of 0.57 dex compared to 2.71 and 0.34 respectively for the lbas .
all six of the lbas lie above the mean value for the type 2 agn .
this suggests that if the lbas are indeed starburst / agn composite systems , the starburst dominates over the agn in the mid - ir in most cases . in a future paper
we will examine this issue in more detail spectroscopically using _ spitzer _ irs data .
note that agn emission from the torus peaks in the mid - ir , while the emission from dusty starbursts peaks in the far - ir ( as is also the case for the lbas ) .
we therefore conclude that the bolometric luminosity of the lbas is primarily due to the starburst .
as noted above , it is possible that the [ o iii ] emission in these lbas is primarily produced by the starburst and that the hard x - ray source suffers little attenuation . in this case , the agn contribution to the bolometric luminosity will be even smaller . as discussed in the introduction
, these lba - composites may offer us an unparalleled opportunity to study the processes involved in the formation of supermassive black holes in the early universe ( e.g. , * ? ? ?
* ) . in this section we will infer the properties of the supermassive black holes in our sample if galaxies using standard methods .
we will consider two possibilities .
the first is the minimal case in which the hard x - ray source is not obscured . in this case
, the bolometric correction to the hard x - ray luminosity in @xcite leads to a range of agn bolometric luminosities from 1 to 6 @xmath55 l@xmath8 .
the implied black hole masses then range from @xmath56 to @xmath57 m@xmath8 .
if these black holes are indeed radiating near the eddington limit , the implied masses are then in the range of the intermediate mass " black holes found in some nearby low - mass galaxies ( e.g. * ? ? ?
* ) . in the context of the high redshift universe
, similar objects could represent the seeds for the subsequent growth of more massive systems .
the second possibility is that the hard x - rays are significantly obscured in these lbas and that the [ o iii]@xmath35007 emission - line is powered by a type 2 agn . adopting the bolometric correction to the extinction - corrected [ o iii ] luminosity from @xcite then yields agn bolometric luminosities of @xmath58 to @xmath59 l@xmath8 and black hole masses of @xmath60 to @xmath61 m@xmath8 . in this case
, an eddington - limited black hole would have a mass of - order 10@xmath62 of the dco stellar mass @xcite .
this is similar to the mass ratio seen in present - day galactic bulges @xcite .
however , if the [ o iii ] line is contaminated or dominated by the starburst , the agn bolometric luminosity and implied black hole mass would be overestimated .
we present in this paper _ _
x__mm observations of 6 lyman break analogs ( lbas ) : rare members of a local population that strongly resemble lyman break galaxies at high - redshift in stellar mass , star formation rate , metallicity , size , dust extinction , and velocity dispersion .
our sample consists of lbas whose optical spectra are consistent with a composite system consisting of both an intense starburst and a type 2 ( obscured ) agn
. however , other interpretations of the optical spectra are possible . to further elucidate the nature of these objects we combine our new x - ray data with sdss optical spectra and _ spitzer _ mid- and far - ir photometry .
we summarize the key results as follows : \(2 ) we found that the ratios of x - ray to fir luminosity of the lbas are higher than the typical starburst values by factors of three to thirty .
this implies a source of x - ray emission in addition to what is seen in typical starbursts .
we suggest that the most likely candidate would be an agn .
\(3 ) the ratios of hard x - ray ( 2 - 10 kev ) to [ o iii ] luminosity are about an order - of - magnitude smaller than in unobscured ( type 1 ) agn but consistent with the broad range of values seen in type 2 agn .
this might suggest that the hard x - rays suffer a significant amount of attenuation ( as in the case of typical type 2 agn ) .
this would imply that the fitted column densities derived from the simple absorbed power - law model underestimate the true attenuation ( as is often the case in type 2 agn ) .
an alternative interpretation is that the hard x - rays are relatively unobscured and that the [ o iii ] line is primarily produced by an extreme starburst .
\(4 ) to discriminate between these possibilities ( and provide a confirmation of the presence of an obscured agn ) , we searched for the presence of the 6.4 kev fe k@xmath23 emission line in our lba sample .
however , the limited photon detections and low signal - to - noise ratio do not allow us to come to a strong conclusion .
the mcmc simulation gives a @xmath54 significance level for the iron line and a 90% confidence upper limit of 2.5 kev for the equivalent width .
\(5 ) we find that the ratio of mid - ir ( 24@xmath44 m ) and [ o iii]@xmath35007 luminosities of the lbas are higher than the values seen in type 2 agn ( by a difference of 0.8 dex in the mean ) .
this would imply that the bolometric luminosity of these systems is primarily due to the starburst rather than an agn . if the [ o iii ] emission is primarily due to a starburst , this conclusion is even stronger .
\(6 ) given that these objects have significantly higher ratios of x - ray to far - ir luminosity than starbursts , we conclude that the most likely interpretation is that these galaxies are indeed composites of an intense starburst and an agn ( consistent with their optical emission - line spectra ) . if so , the implied black hole masses are of - order @xmath64 m@xmath8 for eddington - limited accretion in the case where the hard x - rays are unobscured and roughly an order - of - magnitude larger for an obscured agn that produces the observed [ oiii ] emission - line
. these may be good analogs to black holes forming in the dense , stellar clumps in galaxies in the early universe .
calzetti , d. , sheth , k. , churchwell , e. , & jackson , j. 2009 , the evolving ism in the milky way and nearby galaxies , fourth spitzer science center conf . , ed .
k. sheth et al .
( pasadena , ca : caltech ) , http://web.ipac.caltech.edu/staff/ismevol/proceedings/calzettid_v1.pdf c c c c c c c c c c
name & observation i d & date & exp .
time@xmath65 & @xmath66 & @xmath67 & mode & filter + ( sdss j ... ) & & ( dd / mm / yy ) & ( ks ) & & & & & + 080844.26 + 394852.3 & 0553790101 & 03/10/08 & 8.7/11.5/11.5 & 5.05 & 0.091 & ff@xmath68 & medium + [email protected] & 0553790201 & 18/05/08 & 8.7/11.5/11.5 & 5.41 & 0.137 & ff & thin + 143417.15 + 020742.3 & 0553790301 & 20/07/08 & 13.7/22.1/22.1 & 2.80 & 0.180 & ff & thin + 021348.53 + 125951.4 & 0553790401 & 08/01/08 & 19.4/23.6/23.7 & 7.06 & 0.219 & ff & thin + 092159.38 + 450912.3 & 0553790501 & 27/04/09 & 3.7/20.6/22.4 & 1.55 & 0.235 & ff & medium + 080232.34 + 391552.6 & 0553790701 & 24/03/09 & 1.3/6.4/6.3 & 5.15 & 0.267 & ff & thin + cccccc sdss j0808 + 3948 & uvw1 & @xmath69 & @xmath70 & @xmath71 & 2.70 + sdss j2103@xmath420728 & uvw1 & @xmath72 & @xmath73 & @xmath74 & 2.74 + sdss j1434 + 0207 & uvw1 & @xmath75 & @xmath76 & @xmath77 & 2.09 + & uvm2 & @xmath78 & @xmath79 & @xmath80 & + sdss j0214 + 1259 & uvw1 & @xmath81 & @xmath82 & @xmath83 & 3.85 + & uvm2 & @xmath84 & @xmath85 & @xmath86 & + sdss j0922 + 4509 & uvw1 & @xmath87 & @xmath88 & @xmath89 & 7.93 + & uvm2 & @xmath90 & @xmath91 & @xmath92 & + & uvw2 & @xmath93 & @xmath94 & @xmath95 & + sdss j0802 + 3915 & u & @xmath96 & @xmath97 & @xmath98 & + & uvw1 & @xmath99 & @xmath97 & @xmath100 & 3.03 + & uvm2 & @xmath101 & @xmath102 & @xmath103 & + & uvw2 & @xmath104 & @xmath105 & @xmath106 & + c c c c c i d & @xmath22,obs}}$ ] & @xmath22,corr}}$ ] & @xmath26 & @xmath107 + & ( @xmath39 ) & ( @xmath39 ) & ( @xmath108 ) & ( @xmath108 ) + sdss j0808 + 3948 & @xmath109 & 0.66 & 0.76 & @xmath110 + sdss j2103@xmath420728 & @xmath111 & 12.03 & 4.8 & @xmath112 + sdss j1434 + 0207 & @xmath113 & 2.84 & 0.48 & @xmath114 + sdss j0214 + 1259 & @xmath115 & 1.22 & 1.9 & @xmath116 + sdss j0922 + 4509 & @xmath117 & 4.75 & 2.4 & @xmath118 + sdss j0802 + 3915 & @xmath119 & 4.47 & 1.2 & @xmath120 + ccccccccc sdss j0808
+ 3948 & 75/34/44 & 1.66 ( 1.41 - 1.92 ) & 116.11/138 & @xmath121 & 115.82/138 & 2.11 ( 1.55 - 2.80 ) & 4.01 ( 2.94 - 5.33 ) & 6.08 + sdss j2103@xmath420728 & 76/26/17 & 2.48 ( 2.08 - 2.94 ) & 112.78/100 & @xmath122 & 118.60/100 & 2.93 ( 2.14 - 3.95 ) & 4.83 ( 3.53 - 6.51 ) & 0.40 + sdss j1434 + 0207 & 86/41/43 & 1.46 ( 0.88 - 2.04 ) & 134.41/156 & @xmath123 & 135.80/156 & 2.65 ( 1.53 - 4.28 ) & 4.41 ( 2.55 - 7.13 ) & 1.55 + sdss j0214 + 1259 & 83/25/28 & 2.72 ( 1.91 - 3.68 ) & 133.53/122 & @xmath124 & 136.31/122 & 2.39 ( 1.48 - 3.68 ) & 4.30 ( 2.66 - 6.62 ) & 3.51 + sdss j0922 + 4509 & 54/70/62 & 1.89 ( 1.50 - 2.30 ) & 172.57/165 & @xmath125 & 171.02/165 & 12.35 ( 6.64 - 19.94 ) & 22.70 ( 12.21 - 36.66 ) & 4.78 + sdss j0802 + 3915 & 15/8/19 & 1.71 ( -0.58 - 4.17 ) & 28.56/38 & @xmath126 & 28.62/38 & 9.29 ( 0.81 - 31.96 ) & 16.80 ( 1.47 - 57.81 ) & 3.76 + ccccc sdss j0808 + 3948 & 1.71 ( 1.33 - 2.13 ) & 1.01 & @xmath127 & 0.553 + sdss j2103@xmath420728 & 3.19 ( 2.40 - 4.28 ) & 0.19 & @xmath128 & 2.15 + sdss j1434 + 0207 & 1.24 ( 0.40 - 2.06 ) & 1.09 & 0.16 ( 0 - 2.44 ) & 1.17 + sdss j0214 + 1259 & @xmath129 & 0.79 & @xmath130 & 1.22 + sdss j0922 + 4509 & 2.83 ( 2.01 - 4.18 ) & 0.72 & @xmath131 & 0.99 + ccccc sdss j0808 + 3948 & @xmath36 & 1.85 ( 1.22 - 3.05 ) & @xmath132 & 0.69 + & @xmath28 & 1.79 ( 1.35 - 2.50 ) & @xmath133 & 0.86 + sdss j1434 + 0207 & @xmath36 & 1.23 ( 0.02 - 3.28 ) & @xmath134 & 1.18 + & @xmath28 & 1.41 ( 0.713 - 2.47 ) & @xmath135 & 0.87 + sdss j0922 + 4509 & @xmath36 & 4.56 ( 1.87 - 10.0 ) & 0.59 ( 0 - 2.91 ) & 0.71 + & @xmath28 & 3.35 ( 1.94 - 6.56 ) & 0.42 ( 0.05 - 1.41 ) & 1.01 + ccccccc sdss j0808 + 3948 & 500 ( fixed ) & 0.10 ( 0.01 - 0.41 ) & 0.29 ( 0.15 - 12.69 ) & 1.78 ( 1.09 - 2.74 ) & 3.76 ( 2.31 - 5.79 ) & 114.8/133 + sdss j2103@xmath420728 & 422 ( 276 - 755 ) & 0.17 ( 0 - 1.03 ) & 1.74 ( 0.16 - 3.50 ) & 1.45 ( 0.56 - 2.63 ) & 3.85 ( 1.49 - 6.75 ) & 101.9/96 + sdss j1434 + 0207 & 339 ( 217 - 689 ) & 1.07 ( 0.18 - 2.55 ) & 1.48 ( 0.39 - 2.98 ) & 0.86 ( 0.19 - 1.75 ) & 6.61 ( 1.47 - 13.44 ) & 129.3/152 + sdss j0214 + 1259 & 570 ( 326 - 844 ) & 266.5 ( 13.9- ) & 2.23 ( 1.05 - 3.84 ) & 1.95 ( 1.0 - 4.73 ) & 6.20 ( 3.10 - 15.03 ) & 125.7/118 + sdss j0922 + 4509 & 500 ( fixed ) & 0.16 ( 0 - 0.65 ) & 2.98 ( 1.50 - 11.10 ) & 6.04 ( 1.13 - 12.51 ) & 21.84 ( 4.10 - 45.23 ) & 169.4/162 + sdss j0802 + 3915 & 500 ( fixed ) & 1.96 ( 0.81 - 50.35 ) & 12.64 ( 1.97 - 31.67 ) & 14.23 ( 0.23 - 166.63 ) & 40.58 ( 0.66 - 475.17 ) & 24.42/35 + ccccc sdss j0808 + 3948 & 0.14 ( 0.07 - 0.28 ) & 0.39 ( 0.13 - 0.75 ) & 2.18 ( 1.33 - 3.36 ) & 4.87 ( 2.98 - 7.50 ) + sdss j2103@xmath420728 & 0.35 ( 0.05 - 1.17 ) & 1.83 ( 0.60 - 3.48 ) & 0.97 ( 0.37 - 1.76 ) & 3.31 ( 1.28 - 6.00 ) + sdss j1434 + 0207 & 0.59 ( 0.06 - 1.86 ) & 1.12 ( 0.37 - 2.12 ) & 1.57 ( 0.35 - 3.19 ) & 6.91 ( 1.53 - 14.06 ) + sdss j0214 + 1259 & 769 ( 45 - 1240 ) & 1.94 ( 0.64 - 3.69 ) & ... & 9.11 ( 4.67 - 22.1 ) + sdss j0922 + 4509 & 0.19 ( 0.05 - 0.52 ) & 1.27 ( 0.42 - 2.42 ) & 8.29 ( 1.55 - 17.17 ) & 19.43 ( 3.64 - 40.24 ) + sdss j0802 + 3915 & 2.94 ( 0.50 - 9.30 ) & 6.34 ( 2.08 - 12.05 ) & 2.31 ( 0.04 - 27.05 ) & 63.49 ( 1.03 - 743.45 ) + | we have used _ _ x__mm - newton to observe six lyman break analogs ( lbas ) : members of the rare population of local galaxies that have properties that are very similar to distant lyman break galaxies .
our six targets were specifically selected because they have optical emission - line properties that are intermediate between starbursts and type 2 ( obscured ) agn .
our new x - ray data provide an important diagnostic of the presence of an agn .
we find x - ray luminosities of order 10@xmath0 erg s@xmath1 and ratios of x - ray to far - ir luminosities that are higher than values in pure starburst galaxies by factors ranging from @xmath2 3 to 30 .
this strongly suggests the presence of an agn in at least some of the galaxies .
the ratios of the luminosities of the hard ( 2 - 10 kev ) x - ray to [ o iii]@xmath35007 emission - line are low by about an order - of - magnitude compared to type 1 agn , but are consistent with the broad range seen in type 2 agn . either the agn hard x - rays are significantly obscured or the [ o iii ] emission is dominated by the starburst .
we searched for an iron emission line at @xmath2 6.4 kev , which is a key feature of obscured agn , but only detected emission at the @xmath2 2@xmath4 level .
finally , we find that the ratios of the mid - infrared ( 24@xmath5 ) continuum to [ o iii]@xmath35007 luminosities in these lbas are higher than the values for type 2 agn by an average of 0.8 dex .
combining all these clues , we conclude that an agn is likely to be present , but that the bolometric luminosity is produced primarily by an intense starburst .
if these black holes are radiating at the eddington limit , their masses would lie in the range of 10@xmath6 to 10@xmath7 m@xmath8 .
these objects may offer ideal local laboratories to investigate the processes by which black holes grew in the early universe . |
alignment ( ia ) is a promising technique to mitigate interference in wireless communication systems .
it was shown that @xmath0 dof is achievable per time , frequency or antenna dimension in a @xmath1-user interference channel ( ic ) @xcite . for a @xmath1-user constant mimo ic ,
ia based schemes were introduced in @xcite-@xcite , where it was shown that more dof is achievable than that of conventional schemes . for a constant cellular interfering network ,
it was shown in @xcite that their scheme provides respectable gain for a 19 hexagonal wrap - around - cell layout . however , interference - free dof is only achievable for a two - cell layout .
it was shown in @xcite that optimal dof is achievable when @xmath2 for a two - cell mimo interfering broadcast channel , where each transmitter is equipped with @xmath3 antennas and each receiver is equipped with @xmath4 antennas . in this letter
, we focus on a three - cell constant cellular interfering network by using a new property of alignment , i.e. , _ ia solution obtained in an user - cooperation scenario can also be applied in a non - cooperation environment_. we assume that each base station ( bs ) is equipped with @xmath3 antennas and each mobile station ( ms ) is equipped with @xmath4 antennas , and @xmath5 which is most possible in a practical environment .
we also assume there are @xmath1 cell - edge users per cell where @xmath6 , and each user sends @xmath7 streams to its served bs simultaneously .
we show that totally @xmath8 dof is achievable if @xmath9 and @xmath10 or if @xmath11 and @xmath12 . numerical results show that more dof can be achieved compared with conventional schemes in most cases .
for a three - cell constant cellular interfering network ( an example is shown in fig .
[ fig1 ] ) , we assume that each bs is equipped with @xmath3 antennas , each ms is equipped with @xmath4 antennas , and there are @xmath1 cell - edge users per cell . for notation convenience ,
we refer to the @xmath13-th user in the @xmath14-th cell as user @xmath15 $ ] . for ease of analysis
, we consider an uplink scenario , and assume that each user tries to convey @xmath7 data streams to its served bs by using a normalized precoding matrix @xmath16}$ ] .
then we have @xmath17 } = \textbf{w}^{[ij ] } \textbf{s}^{[ij]},\ ] ] where @xmath18}$ ] is a @xmath19 vector , which denotes the transmitted data streams from user @xmath15 $ ] , and satisfies an average power constraint , i.e. , @xmath20 } \|^2 \right ] \leq p$ ] .
the received signal at the @xmath14-th bs is represented as @xmath21 } = \sum^3_{k=1 } \sum^k_{j=1 } \textbf{h}^{[kj]}_{i}\textbf{w}^{[kj ] } \textbf{s}^{[kj ] } + \textbf{n}^{[i]},\ ] ] where @xmath22 } \sim \mathcal{n}(\textbf{0 } , \sigma^2\textbf{i})$ ] is the @xmath23 additive white gaussian noise , and @xmath24}_{i}$ ] is the @xmath25 channel matrix from user @xmath26 $ ] to the @xmath14-th bs .
the channel is assumed to be constant over time , and perfect channel state information ( csi ) is available at all bss and mss .
the @xmath14-th bs decodes the desired signal for user @xmath15 $ ] by multiplying the cascaded receive beamforming matrices , @xmath27}$ ] and @xmath28}$ ] , and we obtain the desired signal for user @xmath15 $ ] @xmath29 } = \textbf{p}_j^{[i]\dag } \textbf{v}^{[i]\dag } \sum^3_{k=1 } \sum^k_{j=1 } \textbf{h}^{[kj]}_{i}\textbf{w}^{[kj ] } \textbf{s}^{[kj ] } + \widetilde{\textbf{n}}^{[ij]},\ ] ] where @xmath27}$ ] is the normalized inter - cell interference ( ici ) elimination matrix , @xmath28}$ ] is the normalized inter - user interference ( iui ) elimination matrix , and @xmath30 } = \textbf{p}_j^{[i]\dag } \textbf{v}^{[i]\dag } \textbf{n}^{[i]}$ ] is the effective noise vector .
the notation @xmath31 stands for conjugate transpose . let @xmath32 } = \{\textbf{p}_1^{[i ] } , \dots , \textbf{p}_k^{[i]}\}$ ] denote the combined iui elimination matrix at the @xmath14-th bs . we define the dof region as the following @xcite :
@xmath33 } , \dots , d^{[3k ] } ) \in \mathbb{r}^{3k}_+ | \forall ( \omega_{11 } , \dots , \omega_{3k } ) \in \mathbb{r}^{3k}_+ , \\
\sum^3_{i=1 } \sum^k_{j=1 } \omega_{ij } d^{[ij ] } \leq \limsup_{\textrm{snr } \rightarrow \infty } \big [ \sup_{\textbf{r } \in \mathcal{c } } \frac{1}{\log \textrm{snr } } \sum^3_{i=1 } \sum^k_{j=1 } \omega_{ij } r^{[ij ] } \big]\bigg\},\end{gathered}\ ] ] where @xmath34 is the capacity region , @xmath35 , and @xmath36}$ ] is the rate of user @xmath37 $ ] .
let @xmath38}\ ] ] be the total dof in the network .
in this section , an ia based scheme is introduced for the three - cell constant cellular interfering network .
a motivating example is given first ( as is shown in fig .
[ fig1 ] ) , where @xmath39 , @xmath40 , @xmath41 , and @xmath42 .
we show that totally 18 dof is achievable in this scenario .
we divide our scheme into two phases .
first , ici is aligned into a smaller vector space at each bs by joint design of all the precoding matrices .
second , iui is eliminated through cascaded receive beamforming matrices at each bs . * phase i : ici alignment . * by applying the ia solution obtained in an mimo ic to a cellular environment , which is presented in fig
. [ fig1 ] , we show that 12 ici streams at each bs can be aligned into a vector space of 9 dimensions simultaneously , i.e. , @xmath43 } \textbf{w}^{[21 ] } ~ \textbf{h}_1^{[22 ] } \textbf{w}^{[22 ] } ~\\ \textbf{h}_1^{[31 ] } \textbf{w}^{[31 ] } ~ \textbf{h}_1^{[32 ] } \textbf{w}^{[32 ] } ] \big ) \big\}= 9,\end{gathered}\ ] ] @xmath44 } \textbf{w}^{[11]}~ \textbf{h}_2^{[12 ] } \textbf{w}^{[12 ] } ~\\ \textbf{h}_2^{[31 ] } \textbf{w}^{[31 ] } ~ \textbf{h}_2^{[32 ] } \textbf{w}^{[32 ] } \big ] \big ) \big\ } = 9,\end{gathered}\ ] ] @xmath45 } \textbf{w}^{[11 ] } ~ \textbf{h}_3^{[12 ] } \textbf{w}^{[12 ] } ~\\ \textbf{h}_3^{[21 ] } \textbf{w}^{[21 ] } ~ \textbf{h}_3^{[22 ] } \textbf{w}^{[22 ] } \big ] \big ) \big\}= 9.\end{gathered}\ ] ] let @xmath46 } = \big[\textbf{w}^{[i1]\dag}~ \textbf{w}^{[i2]\dag } \big]^{\dag}$ ] be the combined transmit precoding matrix of all cell - edge users in the @xmath14-th cell .
let @xmath47 } = \big [ \textbf{h}_i^{[j1]}~ \textbf{h}_i^{[j2 ] } \big]$ ] represent the combined channel matrix .
then the effective channel is a three - user mimo ic where each node is equipped with @xmath39 antennas . following the analysis in @xcite ,
there exists a @xmath48 @xmath46}$ ] , @xmath49 , satisfies ( [ eq123 ] ) - ( [ eq125 ] ) .
@xmath50 } \underline{\textbf{w}}^{[2 ] } \right ] = \text{span } \left [ \textbf{g}^{[13 ] } \underline{\textbf{w}}^{[3 ] } \right],\ ] ] @xmath51 } \underline{\textbf{w}}^{[1 ] } \right ] = \text{span } \left [ \textbf{g}^{[23 ] } \underline{\textbf{w}}^{[3 ] } \right],\ ] ] @xmath52 } \underline{\textbf{w}}^{[1 ] } \right ] = \text{span } \left [ \textbf{g}^{[32 ] } \underline{\textbf{w}}^{[2 ] } \right].\ ] ] let @xmath53 } ) ^{-1 } \textbf{g}^{[32 ] } ( \textbf{g}^{[12 ] } ) ^{-1 } \textbf{g}^{[13 ] } ( \textbf{g}^{[23 ] } ) ^{-1 } \textbf{g}^{[21]},\ ] ] @xmath54 } ) ^{-1 } \textbf{g}^{[31]},\ ] ] @xmath55 } ) ^{-1 } \textbf{g}^{[21]}.\ ] ] then @xmath56 } ) = \text{span } ( \textbf{e } \underline{\textbf{w}}^{[1 ] } ) , \ ] ] @xmath57 } = \textbf{f } \underline{\textbf{w}}^{[1]},\ ] ] @xmath58 } = \textbf{c } \underline{\textbf{w}}^{[1 ] } .\ ] ] , @xmath40 , @xmath41 , and @xmath42.,width=283 ] so , if we allow full user - cooperation , 24 dof is achievable in this scenario . however , if user - cooperation is not allowed , we show that ( [ eq4 ] ) - ( [ eq6 ] ) are also satisfied in the following . _
remark : _ when the channels are generic , @xmath74 will be a random vector , so are @xmath75}$ ] and @xmath76}$ ] , where @xmath77\dag}~\textbf{w}_k^{[i2]\dag}]^{\dag}$ ]
. then @xmath16}=[\textbf{w}_1^{[ij]}~\textbf{w}_2^{[ij]}~\textbf{w}_3^{[ij ] } ] $ ] will be full rank , i.e. , @xmath78})=3 $ ] , with probability 1 .
[ fig55 ] shows rank distribution of @xmath16}$ ] at user @xmath15 $ ] .
there are 7 interference - free dimensions left at each bs , and each bs can decode 6 streams sent from its cell - edge users
. then we choose @xmath79 } \subseteq \text{null}\big ( \big [ \textbf{h}_1^{[21 ] } \textbf{w}^{[21 ] } ~~ \textbf{h}_1^{[22 ] } \textbf{w}^{[22 ] } ~~ \textbf{h}_1^{[31 ] } \textbf{w}^{[31 ] } ~~ \textbf{h}_1^{[32 ] } \textbf{w}^{[32 ] } \big ] \big),\ ] ] @xmath80 } \subseteq \text{null}\big ( \big [ \textbf{h}_2^{[11 ] } \textbf{w}^{[11]}~~ \textbf{h}_2^{[12 ] } \textbf{w}^{[12 ] } ~~ \textbf{h}_2^{[31 ] } \textbf{w}^{[31 ] } ~~ \textbf{h}_2^{[32 ] } \textbf{w}^{[32 ] } \big ] \big),\ ] ] @xmath81 } \subseteq \text{null}\big ( \big [ \textbf{h}_3^{[11 ] } \textbf{w}^{[11 ] } ~~ \textbf{h}_3^{[12 ] } \textbf{w}^{[12 ] } ~~ \textbf{h}_3^{[21 ] } \textbf{w}^{[21 ] } ~~ \textbf{h}_3^{[22 ] } \textbf{w}^{[22 ] } \big ] \big).\ ] ] * phase ii : iui elimination .
* we finally obtain an ici - free channel , i.e. , @xmath82 } = \textbf{v}^{[j]\dag } \textbf{h}_j^{[jk ] } \textbf{w}^{[jk]}$ ] , @xmath83 and @xmath84 .
the @xmath13-th bs calculates @xmath85 } = \text{null}(\overline{\textbf{h}}_j^{[j2]})$ ] , @xmath86 } = \text{null}(\overline{\textbf{h}}_j^{[j1]})$ ] , and iui is eliminated .
then 18 streams can be sent simultaneously without any interference . if conventional schemes , such as orthogonal schemes are used , at most 16 dof is achievable . for general antenna configuration and general number of users per cell in a three - cell multi - user mimo environment
when @xmath87 , we have the following theorem . in a three - cell constant interfering network ,
we assume that each bs is equipped with @xmath3 antennas , each ms is equipped with @xmath4 antennas , @xmath1 cell - edge users are served simultaneously per cell , and each user sends @xmath7 streams to its served bs . if @xmath9 and @xmath88 or if @xmath11 and @xmath89 , then @xmath90 dof is achievable . }
$ ] at user @xmath15 $ ] for the motivating example , where 10000 monte carlo tests are performed.,width=245 ] @xmath91 } & \dots & \textbf{h}_1^{[2k ] } & \textbf{h}_1^{[31 ] } & \dots & \textbf{h}_1^{[3k ] } \\ & \textbf{h}_2^{[11 ] } & \dots & \textbf{h}_2^{[1k ] } & 0 & \dots & 0 & \textbf{h}_2^{[31 ] } & \dots & \textbf{h}_2^{[3k]}\\ & \textbf{h}_3^{[11 ] } & \dots & \textbf{h}_3^{[1k ] } & \textbf{h}_3^{[21 ] } & \dots & \textbf{h}_3^{[2k ] } & 0 & \dots & 0\\ \end{bmatrix}\ ] ] when @xmath114 and @xmath115 , we combine all the interference signals into a matrix @xmath116 which is defined as ( [ eq77 ] ) .
let @xmath117\dag}~ \dots~ \textbf{w}^{[1k]\dag}~ \textbf{w}^{[21]\dag}~ \dots~\\ \textbf{w}^{[2k]\dag}~
\textbf{w}^{[31]\dag}~ \dots~ \textbf{w}^{[3k]\dag } \big]^{\dag},\end{gathered}\ ] ] and let @xmath118 the dimension of @xmath119 is @xmath120 , and the dimension of the null space of @xmath119 is @xmath121 .
let @xmath122 , then there exists a @xmath123 @xmath124 satisfies ( [ eq237 ] )
. then ( [ eq84 ] ) - ( [ eq86 ] ) are also satisfied . by using the same argument as in the motivating example , @xmath16}$
] will be full rank with probability 1 .
the dimension of the interference space at each bs is decreased to @xmath106 , then each bs needs a vector space of @xmath109d$ ] dimensions to decode @xmath110 streams sent from its @xmath1 cell - edge users , i.e. , @xmath111 should be satisfied when @xmath3 and @xmath1 are given .
ici elimination matrix @xmath27}$ ] and iui elimination matrix @xmath32}$ ] , @xmath99 , can be calculated accordingly .
then totally @xmath8 streams can be sent simultaneously , i.e. , when @xmath125 and @xmath126 , @xmath8 dof is achievable .
for example , when @xmath127 , @xmath128 , and @xmath129 , one stream can be sent from each user simultaneously .
then @xmath130 is achievable , while if orthogonal schemes are used , at most 8 interference - free streams can be sent simultaneously .
the comparison of dof achievable between our scheme and orthogonal schemes is presented in fig .
[ fig33 ] when @xmath131 and @xmath132 where @xmath1 varies from @xmath133 to @xmath134 when @xmath135 .
it is shown that our scheme can achieve more dof compared with orthogonal schemes in most cases . however , achievable dof is less than orthogonal schemes when @xmath136 , 13 , or 19 , as some dimensions are wasted at bss .
if symbol extensions are allowed , even with constant channel , it is expected that more dof can be achieved than that of orthogonal schemes by using similar scheme in @xcite , and we leave it for future work .
r. tresch , m. guilland , and e. riegler , on the achievability of interference alignment in the @xmath1-user constant mimo interference channel , " in _ proc .
ieee workshop stat . signal process _
277 - 280 , cardiff , u.k .
m. razaviyayn , g. lyubeznik , and z .- q .
luo , on the degrees of freedom achievalbe through interference alignment in a mimo interference channel , " _ ieee trans .
signal process .
60 , no.2 , pp .
812 - 821 , feb . 2012
. w. shin , n. lee , j .- b .
lim , c. shin , and k. jang , on the design of interference alignment scheme for two - cell mimo interfering broadcast channels , " _ ieee trans . on wireless communications _ ,
437 - 442 , feb . 2011 .
p. mohapatra , k. e. nissar , and c. r. murthy , interference alignment algorithms for the @xmath1 user constant mimo interference channel , " _ ieee trans .
signal process .
5499 - 5508 , nov . | for a three - cell constant cellular interfering network , a new property of alignment is identified , i.e. , interference alignment ( ia ) solution obtained in an user - cooperation scenario can also be applied in a non - cooperation environment . by using this property , an algorithm
is proposed by jointly designing transmit and receive beamforming matrices .
analysis and numerical results show that more degree of freedom ( dof ) can be achieved compared with conventional schemes in most cases .
interference channel , interference alignment , degrees of freedom , multi - user mimo . |
crl 618 is one of the few clear examples of an agb star in the transition phase to the planetary nebula stage : a proto planetary nebula ( _ ppn _ ) .
it has a compact hii region created by a hot central c - rich star , and is observed as a bipolar nebula at optical , radio and infrared wavelengths .
the expansion velocity of the envelope is around 20 , but co observations show the presence of a high - velocity outflow with velocities up to 300 ( cernicharo 1989 ) .
high - velocity emission in is also detected ( burton & geballe 1986 ) .
the high velocity wind and the uv photons from the star perturb the circumstellar envelope producing shocks and photodissociation regions ( pdrs ) which modify the physical and chemical conditions of the gas ( cernicharo 1989 , and neri 1992 ) .
clumpiness within the visible lobes and low - velocity shocks being the remnant of the agb circumstellar envelope are also proposed by latter ( 1992 ) .
h@xmath1 excitation mechanisms can be investigated through our ir observations .
we also propose to study the possible variations of the otp ratio from the classical value of 3 .
more details are given in herpin ( 2000 ) .
many lines were detected ( see fig .
1 ) with these iso short - wavelength spectrometer ( sws , de grauuw 1996 ) observations : 0 - 0 s(j=0 - 15 ) , 1 - 0 q(j=1 - 7 ) , 2 - 1 q(j=1 - 7,9,10,13,14,15 ) , 3 - 2 q(j=1,3,4,5,7 ) , 4 - 3 q(j=2,3,4,7,9,10 ) , and 1 - 0 o(j=2 - 7 ) .
the fluxes were obtained assuming the source punctual relative to the sws beam .
a sws spectral resolution of @xmath2 was taken .
because of the large number of data and thus the large energy covering , we based our global data analysis on the @xmath3 s transitions . moreover , 0 - 0s(0 ) to 0 - 0s(6 ) emissions are not very sensitive to the extinction and thus are very reliable . a brief look on the plot of the column densities ( derived from the line intensities ) versus the energies of the upper levels for the uncorrected data ( no absorption correction , and otp ratio of 3 ( fig.[figuretotal ] , left ) shows first that an otp ratio ( @xmath4 ) of 3 seems inappropriate as the ortho observations are systematically underestimated .
moreover , correction of the entire set by the same value of the otp ratio seems not possible .
thus we introduced different values for different parts of the plot , i.e. , different energies , and thus probably different regions of excitation in the ppn environment : @xmath5 and 1.87 . but some discontinuities are still present .
this is not an effect of variations in @xmath6 because in this case slopes would vary .
this can only be explained by different regions of emission , and consequently can be corrected by different absorptions .
we thus applied 4 different corrections of the absorption : @xmath7 and 20 mag ( latter 1992 , thronson 1981 ) .
the result is given in the fig .
[ figuretotal ] ( right captions ) .
this study of the resolution of the observed lines ( see fig .
1 ) shows three different parts of emission .
first of all , the @xmath3 s(0 ) to s(6 ) lines are totally resolved ( with a sws resolution of 200 ) .
a broad line , thus resolved , means that the emission arises from a quite large velocity region , this velocity field being then sufficiently important to broaden the line .
this suggests that this part of the emission may arise from the lobes , where the velocity is larger .
an intermediate zone of excitation , between lobes and torus , appears for the @xmath3 s(7 ) , s(8 ) and s(9 ) emissions , whose half power line widths are around 200 ( third zone on the plot ) on the limit of resolution . finally , all the other lines have small widths ( 50 - 160 ) , and then probably arise from a low - velocity region , certainly the edge of the torus .
a similar study is done for the emissions involving excited vibrational levels .
the @xmath8 q lines are unresolved .
we think that these emissions stem from the same intermediate zone as are the @xmath3 s(7)-s(9 ) emissions .
moreover the `` intermediate '' slope derived from the column densities is another indication .
the same conclusion applies for the @xmath8 o lines , even if their origin is not exactly the same ; this emission is probably closer to the lobes , as the derived temperature ( cf .
fig.[figuretotal ] ) is similar as what is found for the @xmath3 s emission in this area .
all the lines involving vibrational levels with @xmath9 are completely unresolved and thus arise from a region close to the torus .
we attempted to fit the data with a very simple model based on a boltzman distribution of the populations ( cf .
2 ) . for the @xmath3 s emissions ,
the fitted temperatures and column densities are in agreement with the hypothesis of low - energy emission from shocks in the lobes ( low temperatures , and high column densities , corresponding to @xmath10 @xmath11 ) , and high - energy emissions coming from the torus with higher temperature and weaker density . for the transitions from excited vibrational levels
, we admitted that 0 - 0 s(j ) , 1 - 0 q(j+2 ) and o(j+4 ) emissions are produced in the same region , and have the same extinction and otp ratio .
the same argument was used for the other excited vibrational emissions .
we obtained different fits with slow slopes ( temperatures are larger ) for different @xmath12 emissions , which may suggest fluorescent emission .
concerning the @xmath8 emissions , we compared our intensities to theoretical results of sternberg ( 1989 ) and black ( 1987 ) .
moreover the temperatures are moderate , as the rot - vibrational column densities .
this suggests the thermal contribution in the excitation . a clumpy structure , or one which partly shields the emitting region from the stellar uv flux may explain a low level of fluorescent emission . in thermal sources these rotational and vibrational temperatures are similar , while fluorescent sources are characterized by a much higher vibrational temperature and a much lower rotational temperature ( tanaka 1989 ) .
that is why emissions 1 - 0 o and q are probably more or less of thermal nature ( @xmath13k and @xmath14 k for @xmath15 ) , but a component of fluorescent emission is probably also present .
the situation seems more complicated for the other vibrational excited transitions .
the only case in sternberg ( 1989 ) and also in black ( 1987 ) with all these emissions is the radiative fluorescent model .
this strongly suggests that the emissions involving @xmath16 levels are fluorescent .
we resume on fig.[energie ] what we deduced about the excitation of the lines .
we attempt to explain the low otp ratio by different mechanisms .
shocks breaking molecules and photodissociation , ionization from the central star can produce h , h@xmath17 and h@xmath18@xmath17 who will react with and altere the otp ratio .
on the other hand , in a shocked region , the shock velocity may be important in the otp conversion process .
we think that there is in fact no otp ratio gradient in our data , but large uncertainties , and consequently that the otp ratio is around 1.76 - 1.87 , the first value of 1 remaining without explanation .
we thus think that the otp ratio is fixed near the central star where low - velocity shocks occur and where atoms and ions can spoil this ratio .
part of this material is swept out into the lobes with this low otp ratio .
shocks in the lobes produce the low - energy 0 - 0 s emission , while shocks and perhaps fluorescence induce emission in a transition area . the high - energy emission may be produced by fluorescence in the torus .
a mix of fluorescence and collisions imay be on the origin of the 1 - 0 o and q lines , while all the other emissions are dominated by fluorescence . concerning the low otp ratio , probably around 1.76 - 1.87
, we think that it is fixed near the central star , the material being after swept out into the lobes , with a very non - classical value .
1987 _ astrophys . journal _ * 322 * , 412 .
1986 _ monthly notice of royal astron .
* 223 * , 13p .
1989 _ astron . &
_ * 222 * , l1 .
1996 _ astron . &
_ * 315 * , l49 .
2000 _ to be submitted to astrophys . journal _ 1992 _ astrophys .
journal _ * 389 * , 347 . 1992
_ astron . &
_ * 262 * , 544 . 1989
. journal _ * 347 * , 863 . 1989
_ astrophys . journal _ * 336 * , 207 .
1981 _ astrophys . journal _ * 248 * , 984 . | we present a complete study of the infrared emission , including the pure rotational lines , of the proto planetary nebulae crl 618 with the iso sws .
a large number of lines are detected .
the analysis of our observations shows : ( i ) an otp ratio very different from the classical value of 3 , probably around 1.76 - 1.87 ; ( ii ) a stratification of the emitting region , and more precisely different regions of emission , plausibly located in the lobes , in an intermediate zone , and close to the torus ; ( iii ) different excitation mechanisms , collisions and fluorescence .
1200 -0.5pc -0.5pc # 1@xmath0he # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#2 # 1#1 0123456789 abcdef @mtlplain@loaded = cmssi10 = cmssi10 = cmssi10 = cmmib10 = mtssi10 = mtssbx10 = mtssbi10 = mtmib10 @mtlplain@loaded |
the study of nonlinear optical phenomena arising from the collective motion of atoms in dynamic optical fields has been an active field of predominantly theoretical research over the last decade .
a large fraction of this work has been concerned with the collective atomic recoil laser ( carl ) @xcite . as part of the continuing progress in the production and investigation of ultracold atomic gases and bose - einstein condensates ( becs ) ,
there have been several experiments which have demonstrated the validity of these models and realized some of their predictions .
examples include the observation of collective atomic recoil lasing by a cold thermal gas in a high - finesse cavity @xcite and the observation of superradiant rayleigh scattering by a bec @xcite .
superradiant rayleigh scattering involves the production of pulses of coherently scattered radiation simultaneous with the splitting of the condensate into discrete momentum groups due to atomic recoil .
several theoretical models have been used to describe the evolution of the superradiant scattering process @xcite , including an extension of the original classical carl model @xcite to include a quantum treatment of the atomic dynamics @xcite .
recent experimental work by schneble _
@xcite has shown that in addition to the superradiant rayleigh scattering process originally observed in @xcite , there is a second scattering regime termed kapiza - dirac scattering . during superradiant rayleigh scattering ,
the scattering process involves only emission of scattered photons i.e. absorption of pump photons and emission of scattered ( probe ) photons . in contrast , during kapiza - dirac scattering the scattering process involves both emission _ and absorption _ of scattered ( probe ) photons i.e. absorption of probe photons and emission of pump photons . in this paper it is shown that many features of the recent experiments of schneble _ et al .
_ @xcite can be described using a one - dimensional mean - field quantum carl model , where optical amplification occurs simultaneously with the production of a periodic density modulation in the atomic medium . using this model ,
we demonstrate that the two regimes of `` kapiza - dirac scattering '' and `` superradiant rayleigh scattering '' observed in @xcite , can be interpreted as the semiclassical and quantum limits respectively of carl lasing .
it will be shown that the two regimes are distinguished by the relative size of the gain of the superradiant scattering process and the frequency separation of the absorption and emission peaks . a significant difference between the results presented here and those of other theoretical models of the experiments of schneble _ et al . _
@xcite is that in this model the regime of scattering is not determined by the pump pulse duration , so the scattering process does not evolve in time from one to the other .
the model used to describe the bec - light interaction is the mean - field quantum - carl model originally derived in @xcite .
the model is one - dimensional and describes the evolution of a backscattered ( probe ) field arising from scattering of a pump laser field ( assumed to be of constant amplitude ) by an elongated bec . in the experiments of schneble
@xcite , the geometry of the experiment is essentially two - dimensional , as illustrated schematically in fig .
[ fig1 ] with emission of two endfire modes from each end of the long axis of the condensate , propagating transversely to the pump laser .
if we assume that coupling between the endfire modes ( which are much weaker than the pump ) is negligible , each endfire mode can be assumed to evolve independently and the atomic motion is one - dimensional .
[ fig1 ] when the pump laser is sufficiently detuned from the atomic resonance , it leaves the atoms in the internal ground state .
consequently radiation pressure due to absorption and subsequent random incoherent , isotropic emission of a photon , can be neglected . in this detuned regime
, coherent scattering of the pump laser is the dominant process .
the atoms interact with a laser beam of wave vector @xmath0 and scatter photons of wave vector @xmath1 , recoiling with a momentum @xmath2 .
the atoms , initially scattered randomly into various momentum states , interfere with the atoms in the original momentum state .
this creates a matter wave grating having the correct periodicity to further scatter the laser beam in the direction @xmath1 . in an elongated condensate a preferential direction for the scattered photons
emerges causing superradiant rayleigh scattering .
both the matter wave grating and the scattered light are coherently amplified @xcite . in a simplified 1d description of the process along the direction of the atomic recoil momentum @xmath3 , the evolution of the matter wave field @xmath4 and of the dimensionless amplitude @xmath5 of the scattered radiation is determined by the following quantum mean - field carl model @xcite : @xmath6\psi+\beta|\psi|^2\psi \label{psi}\\ \frac{da}{dt}&=&gn \int d\theta|\psi|^2e^{-i(\theta+\delta t)}-\kappa a. \label{a}\end{aligned}\ ] ] where @xmath7 ( with @xmath8 ) , @xmath9 is the dimensionless electric field amplitude of the scattered beam with frequency @xmath10 , @xmath11 is the two - photon recoil frequency , @xmath12 is the coupling constant , @xmath13 is the rabi frequency of the laser field with constant amplitude @xmath14 and frequency @xmath15 , @xmath16 is the detuning from atomic resonance @xmath17 , @xmath18 is the electric dipole moment of the atom along the polarization direction @xmath19 of the laser , @xmath20 is the volume of the condensate , @xmath21 is its cross - sectional area , @xmath22 is its length , @xmath23 is the total number of atoms , and @xmath24 .
the matter wave field is normalized such that @xmath25 .
the second term on the right hand side of eq . ( [ psi ] ) is the self - consistent optical lattice , resulting from the interference between the laser and the scattered radiation , whose amplitude is amplified by the matter wave grating described by the first term on the right hand side of eq .
( [ a ] ) .
the third term on the right hand side of eq .
( [ psi ] ) describes the mean - field effect of the atom - atom interaction due to binary collisions , where @xmath26 and @xmath27 is the scattering length .
( [ a ] ) has been written in the mean - field limit , which models propagation of light with respect to the atoms by replacing the non uniform amplitude by its average value and by adding to the equation a damping term with a decay rate @xmath28 of the order of the inverse of the photon flight time through the condensate .
notice that the interaction time @xmath29 appearing in eqs .
( [ psi ] ) and ( [ a ] ) is the pump field duration , so that our model is suitable to explore both superradiant rayleigh scattering ( long pump pulse duration ) and self - stimulated kapitza - dirac diffraction regime ( short pump pulse duration ) investigated in ref .
it will be shown here that the relevant parameter in these experiments is not the pump pulse duration but the characteristic time of the optical / matter - wave amplification process , given by the inverse of the gain rate .
if we assume that the atomic wave function is periodic on the scale of the optical potential , with spatial period @xmath30 which corresponds to a period in @xmath31 of @xmath32 , then we can expand the atomic wave function in a fourier series @xmath33 furthermore , we assume that the condensate is sufficiently dilute such that @xmath34 , where @xmath35 is the average atomic density , so that the atom - atom interaction term in eq .
( [ psi ] ) may be neglected .
the effect of the atom - atom term on the collective recoil lasing has been investigated in ref .
@xcite . substituting for @xmath4 using eq .
( [ expansion ] ) , it can be shown @xcite that eqs .
( [ psi ] ) and ( [ a ] ) can be rewritten as @xmath36 in this quantum description , the fourier expansion ( [ fourier ] ) is equivalent to expanding the wave function @xmath4 in the set of momentum eigenstates with eigenvalues @xmath37 and @xmath38 is the probability for an atom to have a momentum @xmath39 . in the superradiant regime
explored in the experiments of ref.@xcite , the radiation damping rate @xmath40 is always much larger than the gain rate and/or the recoil frequency @xmath41 , so that the field amplitude follows the atomic motion adiabatically .
hence , neglecting the time derivative , eq.([cn2 ] ) yields @xmath42
in order to obtain the gain coefficient of the superradiant process in the semiclassical and quantum limits from the dynamical equations , let us consider the initial equilibrium state with no field , @xmath43 , and all the atoms at rest , i.e. in the momentum state @xmath44 , with @xmath45 and @xmath46 for all @xmath47 .
linearizing around this equilibrium solution and neglecting the small detuning @xmath48 between the laser and scattered frequencies , eqs .
( [ cn1 ] ) and ( [ adia ] ) reduce to the single linear equation for @xmath49 : @xmath50 where @xmath51 is the superradiant gain @xmath52 @xmath53 is the natural linewidth , @xmath54 is the rayleigh scattering rate and @xmath55 is the saturation intensity .
( [ g1 ] ) coincides with the superradiant gain coefficient reported in eq.(4 ) of @xcite .
it is easy to show that eq.([b ] ) has an unstable solution @xmath56 $ ] , with @xmath57 and @xmath58 .
furthermore , it is easy to show , from eq.([b ] ) , that @xmath59 these expressions show that the instability may have a semiclassical or quantum character , depending on the ratio between the superradiant gain @xmath51 and the recoil frequency @xmath41 .
in fact , in the limit where @xmath60 , eqs .
( [ gain ] ) and ( [ ratio ] ) give @xmath61 , @xmath62 and @xmath63 . in this limit ,
the sr process is quantum in nature , with only the lower state @xmath64 being populated , and the sr gain is @xmath51 @xcite . in the opposite limit in which @xmath65 , eqs .
( [ gain ] ) and ( [ ratio ] ) give @xmath66 and @xmath67 . in this limit , the sr process is semiclassical in nature , with the states @xmath68 and @xmath64 almost equally populated . in this case , the superradiant gain is given by @xcite @xmath69 the sr gain in the semiclassical limit is always lower than the sr gain in the quantum limit , since @xmath70 .
notice also the different dependence of the sr gain on @xmath23 in the quantum and the classical limits .
in order to observe the behavior of the collective scattering process in the nonlinear regime , eqs .
( [ cn1 ] ) and ( [ cn2 ] ) were integrated numerically with initial conditions @xmath43 , @xmath71 , @xmath72 , and @xmath73 when @xmath74 and with parameters corresponding to the semiclassical and quantum regimes of evolution .
these parameters correspond to the experiments of @xcite i.e. a @xmath75rb condensate illuminated by a pump beam of wavelength @xmath76 and an intensity of @xmath77 .
the pump couples to the @xmath78 transition which has a natural width @xmath79 s@xmath80 , dipole moment @xmath81 c@xmath82 m , saturation intensity @xmath83 and recoil frequency @xmath84 s@xmath80 .
the condensate had a cigar - shaped form , @xmath85 in diameter and @xmath86 in length , so that @xmath87 s@xmath80 . using these parameters , the quantum and classical superradiant gain coefficients of eqs .
( [ g1 ] ) and ( [ g2 ] ) are @xmath88 and @xmath89 , respectively , where @xmath90 is the pump - atom detuning in mhz .
we assume that @xmath91 and that half of the atoms in the condensate participate in each of two superradiant emissions along the main axis of the condensate and that the number of atoms participating in the collective scattering process is @xmath92 rather than @xmath93 as quoted for the number of atoms in the condensate in @xcite .
qualitative support for this assumption is provided by fig.3a of @xcite , which shows a large fraction of the condensate atoms do not participate in the coherent superradiant scattering process . from @xcite ,
the kapiza - dirac experiment was carried out using @xmath94 , so that @xmath95 s@xmath80 and the quantum superradiant gain is @xmath96 s@xmath80
. consequently the collective scattering process is semiclassical in nature , since @xmath97 . for the experiments in the superradiant rayleigh scattering regime , @xmath98 , @xmath99 s@xmath80 and
the quantum superradiant gain coefficient is @xmath100 s@xmath80 , so that the collective scattering process is quantum - mechanical in nature , since @xmath101 .
[ kapiza_snaps ] shows snapshots of the atomic momentum distribution at different times for the semiclassical case with @xmath102 .
it can be seen from fig .
[ kapiza_snaps](a ) that , in agreement with our theory , at the beginning of the interaction momentum states @xmath68 _ and _ @xmath64 are the only non - zero momentum states to have significant population , with @xmath103 , where @xmath104 .
( [ ratio ] ) predicts a ratio @xmath105 . as time progresses , many momentum states with both positive _ and _ negative @xmath106 are populated .
however , the momentum distribution is not symmetric about @xmath44 . the average momentum is less than zero , so from momentum conservation
there is a net gain of the scattered radiation field , shown in fig [ fielda ] , due to the difference in photon absorption and emission rates .
[ fielda ] shows that amplification of the scattered field in the semiclassical case is simultaneous with the growth of a strong density modulation in the condensate , as represented by the bunching parameter , @xmath107 .
the atomic momentum distributions shown in fig .
[ kapiza_snaps ] are similar to the time of - flight images observed for the so - called `` kapiza - dirac scattering '' observed in @xcite , where population of momentum states due to absorption _ and _ emission of radiation was observed . ,
when ( a ) @xmath108s , ( b ) @xmath109s and ( c ) @xmath110s .
, width=226 ] , and bunching factor , @xmath111 , as a function of time in the semiclassical regime with @xmath112.,width=302 ] fig .
[ srs_snaps ] shows snapshots of the atomic momentum distribution at different times for the quantum case with @xmath113 .
it can be seen from fig .
[ srs_snaps](a ) that , in agreement with our model , at the beginning of the interaction momentum state @xmath64 is the only non - zero momentum state to have significant population , and that the population @xmath114 . eq .
( [ ratio ] ) predicts a ratio @xmath115 .
as time progresses , the atomic population moves sequentially from @xmath116 and states with @xmath117 are never populated .
this sequential decrease in atomic momentum gives rise to amplification of the scattered radiation field and again occurs simultaneously with the development of a strong density modulation in the condensate , as shown in fig .
[ fieldb ] .
in contrast to the semiclassical case , however , this amplification of the scattered field occurs due to emission of scattered ( probe ) photons _ only _ , and the spread in atomic momenta is much smaller than in the semiclassical case .
the atomic momentum distributions shown in fig .
[ srs_snaps ] are similar to the time of - flight images observed for `` superradiant rayleigh scattering '' in @xcite , where the atoms attain momentum _ only _ in the direction of the pump beam in discrete units of @xmath118 .
when ( a ) @xmath119ms , ( b ) @xmath120ms and ( c ) @xmath121ms.,width=226 ] , and bunching factor , @xmath111 , as a function of scaled time in the quantum regime with @xmath122.,width=302 ]
in @xcite , the kapiza - dirac scattering regime and the superradiant rayleigh scattering regime are described as the `` short - pump pulse '' and `` long - pump pulse '' limits respectively .
a problem with this classification is that it implies that for sufficiently long pump pulses the collective scattering makes a transition from kapiza - dirac scattering to superradiant rayleigh scattering , or in the terminology of this paper a transition from the semiclassical to the quantum regime of carl lasing .
the results presented here however do not support this interpretation .
they suggest that the distinguishing feature between the two experiments in @xcite is not the pump pulse duration , but the atom - field detuning ( which differs by an order of magnitude in the experiments of @xcite ) , which determines the atom - field coupling and consequently the timescale of the superradiant scattering process . rather than the ratio of the pump pulse duration relative to the two - photon recoil time ( the inverse of the two - photon recoil frequency ) ,
it is the size of the superradiant decay time relative to the two - photon recoil time which determines whether or not the scattering consists of a sequence of emission processes , as observed in the `` superradiant rayleigh scattering '' regime @xcite , or simultaneous emission and absorption processes , as observed in the `` kapiza - dirac regime '' .
a simple justification for this argument is as follows : the atoms are assumed to all be in the condensate initially , so that @xmath123 .
the frequencies at which the probability of absorption ( @xmath124 ) and emission ( @xmath125 ) are maximum are non - degenerate and differ by a frequency of @xmath126 .
a transition from ( @xmath125 ) will initiate superradiant or superfluorescent decay .
the characteristic time of the superradiant decay is @xmath127 for @xmath128 @xcite , so the spectral width of the superradiant pulse is @xmath129 .
if the spectral width of the sr pulse is much less than the absorption - emission frequency shift i.e. @xmath130 , then absorptive transitions will not occur because they are non - resonant and only ( @xmath125 ) transitions , i.e. absorption of pump photons and emission of probe photons , will occur .
consequently the system evolves in the quantum carl limit when @xmath131 .
in contrast , if the spectral width of the sr pulse is sufficiently large ( i.e. sr decay is sufficiently rapid ) that it is much larger than the absorption - emission frequency difference , so that @xmath132 , then absorptive transitions ( @xmath124 ) will be resonant and both emission and re - absorption of probe photons will occur .
consequently the system evolves in the semiclassical carl limit when @xmath65 .
it should be noted that in contrast to previous explanations @xcite of the experimental results in @xcite , in our argument the duration of the pump pulse is not a significant factor .
the reason that the semiclassical carl or `` kapiza - dirac '' regime can be observed using a short pump pulse in @xcite is because the timescale of semiclassical superradiant decay , @xmath133 ( see eq.([g2 ] ) ) is much shorter than in the `` superradiant rayleigh scattering '' example as a result of the decreased pump - atom detuning . in ref .
@xcite a suppression of quantum sr gain @xmath51 of eq .
( [ g1 ] ) by around two orders of magnitude was observed and attributed in @xcite to the short duration of the pump pulse .
the carl model described here explains the reduced gain observed in @xcite as a result of the fact that the sr scattering process is evolving semiclassically .
consequently the sr gain is not given by @xmath51 as given by eq .
( [ g1 ] ) but the semiclassical sr gain @xmath134 as given by eq .
( [ g2 ] ) .
the semiclassical gain , @xmath134 , gives a value consistently smaller than the one predicted by the quantum superradiant gain , @xmath51 .
using the same parameters as those used in figs [ srs_snaps ] and [ fielda ] , the ratio is @xmath135 .
it has been shown that many features of the recent experiments of schneble _ et al . _
@xcite , which show two different regimes of light scattering by the bec , can be described using a one - dimensional mean - field quantum carl model .
the two regimes of light scattering observed in @xcite , described as `` kapiza - dirac scattering '' and `` superradiant rayleigh scattering '' in @xcite , can be interpreted as the semiclassical and quantum limits respectively of carl lasing . in the semiclassical limit ,
when @xmath136 , the collective scattering process involves both absorption and emission of probe photons , and for sufficiently long times many momentum states are populated simultaneously . in the quantum limit , however , when @xmath137 , the collective scattering process involves emission of probe photons only , and a maximum of two momentum states are populated at any time .
we provide a simple explanation of these results in terms of a comparison between the frequency separation of probe emission and absorption events and the spectral width of superradiant decay between the initial and recoiling condensates .
in contrast to previous models of the experiments in @xcite , the pump pulse duration is not a significant factor in our interpretation .
the results presented here support the view of the bec - light interaction as a strongly coupled atom - optical system where the atoms and light ( and consequently the matter - wave and optical gratings ) evolve dynamically and self - consistently .
this picture gives a more correct and complete description of the interaction than models in which the dynamics of either the atoms ( matter - wave grating ) or the optical fields ( optical grating ) are neglected .
99 r. bonifacio and l. de salvo , nucl . instrum . and meth .
res . a * 341 * , 360 ( 1994 ) .
r. bonifacio , l. de salvo souza , l.m .
narducci and e.j .
dangelo , phys . rev.a * 50 * , 1716 ( 1994 ) .
r. bonifacio , g.r.m . robb and b.w.j .
mcneil , phys .
a * 56 * , 912 ( 1997 ) .
n. piovella , r. bonifacio , b.w.j .
mcneil and g.r.m .
robb , optics comm .
* 187 * , 165 ( 2001 ) .
n. piovella , m. gatelli and r. bonifacio , optics comm .
* 194 * , 167 ( 2001 ) .
n. piovella , m. gatelli , l. martinucci , r. bonifacio , b.w.j . m@xmath138neil and g.r.m .
robb , laser phys .
* 12 * , 188 ( 2002 ) .
g. r. m. robb , n. piovella , a. ferraro , r. bonifacio , ph .
w. courteille , and c. zimmermann , phys .
a * 69 * , 041403 ( 2004 ) . c. von cube , s. slama , d. kruse , c. zimmermann , ph .
w. courteille , g. r. m. robb , n. piovella , and r. bonifacio , phys .
* 93 * 083601 ( 2004 ) .
s. inouye , a. p. chikkatur , d. m. stamper - kurn , j.stenger , d. e. pritchard and w. ketterle , science * 285 * , 571 ( 1999 ) .
mikio kozuma , yoichi suzuki , yoshio torii , toshiaki sugiura , takahiro kugam , e.w .
hagley , l. deng , science * 286 * , 2309 ( 1999 ) .
r. bonifacio , f.s .
cataliotti , m. cola , l. fallani , c. fort , n. piovella , m. inguscio , optics comm . * 233 * , 155 ( 2004 ) .
moore and p. meystre , phys .
a * 58 * , 3248 ( 1998 ) .
moore , o. zobay and p. meystre , phys .
a * 60 * , 1491 ( 1999 ) .
o.e . mustecaplioglu and l. you , phys .
a * 62 * , 063615 ( 2000 ) .
trifonov , laser phys .
* 12 * , 211 ( 2002 ) .
d. schneble , y. torii , m. boyd , e.w .
streed , d.e . pritchard and w. ketterle ,
science * 300 * , 475 ( 2003 ) .
h. pu , w. zhang and p. meystre , phys .
lett . * 91 * 150407 ( 2003 ) .
n. piovella , l. salasnich , r. bonifacio , nd g.r.m .
robb , laser phys .
* 14 * , 278 ( 2004 ) . | we show that many features of the recent experiments of schneble _ et al . _
[ _ d . schneble , y. torii , m. boyd , e.w .
streed , d.e . pritchard and w. ketterle ,
science * 300 * , 475 ( 2003 ) _ ] , which demonstrate two different regimes of light scattering by a bose - einstein condensate , can be described using a one - dimensional mean - field quantum carl model , where optical amplification occurs simultaneously with the production of a periodic density modulation in the atomic medium .
the two regimes of light scattering observed in these experiments , originally described as `` kapiza - dirac scattering '' and `` superradiant rayleigh scattering '' , can be interpreted as the semiclassical and quantum limits respectively of carl lasing . |
electron clouds formed in circular particle accelerators with positively charged beams are known to degrade the quality of the beam .
they are a concern for future accelerator facilities such as the international linear collider ( ilc ) damping rings , the superkekb , and also upgrade of existing facilities such as the large hadron collider ( lhc ) and the fermilab main injector ( mi ) .
their study is also important for the optimum performance of spallation neutron sources .
the detection of electron clouds has been a topic of study ever since their effects were first observed .
the methods used for such a detection have included retarding field analyzers , clearing electrodes , shielded pickup detectors , te waves and study of the response of the beam in the presence of an electron cloud .
the te wave method involves transmitting microwaves through a section of the beam pipe , and then studying the effect of the cloud on the microwave properties .
the microwaves can be introduced either as traveling or standing waves within a section of the beam pipe .
the method of using microwaves as a probe for investigating the presence of electron clouds was first proposed by f. caspers @xcite , in which experiments were conducted at the sps at cern based on this method .
the measurement technique involves measuring the height of modulation side - bands off the carrier frequency of the microwave .
the electron cloud , constituting a plasma modifies the dispersion relationship of the microwave .
the periodic production and clearing of the cloud , based on the bunch train passage frequency , leads to modulation of the phase advance of carrier wave as it travels .
this modulation gives rise to side bands in the spectrum of the propagated wave .
the side bands are spaced from the carrier frequency by a value equal to integer multiples of the train passage frequency .
the details of the distribution of all the side band heights depends upon the nature of the build up and decay of the cloud .
the study of this paper is restricted to the effect of the wave dispersion from a static cloud under various conditions .
a confirmation of the electron cloud induced modulation was shown in the pep ii low energy ring ( ler ) @xcite . in this experiment , the wave was transmitted across a solenoidal section of the ring and the cloud density was controlled by adjusting the strength of the solenoidal magnetic field .
the estimations of the electron cloud density in this experiment , done based on the formulation given in @xcite , were reasonable when compared to earlier build up simulations .
as shown in this paper , reflections within the beam pipe can alter the signal and thus misrepresent the cloud density in the region being sampled . instead of transmitting the wave at a point and receiving it from another called traveling wave rf diagnostics
, one could also trap the wave within a section called resonant wave rf diagnostics .
this has the advantage that one can sample a known section of the beam pipe .
this method would not be affected by waves being reflected from other segments of the pipe coming back into the section of interest and thus compromising the precision of the measurement .
another advantage of trapping , is that there is an enhancement of the signal as long as the point of measurement is close to a peak of the standing wave .
additionally , at resonance , as demonstrated in ref @xcite , there is improved matching of the signal transfer from the electrodes into the waveguide , which is the beam pipe .
as discussed in ref @xcite , the modulation of the cloud density would result in a modulation of the resonant frequency enabling one to relate the frequency modulation signal to an actual cloud density .
the draw back of this technique is the need for having reflectors at both ends of the desired section .
the electron cloud induced phase shift is known to undergo an enhancement in the presence of an external magnetic field under certain conditions , due to a modification in the dispersion relationship .
this occurs when the wave magnetic field has a component perpendicular to the external magnetic field , and the frequency of the wave is close to the electron cyclotron frequency , corresponding to the value of the external magnetic field .
this effect was demonstrated through simulations @xcite , and was later confirmed through experiments done at pep ii across a chicane @xcite .
further measurements done at the same chicane , now installed in cesrta also confirm this effect .
while it is good to have an enhanced signal , the drawback of such a measurement is that there is no available formulation that relates the enhanced side - band amplitude with the expected electron cloud density .
in addition , as discussed in this paper , the phase shift progressively reduces as the electron cyclotron frequency exceeds the carrier wave frequency and at very high external magnetic fields the signal may not be observable at all .
on can suppress the effect of the external magnetic field by aligning the wave electric field parallel to the external magnetic field , however it will be shown in this paper that this can not be fully eliminated unless the waveguide is rectangular .
the program vsim@xcite , previously vorpal , was used throughout to perform the simulations .
the simulations used electromagnetic particle - in - cell ( pic ) algorithms , consisting of propagating waves through a conducting beam pipe .
the end of the pipe had perfectly - matched layers ( pmls ) [ 7 ] meant to absorb any transmissions , and thus simulate a long , continuous beam pipe .
electrons were uniformly distributed and set to initially have zero velocity ( a cold plasma ) .
waves were excited in the simulations with the help of a vertically pointing current density near one end of the beam pipe , which covered the full cross - sectional area and was two cells thick in the longitudinal dimension . in later simulations the pmls and rf current source were replaced with port boundary conditions that simultaneously absorb rf energy at a single frequency at the ends of the simulation while also launching rf energy into the simulation domain to simulate traveling waves .
the waves were propagated along the channel using the yee [ 5 ] algorithm for solving the electromagnetic field equations .
the computational parameters used did not change much between the various simulations performed in this paper .
all the simulations were three dimensional using a cartesian grid .
the grid sizes were around @xmath0 in each direction , the time steps varied between @xmath1 .
the macro - particles - per - cell used was typically 10 , and they were loaded uniformly in position space , with zero initial velocity , often referred to as a cold start " .
the duration of the simulation was about 140 rf cycles .
overall , the modeling effort related to measuring electron clouds using te waves has served as a useful guide toward better understanding of the physical phenomenon and proper interpretation of the measured data .
this paper provides a comprehensive account of simulations performed for various techniques currently under study .
a derivation of the wave dispersion relationship for propagation through a beam pipe with a cold , uniform electron distribution in a field free region is given in the appendix .
the first experiments using microwaves to assess the cloud densities involved simply transmitting the wave using a beam position monitor ( bpm ) and receiving the transmitted wave from another bpm downstream to the traveling wave @xcite . as shown in ref @xcite , the electron cloud induced phase shift per unit length of transmission in the absence of external magnetic fields and a uniform electron distribution , can be related to the electron density as follows , @xmath2 where @xmath3 is the angular cutoff frequency for a waveguide in vacuum
, @xmath4 is the angular plasma frequency , with @xmath5 the electron number density , @xmath6 the charge of the electron , @xmath7 the speed of light , @xmath8 the electron mass and @xmath9 the free space permitivity . in ref @xcite , this formula was validated through simulations for a square cross section beam pipe .
the derivation of the phase shift given by eq [ phaseshift ] used the dispersion relationship given in ref @xcite , @xmath10 which was proved to be valid for a circular waveguides free of external magnetic fields .
the results shown in this paper validates the formula for a cesr beam pipe geometry .
all these results collectively indicate that the formula is valid for any type of geometry . in the appedix
, we provide an explicit proof that this is indeed the case .
the cesr beam pipe geometry may be represented in the form of two circular arcs ( radius 0.075 m ) connected with flat side planes .
about 0.090 m from side to side and 0.050 m between the apices of the arcs .
the cutoff frequency of lowest mode for this geometry is known to be around @xmath11 ghz from past experiments and calculations related to the beam pipe .
figure [ fig : vorpalpic ] provides a snapshot of the simulation , showing the propagation of a wave through the cesr beam pipe .
calculation of electron induced phase shift was performed through separate simulations of the wave transmission through a vacuum beam pipe and through a beam pipe with electrons respectively . at a certain axial distance @xmath12 from the location at which the wave was launched ,
the variation of the voltage between the midpoints of the top and bottom boundaries of the beam pipe cross section were recorded as a function of time .
after normalizing the amplitudes of the two waves to unity , their difference gives a sinusoidal wave with an amplitude equal to the phase shift between the waves .
suppose that the angular frequency of the wave is @xmath13 and phase shift is @xmath14 .
the phase shift for nominal cloud densities is small enough that @xmath15 .
hence we have @xmath16 .
the amplitudes of all the waves were calculated from their respective numerical rms values . to confirm the relationship between the phase shift and electron cloud density ,
simulations were done with a cesr beam pipe with a length of 0.5 m .
figure [ fig : phaseshift1](a ) shows that simulations agree well with the analytically predicted values given by eq [ phaseshift ] .
this establishes the accuracy of the simulation method as well as the validity of eq [ phaseshift ] for any geometry .
we see that the electron cloud induced phase shift increases as one approaches the cutoff frequency . while this is desirable because it amplifies the modulation side - bands relative to the carrier signal , one will encounter reduced transmission as the carrier frequency approaches the cutoff . due to the presence of several mechanical and electronic components all along the vacuum chamber
, it is not very likely that one can perform phase shift experiments that are entirely free of internal reflections . as discussed in ref @xcite , in an experiment
, these internal reflections would potentially affect the value of the phase shift .
a wave reflected from a device that lies beyond the segment being measured , would be received by the detector when it comes back after reflection . at the same time
, waves could get reflected back and forth within the segment before eventually being received at the detector . in both cases ,
the reflected wave would have sampled a length different from that meant to be sampled and would thereby contaminate the signal , because the phase shift is proportional to the length of transmission . in order to understand this effect ,
the simulations were altered to include two protruding conductors , which would reflect some of the transmitted wave .
the protrusions were slabs in the transverse plane , extending from the bottom to 1 cm above the apex of the lower arc ( see fig .
[ fig : protrusions ] ) .
they were spaced 0.4 meters apart , including the thickness of the protrusions , which was 1 mm . the frequencies used for this study 2.41 ghz and 3.87 ghz , the same as those shown in fig [ fig : phaseshift1](a ) , correspond to the resonant harmonics ( @xmath17 and @xmath18 , respectively ) of a 0.4 meter resonant cavity " .
this was done in order to maximize reflections .
[ fig : phaseshift1](b ) shows the resulting phase shifts from these calculations .
the solid shapes represent the data for no reflections and are the same data that appear in fig .
[ fig : phaseshift1](a ) .
the open shapes represent the phase shifts in the presence of reflection .
these results clearly indicate that internal reflections modify the expected phase shift .
the nature of the alteration of phase shift depends upon the complexities of the transmission - reflection combination , and the instrumentation used for the method .
however , the results show that the linear relationship between phase shift and electron density is always preserved .
while internal reflections may interfere with phase shift measurements , they can also be used to trap a wave .
this trapped wave can be used to measure the electron cloud density , as discussed in ref @xcite .
this section shows the results of numerical simulation of such an experiment .
the geometries used were ( a ) cesr beam pipe also used in the previous section and ( b ) beam pipe with a circular cross section of radius 4.45 cm . both of them had conducting protrusions that were 1 mm thick and 1 cm high from the base as shown in fig [ fig : protrusions ] .
the cutoff frequency for the circular beam pipe can be calculated from the analytic expression , and is 1.9755ghz for the lowest ( @xmath19 ) mode . a trapped mode results when standing waves are induced between the reflectors .
in order to test the effectiveness of inducing such a standing wave using partial reflectors , simulations were done with an empty wave guide over a large range of frequencies .
it should be noted that , since there is only a partial reflection of the wave taking place , there is always a net transmission of energy across the segment between the protrusions .
the wave energy that escapes the partial reflectors gets absorbed into the pml regions .
identification of a resonance was done as follows . at each frequency and each time step , the wave energy flux was computed by integrating the poynting vector across a plane located at the mid point between the two protrusions .
this plane was oriented transverse to the axis and covered the entire cross section . for each of these frequencies ,
the mean of the energy flux was calculated over the period of the simulation .
it is expected that as the frequency approaches that of a standing wave , this averaged flux would reach a local minimum .
this is because of increased `` back and forth '' transmission which does not contribute to the average flux due to cancellation .
figure [ fig : poyntingflux ] shows the average energy flux calculated for a variety of frequencies spanning over several resonance points for ( a ) the cesr beam pipe ( b ) the circular cross section beam pipe .
the local minima seen on these plots correspond to a standing wave mode .
the length of the section , including the width of the protrusions was 0.4 m for the cesr beam pipe .
the length of the circular beam pipe , including the width of the protrusions was 0.88 m showing the linear relationship between @xmath20 and @xmath21 according to eq [ vacstandwave],title="fig : " ] showing the linear relationship between @xmath20 and @xmath21 according to eq [ vacstandwave],title="fig : " ] a standing wave occurs when the wavelength @xmath22 and the length @xmath12 of the segment are related such that for any integer @xmath23 , @xmath24 .
the dipersion relationship of a waveguide with wave frequency @xmath25 and cutoff frequency @xmath26 , is given by @xmath27 .
expressing @xmath22 in terms of @xmath12 , for the standing wave , this then gives @xmath28 this indicates the linear relationship between @xmath20 and @xmath21 , and also the relationship of @xmath12 with the slope , and @xmath26 with the intercept of the straight line .
equation ( [ vacstandwave ] ) was used to confirm that the local minima in fig [ fig : poyntingflux ] correspond to resonance points .
figure [ fig : f2vsn2 ] shows the value of @xmath20 plotted as a function of the corresponding value of @xmath21 for ( a ) the cesr beam pipe and ( b ) beam pipe with a circular cross section . performing a straight line fit on these points for case ( a )
, yielded the relationship @xmath29 .
this gives @xmath30 and @xmath31 .
for case ( b ) , a similar operation gives @xmath32 giving @xmath33 and @xmath34 .
these values are close enough to the expected ones and ascertain the accuracy of such a method in determining standing waves between partial reflectors .
the presence of an electron cloud would result in a shift in the standing wave frequency .
experimentally , it is possible to measure this in the form of frequency modulation side - bands associated with the periodic passage of a train of bunches creating electron clouds .
using , eq ( [ disprel ] ) we can show that the condition for standing waves given by eq ( [ vacstandwave ] ) is modified by an electron cloud as follows , @xmath35 where the wave frequency is now denoted by @xmath36 and @xmath37 is the plasma frequency .
subtracting eq [ vacstandwave ] from eq [ elecstandwave ] , and in the limit of small frequency shifts , we get , @xmath38 , where @xmath39 . on inserting the expression for @xmath37 , this then gives a simple expression relating the shift in resonant frequency as a function of electron density .
@xmath40 in which @xmath41 is the classical electron radius .
this shows that the frequency shift is proportional to the electron cloud density .
an effort is underway at cesrta to use this method to measure the density of the electron cloud within the beam pipe section where the reflections are occurring @xcite .
thus , it became necessary to test this phenomenon with simulations .
simulation of the frequency shift of standing waves was done for a cesr beam pipe cross section as well as a beam pipe with a circular cross section .
all the parameters were the same as before except that the length of the section between the partial reflectors for the cesr beam pipe was modified from 0.4 m to 0.9 m , which was somewhat close to one the setups under study at cesrta .
the length of the circular cross section was retained at 0.88 m .
since the frequency shift induced by electrons is very small , it is required that the resonant frequency be determined accurately . to do this
, a parabolic fit was made to the averaged energy flux in the vicinity of the minimum point , using the available points obtained from simulation .
the expression for the parabola may be obtained from a taylor expansion of the function around the minimum .
this gives the mean energy flux as a function of frequency @xmath42 near the @xmath23th minimum point @xmath43 .
thus , we have @xmath44\ ] ] where @xmath45 is the averaged energy flux , @xmath46 is the second derivative of @xmath47 evaluated at @xmath43 .
the first derivative of @xmath47 at the minimum point vanishes .
using the computed coefficients of the parabola one can solve for @xmath43 .
figure [ resshift ] shows the frequency shift induced by electron clouds for the @xmath48 mode for ( a ) the cesr beam pipe and the ( b ) the circular beam pipe .
for case ( a ) , the electron density used in this calculation was @xmath49 m@xmath50 , which is rather high , but helps validate eq ( [ freqshift ] ) with simulations more accurately .
for these parameters , the @xmath48 resonance occurs at 2.0033 ghz and the expected frequency shift due to electrons is 2 mhz .
simulations show a shift of 2.05 mhz . for case
( b ) , we used two values of electron densities , @xmath49 m@xmath50 and @xmath51 m@xmath50 .
the expected frequency shift for the @xmath49 m@xmath50 case is 2.013mhz .
the simulated frequency shift for this was 2.1mhz , and for an electron density of @xmath51 m@xmath50 , it was 4.2mhz .
thus we were able to establish that reasonably accurate values of frequency shift for such an experiment may be determined from simulations .
it is interesting to note that the error obtained in the resonant frequency itself was around @xmath52 , however the shift induced by electron clouds always had reasonable agreement with eq [ freqshift ] .
the agreement with theory provides confidence in estimating electron densities based on eq [ freqshift ] from measurements .
typical electron cloud densities are of the order of @xmath53m@xmath50 leading to frequency shifts about a factor of 100 smaller than those simulated here . simulating frequency shifts this small would in principle be possible by scaling all numerical parameters appropriately , but this would have been a far more intensive computational process .
however , it is well known that in practice , such small frequency modulations can be easily measured with standard spectrum analyzers .
resonance frequency induced by electron clouds in ( a ) the cesr beam pipe and ( b ) the circular cross section beam pipe.,title="fig : " ] resonance frequency induced by electron clouds in ( a ) the cesr beam pipe and ( b ) the circular cross section beam pipe.,title="fig : " ]
this section discusses the effect of external dipole and wiggler magnetic fields on the phase shift measurements .
simulations done in the past , have revealed that the phase shift is greatly amplified in the presence of an external magnetic field if the electron cyclotron frequency lies in the vicinity of the carrier frequency @xcite . following these results ,
the cyclotron resonance was soon confirmed at an experiment performed at the slac chicane @xcite .
the slac chicane has since been transferred to cesrta , where these studies continue to be made @xcite . in the presence of an external magnetic field and electron clouds ,
the medium is no longer isotropic and the polarization of the transmitted microwave plays an important role in the outcome of the measurement .
when the wave electric field is oriented perpendicular to the external magnetic field , the mode is referred to as an extraordinary wave or simply x - wave . in this situation ,
if the external dipole field corresponds to an electron cyclotron frequency close to the carrier wave frequency , we see an enhanced phase shift . the phenomenon is well understood in the case of open boundaries .
it is usually referred to as upper hybrid resonance .
the dispersion relationship for the open boundary case is given as follows ( see for example ref @xcite ) .
@xmath54 the quantity @xmath55 is the upper hybrid frequency which is given by @xmath56 , @xmath57 being the electron cyclotron frequency for the given magnetic field @xmath58 .
when @xmath59 , it is clear that @xmath60 .
it can also be seen that as @xmath61 , i.e. , for very high magnetic fields , the relationship between @xmath62 and @xmath13 approaches that of propagation through vacuum . in the case of electron clouds in beam pipes ,
the plasma frequency is of the order of a few 10 mhz while the carrier frequency is around 2 ghz . in this regime
, it is reasonable to state that resonance occurs when @xmath63 . since the phase advance is the product of the wave vector @xmath62 and the length of propagation , we see that the electron cloud induced phase shift will theoretically go to infinity .
equation ( [ openbdryres ] ) is not valid for waveguides , which have finite boundaries .
nevertheless , simulations show that the same qualitative features are exhibited also for propagation through waveguides
. figure [ [ xwave ] ] shows the enhanced phase shift for three values of cloud densities when the cyclotron frequency approaches the carrier frequency .
the beam pipe cross section was circular with a radius of 4.45 cm , which leads to a cutoff at 1.9755ghz at the fundamental @xmath19 mode .
these parameters match with the beam pipe geometry of the pep ii / cesrta chicane section .
the wave frequency used in the simulation was 2.17 ghz .
the magnetic field corresponding to this cyclotron frequency is 0.077576 t .
the wave was excited with the help of a sinusoidally varying electric field pointing perpendicular to the external magnetic field . when the wave is polarized with an electric field that is parallel to the external magnetic field ,
the mode is often referred to as the ordinary wave , or the o - wave . in this case
one would not expect any effect created by the external magnetic field
. this would be the case for a rectangular cross - section waveguide , or with open boundaries .
however , with a circular cross section as in our study , the boundary conditions would force a component perpendicular to the magnetic field in the wave electric field even if it is launched with a purely parallel electric field .
thus , it is inevitable that a weak component of a wave with an orthogonal polarization gets excited . additionally , the method employed in launching the wave in the simulations is similar to that performed in experiments , where an electric field wave is excited along a particular direction over a surface area .
the functions describing the wave for a cylindrical geometry are bessel functions involving the radial and azimuthal variables . unless care is taken to excite a wave having the given functional form , the wave is expected to couple itself to two orthogonal modes with varying degrees of intensity .
additionally , the detection system , would receive the effect of the two modes to varying degrees .
disentangling this combination will involve more analysis , guided by simulations . due to these effects
, we see a weak resonance effect even in a wave excited with a purely vertical electric field , that is aligned to the external magnetic field as indicated .
figure [ owave ] shows the presence of such a weak resonance and this effect has been observed at cesrta as well .
figure [ [ dens_scan ] ] shows the variation of phase shift with electron cloud density at different settings of external magnetic fields . in these simulations ,
the wave was excited with an electric field perpendicular to the external magnetic field .
these densities are typical of what is produced in cesrta .
the plots show that the variation of phase shift with density remains linear even when one is close to resonance .
this is expected to be true as long as the plasma frequency is much smaller than the wave frequency , regardless of how complex the dispersion relationship of the wave is .
thus , one could easily amplify the signal with the help of an external magnetic field to monitor relative changes in cloud density , if not the absolute density .
experiments have been done to study the phase shift across the damping wigglers at cesrta .
these experiments correspond to various bunch currents and wiggler field settings . the wiggler field setting influences the measurement in more than one way .
the wiggler field affects the motion of the electrons , which influences the secondary production of the cloud .
the synchrotron radiation flux is determined by the strength of the wiggler field , and this in turn determines the photoemission rate of the cloud .
both these effects determine the density of the cloud .
the electron density is not uniform across the length of the wiggler , as shown in ref @xcite .
as already shown in this paper , the external magnetic field , by itself , alters the phase shift for a given cloud density .
given that the wiggler field is rather complex , along with a cloud density that is longitudinally nonuniform , simulations become particularly important to fully interpret results from such an experiment . in this paper , we examine just the effects of the nonuniform wiggler field on the phase shift . the wiggler chamber cross section in cesrta is close to that of a rectangle .
the height is 5 cm and width is 9 cm .
the corners of the rectangle are chopped , so that the actual width of the base and top is 64.6 mm and the height of the side walls are 24.6 mm .
this which would only moderately alter the results obtained from using a perfect rectangle .
thus , for the sake of simplicity , the simulations were done with a perfect rectangle with the above parameters .
the length of the section simulated is 80 cm , which corresponds to half the length of the wiggler .
this is sufficient to account all the variations in the wiggler magnetic field .
the computed wiggler magnetic field used was based on the formulation given in @xcite .
figure [ wigglerfield ] shows the magnetic field , in three dimensions . for most of the region
, the field is oriented vertically , while in the transition region between poles there is a longitudinal component to the field .
there is almost no magnetic field in the horizontal direction . even before performing simulations with the full wiggler field turned on , some preliminary studies were done with a just a constant dipole field within the same geometry .
since the shape of the vacuum chamber cross section is rectangular , the the cutoff frequency of the wave is determined by the polarization of the te wave .
if the wave electric field is pointing in the vertical direction , the cutoff frequency is 1.66ghz and if the field is horizontal , the cutoff is 3 ghz .
all simulations were done at a frequency 10% above the respective cutoff .
figure [ wiggdipvertwave ] shows the phase shift associated with propagation of a vertically polarized wave under different conditions .
the figure shows that when the wave electric field is polarized along the external magnetic field , the phase shift matches with that predicted by eq ( [ phaseshift ] ) .
this result is expected to be true in the case of a rectangular cross section , where the wave electric field is pointing parallel to the external magnetic field everywhere in the pipe , and is thus unaffected by the external magnetic field . as already shown , this would not be true in the case of a curvature in the cross - section boundary .
the figure also shows that the phase shift is suppressed in the case of the wave electric field pointing perpendicular to the external field , referred to as extraordinary wave .
the cyclotron resonance in this case occurs when the magnetic field is equal to @xmath64 t and the field here was set to a much higher value . in general , the wave electric field perturbs the electrons , causing them to oscillate and thereby alter the wave dispersion relation .
when the external magnetic field is very high , the electrons tend to get locked against any motion transverse to the magnetic field .
since the wave electric field is perpendicular to the external magnetic field they will encounter electrons that tend to be `` frozen '' . as a result
the wave will undergo reduced electron cloud induced phase shift at magnetic fields much higher than that causing cyclotron resonance for the specific carrier frequency .
figure [ wiggdiphorwave ] corresponds to a wave with the electric field pointing horizontally .
this shows similar features as those in figure [ wiggdipvertwave ] .
the wave frequency is 3.3ghz and cyclotron resonance occurs at field of 0.11 t , and so the figure shows enhanced phase shift at a magnetic field setting close to this value .
exciting the chamber at this higher frequency could be less efficient due to poorer matching between the various hardware components .
in addition , one can expect a mixing of modes to take place at higher frequencies because of the presence of various irregularities in a real beam pipe as opposed to a simulated one .
nevertheless , studying this mode is important because the wiggler magnetic field is mostly pointing in the vertical direction .
one could amplify the signal by setting a wiggler field so that a cyclotron resonance is excited .
this effect might prove useful in detecting the presence of very low density electrons in wiggler regions .
the presence of low energy electrons in wiggler and undulators is of particular interest if the device is cryogenic , in which case the electrons are believed to contribute to the heat load of the system @xcite .
the electron cloud in such systems would be produced by electron beams primarily through photoemission , and thus if present , they will occur at very low densities , requiring greater sensitivity in the detection . in the end , we look at the phase shift in the presence of the full wiggler field .
figure [ wiggfullfield ] shows that the phase shift gets suppressed by about 20% when compared that expected in the absence of any fields . in this case
, the wave electric field is pointing in the vertical direction , which is a configuration that should have little effect over the phase shift , since the external magnetic field is largely vertical .
a longitudinal magnetic field would alter the dispersion relationship , where the wave gets split into a left and right circularly polarized components .
this has been analyzed for guided wave propagation in cylindrical geometries in ref @xcite .
while the 20% reduction in phase shift in our result can be attributed to such an effect , a detailed analysis of the same is beyond the scope of this paper .
overall , it is clear that simulations are of prime importance to accurately interpret the observed electron cloud induced phase shifts across such wiggler fields .
in this paper , we provide a comprehensive account of the simulation and analysis effort that has been carried out in conjunction with the experimental effort of using te waves to measure electron clouds in cesrta .
these simulations helped confirm several physical phenomena either in a quantitative or in a qualitative manner .
for example , they helped validate eq [ phaseshift ] that relates the phase shift with cloud density for geometries like the cesrta beam pipe , which does not have a regular shape such as rectangular or circular
. the effect of reflections on phase shift measurements has always been a concern , and simulations show that one must be careful especially of standing waves excited within the beam pipe due to partial reflectors .
the feasibility of using standing waves to measure the cloud density is clearly demonstrated by simulations and has provided valuable guidance to the experimental effort being carried out at cesrta . in the process , we were able to determine a novel method of detecting the presence of standing waves in simulations by averaging the total poynting vector flux across a surface over time for varying frequencies .
simulations of phase shifts in the presence of external magnetic fields were modeled for a variety of cases .
the nature of results vary greatly based on the the parameters present in the system .
for example , the possibility of exciting cyclotron resonances is clearly shown in simulations , which would be possible to produce in dipole fields present in a chicane . however , dipole fields used in bend regions of an accelerator are much higher and they can suppress the electron induced phase shift , depending on the polarization of the wave . in the presence of curved boundaries ,
there is always a mixture of effects from ordinary and extraordinary wave propagation .
since most accelerator vacuum chambers have a curvature , this effect is important to understand .
a direct comparison with the analytic expression eq [ disprel ] would lead to an incorrect interpretation of the measured data .
the results will always be hard to interpret when a system is near the cyclotron resonance , when the phase shift is theoretically infinity .
simulations and experiments would never yield an infinity , and there may be poor agreement between the two in such regimes .
however , an enhancement of the signal that is still predictable can always be obtained by moving reasonably close to a cyclotron resonance point .
electron cloud formation in wiggler fields can be very important in machines such as positron damping rings . given the complexity of such a system , experiments of determining the cloud density using te waves have to be accompanied by careful simulations before interpreting any results obtained from measurements .
it may be noted that the relationship between the phase shift and cloud density is always linear regardless of how complex the system is .
this would be true for very low cloud densities , which would have low plasma frequencies .
thus , this method can be of great utility if one is interested in relative changes in electron cloud densities for example when a machine is undergoing conditioning .
simulations can then be used to obtain a proportionality constant between cloud density and phase shift .
the studies in this paper were always done with an electron density that was cold and uniformly distributed transversely and longitudinally .
simulations have not indicated a dependence on temperatures associated with typical electron cloud densities . in a transverse nonuniform distribution
, there will be a slight enhancement in the phase shift if more electrons are populated in regions with high peak electric fields produced by the wave . as mentioned earlier ,
longitudinal variation of the cloud density becomes important in wiggler fields because it couples with the longitudinal variation of the magnetic field .
these additional complexities could be topics for future studies .
the te wave method is an attractive technique for measuring electron cloud densities that can replace or complement other measurement methods . besides cesrta , this method is being studied at other accelerator facilities @xcite .
the measurement technique and its required instrumentation are simple , the process is noninvasive , and can be kept in operation continuously .
thus , it holds the promise of wide usage wherever it is useful to monitor the electron cloud properties continuously and at all locations of the accelerator .
it is evident that a careful study toward understanding of the physical phenomenon through analysis and simulation are very important toward proper interpretation of the measured data .
in this appendix , we provide a derivation of eq [ disprel ] which is not specific to the geometry of the cross - section of the waveguide .
we also discuss the approximations and assumptions associated with the derivation of this equation .
this dispersion relationship is specific to guided waves propagating through electron clouds in field free regions .
the starting equations for such a system would include the fluid and maxwell s equations .
these are , @xmath65 + e({\bf e } + { \bf v } \times { \bf b } ) = 0 \nonumber \\ & & \frac{\partial n_e}{\partial t } + \nabla \cdot ( n_e{\bf v } ) = 0 \nonumber \\ & & \nabla \cdot { \bf e } = \frac{en_e}{\epsilon_0 } \nonumber \\ & & \nabla \cdot { \bf b } = 0 \nonumber \\ & & \nabla \times { \bf e } = - \frac{\partial { \bf b}}{\partial t } \nonumber \\ & & \nabla \times { \bf b } = \mu_0(-en_e{\bf v } + { \bf j_{ext } } ) + \mu_0 \epsilon_0 \frac{\partial { \bf e}}{\partial t } \label{unperturbeqs}\end{aligned}\ ] ] where @xmath5 is the number density of the electrons , @xmath66 is the velocity of the fluid , and @xmath67 is an external current density .
the other terms have their usual meanings .
the continuity equation is not independent from the rest of the equations as it can be obtained from maxwell s equations .
we perturb all quantities about an equilibrium , so that we have @xmath68 , @xmath69 , @xmath70 , @xmath71 where the zeroth order quantities satisfy the steady state condition @xmath72 . as a result
, we get @xmath73 such an equilibrium state requires for the particles to be confined in a steady state indefinitely .
this is normally associated with neutral plasmas in which the ions may be considered immobile , or charged particles trapped for a long period of time in confinement devices . in this paper ,
the electron cloud density is sustained for the duration of the bunch train passage .
the above equilibrium condition may be considered valid as long as @xmath74 where @xmath75 is time of confinement of the charge and @xmath42 is the frequency of the perturbing wave .
periodic changes in the state occurring over time scales greater than the wave periodicity manifest themselves as modulations of the output signal , while changes occurring over much smaller time scales , remain unresolved by the carrier wave .
thus , the spectrum of the output signal would depend on the variation of the electron cloud associated with its build up and decay . inserting the perturbation expansion into the original fluid and maxwell s equation , and imposing the above equilibrium conditions and ignoring terms of second and higher order , we get , @xmath76 these equations are linear and we can seek all perturbations to have the following form , @xmath77 from the above , it is clear that @xmath78 this form is valid as long as the geometry along @xmath79 " , the longitudinal coordinate is uniform and infinite .
this is not entirely true when there are partial reflectors . in the case of perfect reflectors
, one would obtain discrete values for @xmath62 , representing standing waves .
thus one can expect that the above form to be more accurate when close to a resonance in the presence of partial reflectors .
in general , the reflectors would be small enough so that the above form of solutions can be considered a valid approximation . for the sake of convenience
, we drop the accent @xmath80 and the arguments @xmath81 in the functions given in eq ( [ wavesoln ] ) . to simplify the analysis ,
we make two assumptions , ( 1 ) the fluid is cold and at rest .
so @xmath82 , and ( 2 ) the density is uniform , which means @xmath83 = constant .
we further assume that there is no external magnetic field , which means that @xmath84 . in the absence of any static external magnetic fields , the only contribution to @xmath85 would be the magnetic field produced by the beam .
since the beam is highly relativistic , this would be confined along the length of the bunch .
it is reasonable to disregard this when the gap between the bunches is much larger than the bunch length , in which case the wave would sample mostly a field free region .
inserting the waveform solutions ( eq [ wavesoln ] ) into the perturbed momentum and continuity equations in eq ( [ perturbeqns ] ) , with @xmath84 and using the relationships of eq ( [ waveform ] ) , we have , @xmath86 combining these , we get @xmath87 combining this with the perturbed electrostatic field equation in eq ( [ perturbeqns ] ) gives @xmath88 , which means , up to the first order , there is no perturbation in the charge density due to the wave electric and magnetic fields .
this gives us @xmath89 implying that the wave is purely electromagnetic . by combining the first ( momentum ) , and the last ( ampere s law ) equation of eq ( [ perturbeqns ] ) , and using the relationships of eq ( [ waveform ] ) , we get after some algebra , @xmath90 where @xmath91 .
similarly , it is easy to see that , @xmath92 using eqs [ wavegauss ] , [ waveampere ] and [ wavefaraday ] , along with @xmath93 , and assuming that the boundary conditions are perfectly conducting , one can follow the steps given in ref@xcite . to get
@xmath94 the constant @xmath95 must be nonnegative for oscillatory solutions , and will take on a set of discrete eigenvalues " , corresponding to the different modes associated with the geometry of the cross - section of the waveguide . combining eq ( [ eigenwaveeq ] ) with a similar relationship for a vacuum waveguide where , @xmath96 , one can easily show that , @xmath97 @xmath98 being the angular cutoff frequency for the vacuum waveguide . this relationship is the same as eq ( [ disprel ] ) .
the authors wish to thank john sikora for many useful discussions and for suggesting us to do the simulations with partial internal reflections .
thanks to jim crittenden for helping us in generating the complete the wiggler magnetic field .
we also wish to thank david rubin , mark palmer and peter stoltz for their support and guidance .
this work was supported by the us national science foundation ( phy-0734867 , phy-1002467 , and phy-1068662 ) and the us department of energy ( de - fc02 - 08er41538 and de - sc0006505 ; de - fc02 - 07er41499 as part of the compass scidac-2 project , and de - sc0008920 as part of the compass scidac-3 project ) .
24 f. caspers , w. hofle , j. m. jimenez , j. f. malo , j. tuckmantel , and t. kroyer , in proceedings of the 31st icfa beam dynamics workshop : electron cloud effects ( ecloud04 ) , napa , california 2004 ( cern report no .
cern-2005 - 001 , 2004 ) .
t. kroyer , f. caspers , e. mahner , proceedings of 2005 particle accelerator conference , knoxville , tennessee 2212 - 2214 s. de santis , j. m. byrd , f. caspers , a. krasnykh , t. kroyer , m. t. f. pivi , and k. g. sonnad phys .
100 , 094801 ( 2008 ) kiran sonnad , miguel furman , seth veitzer , peter stoltz and john cary , proceedings of pac07 , albuquerque , new mexico , usa , pp .
thpas008 k g sonnad _ et al _ proceedings of particle accelerator conference 2009 , vancouver , canada , 2009 , pp.th5rfp044 j.p .
et al _ , proceedings of international particle accelerator conference 2011 , san sebastin , spain , pp.tupc170 k.g .
et al _ http://meetings.aps.org/link/baps.2007.dpp.tp8.134 49th annual meeting of the division of plasma physics s. veitzer , doe scientific and technical information , http://www.osti.gov/bridge identifier number 964651 m. t. f. pivi , _ et al _ , proceedings of european particle accelerator conference 2008 , genoa , italy pp .
mopp065 c. nieter and j. r. cary , j. comp .
196 , 448 - 472 ( 2004 ) .
h. s. uhm , k. t. nguyen , r. f. schneider an d j. r. smith , journal of applied physics , vol 64(3 ) , 1988 , pages 1108- 1115 . j. berenger , journal of computational physics 114 , 185 ( 1994 ) k. yee , ieee transactions on antennas and propagation , ap-14 , 302 ( 1966 ) s. de santis , phys . rev .
st accel . beams 13 , 071002 ( 2010 ) john sikora and stefano desantis , http://arxiv.org/abs/1311.5633 s. de santis _
et al _ , proceedings of 2011 particle accelerator conference , new york , ny , usa , pp .
mop228 r j goldstone and p h rutherford , introduction to plasma physics institute of physics publishing , 1995 c. celata phys .
st - accel .
beams 14 , 041003 ( 2011 ) d. sagan , j. a. crittenden , d. rubin and e. forest , proceedings of particle accelerator conference 2003 , portland , or , usa pp.1023 s. casalbuoni , s. schleede , d. saez de jauregui , m. hagelstein , and p. f. tavares , phys . rev .
beams 13 , 073201 ( 2010 ) s. federmann , f. caspers , and e. mahner , phys .
beams 14 , 012802 ( 2011 ) n. eddy , j. crisp , i. kourbanis , k. seiya , b. zwaska , s. de santis proceedings of particle accelerator conference 2009 , vancouver , bc , canada pp . we4grc02 j. c. thangaraj , n. eddy , b. zwaska , j. crisp , i. kourbanis , k. seiya proceedings of the electron cloud workshop 2010 , ithaca , new york , usa pp .
dia00 j d jackson .
classical electrodynamics , wiley , new york , ny , 3rd ed . edition , ( 1999 ) | the use of transverse electric ( te ) waves has proved to be a powerful , noninvasive method for estimating the densities of electron clouds formed in particle accelerators .
results from the plasma simulation program vsim have served as a useful guide for experimental studies related to this method , which have been performed at various accelerator facilities .
this paper provides results of the simulation and modeling work done in conjunction with experimental efforts carried out at the cornell electron storage ring test accelerator " ( cesrta ) .
this paper begins with a discussion of the phase shift induced by electron clouds in the transmission of rf waves , followed by the effect of reflections along the beam pipe , simulation of the resonant standing wave frequency shifts and finally the effects of external magnetic fields , namely dipoles and wigglers .
a derivation of the dispersion relationship of wave propagation for arbitrary geometries in field free regions with a cold , uniform cloud density is also provided . |
as is well known , the deflection of light by a gravitating body was one of the first predictions of einstein s general theory of relativity to be observationally confirmed .
later on , einstein himself predicted what is called today a _
microlens _ : the momentarily increase in apparent brightness of a background star as it passes close to a foreground massive body einstein . both the deflection of light , and the change in apparent brightness of a radiation source by an external gravitational field , are collectively known as a _ gravitational lens_. nowadays , _
gravitational lensing _ is a very active area of research , and it has found applications ranging from the search of extrasolar planets and compact dark matter to estimate the value of the cosmological parameters @xcite . in most of these applications
it is only necessary to assume that the gravitational field is weak and that the deflection angle ( due to a spherically symmetric body of mass @xmath1 can be approximated by : @xmath2 , where @xmath3 is the impact parameter @xcite . on the other hand
, it is well known that for a schwarzschild black hole , the deflection angle diverges as @xmath4 , allowing photons to orbit the black hole many times before reaching the observer .
this gives rise to an infinite set of images at both sides of the black hole @xcite .
notice that in this region the gravitational field is no longer weak , and the above approximation fails .
recently , a paper by virbhadra and ellis have renewed interest in such images , which they called _ relativistic images _ @xcite .
later on , bozza @xcite and eiroa , romero and torres @xcite developed an approximation method for the case of strong _ spherically - symmetric _ gravitational fields .
in fact , by expanding the deflection angle near the point of divergence , these authors were able to find analytic expressions for the positions and magnifications of the resulting relativistic images .
interestingly , such images were only characterized by the number of windings around the black hole .
although several authors have began to study gravitational lenses with a rotating black hole as a gravitational deflector ( see @xcite and references therein ) , most of their approaches to this subject still use the weak field approximation , and focus only in null geodesic motion at the black hole s equatorial plane . in passing ,
we shall mention for the interested reader a nice discussion of a kerr black hole as a gravitational lens by bray @xcite .
unfortunately , in this work there are approximations that are valid only for small deviations from the straight line path , and therefore are not suited for studying relativistic images . if we want to consider the phenomenology of the relativistic images in the strong gravitational field of a kerr black hole , no approximations can be taken and we need to work with the full equations of motion for null rays .
there have been numerous articles about the motion of null rays in the gravitational field of a kerr black hole @xcite .
although some of them have address the gravitational lens problem , most have concentrated in the observational effects on accretion disks and sources orbiting at the equatorial plane of the black hole and have not discussed the phenomenology of the relativistic images .
recently , bozza @xcite studied quasi - equatorial " orbits of photons around a kerr black hole and provided analytical expressions for the positions and magnifications of the relativistic images .
however , these approximations fail when the observer and the source are located far away from the equatorial plane .
moreover , as we show in sec .
v , bozza s procedure begins to fail when the black hole s angular momentum increases near its maximum value , even if the observer and the source are close to the rotating black hole s equatorial plane . in this paper , we shall discuss the phenomenology of the relativistic images by using the exact null equations of motion , with the assumption that the observer and the source are far away from the black hole .
although this new procedure is far more complicated than previous works on this subject , it allow us to calculate , estimate , and discuss for the first time the observational properties of the relativistic images for arbitrary source and observer inclinations in a kerr gravitational lens .
therefore the purpose of this paper is two - fold .
first , to extend the study of relativistic images for the case when the gravitational deflector is a rotating kerr black hole . secondly , since the trajectory of a photon will not always will be confined to a plane , we shall also be concerned with photon trajectories off the black hole s equatorial plane . as a consequence of this undertaking
, we show below that all relativistic images deflected by a rotating kerr black hole are characterized by only two integers numbers : namely , the number of turning points in the polar coordinate @xmath0 , and the number of windings around the black hole s rotation axis . to facilitate reading this paper is divided as follows : in sec .
ii to make this study self - contained a review of the null geodesic s equations of motion in the kerr space - time is briefly discussed . in sec .
iii we explain the gravitational lens geometry , and present a general classification for all images that allow us to define formally what we mean by relativistic images " . in sec .
iv , analytical expressions for the images magnification are derived . in sec .
v , we calculate the positions and magnifications of the relativistic images for several special cases , and compare our results with those in the literature . finally , in sec .
vi a discussion of our results is undertaken .
an appendix is also included to show how the null equations of motion can be solved in terms of elliptic integrals .
in this section we briefly review the equations of motion for light rays in the kerr space - time .
we use the usual boyer - lidquist coordinates which , at infinity , are equivalent to the standard spherical coordinates .
we also discuss about the relevant range of coordinates and constants of motion for the case of an observer and a source located far away from the black hole .
details about the relation between the null geodesic equations , the gravitational lens geometry , and observable quantities are given in the sec .
iii . by a convenient choice of the affine parameter ,
null geodesics in kerr space - time can be described by the following first - order differential system @xcite : @xmath5 where @xmath6 r - a^2 \eta \ ; , \\ \nonumber\\
\label{theta } \theta(\theta ) & = & \eta + a^2 \cos^2\theta - \lambda^2 \cot^2\theta \ ; , \\ \nonumber\\ p & = & ( r^2 + a^2 ) - \lambda a\ ; , \\
\nonumber \\ \sigma & = & r^2 + a^2 \cos^2\theta \ ; , \\
\nonumber\\ \delta & = & r^2 - 2 r + a^2\;.\end{aligned}\ ] ] in the above , @xmath7 is the four - velocity ( @xmath8 is an affine parameter ) , @xmath9 and @xmath10 are the constants of motion , @xmath11 is the black hole s angular momentum per unit mass , and units are chosen such as @xmath12 .
as is well known , for a kerr black hole , the parameter @xmath13 is restricted to @xmath14 . from eqs .
( [ ur ] ) - ( [ ut ] ) it follows that the relevant integrals of motion are @xmath15 ^ 2}\,dr \nonumber \\ & & + \int^\theta \frac{a^2 \cos^2 \theta } { \pm \sqrt{\theta(\theta)}}\,d\theta\;.\end{aligned}\ ] ] the signs of @xmath16 and @xmath17 are those of @xmath18 and @xmath19 respectively .
thus , the positive sign is chosen when the lower integration limit is smaller than the upper limit , and the negative sign otherwise .
for an observer and a source located far away from the black hole , the relevant radial integrals can be written as follows : @xmath20 where @xmath21 is the only turning point in the photon s trajectory , and it is defined by the largest positive root of @xmath22 . for the angular integrals ,
however , there could be more than one turning point . in consequence ,
the most general trajectory is described by @xmath23 the turning points in @xmath24 are defined by @xmath25 and are given by @xmath26^{1/2 } \right\}\;. \nonumber \\\end{aligned}\ ] ] it is easy to show from eq .
( [ utheta ] ) that for any pair of parameters @xmath27 , the motion in @xmath24 is bounded by @xmath28 and @xmath29 .
thus , the space of parameters is restricted because @xmath30 .
this constrain will be implemented in sec .
iii after discussing the lens geometry .
finally , the deflection angle @xmath31 has to be chosen to satisfy a given source - observer geometry ( see sec .
iii for details ) .
now we want to know what region of the parameter space ( @xmath32 ) correspond to photons that after reaching @xmath21 can escape to infinity .
writing eq .
( [ utheta ] ) as @xmath33 we see that @xmath10 could be negative .
however , assuming a photon crosses the equator ( @xmath34 ) , implies that @xmath35 . since in this research
we consider a source behind the black hole and we are mostly interested in images formed by photons that go around the black hole before reaching the observer , we will only consider the case of positive @xmath10 . for a photon to be able to return to infinity ,
we need @xmath36 at @xmath37 . setting @xmath38 we get : @xmath39}{a^2(1 - \underline{r})^2}\;,\end{aligned}\ ] ] where @xmath40 is the lower bound of @xmath21 with @xmath41 , and @xmath42 is the kerr black hole s horizon .
this is a parametric curve in the ( @xmath43 ) space and it is shown in fig .
[ parspace ] for the case @xmath44 .
photons with constants of motion inside the shaded region do not have a turning point outside the horizon so they will fall into the kerr black hole . ) inside the shaded region will fall into the black hole .
the parametric curve is for @xmath45 and is defined by eqs .
( [ lambdamin ] ) and ( [ etamin]).,width=317,height=226 ] this is analogous to the motion of null geodesics around a schwarzschild black hole , where any photon with impact parameter @xmath46 can not escape to infinity @xcite .
this value of the parameter @xmath3 , correspond to @xmath47 which defines the photon sphere " . in the parameter space
( @xmath32 ) , this correspond to a close region bounded by the line @xmath48 and the curve , @xmath49 .
this forbidden region in the parameter space shall be called : _
photon region_. the radial integrals of eqs .
( [ firstint ] ) and ( [ secondint ] ) diverge at the boundary of the photon region ( except at the line @xmath48 ) and take complex value inside of it .
therefore , knowledge of the mapping of this region will allow us to ensure that the equations of motion will remain well behaved .
since kerr space - time is asymptotically flat , an observer far away from the black hole ( @xmath50 ) can set up a reference euclidean coordinate system ( @xmath51 ) with the black hole at the origin ( see fig .
[ lensgeometry ] ) .
the boyer - lidquist coordinates coincide with this reference frame _ only for large @xmath52_. the coordinate system is chosen so that , as seen from infinity , the black hole is rotating around the @xmath53 axis . for @xmath54
the rotation will be assumed to be in the counterclockwise direction as seen from the positive @xmath53 axis . without loss of generality ,
we choose @xmath55 so that the coordinates of the observers in the boyer - lidquist system are ( @xmath56 ) .
similarly , for the source we have ( @xmath57 ) . with the black hole at the origin .
the boyer - lidquist coordinates coincide with this system only at infinity .
the reference frame is chosen so that , as seen from infinity , the black hole is rotating around the @xmath53 axis . in this system ,
the line joining the origin with the observer is normal to the @xmath58-@xmath59 plane .
the tangent vector to an incoming light ray defines a straight line , which intersects the @xmath58-@xmath59 plane at the point ( @xmath60).,width=326,height=289 ] in the observer s reference frame , an incoming light ray is described by a parametric curve @xmath61 , @xmath62 , @xmath63 , where @xmath64 . for large @xmath52
this is just the usual radial coordinate in the boyer - lidquist system . at the location of the observer
, the tangent vector to the parametric curve is given by : @xmath65 this vector describes a straight line which intersect the @xmath66-@xmath67 plane shown in fig . [ lensgeometry ] at ( @xmath68 ) .
a line joining the origin with the observer is normal to this plane .
we call this plane the _ observer s sky_. the point ( @xmath69 ) in this plane is the point ( @xmath70 ) in the ( @xmath51 ) system . changing to spherical coordinates and using the equations of the straight line ,
is easy to show that @xmath71 by using eqs .
( [ ur ] ) - ( [ uphi ] ) in ( [ alphai ] ) and ( [ betai ] ) and further assuming @xmath72 , it is possible to relate the constants of motion @xmath73 and @xmath74 to the position of the images in the observer s sky : @xmath75 these equations can , in turn , be written in terms of the angles ( @xmath76 ) in the observer s sky by : @xmath77 , @xmath78 .
we would like to mention that eqs .
( [ alphai ] ) - ( [ eta ] ) are not equivalent to eqs .
( 7 ) , ( 8) , ( 11 ) and ( 12 ) of ref . @xcite .
however , our equations exactly coincide with eqs .
( 28a ) and ( 28b ) of ref .
we believe that in ref .
@xcite , there is a calculation mistake in the intersection of the tangent to the light ray and the @xmath58-@xmath59 plane .
the author in ref .
@xcite also ignore any contribution of the black hole s spin when expanding @xmath79 and @xmath80 .
as can be seen later , it is also useful for our purposes to write the source s angular coordinates ( @xmath81 ) in terms of its position in the observer s sky . from the geometry of fig .
[ lensgeometry ] , we obtain @xmath82 where ( @xmath83 ) are the angles of the source in the observer s sky . inverting eqs .
( [ xs ] ) and ( [ ys ] ) for the polar and azimuthal angles , it is convenient to consider good source alignments ( e.g. small @xmath84 and @xmath85 ) .
the motivation for this approximation will become clear when we consider the magnifications of the relativistic images ( defined below ) . if the observer is not exactly at @xmath86 , we can approximate : @xmath87 and @xmath88 with @xmath89 and @xmath90 ( an observer exactly at @xmath86 will be considered in sec .
v ) . expanding eqs .
( [ xs ] ) and ( [ ys ] ) to second order in the perturbations , one finds that @xmath91 now we want to know how the restrictions in the parameter space ( @xmath92 ) and in the lens geometry are reflected in the possible values of the image coordinates ( @xmath93 ) .
we begin with the photon region discussed in sec .
this constraint correspond to a closed region in the observer s sky .
it is useful to write its boundary as a parametric curve @xmath94 , where @xmath95 and @xmath96 . inserting eqs .
( [ lambdamin ] ) , ( [ etamin ] ) , ( [ lambda ] ) and ( eta ) in the definition of @xmath97 ( using the small angle approximations @xmath98 , @xmath99 ) , we can solve for @xmath100 and obtain the desired parametric curve . when doing so , one encounters a sixth order polynomial in @xmath40 .
the largest positive root is valid for @xmath101 and the second largest positive root for @xmath102 .
another constraint comes from the polar movement of the light rays . in sec .
ii we pointed out that the motion in @xmath0 was restricted between the turning points @xmath103 defined by eq .
( [ um ] ) . therefore , points in the parameters space where the inequality @xmath104 is not satisfied must be discarded .
now we will prove that if @xmath73 and @xmath74 are given by eqs .
( lambda ) and ( [ eta ] ) , then the inequality is always satisfied for @xmath105 . using the notation : @xmath106 , @xmath107 ( @xmath108 is any subscript ) , the above inequality is equivalent to @xmath109 ( @xmath110 stands for min / max " ) . in this notation ,
the condition for a turning point in @xmath0 , @xmath111 , becomes @xmath112 writing @xmath113 and substituting into eq .
( [ turneq ] ) , we get @xmath114 ^ 2 - 4a^2\beta_i^2 w_o } \ ; \right\}\;.\end{aligned}\ ] ] on the other hand , using eqs .
( [ lambda ] ) and ( [ eta ] ) we have @xmath115 the radius " @xmath116 must be greater or equal than the boundary of the photon region defined by eqs .
( [ lambdamin ] ) and ( [ etamin ] ) and the line @xmath48 .
it is then easy to show that the minimum value of this radius " is reached when @xmath48 and @xmath117 .
the actual minimum value is thus @xmath118 .
therefore , by eq .
( [ rad ] ) we have that @xmath119 , and since @xmath120 , it follows the inequality @xmath121 and hence , @xmath122 .
the equality is satisfied only for @xmath123 .
we then can conclude that @xmath124 .
if @xmath125 , we can ensure that @xmath126 by doing the same construction in the frame of the source : @xmath127 where @xmath128 and @xmath129 are the coordinates in the source s sky . by using eqs .
( [ lambda ] ) , ( [ eta ] ) , ( [ lambdasource ] ) and ( [ etasource ] ) it follows that @xmath130 since for physically relevant situations we must have @xmath131 ( where the equality holds only for @xmath132 ) , the excluded points will be those between the curves @xmath133^{1/2 } \;,\end{aligned}\ ] ] where @xmath134 and @xmath135 .
the forbidden regions in the observer s sky are shown in fig .
[ freg ] for the case of @xmath125 . for @xmath136 ,
only the photon region is present . .
the forbidden regions are shown in shaded gray . in this case
we used @xmath137 , @xmath138 and @xmath139 .
for @xmath136 only the photon region is present.,width=326,height=263 ] next , we focus in relating the null equations of motion considered in sec .
ii with the geometry of the gravitational lens system . in the familiar case of a schwarzschild black hole lens , the movement of light rays
is restricted to a plane .
therefore , without loss of generality we can work at the equatorial plane : @xmath34 .
then , by substituting eq .
( [ firstint ] ) in ( [ secondint ] ) and setting @xmath140 , we obtain the familiar schwarzschild deflection angle : @xmath141 where @xmath48 by the fact that we are working at the equatorial plane [ see eq .
( [ etacarter ] ) ] . this deflection angle can be written in terms @xmath142 using eq .
( [ lambda ] ) .
then , by use of the familiar lens equation " we can solve for the position of the virtual " images @xcite .
however , another approach is to give @xmath143 in terms of the geometry of the observer - source pair and solve eq .
( [ schang ] ) for @xmath142 . in this case
we have @xmath144 where @xmath145 is the number of windings around the @xmath53 axis .
the solutions for @xmath146 are the familiar weak field images , and for @xmath147 we have the relativistic images studied in @xcite .
this last method is more useful when we consider trajectories outside the equatorial plane . in this case ,
( [ firstint ] ) and ( [ secondint ] ) with ( [ dphi ] ) become our lens equations " .
nevertheless , for trajectories outside the equatorial plane , an additional parameter appears : the number of turning points in the polar coordinate @xmath24 ( @xmath110 ) . for the case of the schwarzschild black hole it is easy to prove that @xmath110 is related to @xmath148 as @xmath149 by the fact that photons travel only in a plane .
what about the case of a kerr black hole ?
in this situation , since the movement of light rays is not necessarily restricted to a plane , we must consider @xmath148 and @xmath110 as independent parameters . additionally , in trying to write @xmath150 as in eq .
( [ dphi ] ) one is faced with a possible complication : unlike in the schwarzschild space - time , the kerr geometry admits turning points in @xmath151 which could complicate the analysis . by using eq
( [ uphi ] ) with @xmath152 is easy to show that the possible turning points occur at @xmath153 both solutions are surfaces of revolution around the rotation axis of the black hole .
it is easy to prove that for a given @xmath154 , only one of the surfaces lies outside the horizon . since in our case
we are considering photons that come from infinity , reach a single turning point in @xmath52 and return to infinity , they can cross such a surface at most in two points . in consequence
, there can be at most two turning points in @xmath155 .
however , by using eqs .
( [ uphi ] ) and ( [ ut ] ) for an observer and a source at large @xmath52 , it is easy to prove that : @xmath156 ( conservation of angular momentum ) . therefore , the asymptotic sign of @xmath157 must be unchanged by the gravitational interaction .
this rule out a single turning point in @xmath155 which would obviously change that sign .
we conclude that the number of turning point in @xmath155 must be zero or two . in that case , the most general expression for @xmath158 is still given by eq .
( [ dphi ] ) but with the following modification : the first case is to be use when the right hand side of eq .
( [ secondint ] ) is negative , and the second case otherwise . by using eqs .
( [ lambda ] ) , ( [ eta ] ) , ( [ thetas ] ) and ( [ phis ] ) , the lens equations " ( [ firstint ] ) and ( [ secondint ] ) can be expressed as @xmath159 where @xmath160 and @xmath161 ( @xmath162 and @xmath163 ) are the radial ( angular ) integrals of eqs .
( [ firstint ] ) and ( [ secondint ] ) respectively , and @xmath164 and @xmath165 are the number of windings around the @xmath53 axis and the number of turning points in the polar coordinate @xmath0 respectively . although the integers @xmath148 and @xmath110 should be considered independent , as we will see in sec .
iv , the magnification does not depend directly on @xmath148 .
therefore , in this paper we consider @xmath110 to be the fundamental parameter .
this interpretation is confirmed by our numerical results where we find that the magnification always decreases as we increase @xmath110 , and for a given @xmath110 , it can ever increase for images with larger @xmath148 ( see sec .
v ) . in the familiar schwarzschild gravitational lens
, we always have two images which are formed by light rays which suffer small deviations in their trajectories .
they are called _ weak field images_. additionally , we have an infinite set of faint images at both sides of the black hole @xcite .
these are _ relativistic images _ , and are the result of photons that orbit the black hole several times before reaching the observer . in kerr space - time , where photons trajectories are no longer confined to a plane , the concept of orbiting the black hole several times " can be subtle . for this reason ,
we classify images as follows : images with @xmath166 are called _ direct images _ ( di hereafter ) , and images with @xmath167 are called _ relativistic images _ of order @xmath110 ( ri hereafter ) . as we show in the appendix , eqs .
( [ lens1 ] ) and ( [ lens2 ] ) can be written in terms of elliptic integrals , and are highly non - linear in all arguments with the exception of @xmath148 and @xmath110 .
therefore , their solution require numerical and graphical methods . to solve the lens equations , we use the standard routine findroot
" build in mathematica . to give an initial approximation and to ensure that a solution exist at all
, we use a variety of graphical methods .
since , as we will see in sec .
v , ris appear very near the boundary of the photon region , we found that is useful to write : @xmath168 and @xmath169 , where @xmath170 , @xmath171 and @xmath172 is the boundary of the photon region as discussed earlier in this section .
we then express all functions in terms of @xmath173 and @xmath174 to avoid entering the photon region .
we find that ris form at very small @xmath173 . to find the images we plot the surfaces @xmath175 and @xmath176 for a given source position and image numbers @xmath177 .
the intersection of both surfaces with the plane @xmath178 form various curves in that plane .
if the curves formed by the two surfaces intersect each other , there is a solution .
visual inspection of these curves allow us to give an initial approximation for the numerical routine . as an example
, we plot part of both surfaces in fig .
[ approxmethod ] for @xmath179 and @xmath180 . and @xmath181 for a source at @xmath182 arcmin , an observer at @xmath183 and for @xmath179 , @xmath180 , @xmath184 .
we have chosen @xmath185 kpc as in the numerical examples of section v. the intersection of the surfaces with the @xmath186 plane ( gray ) form various curves in that plane .
the intersection of two such curves marks the position of a solution to the lens equations " .
the range of @xmath53 has been restricted to @xmath187 in order to clarify the intersection of the curves .
, width=326,height=340 ] we usually do this for the two lowest values of @xmath110 where a solution can be found .
the possible values of @xmath148 are bounded from above by the requirement @xmath188 . in consequence , for a given @xmath110 we usually find images for just a few values of @xmath148 .
this procedure is very tedious and time consuming , and for these reasons we are unable to give a complete phenomenological description of the behavior of the images for any given observer - source geometry .
instead , we shall present some numerical examples of the kind of behavior that can be expected by having a kerr black hole as a gravitational deflector and null geodesic motion off the equatorial plane . as pointed out before ,
several authors have already obtained analytic approximations to the radial integrals in the strong field limit " for schwarzschild , reissner - norsdstrn and kerr black holes @xcite .
their approximations are valid for relativistic photons , with @xmath189 close to @xmath40 and for small deviations from the equatorial plane in the case of the kerr black hole . however , since in this article we are interested in orbits that can deviate significantly from the equatorial plane we can no longer use such approximation schemes .
moreover , their whole approach fail for the angular integrals because , in general , there is no relation between @xmath21 and the turning points in @xmath0 . in sec .
v , to make contact with the relevant literature , we consider trajectories close to the equatorial plane to be compared with bozza @xcite procedures .
the magnification of an image is defined as the ratio of the observed flux to the flux of the unlensed source . by liouville
s theorem , the surface brightness is unchanged by the gravitational light deflection @xcite . therefore , the magnification is defined as the ratio of the solid angle subtended by the image to the solid angle of the unlensed source : @xmath190 where @xmath163 is the jacobian of the transformation ( @xmath191 ) @xmath192 ( @xmath76 ) . writing @xmath193 and @xmath194 we can find expressions for @xmath195 , @xmath196 , @xmath197 and @xmath198 by differentiating the lens equations ( [ lens1 ] ) and ( [ lens2 ] ) with respect to @xmath199 and @xmath200 . after some algebraic manipulations , we find that @xmath201 where @xmath202 the above derivatives are very cumbersome and we will not expand it , neither show explicitly them here .
in fact , instead we shall use the equivalent numerical derivatives in all forthcoming calculations .
note that the magnification do not depend on @xmath148 directly , since we have to take the derivative of @xmath31 . because of the complexity of these expressions , we are unable to give a complete description of the caustic structure of the kerr space - time .
thus , we shall limit ourselves to compute the magnification of the images that we found . for a description of the caustic structure of a kerr space - time ,
see @xcite .
in this section we present numerical calculations for the positions and magnifications of the ris for different source - observer geometries .
our purpose is to provide a physical insight of the phenomenology that can be expected from a rotating black hole behaving as a gravitational deflector .
also , our general procedures and numerical solutions can provide a set of test - bed " calculations for more sophisticated gravitational lens models to be developed in the near future . to be able to compare our results with recent published articles @xcite we shall consider a gravitational lens composed of a rotating black hole at the galactic center with a mass of @xmath203 , and located at a distance of @xmath204 kpc @xcite .
we shall also take @xmath205 .
the section is subdivided as follows : we first consider two simple cases involving a schwarzschild black hole ( @xmath140 ) and an observer located at the pole ( @xmath86 ) .
next , as a consistency check , we consider an observer at the equator ( @xmath206 ) and compare our calculations with those of bozza @xcite . finally , we work out a more general case of an observer at say @xmath207 .
the ris are classified as follows : we use a plus ( + ) sign for images with sign @xmath208 sign @xmath85 ( e.g. in the same side " of the source ) and a negative ( @xmath209 ) sign otherwise .
we do all calculations for the two lowest values of @xmath110 where a solution exist .
we find that , in general , images with larger @xmath110 are more demagnified .
the following notation will also be used : @xmath210 , @xmath211 , where @xmath212 are the angular positions in the observer s sky . in all geometric configurations considered in this section
, we found that for @xmath213 we recovered the usual weak field images without any noticeable effect from the black hole s spin .
moreover , we could not find any image with @xmath214 for this kind of geometry . in the appendix
we show that for @xmath140 , the angular integrals can be solved in close form . as expected , we find that the ris are always found in the line joining the source position @xmath215 and origin of the observer s sky regardless of the inclination of the observer @xmath216 .
the separation @xmath217 of the two outermost images is always the same regardless of the source position ( or @xmath216 ) .
they are : @xmath218arcsec ( @xmath219 ) and @xmath220arcsec ( @xmath221 ) for the two lowest order images .
these results are close to those found in @xcite : @xmath222arcsec and @xmath223arcsec respectively .
the magnifications calculated by eq .
( [ mag ] ) are also close to those in the literature .
however , we encountered problems of numerical noise when using the exact expression for @xmath224 .
for that reason , to check that we obtain the right magnifications , we used the expression for @xmath225 [ eq .
( [ jelliptic ] ) ] and set @xmath226 .
the magnifications obtained this way for the outermost image as a function of the source separation agree with the literature @xcite .
for instance , we obtained @xmath227 and @xmath228 for @xmath229 arcsec and @xmath230arcsec respectively . for comparision , in ref .
@xcite we read : @xmath231 and @xmath232 respectively .
although is very unlikely that an observer will be exactly at @xmath234 , this is the simplest case that can be solved quasi - analytically . using eq .
( [ lambda ] ) for @xmath235 we find that @xmath236 .
the minimum value of @xmath74 , corresponding to the photon sphere , is found from eqs .
( [ lambdamin ] ) and ( [ etamin ] ) by setting @xmath237 .
we obtain @xmath238 where @xmath239\ ; , \\ \nonumber \\
\tan \psi = \frac{3\sqrt{3 } a \sqrt{108 - 135 a^2 + 36 a^4 - 4 a^6}}{54 - 81a^2 + 18a^4 - 2a^6}\;.\end{aligned}\ ] ] eq .
( [ etaminpolar ] ) is a decreasing function of @xmath13 satisfying @xmath240 .
since we can relate the position of the images to @xmath74 by eq .
( [ eta ] ) : @xmath241 , eq .
( [ etaminpolar ] ) sets a lower bound to the separation of the images @xmath217 .
the null equations of motion simplify considerable for @xmath242 .
first we note that from eq .
( 6 ) , there are no turning points in @xmath0 since @xmath243 .
therefore , the angular integration limits can be shown to be @xmath244 where @xmath245 is the parity of the image ( @xmath209 for images that are in the same side of the source when @xmath246 and @xmath247 for images in the opposite side ) . here , @xmath148 is the number of loops around the black hole . to relate @xmath248 to the unperturbed position of the source in the observer s sky we can no longer use the approximate expressions ( [ thetas ] ) and ( [ phis ] ) .
instead , it can be shown that @xmath249 where we are assuming that @xmath250 .
now , using eqs .
( [ helliptic ] ) , ( [ lelliptic ] ) and ( [ int2 ] ) given in the appendix , the null equations of motion ( [ firstint ] ) and ( [ secondint ] ) become : @xmath251 \right .
\nonumber \\ & & \;\ ; \;\;\;\;\;\;\ ; \;\;\;\;\ ; \left .
+ 2 ( 2n+1)\ , k \left(\frac{a^2}{r_o^2
\xi_i^2}\right ) \right\ } = 0\;,\end{aligned}\ ] ] @xmath252\;,\end{aligned}\ ] ] where @xmath253 and @xmath254 , @xmath255 , @xmath256 , @xmath257 ( @xmath258 ) are the roots of @xmath259 , and eq .
( [ firstintpolar ] ) is valid only for @xmath260 ( for @xmath261 the right hand side of eq .
( [ secondintpolar ] ) has to be substituted by eq .
( [ lelliptica0 ] ) of the appendix ) .
the procedure to calculate the positions of the ris is the following : for a given source separation @xmath262 , we use eq .
( [ firstintpolar ] ) to calculate the angular separation of the ri ( @xmath217 ) and then insert this value in eq .
( [ secondintpolar ] ) to obtain the offset from the source inclination @xmath263 . in calculating
@xmath217 one has to use eq .
( [ etaminpolar ] ) to determine its minimum value and avoid numerical problems .
note that for @xmath264 , eq .
( [ firstintpolar ] ) gives @xmath265 as expected . in fig .
[ deltavarphi ] we plot @xmath266 for the three outermost images ( @xmath267 ) as a function of @xmath13 for a source located at @xmath268arcsec . ) as a function of the normalized black hole angular momentum @xmath13 . here
we consider an observer located at @xmath234 and a source located at @xmath269arcsec .
the parameter @xmath148 is the number of loops around the black hole .
these plots are for @xmath270 , but photons with @xmath271 have almost the same curves ( indistinguishable within the plot resolution).,width=326,height=229 ] the sign of @xmath266 is that of @xmath13 ( which is what one intuitively expects ) . for a given @xmath148 , both images with opposite parity have almost the same @xmath266 .
however , the greater the number of loops @xmath148 , the greater the deflection @xmath266 for a given @xmath13 . the physical picture emerging is a very simple one : the angular momentum of the black hole just adds a twist " in the direction of the rotation to the usual schwarzschild trajectory . to give an intuition of the full movement of the images , including their separation @xmath217 , in fig .
[ polarmovement ] we have plotted the position of the two lowest order images ( @xmath272 ) as a function of the spin parameter @xmath13 .
and @xmath273 ) as a function of the normalized angular momentum of the black hole @xmath13 .
the arrows in the curves represent the direction of the movement as we increase @xmath13 from 0 to 1 .
the source is located at @xmath268arcsec and @xmath274.,width=321,height=342 ] what about the magnifications ? in this simple case where we have circular symmetry as seen by the observer
the magnification is given by @xmath275 where @xmath276 is the left hand side of eq .
( [ firstintpolar ] ) .
this expression for the magnification can be verified by calculating the jacobian of the transformation @xmath277 .
we find that , as in the schwarzschild case , both images with the same winding number @xmath148 have approximately the same magnification ( to at least three significant figures ) . in fig .
[ magpolarplot ] we plot the magnifications for the three lowest order images ( @xmath278 ) and with a source located at @xmath268arcsec . ) for an observer at @xmath234 and a source at @xmath279arcsec . here
we consider the case of @xmath273 , but the magnifications for @xmath280 are almost the same ( indistinguishable in this plot).,width=329,height=210 ] the net effect of the angular momentum in this case is to enhance the brightness of the images . for the schwarzschild case ( @xmath281 ) , the magnifications obtained with eq .
( [ magpolar ] ) agree perfectly with the results of ref .
@xcite . in this subsection
we consider an observer located exactly at @xmath283 .
our purpose is not to give a complete account of the phenomenology of the ris since , from an astronomical perspective , it is very unlikely that an observer will be exactly at the equator ( also setting @xmath283 does not significantly simplify our analysis ) . rather than that , this special case will serve as a consistency check by allowing us to compare some of our numerical results with those of bozza @xcite . in that article
, the author considered quasi - equatorial " orbits in kerr spacetime .
one of the main approximations employed was that the horizontal position of the ris ( @xmath199 in our notation ) was calculated independently using the familiar lens equation in the equatorial plane .
this , of course , assumed that the motion in @xmath151 was unaffected by the motion in @xmath0 .
therefore , we are forced to consider very small source declinations @xmath284 .
to this end ( and for simplicity ) we fix the source at @xmath285 arcsec and @xmath286arcsec ( @xmath287 ) .
then , we calculate the position and magnification of the lowest order ri in the same side of the source as a function of the spin parameter @xmath13 .
we fixed @xmath288 to be consistent with the reference , where the ris are classified only with the winding number @xmath148 .
we also use @xmath289 for latter convenience .
the results using the approximate equations ( 66 ) , ( 77 ) and ( 90 ) of ref .
@xcite and the equations of this paper are shown in table i. and , in their notation , @xmath290 is equivalent to @xmath289 in our notation . ] [ cols="^,^,^,^,^,^,^,^ " , ] with regard to the magnifications , we observe that as in previous cases , the effect of the black hole angular momentum is to demagnify the images ( although there is a slight increase in the magnification for @xmath291 ) .
also note that the images with @xmath292 in the first quadrant have magnifications very similar to those with @xmath293 ( located in the third quadrant ) .
therefore , these images should be considered dual " to each other since they form the pair of brightest images for a given value of @xmath13 ( @xmath294 ) .
an interesting consequence of the black hole rotation is that for large values of @xmath13 ( @xmath295 ) the two brightest images have different magnifications .
this is to be compared to the case of a schwarzschild black hole where the two brightest ris have exactly the same magnification .
also , for large @xmath13 the ris are very static as in the schwarzschild case ( although for @xmath264 they have @xmath265 ) .
in other words , their positions do not depend on the location of the source .
what is even more surprising is that for @xmath225 and within our numerical precision , it seems that the ratio of the magnifications of the two brightest ris is insensitive to the position of the source .
this raises the possibility of extracting information about the orientation and spin of the black hole by comparing the brightness of this two images .
however , to move this proposal forward we would need to study the behavior of the ratio of their magnifications for different values of @xmath216 .
we do not attempt to carry such analysis this time .
we hope that any new approximation scheme developed in the future can allow us to address that question .
in this article we have explored the phenomenology of strong field gravitational lensing by a kerr black hole .
in particular we have developed a general procedure to calculate the positions and magnification of all images for an observer and source far away from the black hole and at arbitrary inclinations .
we have applied our developed procedure to the case of a black hole at the galactic center with mass @xmath296 and at a coordinate distance of @xmath297 kpc .
we have reproduced the positions and magnifications of the lowest order relativistic images found in the references for a schwarzschild black hole and for quasi - equatorial " trajectories around a kerr black hole .
we have also presented new numerical results for the case of an observer located at @xmath183 . although we have not been able to give a full account of the phenomenology for all possible combination of the source - observer geometry and the spin parameter @xmath13 , our limited results were useful to get a physical insight of the effects of the black hole angular momentum in the strong - field regime of gravitational lensing .
moreover , our findings can serve as a test - bed " calculation for any improved new model of gravitational lens .
there is no doubt that observations of the strong - field regime of gravitational lensing will be an extremely challenging task in the near future .
this is because , as we have seen , the relativistic images are always highly demagnified .
however , if we are able to observe them in any foreseeable future , they will provide one of the best tests of einstein s general theory of relativity in strong gravitational fields .
moreover , as we have seen , they could provide new tools to astrophysics by allowing the measurement of the orientation and/or magnitude of the angular momentum of the black hole .
however , to fully confirm that this is the case , more research is needed toward developing an analytical solution of the strong - field gravitational lens problem in kerr space - time .
s. e. vzquez is very grateful to e. p. esteban for all his support and advise during his undergraduate years .
he would also like to thank nsf for a graduate research fellowship and the university of california at santa barbara for a broida excellence fellowship .
e. p. esteban thanks the support given by upr - humacao and rice university during his sabbatical leave . where @xmath299 is the normal elliptic integral of the first kind .
also , @xmath300 where @xmath254 , @xmath255 , @xmath256 , @xmath257 ( @xmath258 ) are the roots of @xmath259 . for the radial integral of eq .
( [ secondint ] ) we have : @xmath304 \frac{dr}{\sqrt{r(r ) } } \nonumber \\ & = & \frac{g}{r_a - 1}\left\ { \beta^2 \left [ 2 + \frac{\beta^2 ( 2 - \lambda)}{r_a - 1}\right]f(\psi , k ) +
2(1 - \beta^2)\left [ 1 + \frac{\beta^2 ( 2 - \lambda)}{r_a - 1}\right]\pi(\psi,\alpha^2,k ) + \frac{(1 - \beta^2)^2(2 - \lambda)}{r_a - 1 } v \right\ } \ ; , \nonumber \\\end{aligned}\ ] ] we begin by defining @xmath308 , such that @xmath309 since all angular integrals considered in this paper are symmetric around @xmath310 , we can restrict our interval to @xmath311 and discard the sign of eq .
( [ difold ] ) ( remember that all integrals are positive definite ) . however , we need to compensate for the case of a photon that crosses the equator . to this end
we define the operator @xmath312 for any two angles @xmath313 and @xmath314 . is clear that @xmath315 if both angles are at the same hemisphere and negative otherwise .
therefore , is easy to show that we can write @xmath316 \int_0^{\underline{u } } \;,\end{aligned}\ ] ] where @xmath317 and @xmath318 .
now , for a trajectory that encounters @xmath110 turning points ( @xmath324 ) we have @xmath325 \int_0^{u_s } \nonumber \\
+ \int_{u_o}^{u_m } + \left[1 - \textrm{sign}(\theta_o * \theta_{mo})\right ] \int_0^{u_o } \nonumber \\ + 2(m - 1 ) \int_{0}^{u_m}\ ; , \nonumber \\\end{aligned}\ ] ] where @xmath326\ ; , \\ \nonumber \\ \theta_{ms } & \equiv & \left\ { \begin{array}{ll } \theta_{mo}\ ; , & \hbox { $ m$ odd ; } \\ \pi - \theta_{mo}\ ; , & \hbox{$m$ even , } \\
\end{array } \right.\end{aligned}\ ] ] with @xmath200 as the ( possible ) position of the image . in deriving eq .
( [ thetaintegrals ] ) we have use the fact that , as discussed in section iii , @xmath327 . for @xmath328 we can write @xmath329 \int_0^{\underline{u } } \nonumber \\ & = & \int_{\underline{u}}^{u_m } - \int_{\overline{u}}^{u_m } + \left[1 - \textrm{sign}(\theta_s * \theta_o)\right ] \int_0^{\underline{u}}\ ; , \nonumber \\\end{aligned}\ ] ] where @xmath330 and @xmath331 . to write the angular integrals of eq .
( [ firstint ] ) as elliptic integrals we change variables to @xmath332 so that , for @xmath333 and @xmath334 in the same hemisphere , we have @xmath335 where @xmath336^{1/2 } \right\}\;,\end{aligned}\ ] ] and @xmath337 ( @xmath338 ) .
on the other hand we have @xmath339 where @xmath340 using eqs .
( [ thetaintegrals ] ) , ( [ m=0 ] ) , ( [ int1 ] ) and ( [ int2 ] ) we can write the left hand side of eq .
( [ firstint ] ) as @xmath341 f(\phi_s,\kappa)\right.\nonumber \\ & & \left .
+ \left[1 - \textrm{sign}(\theta_o * \theta_{mo})\right ] f(\phi_o,\kappa ) \right .
\nonumber \\ & & \left .
+ 2(m - 1)k(\kappa ) \right\}\;,\end{aligned}\ ] ] for @xmath324 .
here , @xmath342 is the compete elliptic integral of the first kind : @xmath343 . for @xmath328 we have @xmath344 f(\underline{\phi},\kappa ) \right\}\;,\nonumber\\\end{aligned}\ ] ] where @xmath345 and @xmath346 are given by eq .
( [ psij ] ) with the substitutions @xmath347 and @xmath348 respectively , and @xmath349 is given by eq .
( [ phij ] ) with the substitution @xmath350 .
now we turn our attention to the angular integrals of eq .
( [ secondint ] ) . making the usual change of variable ,
we get : @xmath351 where @xmath352 the second integral we need is @xmath353 \ ; , \nonumber \\\end{aligned}\ ] ] where @xmath354 @xmath355 + \frac{h\,\lambda}{1 - u_3 } \left\{\left(1 - \textrm{sign}(\theta_s * \theta_{ms } ) \right)\left [ f(\phi_s,\kappa ) - u_3 \pi(\phi_s,\rho^2,\kappa ) \right ] \right .
\nonumber \\ & & \left .
+ \left(1 - \textrm{sign}(\theta_o * \theta_{mo } ) \right)\left [ f(\phi_o,\kappa ) - u_3 \pi(\phi_o,\rho^2,\kappa ) \right ] + 2(m-1)\left [ k(\kappa ) - u_3 \pi(\rho^2,\kappa ) \right ] \right\}\;,\end{aligned}\ ] ] @xmath356 + \frac{h\,\lambda}{1 - u_3 } \left\{\left[1 - \textrm{sign}(\theta_s * \theta_{o } ) \right]\left [ f(\underline{\phi},\kappa ) - u_3 \pi(\underline{\phi},\rho^2,\kappa ) \right ] \right\}\ ; , \nonumber \\\end{aligned}\ ] ] for the case of a schwarzschild black hole ( @xmath264 ) the angular integrals can be solved in close form since the polynomial in the square root become one of second order : @xmath357\;,\end{aligned}\ ] ] where @xmath358 similarly , @xmath359\;.\end{aligned}\ ] ] therefore , using eqs .
( [ thetaintegrals ] ) and ( [ m=0 ] ) , the right hand side of eq .
( [ firstint ] ) becomes @xmath360\;,\nonumber \\\end{aligned}\ ] ] for @xmath324 , and @xmath361 \right\}\ ; , \nonumber \\\end{aligned}\ ] ] for @xmath328 . using the expressions for @xmath160 , @xmath162 , @xmath161 and @xmath163 given above
, the lens equations " ( [ firstint ] ) and ( [ secondint ] ) become : @xmath368 with @xmath369 where @xmath370 is the number of windings around the @xmath53 axis , and @xmath73 , @xmath74 and @xmath248 can be written in terms the observer s sky coordinates by using eqs .
( [ lambda ] ) , ( [ eta ] ) , ( [ xs ] ) and ( [ ys ] ) . | we consider a kerr black hole acting as a gravitational deflector within the geometrical optics , and point source approximations .
the kerr black hole gravitational lens geometry consisting of an observer and a source located far away and placed at arbitrary inclinations with respect to the black hole s equatorial plane is studied in the strong field regime . for this geometry the null geodesics equations of our interest can go around the black hole several times before reaching the observer .
such photon trajectories are written in terms of the angular positions in the observer s sky and therefore become lens equations " . as a consequence
, we found for any image a simple classification scheme based in two integers numbers : the number of turning points in the polar coordinate @xmath0 , and the number of windings around the black hole s rotation axis .
as an application , and to make contact with the literature , we consider a supermassive kerr black hole at the galactic center as a gravitational deflector . in this case , we show that our proposed computational scheme works successfully by computing the positions and magnifications of the relativistic images for different source - observer geometries . in fact , it is shown that our general procedure and results for the positions and magnifications of the images off the black hole s equatorial plane , reduce and agree with well known cases found in the literature . |
when cracks develop slowly via the nucleation of damage ahead of the tip , the crack surfaces left behind appear to be rough . a question of great interest is what is the scaling exponent that characterizes the roughening of such surfaces and how to relate the value of the exponent to the physical phenomena that govern the crack propagation . for cracks surfaces in 2 + 1 dimensions the anisotropy of the fracture experiment results in a number of scaling exponents , making the attainment of a satisfactory theory quite difficult @xcite . on the other hand , for cracks in quasi 2-dimensional samples , where the resulting surfaces are 1 + 1 dimensional graphs ( rupture lines ) , the issues are clear at least in the sense that there exists one well - defined scaling exponent
this is conveniently defined by measuring @xmath6 where @xmath7 is the height of the graph above the euclidean coordinate @xmath8 that defines the crack direction and then defining some measure of the height fluctuations , for example @xmath9 for self - affine graphs the scaling exponent @xmath10 is defined via the scaling relation @xmath11 it is well known that random graphs are consistent with @xmath12 , whereas positively ( negatively ) correlated graphs are characterized by @xmath13 ( @xmath14 .
experiments on 2-dimensional samples tend to report scaling exponents in the range @xmath15 @xcite , indicating the existence of positive correlations between successive crack segments . in recent work a model was proposed in 1 + 1 dimensions for such slow crack propagation via damage nucleation and coalescence ahead of the crack tip @xcite .
a crucial aspect of this model is the existence of a typical length scale @xmath1 ahead of the crack tip where damage nucleation can take place @xcite . a physical picture that might support such a scenario ( though definitely not a unique one ) is that a small plastic zone of linear dimension @xmath1 forms around the crack tip and the relevant damage units involved in the process are plastic voids .
the idea is that since plastic deformation is typically associated with a limiting stress level ( denoted the yield stress " ) @xmath16 , the purely linear elastic divergent stresses are cut off such that they can not reach a critical level required for the nucleation of voids _ at the crack tip_. it is claimed that such critical levels of stress can be attained approximately near the outer boundary of the plastic zone , i.e. the elastic - plastic boundary .
whenever a void is nucleated in this region , it evolves and eventually coalesces with the current crack tip to form a new crack configuration .
the crack then evolves by successive applications of such nucleation and coalescence events .
@xcite demonstrated that the rupture lines generated by this model are self - affine rough graphs with a _ correlated _ scaling exponent @xmath17 .
not only the value of the roughness exponent is found to be significantly above the random walk exponent @xmath18 , it also appears close to the measured one @xcite .
the aim of this paper is to gain a deeper understanding of the origin of this result .
an essential question asked in the context of any such fracture growth model is how to represent and incorporate the effect of material disorder .
this issue is important since it asks how small scale features affect large scale properties , for example the power - law scaling of eq .
( [ power_law ] ) that indicates a lack of characteristic length scale . to our knowledge
there had been no systematic study of the role of the nature of material disorder in determining the roughness of cracks in 1 + 1 dimensions .
our aim here is to shed some light on this issue by demonstrating that different views of material disorder and the associated length scales have a qualitative effect on the scaling properties of rupture lines .
to this aim we elaborate on the type of material disorder adopted in the model described briefly above - referred to below as model a - and present a new model - referred to below as model b - that incorporates a different picture of material disorder . in model
b the disorder is quenched , and the stochasticity associated with the material heterogeneities is fixed a - priori in space and time . in model
a the disorder is annealed " in the sense that the stochasticity depends on the actual state of the system . whenever the disorder has some spatial characteristic scale we denote it by @xmath0 and call it the `` disorder length '' that should be compared to the previously introduced length @xmath1 .
thus model a is characterized by annealed disorder and @xmath3 , while model b is characterized by quenched disorder and @xmath19 .
one of the points of this paper is that this change in material disorder in model b is sufficient to destroy the positive correlations between successive crack segments observed in model a , changing the universality class of the model and ending up with a random graph with an asymptotic scaling exponent @xmath20 . in sect .
[ models ] we present in more detail model a and recall its results , elaborating on the way in which material disorder is incorporated into the model .
we then explain the modifications leading to model b. in sect .
[ results ] we present the new results for model b , compare them with model a and clarify the origin of the qualitative differences between them .
[ sum ] offers a summary and some concluding remarks .
the mathematical difficulty in developing a theory for the morphology of fracture surfaces is the necessity of calculating the linear - elastic stress fields for highly non - regular crack paths .
typically , the stress conditions near the crack tip depend _ non - linearly _ on the crack path @xmath6 .
formally , one has to solve the bi - laplace equation for the airy stress potential @xmath21 @xcite @xmath22 in the infinite plane with traction - free boundary conditions on the crack surfaces @xmath23 here @xmath24 is the arc - length parametrization of the crack shape and the @xmath25 denotes the stress acting in the @xmath26-th direction on a segment whose out - ward normal is the normal to the crack face at @xmath24 .
the stress tensor field @xmath27 is derivable form the airy stress potential @xmath21 according to @xmath28 the relevant experimental configuration for our purpose is that of global mode i fracture in which a system containing _ initially _ a straight crack in the @xmath8 direction , subjected to a stress applied in the direction @xmath7 , perpendicular to the crack . at infinity
we write the boundary conditions @xmath29 where @xmath30 is assumed to be constant .
note that even though the initial configuration is that of a straight crack , with material disorder the crack might deviate from the straight path , attaining an arbitrary rough shape .
solving the bi - laplace equation with boundary conditions on such an arbitrary boundary is quite a formidable task .
recently , we have developed a general method of solution based on iterated conformal maps @xcite . in this method ,
one starts with a crack for which the conformal map from the exterior of the unit circle to the exterior of the crack is known .
for example , in our case we start with a long straight crack in the form of a mathematical branch - cut , representing the common experimental practice of introducing the sample with a notch in order to localize the fracture process in a controlled way .
we can then grow the crack by little steps in the desired directions , computing at all times the conformal map from the exterior of the unit circle to the exterior of the resulting crack .
having the conformal map makes the _ exact _ calculation of the stress field straightforward in principle @xcite and highly affordable in practice .
the details of the method and its machine implementations are described in full detail in @xcite .
we consider a crack evolving under quasi - static conditions by the nucleation and coalescence of damage ahead of the crack tip .
these damage elements can be voids or micro - cracks .
we focus on situations where only one damage element nucleates before the process of coalescence .
this process is associated with a length scale @xmath1 characteristic of the distance of the damage element from the tip . a plausible physical picture for such a process was proposed in @xcite .
the idea is to identify @xmath1 with the size of the plastic zone that develops near the crack tip due to the large stresses concentrated there .
more specifically , it was assumed that the material flows plastically such as to reduce the stress field near the crack tip to a level determined by the yield stress @xmath16 .
mathematically , the statement is that the distortional energy @xmath31 , with @xmath32 , satisfies the relation @xcite @xmath33 inside the plastic zone . outside this region
, the stress field behaves linear - elastically . to find the outer boundary of the plastic zone , which has a characteristic length @xmath1
, we use the iterated conformal mapping solution of the linear - elastic problem .
we calculate the spatial curve for which eq .
( [ mises ] ) is satisfied when approaching the crack tip .
this curve defines the elastic - plastic boundary .
it was further shown @xcite that the hydrostatic tension @xmath34 , defined as @xmath35 attains a larger value near the elastic - plastic boundary than inside the plastic zone . under the physically plausible assumption that damage will nucleate in regions where @xmath34 exceeds some threshold value @xmath36
, we expect damage to nucleate near this boundary . as was explained before , after damage nucleates it evolves such that it coalesces with the tip , generating a new plastic zone under the influence of the liner - elastic fields and so on
. note that in this physical interpretation the damage elements are plastic voids and the coalescence process is assumed plastic as well ( for example , necking of the ligament between the crack tip and the void ) . in fact , as we are not resolving the processes by which the crack tip coalesces with the void ahead of it , using the nucleation site only as a pointer for the advance of the crack .
therefore , we are only interested in the roughness of the crack on scales larger than @xmath1
. naturally , the precise location of the nucleating damage may be stochastic due to material disorder . to quantify this ,
assume that nucleation occurs only at locations @xmath37 in which the hydrostatic tension @xmath34 exceeds some threshold value @xmath36 . given the distribution @xmath38 we consider a probability density function @xmath39 .
one has in mind an activation process for the nucleation of damage , and this activation is more efficient when @xmath40 is large .
the probability for activation vanishes for @xmath41 .
this activation may be due to stress corrosion in one case , or due to other mechanism in another case , but the important thing to note is that @xmath38 is a long - ranged functional of the history of crack evolution , potentially leading to the long range correlations implied by @xmath13 . the fact that damage nucleates depending on the stress field @xmath27 through @xmath34 without reference to any pre - determined distribution of disorder implies that the model is characterized by _
annealed disorder_. note that the same formulation describes equally well a situation in which damage nucleates at points where the material is weak , if the _
random _ weak points are dense enough such that the typical scale @xmath0 separating them is much smaller than @xmath1 and that the distribution of nucleation thresholds is immaterial , characterized only by @xmath36 .
the relation @xmath42 allows us to take the continuum limit to define a probability distribution function .
model a was studied in refs .
@xcite . in the absence of precise knowledge of the activation process we adopted reasonable probability distribution functions @xmath39 and demonstrated that the cracks generated by the model were self - affine with @xmath43 _ irrespective _ of the specific form of @xmath39 .
this model should be contrasted with the more common mathematical representation of stochastic growth models via a langevin type equation . in this case
a _ deterministic _
equation is supplemented with an additive noise term whose statistics are independent of the deterministic part . in model a the randomness can not be represented by an additive independent noise .
we now turn to model b to test the influence of the type of randomness employed on the roughness of cracks . in model b the crack is still assumed to propagate by the nucleation and coalescence of damage ahead of its tip .
the linear - elastic stress fields are still calculated using the powerful method of iterated conformal mapping described in sec .
[ conformal ] .
the main difference between the two models stems from a different way of incorporating material disorder into the crack evolution process . in model b the disorder is assumed to be _ quenched _ , represented by an a - priori random distribution of identical weak points .
the random weak points have a prescribed density such that the average distance between them is @xmath0 .
physically , the weak points can be realized by density fluctuations in an otherwise homogeneous material or by small particles that have a lower breaking threshold than the matrix in which they are embedded , but do not change significantly the elastic properties of the system .
as in model a , the damage nucleation process near the crack tip is characterized by a length scale @xmath1 .
a second point of departure from model a is that in this model we assume @xmath19 .
this relation can be realized in different physical situations .
for example , @xmath1 can be still identified ( as in model a ) with the linear dimension of the plastic zone , where the independent scale @xmath0 just happens to be of the order of magnitude ; in that case the damage elements can still be plastic voids .
alternatively , if plastic process are not dominant , one could imagine the crack pinned to a weak point until a micro - crack nucleates at another weak point to propagate the crack by coalescence . in this interpretation @xmath1
is by definition of the order of @xmath0 . to complete the model
we need a growth rule .
as was mentioned before , the weak points are assumed identical in the sense that they have the same breaking threshold that is significantly smaller than the ordinary material points @xcite . since the disorder in this model
is quenched , we can define a _
growth rule stating that the crack advances to the weak point where the hydrostatic tension @xmath34 is maximal .
note that even though the weak points are spread randomly and independently of @xmath34 , the selection of a weak point to be a pointer for the next crack growth depends crucially on its spatial proximity to the maximal hydrostatic tension @xmath34 ; the closest weak point to the maximal hydrostatic tension @xmath34 is most likely to be chosen at each growth step .
it is worthwhile mentioning that if one could fix @xmath1 and decrease @xmath0 such that @xmath44 ( a limit that is not realized in our model where @xmath19 ) , one would obtain the deterministic limit of the model since the maximal hydrostatic tension @xmath34 would almost inevitably coincide with a weak point and the crack would advance almost always to the point of maximal hydrostatic tension .
thus , the ratio @xmath45 is a measure of the width of the statistical distribution in this model . in the next section
we analyze the new model and compare its results to the results of mode a.
we have simulated model b and obtained several crack realizations each of about @xmath46 growth steps .
an example of a resulting crack is shown in fig .
[ crack ] .
we have measured the roughness of the cracks in the model with both the variable bandwidth max - min method of eq .
( [ maxmin ] ) and the variable bandwidth rms method @xcite . in order to avoid strong finite size effects
we have used the results of ref .
@xcite to calibrate our results for the different measurement methods .
this procedure turned up to be consistent in the sense that the variance in the results obtained by the different methods pointed to a single well - defined exponent according the finite size effects predicted in @xcite .
finally , we averaged over different realizations , obtaining @xmath47 .
an example a single roughness measurement is shown in fig .
[ roughness ] .
this result indicates that the cracks in model b exhibit _ random _ roughness as the roughness exponent is not significantly different from the random walk exponent @xmath12 .
this result is qualitatively different from the results of model a in which @xmath43 . at this point
we must conclude that even though the two models share many features and on the face apparently one should not expect dramatically different scaling properties , the two models belong to different universality classes .
model a exhibits _ correlated _ roughening , while model b exhibits _ random _ roughening .
we turn now to further clarifying the origin of the qualitatively different results .
the first thing to point out is that both models a and b deviate from many models common in the literature in a way that might affect significantly the scaling properties . in both models
the crack propagates via the nucleation and coalescence of damage at a _
finite _ distance @xmath1 ahead of its tip , essentially under the influence of a linear - elastic stress field . in ideal linear elasticity
, the stress tensor field @xmath27 attains the following asymptotic form approaching the crack tip @xmath48 where @xmath49 is a polar coordinates system located at the crack tip , @xmath50 and @xmath51 are known universal functions and @xmath52 and @xmath53 are the stress intensity factors corresponding to opening ( mode i ) and shearing ( mode ii ) stresses @xcite .
( [ sifs ] ) describes well the stress fields on a scale @xmath37 relative to the crack tip that is much smaller than any other length scale in the problem .
naturally , we can not expect this formula to be precise for @xmath37 of the order of @xmath1 .
in particular , the hydrostatic tension @xmath38 should be sensitive to significant corrections to the ideal law eq .
( [ sifs ] ) .
we can expect that the stress fields on a scale @xmath1 away from the tip in both models is described by eq .
( [ sifs ] ) with @xmath54 _ plus _ additional terms . to verify this expectation we calculated the hydrostatic tension @xmath34 on an arc a distance @xmath1 from the tip and fitted to the form @xmath55 , \label{angular_fit}\ ] ] predicted by eq .
( [ sifs ] ) @xcite .
the results support our expectation , showing that @xmath54 and an additional contribution of 10 - 15% from other non - universal terms at a distance @xmath1 away from the tip for both models .
an example is shown in fig . [ fit ] .
the conclusion is that both models are in the non - linear regime ( in terms of @xmath56 ) .
in addition one can not neglect the influence of contributions on top of eq .
( [ sifs ] ) .
we believe that this is an important aspect in the success of model a in reproducing a _ correlated _ roughening , quantitatively close to the experimental observations
. it also explains why models which used the linear approximation of a straight crack plus a perturbed @xmath53 achieved scaling exponents different from those observed in experiments @xcite
. on the other hand , this can not be the only factor explaining the results of model a since this feature is common to model b which fails to produce _ correlated _ roughening .
we propose that the differences in randomness are responsible for the different universality classes of model a and b. recall that the quenched disorder length @xmath0 in model b is of the order of the nucleation length @xmath1 . as a result , there are only _ few _ weak points available for damage nucleation on a scale @xmath1 . in this situation
the crack `` selects '' a damage nucleation site out of a small number of possibilities and in fact there is a sizeable probability of having a growth step chosen uncorrelated with the deterministic field @xmath34 that carries the history of the evolution . we thus expect this randomization effect of uncorrelated growth steps to accumulate gradually and reduce the correlated roughening exponent observed in model a. the situation is fundamentally different in model a where the probability of having completely random growth steps is small . to support this explanation we have measured the roughness exponent @xmath10 ( see eq . [ power_law ] ) as a function of the crack length in terms of growth steps . in fig .
[ zeta_n ] we show the dependence of @xmath10 on the number of growth steps .
it is observed that indeed roughness exponent @xmath10 decreases monotonically from @xmath57 to @xmath58 , presumably approaching asymptotically @xmath20 .
this finding is consistent with the suggested explanation .
we have studied the role of disorder in generating correlated roughening exponent in cracks of @xmath59 dimensions .
the main conclusion is that adding additive material disorder to linear elasticity is not sufficient to generate correlated crack graphs with exponents larger than 0.5 .
this is due to the destructive events or uncorrelated steps which accumulate and gradually produce a random graph .
the generation of correlated graphs ( @xmath13 ) is due to the correlations between the deterministic field ( the hydrostatic pressure ) and the pdf carrying the randomness , like the annealed disorder of model a. we reiterate the result concerning the magnitude of the stress intensity factors , and specifically @xmath53 .
the measurement of @xmath53 and higher order terms of the pressure field tells us that linear approximations of @xmath53 , or the principle of local symmetry , do not represent properly the stress field at the vicinity of @xmath1 away from the rough crack tip .
therefore , one can view this feature as an additional test for models of crack growth which produce the observed correlated roughness .
this way of representing disorder is different , for example , from lattice models in which a threshold taken from an uncorrelated distribution is assigned to each lattice point . in that case
the disorder is not introduced by the fluctuations in the locations of weak points ( as in our model b ) , but by the fluctuations in the breaking thresholds . for more information on lattice models , see h. j. herrmann and s. roux ( eds . ) , _ statistical models for the fracture of disordered media _ , ( north - holland , amsterdam , 1990 ) . | we address the role of the nature of material disorder in determining the roughness of cracks which grow by damage nucleation and coalescence ahead of the crack tip .
we highlight the role of quenched and annealed disorders in relation to the length scales @xmath0 and @xmath1 associated with the disorder and the damage nucleation respectively . in two related models , one with quenched disorder in which @xmath2 , the other with annealed disorder in which @xmath3
, we find qualitatively different roughening properties for the resulting cracks in 2-dimensions .
the first model results in random cracks with an asymptotic roughening exponent @xmath4 .
the second model shows correlated roughening with @xmath5 .
the reasons for the qualitative difference are rationalized and explained . |
let @xmath0 be a finitely connected domain in the riemann sphere @xmath6 . a classical theorem of koebe states that @xmath0 is conformally equivalent to the complement of a finite set of pairwise disjoint closed discs and points .
such a domain is called a finitely connected _
circle domain_. koebe s theorem was extended by he and schramm @xcite to apply to domains with countably many complementary components .
schramm later gave a different proof of this result using transboundary extremal length @xcite .
* theorem 0.1)[t : hs1 ] let @xmath0 be a domain in @xmath6 such that boundary @xmath7 has at most countably many components .
then @xmath0 is conformally homeomorphic to a circle domain @xmath8 in @xmath6 .
moreover , @xmath8 is unique up to mbius transformations and every conformal automorphism of @xmath8 is the restriction of a mbius transformation . if @xmath0 is a domain in @xmath6 that has at least three complementary components , or at least one complementary component that is not a puncture , then we say that @xmath0 is _
the reason is that there is then an unbranched analytic covering map from the unit disc @xmath9 onto @xmath0 , which can be used to transfer the poincar metric on @xmath9 to a complete conformal metric of constant curvature @xmath10 on @xmath0 , which is called the hyperbolic metric .
every ring domain is conformally equivalent to a round annulus , a punctured disc , or the punctured plane . for a round annulus @xmath11 ,
each complementary component @xmath12 is a spherical disc , i.e. a closed ball in the spherical metric .
observe that @xmath12 is also a closed ball with respect to the hyperbolic metric belonging to the domain @xmath13 , which is a larger spherical disc . in this paper
we generalize this property to obtain a new canonical form for multiply connected domains .
[ t : main ] let @xmath0 be a finitely connected domain in the riemann sphere @xmath6 with complementary components @xmath14 .
suppose that each domain @xmath15 is hyperbolic
. then @xmath0 is conformally equivalent to a domain @xmath16 in @xmath6 with complementary components @xmath17 such that for each @xmath18 , @xmath19 is hyperbolic and either @xmath20 is a puncture or @xmath20 is a closed disc with respect to the hyperbolic metric of @xmath19 .
moreover , @xmath16 is unique up to mbius transformations .
we call the canonical domains given by this theorem _ intrinsic circle domains_. theorem [ t : main ] has a generalization to positive genus ambient surfaces in place of @xmath6 , which we will now explain . in any riemann surface @xmath1
, a _ closed geometric disc _ will mean a closed ball of some positive radius with respect to the appropriate complete conformal metric of constant curvature on @xmath1 , ( i.e. hyperbolic , euclidean or spherical ) , with the extra condition that it must be homeomorphic to the closed unit disc .
equivalently , the radius of the ball must be strictly less than the injectivity radius of the metric at the center of the disc .
a circle domain @xmath16 in @xmath1 is a connected open subset of @xmath1 for which each complementary component is either a point or a closed geometric disc in @xmath1 . via the uniformization theorem , he and schramm extended theorem [ t : hs1 ] to this setting , as follows .
* theorem 0.2)[t : hs2 ] let @xmath0 be an open riemann surface with finite genus and at most countably many ends .
then there is a closed riemann surface @xmath21 such that @xmath0 is conformally homeomorphic to a circle domain @xmath16 in @xmath21 .
moreover the pair @xmath22 is unique up to conformal homeomorphisms .
likewise , we can extend theorem [ t : main ] to deal with the case of arbitrary finite genus . for the moment we also restrict ourselves to finite connectivity
; we will relax this condition in
[ s : countable ] .
[ t : general ] let @xmath0 be a riemann surface of finite genus and finite connectivity . then there is a conformal embedding @xmath23 of @xmath0 into a compact riemann surface @xmath1 , of the same genus as @xmath0 , so that @xmath24 is the union of disjoint closed , connected and simply connected sets @xmath17 , and for each @xmath25 , @xmath20 is either a single point or a closed geometric disc with respect to the riemann surface @xmath26 .
moreover , the pair @xmath27 is unique up to conformal homeomorphisms .
the condition that @xmath0 and @xmath1 have the same genus means that @xmath1 has the minimal possible genus among all compact riemann surfaces into which @xmath0 may be embedded .
we will refer to the domain @xmath28 as an _ intrinsic circle domain _ in @xmath1 .
note that theorem [ t : general ] includes theorem [ t : main ] as the special case of genus @xmath29 , so this terminology is consistent ; we shall only give a proof of theorem [ t : general ] .
there are some simple special cases .
for example , there is precisely one case in which the appropriate geometry of @xmath26 is spherical ; this occurs when @xmath0 is simply connected , in which case the statement reduces to the riemann mapping theorem .
the doubly connected genus @xmath29 cases occur when @xmath0 is conformally equivalent to the punctured plane @xmath30 , the punctured disc @xmath31 , or a round annulus .
note that in the case of the punctured disc , the appropriate geometry for @xmath26 is euclidean for one complementary component and hyperbolic for the other .
another case in which more than one type of geometry must be considered occurs when @xmath0 is a simply connected domain with two points removed ; then the resulting canonical domain is either a triply - punctured sphere or it is @xmath32 , where @xmath33 is the image under the exponential map of a closed disc of radius less than @xmath34 . if @xmath0 is a riemann surface of genus @xmath35 with one end , then @xmath16 is the complement of a point or of a euclidean disc in @xmath36 , where @xmath37 is a lattice in @xmath38 . in all other cases
the natural geometry of @xmath26 is hyperbolic for every complementary component @xmath20 .
our definition of intrinsic circle domains was motivated by an observation about extremal cases in the grtzsch or plya - chebotarev problem , which asks for the minimizer of the logarithmic capacity among compact connected sets containing a given finite set of points in @xmath38 .
this problem often appears in the course of studying other extremal problems in geometric function theory .
for example , there are recent applications to the bloch - landau constant @xcite and to smale s mean value conjecture @xcite .
it is a classical result of lavrentiev and goluzin that the extremal continuum @xmath39 is unique and is the union of finitely many analytic arcs @xmath40 , which are trajectories of a certain rational quadratic differential ( see for example @xcite ) .
it is not hard to show that each arc @xmath40 is a geodesic arc in the hyperbolic metric of the domain @xmath41 .
a similar condition arises for local minima of the condenser capacity for domains separating one finite set of points from another . in @xcite
it is observed that the extremal continuum @xmath39 enjoys _
harmonic symmetry_. this means that for any subarc @xmath42 of @xmath39 , the harmonic measure of the complementary domain @xmath43 with respect to the point @xmath44 assigns equal masses to each side of @xmath42 .
note that both the geodesic arc condition and the harmonic symmetry condition have the property that they may be verified by checking each arc @xmath40 separately . to check @xmath40
, we need to know only the conformal class of the pair @xmath45 .
our notion of intrinsic circle domains arose by analogy with this property .
we now outline the rest of the paper .
section [ s : geometric lemma ] gives a simple qualitative distortion bound for quasiconformal extensions of conformal maps between ring domains , which we later use several times .
section [ s : uniqueness ] gives the proof of uniqueness in theorem [ t : general ] , and section [ s : existence ] gives the existence proof . in section [ s : countable ] , we extend theorem [ t : general ] to include some cases with countably infinitely many complementary components , subject to a geometric constraint .
section [ s : mixed ] deals with a mixed condition , in which some boundary components are required to be circles in the spherical metric while the others are required to be intrinsic circles in the sense of theorem [ t : main ] . in section [
s : pictures ] we discuss the use of circle packings for the numerical approximation of intrinsic circle domains , and illustrate with some examples .
[ l : distortion bound ] suppose @xmath46 and @xmath47 are ring domains in @xmath38 whose inner boundaries are circles @xmath48 and @xmath49 respectively .
suppose there is a conformal homeomorphism @xmath50 under which @xmath48 corresponds to @xmath49 .
then the induced homeomorphism @xmath51 has a @xmath33-quasiconformal extension between the interior discs of @xmath48 and @xmath49 , where the constant @xmath33 depends only on the conformal modulus of @xmath46 .
lemma [ l : distortion bound ] is a consequence of ( * ? ? ?
* thm 1.4 ) , but it can also be proved simply as follows . after applying similarities
, we may assume that @xmath48 and @xmath49 are both the unit circle @xmath52 .
let @xmath53 be the conformal modulus of @xmath46 .
apply schwarz reflection across @xmath52 to extend @xmath54 to a conformal homeomorphism @xmath55 between ring domains @xmath56 and @xmath57 .
if @xmath58 then @xmath55 is a conformal homeomorphism from the punctured plane to itself , so is in fact a similarity , and its restriction to @xmath59 is an isometry . otherwise , the doubled ring domains @xmath60 are hyperbolic , @xmath55 is a hyperbolic isometry , and the hyperbolic length of @xmath52 in each domain depends only on @xmath53 .
we claim that the density of the hyperbolic metric of @xmath56 on @xmath52 is bounded above and below in terms of @xmath53 . indeed , the hyperbolic length of @xmath52 is a function of @xmath53 so we get such a bound by applying the koebe distortion theorem to a single - valued lift of @xmath61 to the universal cover of @xmath56 .
it follows that @xmath62 is bi - lipschitz , with constants that depend only on @xmath53 .
therefore the radial interpolation of the boundary correspondence provides a quasiconformal extension with dilatation bounded in terms of the modulus of @xmath46 .
suppose that @xmath0 and @xmath63 are two intrinsic circle domains in closed riemann surfaces @xmath1 and @xmath64 respectively .
let @xmath65 be a conformal homeomorphism .
we have to show that @xmath66 is the restriction of a conformal isomorphism @xmath67 .
first , suppose that @xmath0 is simply connected or doubly connected .
if @xmath0 is simply connected or is doubly connected with infinite conformal modulus , then @xmath1 is the riemann sphere and @xmath0 is a circle domain .
@xmath63 has the same connectivity as @xmath0 since they are homeomorphic , so @xmath63 is also a circle domain in the riemann sphere @xmath1 .
the uniqueness part of koebe s theorem then shows that @xmath66 is the restriction of a mbius map . otherwise , @xmath0 is a ring domain of finite modulus , as is @xmath63
. then @xmath1 and @xmath64 are both of genus @xmath29 and we have to show that @xmath66 extends to a mbius map .
consider a complementary component @xmath68 of @xmath0 , and let @xmath69 be the corresponding complementary component of @xmath63 .
let @xmath70 and @xmath71 be any riemann maps . then by hypothesis @xmath72 and @xmath73 are closed hyperbolic discs in @xmath9 , which are bounded by euclidean circles .
thus @xmath74 and @xmath75 are round annuli .
any conformal homeomorphism between round annuli is the restriction of a mbius map , so we can extend @xmath76 to a mbius map @xmath77 .
now define @xmath78 .
then @xmath79 is a conformal homeomorphism extending @xmath66 .
similarly , @xmath66 extends to a conformal homeomorphism @xmath80 .
gluing @xmath81 and @xmath82 together we obtain a conformal homeomorphism from the riemann sphere to itself which extends @xmath66 .
this must be a mbius map . since @xmath0 is conformally equivalent to some round annulus , which is an intrinsic circle domain , this in fact shows that @xmath0 is a round annulus .
now suppose that @xmath0 is at least triply connected .
consider any complementary component @xmath12 of @xmath0 , and let @xmath83 be the corresponding complementary component of @xmath63 .
this makes sense since the homeomorphism @xmath66 induces a bijection between the ends of @xmath0 and the ends of @xmath63 ; for an intrinsic circle domain each end corresponds to precisely one complementary component since the complementary components are all contractible . since @xmath0 and @xmath63 are at least triply connected , @xmath13 and @xmath84 are not simply connected
. therefore we must consider their universal covers , @xmath85 and @xmath86 respectively , in order to understand their hyperbolic metrics .
let @xmath87 and @xmath88 be unbranched analytic covering maps , with deck transformation groups @xmath89 and @xmath90 respectively . since @xmath12 and @xmath83 are contractible subsets of @xmath91 and @xmath84 respectively , the homeomorphism @xmath66 induces a homotopy equivalence @xmath92 , which in turn induces an isomorphism @xmath93 .
then @xmath66 lifts to a conformal homeomorphism @xmath94 note that @xmath95 is @xmath96-equivariant : for any element @xmath97 we have @xmath98 the connected components of @xmath99 and of @xmath100 are disjoint closed discs because @xmath0 and @xmath63 are intrinsic circle domains . therefore @xmath95 is a conformal homeomorphism between circle domains .
hence by the uniqueness part of theorem [ t : hs2 ] it is the restriction of a mbius map @xmath101 .
we find that @xmath101 takes @xmath85 onto @xmath86 since each of the two circle domains has only one non - isolated boundary component .
in particular , @xmath102 . moreover , the map @xmath103 is also @xmath104-equivariant , since @xmath105 is a mbius map extending @xmath95 and is therefore equal to @xmath101 .
it follows that @xmath101 descends to a conformal homeomorphism @xmath106 that extends @xmath66 .
gluing the extensions @xmath107 together for @xmath18 , we obtain the desired conformal homeomorphism @xmath67 extending @xmath66 .
this completes the proof of uniqueness .
is hyperbolic . ]
the first step is to apply theorem [ t : hs2 ] to map @xmath0 via a conformal homeomorphism onto a circle domain @xmath16 in some compact riemann surface @xmath108 .
our goal is to construct a quasiconformal homeomorphism @xmath66 of @xmath108 onto another riemann surface @xmath1 so that @xmath66 is conformal on @xmath16 and the image @xmath109 is an intrinsic circle domain in @xmath1 .
in fact we will construct a beltrami coefficient @xmath110 on @xmath108 with @xmath111 ; then by the measurable riemann mapping theorem we will obtain a riemann surface @xmath1 and a quasiconformal homeomorphism @xmath112 such that @xmath113 a.e . on @xmath108 . by construction
, @xmath110 will be identically zero on @xmath16 , so that @xmath66 is conformal there .
consider any connected component @xmath2 of @xmath114 that is not a single point .
our aim is to construct @xmath110 on @xmath108 so that @xmath115 will be a closed geometric disc with respect to the riemann surface @xmath116 . since this is a condition on the conformal structure of @xmath116 , it depends only on the restriction of @xmath110 to @xmath117 , which by construction will be non - zero only on @xmath2
this is a key point , for it means that we can correct the conformal structure on each complementary component separately and the corrections will not interfere with each other .
this is in contrast to the proof of koebe s theorem by iterated riemann mapping , where each complementary component has to be corrected infinitely many times and the required mapping is obtained in the limit . by hypothesis @xmath16 has only finitely many complementary components in @xmath108 , so to ensure that @xmath111 it will suffice that @xmath118 for each complementary component @xmath2 .
let @xmath119 be an unbranched analytic covering map , where @xmath85 is one of @xmath9 , @xmath38 , or @xmath6 .
we rule out the case @xmath120 , since in this case @xmath16 is an open disc in @xmath6 , so it is already an intrinsic circle domain .
let @xmath121 be the deck transformation group of @xmath34 , i.e. the ( infinite ) group of conformal automorphisms @xmath122 such that @xmath123 .
the preimage @xmath124 has infinitely many connected components , each of which is a topological closed disc , bounded by an analytic jordan curve .
apply theorem [ t : hs1 ] to the domain @xmath125 .
this provides a conformal mapping @xmath126 , where @xmath127 is a circle domain .
each component of @xmath124 corresponds under @xmath128 to a component of @xmath129 that is a closed disc , not a singleton , since it is isolated and can not be separated from the other components by a ring domain of arbitrarily large modulus contained in @xmath130 .
the remaining end of @xmath130 may correspond to either a point or a disc in the complement of @xmath127 ; this component is distinguished because it is not isolated .
let @xmath131 be the union of @xmath127 and all of its isolated complementary components . for any non - trivial element @xmath132 ,
the map @xmath133 is also a conformal mapping of @xmath130 onto the circle domain @xmath127 , so by the uniqueness part of theorem [ t : hs1 ] we have @xmath134 for a unique mbius map @xmath135 , which restricts to an automorphism of @xmath127 with no fixed points since @xmath136 has no fixed points in @xmath85 .
the map that sends @xmath136 to @xmath135 is therefore an injective homomorphism @xmath137 , where @xmath138 is a subgroup of @xmath139 . since @xmath138 acts simply transitively on the components of @xmath140 and has no fixed points in @xmath127 , it acts freely on @xmath131 .
since @xmath141 acts properly discontinuously on @xmath127 , and each compact subset of @xmath131 intersects only finitely many components of @xmath140 , we also conclude that @xmath138 acts properly discontinuously on @xmath131 .
let @xmath142 be a connected component of @xmath143 , and let @xmath144 be the corresponding component of @xmath145 , so that @xmath144 is bounded by a circle @xmath146 .
let @xmath147 be an analytic universal covering of @xmath108 , so that @xmath148 is a circle domain .
let @xmath149 be a connected component of @xmath150 , bounded by a circle @xmath151 .
we will examine the behavior of the maps @xmath152 and @xmath153 in the neighborhood of the curves @xmath154 and @xmath151 .
for the present proof the aim is merely to show that @xmath128 may be extended continuously to a quasiconformal homeomorphism @xmath155 that is equivariant with respect to @xmath121 and @xmath138 .
in fact we will work slightly harder , using lemma [ l : distortion bound ] to show that the quasiconformal dilatation of this extension can be made to depend only on a conformal invariant of @xmath0 .
this will be useful in the next section .
let @xmath156 be the simple closed geodesic in the hyperbolic metric of @xmath16 that separates @xmath2 from all the other components of @xmath114 .
let @xmath39 be the ring domain bounded between @xmath156 and @xmath2 .
it has modulus @xmath157 , where @xmath158 is the length of @xmath156 in the hyperbolic metric of @xmath16 ; note that this length is a conformal invariant of @xmath0 .
the restriction of @xmath153 to this circle domain is an unbranched analytic covering of @xmath16 and is therefore an isometry from the hyperbolic metric of @xmath148 to the hyperbolic metric of @xmath16 .
thus the connected components of @xmath159 are simple closed geodesics in the hyperbolic metric of @xmath160 and are therefore disjoint , each separating a connected component of @xmath150 from all of the other components of @xmath161 .
it follows that @xmath153 maps each connected component of @xmath162 bijectively onto @xmath39 .
one of these is a ring domain @xmath163 surrounding @xmath149 .
likewise , the restriction of @xmath164 to @xmath127 is an unbranched analytic covering of @xmath16 , so @xmath164 maps each connected component of @xmath165 bijectively onto @xmath39 .
one of these connected components is a ring domain @xmath166 surrounding @xmath144 .
consider the branch of @xmath167 that maps @xmath166 to @xmath163 .
it is a homeomorphism of ring domains taking the inner boundary circle @xmath168 onto the inner boundary circle @xmath151 .
lemma [ l : distortion bound ] gives an extension to a quasiconformal homeomorphism @xmath169 from @xmath170 to @xmath171 .
now @xmath172 gives us a quasiconformal homeomorphism @xmath173 that continuously extends the boundary correspondence induced by of @xmath174 .
we define @xmath175 to be the beltrami coefficient @xmath176 .
when we solve the beltrami equation to obtain @xmath112 , the covering map @xmath177 such that @xmath178 will be an unbranched analytic covering , mapping the disc @xmath144 onto @xmath115 , as required .
it is natural to ask whether the notion of intrinsic circle domains can be extended to domains of countably infinite connectivity .
first , we deal with the case of a domain @xmath0 such that every end of @xmath108 is _ isolated _ , i.e. has a neighborhood that meets no other end .
if @xmath0 is embedded in a compact ambient riemann surface then this implies that @xmath0 has only finitely many ends and finite genus , and we already understand this case . for a more interesting example , let @xmath108 be a @xmath179-cover of a compact genus @xmath180 riemann surface , so that @xmath108 has infinite genus and two ends
. then let @xmath0 be a domain obtained by removing countably many disjoint closed topological discs with no accumulation point from @xmath108 . we could hope to modify the structure of @xmath108 on a neighborhood of each disc in order to make the complementary components be intrinsic discs
. however , there is no way to fix the two ends of @xmath108 so that they are represented by intrinsic discs , for the resulting ambient riemann surface would be compact and therefore have only finite genus .
suppose that @xmath0 is infinitely connected or has infinite genus .
we call an end of @xmath0 _ fixable _ if it has a neighborhood that is a ring domain . for any riemann surface @xmath0
, there is a conformal embedding of @xmath0 into another riemann surface @xmath108 such that every connected component of @xmath181 is an intrinsic disc and no end of @xmath108 is fixable .
we can find disjoint open neighborhoods of all the fixable ends , for example by cutting along closed geodesics or horocycles around each end .
in the existence proof for theorem [ t : general ] we only used a local surgery to modify the ambient riemann surface in a neighborhood of each complementary component .
the initial step of passing to a circle domain was technically convenient ( and will be needed later ) , but was not really necessary .
all we really needed to know was that for each individual end @xmath39 , we can embed @xmath0 in a riemann surface @xmath182 such that @xmath183 is a geometric disc in @xmath182 .
this is true for any fixable end @xmath39 , since it has a neighborhood that is a ring domain and therefore it has a neighborhood @xmath184 conformally equivalent to a round annulus latexmath:[$a = \ { z \in \mathbb{c}\,|\ , a < open unit disc to @xmath0 , identifying @xmath186 with @xmath184 , to obtain a riemann surface @xmath187 in which @xmath0 is embedded so that the end @xmath39 corresponds to a connected component @xmath33 of @xmath188 .
then we can apply theorem [ t : hs1 ] to the preimage of @xmath0 in the universal cover of @xmath187 , to find out how to modify the conformal structure on @xmath33 to obtain an embedding of @xmath0 in a new riemann surface @xmath182 so that the end @xmath39 corresponds to a connected component of @xmath189 that is an intrinsic disc . to construct @xmath108
we glue together all the riemann surfaces @xmath182 corresponding to fixable ends @xmath39 by identifying the embedded copies of @xmath0 .
the resulting @xmath108 is locally a riemann surface , is connected , and is second countable because @xmath0 can only have countably many fixable ends .
we now return to the case of a subdomain @xmath0 of a compact riemann surface , but assume that @xmath0 is countably infinitely connected .
it must now be the case that some complementary components are not isolated .
there is a potential topological obstruction associated with complementary components that are not isolated : they can not be represented by intrinsic discs of finite radius .
we might attempt to salvage something by allowing complementary components to be horodiscs with respect to the hyperbolic metric , but this would not help in the case of a circle domain @xmath190 in which some circular complementary component @xmath2 has precisely two points on its boundary that are accumulation points of other complementary circles . therefore in the countably infinitely connected case , we define an intrinsic circle domain @xmath0 to be a subdomain of a compact riemann surface @xmath108 such that any complementary component that is not isolated is a singleton and every non - singleton complementary component @xmath191 is a closed geometric disc with respect to @xmath192 .
this definition is intended to be analogous to the definition of a circle domain in the countably connected case . in order to obtain a positive theorem ,
we can place a simple conformal geometric constraint on the domain .
we will say that a countably connected domain @xmath0 in a compact riemann surface @xmath108 is _ uniformly separated _ if there exists @xmath193 such that each connected component of @xmath181 either is a single point or is separated from all the remaining complementary components by a ring domain of modulus at least @xmath194 embedded in @xmath181 .
[ l : no cavitation ] suppose @xmath0 is a countably - connected uniformly separated domain in a compact riemann surface @xmath108 .
let @xmath23 be any conformal embedding of @xmath0 into a compact riemann surface @xmath195 of the same genus as @xmath108 .
then the non - singleton complementary components of @xmath195 correspond to the non - singleton complementary components of @xmath108 , so @xmath28 is uniformly separated in @xmath195 . to explain the significance of this , we first note that a complementary component @xmath2 is isolated if and only if @xmath0 contains a ring domain with @xmath2 as one of its complementary components ; this is a topological condition on @xmath0 , so the corresponding complementary component in @xmath28 is also not isolated .
if @xmath0 is uniformly separated , then the connected components of @xmath181 that are not singletons must be isolated .
however , the remaining complementary components need not be punctures ; some or all of them could be accumulation points of sequences of other complementary components . for a general open riemann surface @xmath0 of countable connectivity and finite genus , it is possible for an end of @xmath0 to be represented in one conformal embedding by a complementary component that is a non - isolated singleton , yet in some other conformal embedding to be represented by a non - singleton .
we can construct an example of this _ cavitation _ behavior as follows . consider a subdomain @xmath196 of @xmath197 obtained by removing countably many complex conjugate pairs of circular arcs , where one arc of the @xmath198 pair is @xmath199 the domain @xmath200 is conformally equivalent via a conformal map that fixes @xmath201 to a slit domain @xmath202 , which may be obtained from @xmath203 by removing finitely many pairs of intervals @xmath204
, @xmath205 contained in the imaginary axis , arranged symmetrically about @xmath29 , together with an interval @xmath206 .
see figure [ f : cavitation ] for a schematic illustration
. one can show that if the angles @xmath207 are chosen to decrease to @xmath29 sufficiently fast then the domains @xmath202 converge in the carathodory topology to a limit domain @xmath1 in which @xmath29 is a non - isolated singleton complementary component , but every other complementary component is an isolated interval of the imaginary axis .
then @xmath196 is conformally equivalent to @xmath1 and we have an example of cavitation : the singleton complementary component @xmath208 of @xmath1 corresponds to the complementary component @xmath209 of @xmath196 .
the idea is that by making @xmath207 small we can ensure that the extremal length of the family of curves in @xmath196 joining @xmath210 to @xmath211 is as small as we like .
this means we can make the gap between the intervals @xmath204 and @xmath205 as short as we like in comparison to the length of @xmath204 .
in fact to ensure that the end represented in @xmath196 by the unit circle is represented in @xmath1 by a singleton , it suffices to take @xmath212 .
we will now prove lemma [ l : no cavitation ] , which says that uniform separation prevents cavitation .
suppose each non - singleton connected component @xmath12 of @xmath213 is separated from the remaining complementary components by a ring domain @xmath214 of modulus @xmath194 .
then the core curve @xmath215 of the ring domain @xmath40 has length @xmath216 with respect to the hyperbolic metric of @xmath40 . by the schwarz - pick lemma ,
the length of @xmath215 with respect to the hyperbolic metric of @xmath0 is also at most @xmath216 .
consider the simple closed geodesic @xmath217 ( with respect to the hyperbolic metric of @xmath0 ) that lies in the free homotopy class of @xmath215 . since @xmath217 is a length minimizer in its free homotopy class , it is no longer than @xmath215 in the hyperbolic metric of @xmath0 .
it follows that the ring domain @xmath218 bounded between @xmath217 and @xmath12 has modulus at least @xmath219 .
our reason for passing from the ring domains @xmath40 to the ring domains @xmath218 is that the closures of the @xmath218 in @xmath108 are pairwise disjoint , which was not necessarily true of the @xmath40 , and moreover the @xmath218 are intrinsically - defined subdomains of @xmath0 .
we now cut @xmath0 along each @xmath217 to leave a domain @xmath220 .
we claim that for any conformal embedding @xmath23 of @xmath0 into a compact riemann surface @xmath195 of the same genus as @xmath108 , there exists a quasiconformal homeomorphism @xmath221 such that @xmath222 .
to prove this , consider a component @xmath223 of @xmath224 .
it corresponds to a component @xmath225 of @xmath226 , where @xmath83 is a component of @xmath227 .
we claim that there is a @xmath33-quasiconformal homeomorphism @xmath228 whose restriction to the curve @xmath217 agrees with @xmath23 . here
@xmath33 depends only on @xmath194 . to see this ,
let @xmath229 and @xmath230 be riemann mappings , and note that @xmath231 and @xmath232 are analytic jordan curves .
apply lemma [ l : distortion bound ] to the conformal homeomorphism @xmath233 and apply a schwarz reflection to obtain a @xmath33-quasiconformal homeomorphism @xmath234 whose boundary correspondence agrees with that of @xmath235 , where @xmath33 depends only on @xmath194 .
now @xmath236 is a @xmath33-quasiconformal homeomorphism agreeing with @xmath23 on the boundary , as required .
gluing together @xmath23 and the @xmath33-quasiconformal extensions to the components @xmath237 , we obtain a @xmath33-quasiconformal homeomorphism @xmath238 since @xmath0 is countably connected , the complement @xmath239 is a countable union of singletons , and therefore is a removable set for the solution of the beltrami equation .
thus @xmath240 extends to a homeomorphism @xmath241 , and hence the complementary components of @xmath195 that are not among the @xmath12 are singletons .
[ t : infinite ] suppose @xmath0 is a domain in a compact riemann surface @xmath108 such that @xmath0 has countably infinite connectivity , has the same genus as @xmath108 , and is uniformly separated . then there is a conformal embedding @xmath23 of @xmath0 in a compact riemann surface @xmath1 so that @xmath28 is an intrinsic circle domain in @xmath1 .
the pair @xmath27 is unique up to conformal isomorphism .
the existence proof is similar to that of theorem [ t : general ] .
lemma [ l : no cavitation ] shows that when we apply theorem [ t : hs2 ] , the resulting countably connected circle domain @xmath242 satisfies the hypotheses of theorem [ t : infinite ] , in addition to having the property that its non - singleton complementary components are geometric discs . then we note that by lemma [ l : distortion bound ] the quasiconformal distortion of the extension @xmath243 can be bounded in terms of @xmath194 .
then the resulting beltrami coefficient @xmath110 has @xmath244 , so we may still apply the measurable riemann mapping theorem to obtain the desired conformal structure on @xmath1 .
then lemma [ l : no cavitation ] ensures that the non - isolated complementary components of @xmath109 are singletons , so that @xmath109 is an intrinsic circle domain .
the uniqueness proof goes through as before , but with one additional step .
because of the separation hypothesis there is no difficulty in extending a conformal homeomorphism @xmath65 between two countably - connected intrinsic circle domains across each non - singleton complementary component .
we thus obtain a conformal homeomorphism between domains whose complements consist of countably many points , which in turn extends to a conformal homeomorphism @xmath67 between the ambient riemann surfaces .
it is likely that the condition of uniform separation can be weakened by making use of results on existence and uniqueness of solutions to the beltrami equation in the case where the beltrami coefficient has norm @xmath35 but the measure of the set on which the distortion is large is controlled .
now we present a generalization of theorems [ t : general ] and [ t : infinite ] in which the conditions on the complementary components of @xmath0 in @xmath1 are mixed .
that is , some components are punctures or geometric discs in the natural geometry of the ambient compact riemann surface @xmath1 , while each of the others is an intrinsic disc , i.e. a geometric disc with respect to its own union with @xmath0 .
[ t : mixed ] let @xmath0 be an open riemann surface with finite genus and at most countably many ends .
let @xmath245 , @xmath246 be some of the ends of @xmath0 , none of them punctures , where the index set @xmath42 may be either finite or countably infinite .
suppose that for some @xmath193 and for each @xmath246 there is a ring domain @xmath40 contained in @xmath0 , with modulus at least @xmath194 , that separates the end @xmath245 from the other ends of @xmath0 .
then @xmath0 is conformally equivalent to a domain @xmath28 in a compact riemann surface @xmath1 , with complementary components @xmath20 corresponding to @xmath245 , such that @xmath247 is a circle domain in @xmath1 and for each @xmath246 , @xmath20 is either a singleton or a closed geometric disc with respect to the domain @xmath26 .
such a pair @xmath248 is unique up to conformal isomorphism .
we begin by replacing @xmath0 by a conformally equivalent circle domain @xmath16 in a compact riemann surface @xmath108 , as in theorem [ t : hs2 ] .
apply the construction of section [ s : existence ] to each of the components @xmath245 , @xmath246 , to obtain a beltrami differential @xmath249 supported on @xmath250 . in the case where @xmath42 is infinite
, the separation condition guarantees that @xmath244 , as it did for theorem [ t : infinite ] .
solving the beltrami equation @xmath251 on @xmath108 gives a new compact riemann surface @xmath64 with a homeomorphism @xmath252 , conformal away from the preimages of the @xmath245 , such that @xmath253 is a closed hyperbolic disc with respect to the hyperbolic metric of @xmath254 .
the domain @xmath255 is countably connected so by theorem [ t : hs2 ] there is a conformal homeomorphism @xmath256 , where @xmath16 is a circle domain in a compact riemann surface @xmath1 .
the image @xmath257 is the required domain , and @xmath258 .
indeed , @xmath77 is conformal on @xmath259 , so it is an isometry from the hyperbolic metric of @xmath259 to the hyperbolic metric of @xmath19 . to prove uniqueness ,
suppose that @xmath65 is a conformal homeomorphism between two domains @xmath260 and @xmath261 , each satisfying the conditions of the theorem .
let the complementary components of @xmath63 be @xmath262 , @xmath246 .
apply the argument of
[ s : uniqueness ] to extend @xmath66 to a conformal homeomorphism of countably connected circle domains @xmath263 .
this must be the restriction of a conformal map from @xmath1 to @xmath64 by the final part of he and schramm s theorem [ t : hs2 ] .
the rest of this paper concerns the practical numerical approximation of finitely - connected intrinsic circle domains in the sphere .
the idea is roughly to follow the steps of the existence proof , performing each step numerically .
firstly , we need a way to approximate circle domains having fuchsian symmetry .
this already presents us with a choice of methods , as there are several numerical methods for computing circle domains .
finitely connected circle domains can be approximated using iterated riemann mapping , the individual steps of which can be carried out using a number of different methods for computational conformal mapping . however , we must deal with an infinitely connected circle domain , so it seems at first sight that we must make decisions about which complementary components to ignore at any particular stage of the approximation .
we will see that this truncation can in fact be avoided .
secondly , if we follow the existence proof directly , it seems that we need to compute a quasiconformal extension numerically , find its dilatation and then solve a beltrami equation on a larger domain . in practice ,
these steps can be combined into a single _ conformal welding _ step .
fortunately , circle packing is a numerical approximation method that offers solutions to all of these challenges .
although circle packing is not often used for high - precision conformal mapping problems on account of its relatively slow convergence , it is particularly well suited to the approximation of circle domains , especially those with fuchsian symmetry .
it is also a practical tool for numerical conformal welding .
we were able to carry out all of the steps of an approximation procedure for computing intrinsic circle domains using the ` circlepack ` software @xcite written by ken stephenson et al .
for an introduction to circle packing and its use as a computational tool , we refer the reader to stephenson s monograph @xcite .
let @xmath264 be a finite graph embedded in the plane .
a circle packing of @xmath264 in the plane is a collection @xmath265 of circular discs @xmath266 with disjoint interiors , one corresponding to each vertex @xmath267 of @xmath264 , such that whenever vertices @xmath267 and @xmath268 are adjacent in @xmath264 the discs @xmath266 and @xmath269 are tangent .
we call @xmath264 the _ nerve _ of @xmath265
. we may also circle pack in the sphere , using discs in the spherical metric , or in the hyperbolic plane , using hyperbolic circles and possibly also horocircles on the boundary .
the _ carrier _ of @xmath265 is the geometric complex formed by connecting the centers of neighboring circles by geodesic segments . according to the koebe - andreev - thurston theorem ,
when @xmath264 is any triangulation of the sphere , there exists a circle packing of @xmath264 , and it is unique up to mbius maps and reflection . in this case , the carrier of @xmath265 is the entire sphere . since we want to use circle packings to approximate conformal structures , we remove the reflection ambiguity by imposing a fixed orientation .
there is a beautiful algorithm due to bill thurston for computing a circle packing from a given nerve .
it works by removing one vertex and packing the remaining triangulation into the unit disc .
this is achieved by computing the _ hyperbolic packing label _ for the packing ; this is the function which assigns a hyperbolic radius to each vertex of the triangulation , in such a way that the boundary circles are given infinite radius , so that they correspond to horocircles , internally tangent to the unit circle , and all interior circles have finite hyperbolic radius .
there is a unique packing label that results in an angle sum of @xmath270 at each interior vertex .
algorithm approximates the correct label as the limit of a pointwise increasing sequence of labels .
once the correct packing label is known , the circles of the given radii can be laid out in the hyperbolic plane iteratively so that each satisfies the appropriate tangency conditions . here
we outline a numerical method for approximating intrinsic circle domains .
let @xmath0 be a given finitely connected domain in the riemann sphere . begin by taking a circle packing @xmath265 such that the interiors of all the circles of @xmath265 are contained in @xmath0 and the complementary components of @xmath0 are separated by the carrier of @xmath265 .
all subsequent calculations depend only on the nerve of @xmath265 , so it is in this step that we have captured a discrete approximation to the conformal equivalence class of @xmath0 .
the quality of the final approximation of the intrinsic circle domain @xmath16 conformally equivalent to @xmath0 will depend on the mesh of @xmath265 ( the size of its largest circle ) and the maximum distance of @xmath271 from the carrier of @xmath265 .
we perform a sequence of circle packing computations in order to construct a triangulation of the sphere together with an embedding of the nerve of @xmath265 as a subcomplex .
when we compute the circle packing @xmath272 of the spherical triangulation , the carrier of the embedded sub - packing gives an approximation to the intrinsic circle domain @xmath0 conformally equivalent to @xmath273 .
we can interpolate the mapping of circle centers of @xmath265 to the corresponding circle centers in @xmath272 to give a _ polyhedral _ embedding of the carrier of @xmath265 into the carrier of @xmath272 .
since we are mapping from the plane to the sphere , we can not map in a piecewise affine fashion , so instead we map in a piecewise affine fashion to the polyhedron in @xmath274 whose vertices are the vertices of @xmath272 in @xmath275 , and then project radially outwards to @xmath275 .
this map gives a homeomorphism which approximates the conformal map from @xmath273 to @xmath0 and is locally quasiconformal .
we will explain informally why each step of the computation provides an arbitrarily good approximation to a conformal map occurring at the corresponding step in the constructive proof of theorem [ t : main ] .
note that the computation consists of finitely many circle packing steps . in
each step we are approximating the solution of a problem whose solution depends continuously on its data .
this should suffice to prove local uniform convergence of the approximations to the conformal map from @xmath0 to @xmath16 , but we will not attempt to estimate the rate of convergence .
suppose we are given a bounded finitely connected domain @xmath0 in the complex plane .
we find a sequence of circle packings @xmath276 in @xmath0 whose carriers exhaust @xmath0 .
to construct @xmath276 we _ cut out _ a portion of the regular hexagonal circle packing with circles all of radius @xmath277 .
this means that we keep only those circles whose interiors are entirely contained in @xmath0 .
stephenson @xcite describes this as using @xmath7 as a ` cookie cutter '
. we have to take some care at the boundary : we retain only the largest connected component of the hexagonal packing that remains , discarding any peripheral islands and iteratively removing vertices of degree one until we are left with a connected triangulation .
when we take @xmath278 large enough , the boundary components of this triangulation will be in one - to - one correspondence with the complementary components of @xmath0 .
next , to each cycle of boundary vertices of @xmath276 we adjoin a new ` ideal ' vertex , adjacent to all of the vertices in the cycle , and in this way we obtain a triangulation of the sphere . by the koebe - andreev - thurston theorem there is a circle packing whose nerve is this triangulation , and it is unique up to mbius transformations . consider what happens to this spherical packing as we let @xmath278 tend to @xmath44 .
using quasiconformal distortion estimates for packings of bounded vertex degree , one can show that after a suitable mbius normalization , the circles corresponding to the added ideal vertices converge to limit circles of positive radius , while the maximum radius of the all the other circles tends to zero uniformly .
let @xmath279 be the polyhedral map from the hexagonal cut - out packing @xmath276 to the carrier of the spherical packing , normalized so that three chosen points in @xmath0 are fixed by @xmath279 .
it turns out that as @xmath280 , @xmath279 converges locally uniformly on @xmath0 to a conformal map from @xmath0 to a circle domain . to prove this
, we can use the hex packing lemma of rodin and sullivan @xcite .
this says that the quasiconformal dilatation of @xmath279 converges locally uniformly to @xmath29 on @xmath0 .
the rodin - sullivan bound for the quasiconformal dilation on a given face of the triangulation depends only on the number of layers of hexagonal packing surrounding it , and tends to @xmath29 as the number of layers tends to infinity .
let @xmath63 be any subdomain of @xmath0 bounded by jordan curves in @xmath0 , such that @xmath63 is homeomorphic to @xmath0 .
the dilatation bound shows that the images of the maps @xmath279 restricted to @xmath63 remain within a compact subset of the teichmller space of multiply connected domains homeomorphic to @xmath0
. any subsequential limit of the sequence @xmath279 must be a conformal homeomorphism , and its image must be a circle domain . given the normalization , there is a unique such homeomorphism .
it follows that the sequence @xmath279 converges locally uniformly as @xmath281 .
we can augment the image packings by inverting across all the ` ideal ' circles , adding circles of degree @xmath282 in the resulting four - sided interstices to maintain a triangulation . by considering moduli of ring domains in the resulting packing it is possible to show that the maximum radius of any non - ideal circle in the image packing also converges to @xmath29 as @xmath281 .
suppose @xmath2 is one of the complementary components of @xmath16 .
let @xmath283 be an analytic universal covering map .
to compute an approximation to the circle domain conformally equivalent to @xmath284 , we use the same trick again .
we have a triangulation of @xmath16 , to which we add just one ideal vertex @xmath285 corresponding to the complementary component @xmath2 .
now we apply thurston s algorithm to compute the _ hyperbolic _
packing label for the resulting multiply connected triangulation .
if we lay the circles out according to this packing label , we get a fuchsian monodromy group .
call the resulting packing @xmath286 .
the circle corresponding to @xmath285 and its translates under the monodromy group approximate the bounded components of the required circle domain . for the conformal welding step that follows we will only need to know the hyperbolic radii for the circles corresponding to the vertex @xmath285 and its neighbors .
it is not necessary to perform the layout routine . given a jordan curve @xmath136 in the riemann sphere , we can compute the riemann mappings of the unit disc onto the interior and the exterior of @xmath136 .
the boundary values of these riemann mappings give two different continuous maps of the unit circle onto @xmath136 , @xmath287 and @xmath288 .
the shape of @xmath136 is encapsulated in the boundary correspondence @xmath289 , which is an orientation - reversing homeomorphism of the unit circle .
conformal welding is the process of recovering @xmath136 ( up to a mbius map ) from the boundary correspondence .
the ` circlepack ` software includes a package for performing approximate conformal welding , following the procedure described by williams @xcite .
the idea is to paste two combinatorial closed discs together along their boundaries , according to a best possible combinatorial approximation of the given boundary correspondence , to produce a triangulation of the sphere .
we then compute the circle packing for this triangulation .
for example , we may start with two maximal packings of the unit disc ( meaning that the boundary circles are internally tangent to the unit circle ) , and use the arc length around the unit circle as a guide for pasting together the two triangulations , introducing new vertices where necessary to keep control of the vertex degrees . in our application to intrinsic circle domains ,
we weld a combinatorial closed disc to the boundary cycle corresponding to each complementary component @xmath2 of @xmath0 , using the arc length around the circle corresponding to @xmath285 in the packing @xmath286 and the trivial boundary correspondence @xmath290 . a suitable choice of combinatorial closed disc would be a large section of the regular hexagonal packing , for then the hex packing lemma of rodin and sullivan can be applied to show that the desired welding and the computed welding differ by a quasiconformal map which has small dilatation except on an annulus of small modulus and small area covering the cycle of edges along which the welding is performed . after performing a welding for each boundary component , we simply apply thurston s algorithm to compute a circle packing of the resulting spherical triangulation . in the case
where the ambient riemann surface @xmath108 has positive genus , we would compute a packing of the universal cover of the welded complex at this stage to get a circle packing approximation to the riemann surface @xmath1 .
the finished spherical circle packing gives an approximation to the map @xmath23 and to the desired conformal structure on the interior of each complementary component of @xmath28 , and on the interior of the image of @xmath28 .
we sketch a proof that the approximation scheme described above does converge locally uniformly on the domain @xmath16 to the conformal mapping to an intrinsic circle domain , after suitable normalization .
the strategy is to show that every map appearing in figure [ f : existence proof ] is well approximated by the corresponding polyhedral map between circle packings .
then since each step depends continuously on its input data , we will find that the polyhedral map from @xmath291 to @xmath275 converges locally uniformly on @xmath16 to the desired conformal mapping . on the subcomplex of the packing representing the intrinsic circle domain ,
the hex - packing distortion estimate of rodin and sullivan shows that the polyhedral mapping is quasiconformal with dilatation that converges locally uniformly to zero .
we can apply a similar estimate on the interior of each complementary component .
it remains to show that the conformal welding step is well approximated by the discrete conformal welding .
to do this we have to control the conformal modulus of the image of a narrow annular neighborhood of each boundary component , where the circles are not deep enough in the hex packing for the rodin - sullivan estimate to give the bound we need .
to do this it suffices to bound the maximum degree of the vertices appearing in the final triangulation .
this is because the rodin - sullivan ring lemma allow the quasiconformal distortion of polyhedral maps between circle packings to be bounded in terms of the maximum degree .
we can arrange that all vertices have degree @xmath292 except those involved in the combinatorial welding . the largest degree arising when we weld using the hyperbolic radii of the neighbors of @xmath285 in the packing @xmath286
can be bounded in terms of the ratio between the radii of the largest and smallest circles adjacent to @xmath285 .
since we began with a hex - packing of @xmath0 with circles all of equal radius , this ratio is controlled in the limit by the maximum and minimum value of the derivative on the boundary @xmath293 of the conformal map @xmath174 .
examining the figures , the reader will note that in our numerical experiments we have not restricted ourselves to using hex packings as our starting point .
instead we have used packings with few layers of degree @xmath294 vertices near the boundary , so that there can are many very small circles on the boundary without needing a very large number of circles in total .
this was done to reduce the errors from the combinatorial welding step .
we pay for this by having larger circles far from the boundary , and losing the rigorous distortion bounds , but this seems a worthwhile trade since we expect the schwarzian derivative of the map that we are approximating to be small on this region . in figures
[ f : circle domain ] and [ f : intrinsic circle domain ] there is a circle of degree @xmath295 fixed by a rotational symmetry of the packing of order @xmath296 .
the schwarzian derivative of the map we are approximating must certainly vanish at the center of this circle , by symmetry considerations .
apart from two circles of this type , the maximum degree in the packing is @xmath297 . in the packings in figures
[ f : circle domain 2 ] and [ f : intrinsic 2 ] , the maximum degree is @xmath294 .
finally we comment that it may be more satisfactory from the point of view of computation , and also to simplify the convergence proof , to perform the discrete conformal welding using packings with specified overlap angles rather than restricting ourselves to tangency packings .
however , we have not implemented this in practice .
aaa t. carroll and j. ortega - cerd , _ the univalent bloch - landau constant , harmonic symmetry and conformal glueing _ , j. math .
pures appl .
( 9 ) * 92 * ( 2009 ) , no .
4 , 396406 . e. crane , _ a bound for smale s mean value conjecture for complex polynomials _ , bull . lond .
* 39 * ( 2007 ) , no . 5 , 781791 . g. m. goluzin , _ geometric theory of functions of a complex variable _ , transl .
monogr . , vol .
26 , ams ( 1969 ) .
z. x. he and o. schramm , _ fixed points , koebe uniformization and circle packings_. ann . of math .
( 2 ) * 137 * ( 1993 ) , 369406 . j.
ortega - cerd and b. pridhnani , _ the plya - tchebotarv problem _ , harmonic analysis and partial differential equations , contemp .
* 505 * ( 2010 ) , 153170 .
b. rodin and d. sullivan , _ the convergence of circle packings to the riemann mapping _ , j. differential geom .
* 26 * ( 1987 ) , no .
2 , 349-360 . o. schramm , _ transboundary extremal length _
, j. anal .
* 66 * ( 1995 ) , 307329 .
k. stephenson , _ introduction to circle packing : the theory of discrete analytic functions _
, cambridge university press , 2005 .
k. stephenson et al .
circlepack , software for circle packing . | using quasiconformal mappings , we prove that any riemann surface of finite connectivity and finite genus is conformally equivalent to an intrinsic circle domain @xmath0 in a compact riemann surface @xmath1 .
this means that each connected component @xmath2 of @xmath3 is either a point or a closed geometric disc with respect to the complete constant curvature conformal metric of the riemann surface @xmath4 .
moreover the pair @xmath5 is unique up to conformal isomorphisms .
we give a generalization to countably infinite connectivity .
finally we show how one can compute numerical approximations to intrinsic circle domains using circle packings and conformal welding . _
accepted for publication in conformal geometry and dynamics _ |
any code designed to evolve a general relativistic system will start with a solution to the initial value problem corresponding to the astrophysical situation of interest . in the 3 + 1 formulation of general relativity ,
the problem is defined by the hamiltonian and momentum constraints @xcite r+ k^2 - a_i^ j a_j^ i = 0 +
_ j a_i^ j - _ i k = 0 .
[ constraints ] above , @xmath0 is the 3-dimensional ricci scalar constructed from the spatial 3-metric @xmath1 , and @xmath2 and @xmath3 are the trace and trace - free parts of @xmath4 such as k_ij = a_ij + g_ij k .
the problem of solving the hamiltonian and momentum constraints for two black holes has been addressed in the past by several groups ( see @xcite and references therein ) . for us , it is inherently 3-dimensional since we want to be able to specify arbitrary boost and spin directions for each hole .
the methods in most frequent use @xcite are based on an approach which chooses maximal spatial hypersurfaces ( @xmath5 ) and takes the spatial 3-metric to be conformally flat ( @xmath6 , @xmath7 being the conformal factor ) . under these conditions ,
the hamiltonian constraint can be decoupled from the momentum constraint .
analytical solutions for the extrinsic curvature @xmath4 for holes with specific linear momenta can be found @xcite .
this simplification leaves only one elliptic equation for @xmath7 , which is derived from the hamiltonian constraint .
the inner boundaries ( the throats of the black holes ) are usually dealt with by imposing an isometry condition between two identical asymptotically flat spatial slices , joined by an einstein - rosen bridge , though other boundary conditions are sometimes used .
unfortunately , the numerical solution of the equation for @xmath7 presents a technical challenge at the inner boundaries .
brandt and brgmann @xcite simplified this problem by compactifying the internal asymptotically flat regions to obtain a domain without inner boundaries .
however , the main disadvantage of these methods is not numerical , but related to the physical interpretation of black - hole spaces described through a conformally flat 3-metric .
there are no space slices for which the spatial 3-metric of a single kerr ( non - zero spin ) black hole can be written in a conformally flat way @xcite , and recent work on sequences of initial data sets for circular orbits casts some doubt on the physical realism of conformally flat approaches to black hole binaries @xcite .
one way to overcome this problem is to use a kerr - schild slicing of spacetime @xcite .
matzner , huq , and shoemaker @xcite proposed a method that bases the initial data on a background metric and extrinsic curvature that is a superposition of kerr black hole metrics written in ingoing eddington - finkelstein coordinates .
thus , for a single kerr black hole one obtains the exact solution to the problem .
@xcite added to this method a variation on the background fields that eliminates the inner boundary problem , greatly simplifying the numerical treatment of the elliptic equations .
the _ background _ fields generated in this way are also very good approximate solutions to the initial value problem ( i.e. , the violation of the constraints ( [ constraints ] ) is small enough to fall below the numerical truncation error for a wide range grid spacings ) .
we present here the first solution to the initial value problem for black - hole binaries based on physically realistic background 3-metric and extrinsic curvature .
we study the particular case of a hyperbolic encounter of two kerr black holes of mass @xmath8 , separated by a distance of @xmath9 .
the method begins by specifying a conformal spatial metric which is a straightforward superposition of two kerr - schild single hole ( spatial ) metrics : _ ij & = & _ ij + 2 _ 1b _ 1h _ 1l_i 1l_j + 2 _
2b _ 2h _ 2l_i 2l_j , + & = & _ 1b _ 1k_i^ i+_2b _
2k_i^ i , + _ ij & = & _ n(i ( _ 1b _ 1k_j)^ n + _ 2b
_ 2k_j)^ n - _
j)^ n ) , [ att_fields ] where the parenthesis in the subscripts denote symmetrization in @xmath10 and @xmath11 .
the scalar function @xmath12 has a known form @xcite , and @xmath13 is an ingoing null vector congruence associated with the solution .
the fields marked with the pre - index @xmath14 ( @xmath15 ) correspond to an isolated black hole with specific angular momentum @xmath16 ( @xmath17 ) and boosted with velocity @xmath18 ( @xmath19 ) .
the trace - free part of the extrinsic curvature tensor @xmath20 is also constructed from the superposition of the curvatures @xmath21 and @xmath22 associated with the black holes 1 and 2 respectively . the attenuation function @xmath23 ( @xmath24 ) is unity everywhere except in the vicinity of hole 2 ( hole 1 ) where it rapidly vanishes , so the fields there are effectively those of a single black hole , thus providing an exact solution to the constraints for distances arbitrarily close to the singularities @xcite . following the conformal decomposition presented by york and collaborators @xcite , we relate the physical metric @xmath1 and the trace - free part of the extrinsic curvature @xmath3 to the background fields through a conformal factor : g_ij & = & ^4 _ ij + a^ij & = & ^-10 ^ij .
[ conf_fields ] in order to find a solution to the four constraint eqs .
( [ constraints ] ) , we add a longitudinal part to the extrinsic curvature @xmath25 : a^ij ^-10 ( ^ij + ^ij ) , where @xmath26 is a vector potential to be solved for and ^ij ^i w^j + ^j w^i - ^ij w^k .
[ lw ] plugging eqs .
( [ att_fields]-[lw ] ) , into the hamiltonian and momentum constraints ( [ constraints ] ) , we obtain four coupled elliptic equations for the fields @xmath7 and @xmath26 @xcite : ^2 & = & ( 1/8 ) ( + ^2 ^ 5 - + & & ^-7 ( + ( ) ^ij ) ( _ ij + ( ) _ ij ) ) + _ j()^ij & = & ^ij ^6 _ j k - _ j [ ell_eqs ] to solve the elliptic eqs .
( [ ell_eqs ] ) we use an adaptation of an multigrid elliptic solver developed for the solution of the initial value problem of neutron - star binaries @xcite , which uses second order stencils . to optimize the use of memory ,
this solver was designed to handle only identical black holes ; only one object is placed on the numerical grid , the second is accounted for using reflective boundary conditions at the corresponding surface of the grid cube ( see @xcite for details ) .
the four elliptic equations are solved consecutively in an iteration loop .
the iteration starts with a trivial initial guess , namely ( x^i)&=&1 + w^i(x^i)&=&(0,0,0 ) x^i , [ i_guess ] and it relaxes the fields until the variation from one iteration step to the next falls below some fraction of the truncation error .
this typically takes around 10 to 20 iteration steps .
the sources for these elliptic eqs .
consist mostly @xcite of the residuals of the constraint equations as defined in @xcite .
the use of attenuation functions eliminates the singular behavior of these sources near the ring singularities , simplifying the numerical implementation of the elliptic solver . instead of using excision techniques around the singularities @xcite
, we modified a standard elliptic solver to `` ignore '' ( i.e. leave the fields with the initial values ( [ i_guess ] ) ) the grid points in a small volume embedding the singularity .
this `` inner '' region is defined monitoring the value of the background 3-dimensional ricci scalar @xcite , and is chosen to be smaller than the excision volume used in current numerical simulations @xcite , making the initial value data suitable for the evolutionary codes .
the choice of the `` inner '' region values ( [ i_guess ] ) is based on the fact that arbitrarily close to the ring singularity , the background fields recover the single - hole values which are an exact solution to eqs .
( [ constraints ] ) .
we use robin conditions for the outer boundaries @xcite , guaranteeing that @xmath7 and @xmath26 fall off as @xmath27 at infinity .
the calculations presented in this paper were performed using 16 processors of an origin2000 system located at the national computational science alliance center in urbana - champaign , taking close to 200 cpu hours .
the method described above was applied to the problem of two black holes with mass @xmath8 in a hyperbolic encounter configuration with parameters @xmath28 , @xmath29 , and @xmath30 corresponding to black hole 1 coordinate position , velocity , and specific angular momentum .
the parameters for black hole two are @xmath31 , @xmath32 , and @xmath33 . to test the convergence of the code ,
two grid resolutions were used : one with grid spacing @xmath34 ( low ) and another with @xmath35 ( high ) .
the number of grid points was 60,750 and 536,238 respectively and the bounding box of the numerical grid covers a rectangle from @xmath36 to @xmath37 .
we estimated the adm mass corresponding to the background fields and to the full numerical solution , obtaining @xmath38 ( note that @xmath39 ) and @xmath40 respectively .
this was done by integrating over the boundary surface of the computational grid .
the difference is accounted for by the binding energy .
we also obtained a background total angular momentum estimation of @xmath41 , consistent with analytical estimates .
when differenced under our numerical scheme , even analytically exact single - hole solutions show inevitable nonzero residual errors in the hamiltonian and momentum constraints , which increase near the hole , due to the larger gradients there . in figures
[ hc ] and [ mcx ] these are plotted as points ( filled circles for resolution @xmath42 , filled squares for resolution @xmath43 ) .
the decrease of this truncation error with increasing resolution shows the convergent behavior of the differencing scheme .
also in figures 1 - 2 we plot as curves the computed constraint residuals of the rightmost hole in our two - hole solution .
except for the innermost points in the solution , the residuals are effectively identical to those of the analytical single - hole solution at the same resolution .
the inset in figure [ hc ] shows this discrepancy at the innermost points for the high resolution curve ( empty squares mark the grid points and the singularity is represented by a gray vertical band ) .
we thus confirm our two - hole solution at the level of truncation error except for these innermost points , which will anyway be excised in the evolution scheme @xcite .
figure [ phi ] is a contour plot of the conformal factor @xmath7 .
the inner thick lines limit the `` inner '' region where the initial value @xmath44 is preserved .
the apparent horizons of the holes were calculated using a code developed by huq _
_ @xcite and are represented by the outer thick lines .
the resulting horizon areas for the background system and the full solution were @xmath45 and @xmath46 respectively , which are consistent with the value of @xmath47 obtained for a single - hole configuration .
table [ tablei ] shows the @xmath48 ( average of the absolute value ) and @xmath49 ( maximum value ) norms for the calculated fields .
the final fields @xmath7 and @xmath26 depart very little from the initial guess ( [ i_guess ] ) all across the numerical grid , indicating the great degree of accuracy of the background fields as approximate solutions to the constraints @xcite .
the attenuation method , while providing a clear numerical advantage , also raises the question of physical interpretation of these results .
one would expect the fields around each hole to be tidally distorted by the presence of the second hole , and the attenuation functions partially occlude this effect .
however , we notice that most of this effect is confined to the regions very near the ring singularities : i.e. regions that will be excised from the computational domain by the time evolution code . at the same time , we see from figure [ phi ] that the tidal interaction is still ( at least partially ) present at the location of the apparent horizons
. the influence of the attenuation functions can be interpreted as an addition of gravitational radiation that conveniently ( from the numerical point of view ) eliminates the singular behavior from the elliptic equations near the singularities .
this solution to the initial value problem presents a different approach than the conformally flat / maximal slicing methods , and allows us to specify more directly the physical content of the data .
it also entails a computational simplification on the treatment of the singularities on the working grid .
however , the physical aspects of these data sets can be further refined , for instance by providing background fields based on post - newtonian expansions or with a more physical prescription for the background fields for circular orbits @xcite .
this work was supported by nsf grant phy9800722 and phy9800725 to the university of texas at austin .
most of the elliptic solver was developed as part of the thesis work of one of the authors ( pm ) at the university of notre dame .
pm wants to express his gratitude to grant j. mathews and james r. wilson for their guidance and encouragement .
special thanks to l. lehner for his help with the apparent horizons calculations and to m. huq , p. laguna , d. nielsen , d. shoemaker , and g. b. cook for very helpful discussions .
g. b. cook , m. w. choptuik , m. r. dubal , s. klasky , r. a. matzner and s. r. oliveira , phys .
rev . * d47 * , 1471 ( 1993 ) .
g. b. cook , phys .
rev . * d50 * , 5025 ( 1994 ) [ gr - qc/9404043 ]
. t. w. baumgarte , phys .
rev . * d62 * , 024018 ( 2000 ) [ gr - qc/0004050 ] .
j. w. york and t. piran `` the initial value problem and beyond '' , _ spacetime and geometry : the alfred schild lectures _
, r. a. matzner and l. c. shepley eds .
university of texas press , austin , texas . ( 1982 ) ; g. b. cook , `` initial data for the two - body problem of general relativity '' , ph.d .
dissertation , the university of north carolina at chapel hill ( 1990 ) .
p. marronetti , g. j. mathews and j. r. wilson , phys .
* d58 * , 107503 ( 1998 ) [ gr - qc/9803093 ] .
marronetti , g. j. mathews and j. r. wilson , proceedings of 19th texas symposium on relativistic astrophysics : texas in paris , paris , france .
j. paul , t. montmerle , and e. aubourg eds .
nuclear physics b , proc .
* 80 * ( 2000 ) .
[ gr - qc/9903105 ] .
p. marronetti , g. j. mathews and j. r. wilson , phys .
rev . * d60 * , 087301 ( 1999 ) [ gr - qc/9906088 ] . in the numerical implementation of the solver , the terms on the rhs . of eqs .
( [ ell_eqs ] ) that include the background affine connection coefficients are passed to the sources .
these terms still have the `` singularity '' embedded in them , thus part of the singular behavior remains .
fortunately , the numerical experiments showed that their influence in the convergence of the algorithm is very small and confined to the points closest to the inner region limit .
the numerical code uses domain decomposition to split the grid in eighteen cubes .
the elliptic equations are solved independently in each cube and then the new field values are used to update the source terms of the equations and the boundaries of each cube for the next iteration step .
the cubes overlap each other with one buffer grid point that allows the calculation of the cube - surface boundary values . while this scheme worked remarkably well in the case of neutron - star binaries @xcite
, it seems somewhat inadequate for the present case ; the convergence is degraded to first order for the grid points adjacent to the cube s sides . here , we concentrated on the study of the convergence near the singularities to better test the attenuation method .
since these regions are well inside the center grid cube , they do not present this problem and allowed us to observed the second order convergence clearly .
this problem could be solved by increasing the number of overlapping buffer zones or discarding the grid decomposition altogether ( working with only one grid ) , and it will be addressed in future versions of the code . | we solve the elliptic equations associated with the hamiltonian and momentum constraints , corresponding to a system composed of two black holes with arbitrary linear and angular momentum .
these new solutions are based on a kerr - schild spacetime slicing which provides more physically realistic solutions than the initial data based on conformally flat metric / maximal slicing methods .
the singularity / inner boundary problems are circumvented by a new technique that allows the use of an elliptic solver on a cartesian grid where no points are excised , simplifying enormously the numerical problem . |
it took more than eighty years from its discovery till it was possible to experimentally determine and visualize the most fundamental object in quantum mechanics , the wave function . the forward route from quantum state
to probability distribution of measurement results has been the basic stuff of quantum mechanics textbooks for decennia . that the corresponding mathematical inverse problem had a solution , provided ( speaking metaphorically ) that the quantum state has been probed from a sufficiently rich set of directions , had also been known for many years .
however it was only with @xcite , that it became feasible to actually carry out the corresponding measurements on one particular quantum system in that case , the state of one mode of electromagnetic radiation ( a pulse of laser light at a given frequency ) .
experimentalists have used the technique to establish that they have succeeded in creating non - classical forms of laser light such as squeezed light and schrdinger cats .
the experimental technique we are referring to here is called quantum homodyne tomography : the word homodyne referring to a comparison between the light being measured with a reference light beam at the same frequency .
we will explain the word tomography in a moment .
the quantum state can be represented mathematically in many different but equivalent ways , all of them linear transformations on one another .
one favorite is as the wigner function @xmath0 : a real function of two variables , integrating to plus one over the whole plane , but not necessarily nonnegative .
it can be thought of as a `` generalized joint probability density '' of the electric and magnetic fields , @xmath1 and @xmath2 .
however one can not measure both fields at the same time and in quantum mechanics it makes no sense to talk about the values of both electric and magnetic fields simultaneously .
it does , however , make sense to talk about the value of any linear combination of the two fields , say @xmath3 . and
one way to think about the statistical problem is as follows : the unknown parameter is a joint probability density @xmath0 of two variables @xmath4 and @xmath5 .
the data consists of independent samples from the distribution of @xmath6 , where @xmath7 is chosen independently of @xmath8 , and uniformly in the interval @xmath9 $ ] .
write down the mathematical model expressing the joint density of @xmath10 in terms of that of @xmath8 .
now just allow that latter joint density , @xmath0 , to take negative as well as positive values ( subject to certain restrictions which we will mention later ) . and
that is the statistical problem of this paper .
this is indeed a classical tomography problem : we take observations from all possible one - dimensional projections of a two - dimensional density .
the non - classical feature is that though all these one - dimensional projections are indeed bona - fide probability densities , the underlying two - dimensional `` joint density '' need not itself be a bona - fide joint probability density , but can have small patches of `` negative probability '' .
though the parameter to be estimated may look strange from some points of view , it is mathematically very nice from others .
for instance , one can also represent it by a matrix of ( a kind of ) fourier coefficients : one speaks then of the `` density matrix '' @xmath11 .
this is an infinite dimensional matrix of complex numbers , but it is a positive and selfadjoint matrix with trace one .
the diagonal elements are real numbers summing up to one , and forming the probability distribution of the number of photons found in the light beam ( if one could do that measurement ) .
conversely , any such matrix @xmath11 corresponds to a physically possible wigner function @xmath0 , so we have here a concise mathematical characterization of precisely which `` generalized joint probability densities '' can occur .
the initial reconstructions were done by borrowing analytic techniques from classical tomography
the data was binned and smoothed , the inverse radon transform carried out , followed by some fourier transformations . at each of a number of steps ,
there are numerical discretization and truncation errors .
the histogram of the data will not lie in the range of the forward transformation ( from quantum state to density of the data ) .
thus the result of blindly applying an inverse will not be a bona - fide wigner function or density matrix .
moreover the various numerical approximations all involve arbitrary choices of smoothing , binning or truncation parameters .
consequently the final picture can look just how the experimenter would like it to look and there is no way to statistically evaluate the reliability of the result . on the other hand the various numerical approximations tend to destroy the interesting `` quantum '' features the experimenter is looking for , so this method lost in popularity after the initial enthousiasm .
so far there has been little attention paid to this problem by statisticians , although on the one hand it is an important statistical problem coming up in modern physics , and on the other hand it is `` just '' a classical nonparametric statistical inverse problem .
the unknown parameter is some object @xmath11 , or if you prefer @xmath0 , lying in an infinite dimensional linear space ( the space of density matrices , or the space of wigner functions ; these are just two concrete representations in which the experimenter has particular interest ) .
the data has a probability distribution which is a linear transform of the parameter . considered as an analytical problem , we have an ill - posed inverse problem , but one which has a lot of beautiful mathematical structure and about which a lot is known ( for instance , close connection to the radon transform ) .
moreover it has features in common with nonparametric missing data problems ( the projections from bivariate to univariate , for instance , and there are more connections we will mention later ) and with nonparametric density and regression estimation .
thus we think that the time is ripe for this problem to be `` cracked '' by mathematical and computational statisticians . in this paper
we will present some first steps in that direction .
our main theoretical results are consistency theorems for two estimators .
both estimators are based on approximating the infinite dimensional parameter @xmath11 by a finite dimensional parameter , in fact , thinking of @xmath11 as an infinite dimensional matrix , we simply truncate it to an @xmath12 matrix where the truncation level @xmath13 will be allowed to grow with the number of observations @xmath14 .
the first estimator employs some analytical inverse formulas expressing the elements of @xmath11 as mean values of certain functions , called pattern functions , of the observations @xmath10 .
simply replace the theoretical means by empirical averages and one has unbiased estimators of the elements of @xmath11 , with moreover finite variance . if one applies this technique without truncation the estimate of the matrix @xmath11 as a whole will typically not satisfy the nonnegativity constraints .
the resulting estimator will not be consistent either , with respect to natural distance measures . but provided the truncation level grows with @xmath14 slowly enough , the overall estimator will be consistent .
the effect of truncating the density matrix @xmath11 is to project on the subspace generated by the first elements of the corresponding basis , we shall call it the pattern function projection estimator ( pfp ) .
the second estimator we study is sieve maximum likelihood ( sml ) based on the same truncation to a finite dimensional problem .
the truncation level @xmath13 has to depend on sample size @xmath14 in order to balance bias and variance .
we prove consistency of the sml estimator under an appropriate choice of @xmath15 by applying a general theorem of @xcite .
to verify the conditions we need to bound certain metric entropy integrals ( with bracketing ) which express the size of the statistical model under study .
this turns out to be feasible , and indeed to have an elegant solution , by exploiting features of the mapping from parameters ( density matrices ) to distributions of the data .
various distances between probability distributions possess analogues as distances between density matrices , the mapping from parameter to data turns out to be a contraction , so we can bound metric entropies for the statistical model for the data with quantum metric entropies for the class of density matrices . and the latter can be calculated quite conveniently .
our results form just a first attempt at studying the statistical properties of estimators which are already being used by experimental physicists , but they show that the basic problem is both rich in interesting features and tractable to analysis .
the main results so far are consistency theorems for pfp and sml estimators , of both the density matrix and the wigner function .
these results depend on an assumption of the rate at which a truncated density matrix approximates the true one .
it seems that the assumption is satisfied for the kinds of states which are met with in practice .
however , further work is needed here to describe in physically interpretable terms , when the estimators work .
secondly , we need to obtain rates of consistency and to further optimize the construction of the estimator .
thirdly , one should explore the properties of penalized maximum likelihood .
this will make the truncation level data driven .
fourthly , one should try to make the estimators adaptive with respect to the smoothness of the parameter .
we largely restrict attention to an ideal case of the problem where there is no further noise in the measurements . in practice , the observations have added to them gaussian disturbances of known variance .
there are some indications that when the variance is larger than a threshold of @xmath16 , reconstruction becomes qualitatively much more difficult .
this needs to be researched from the point of view of optimal rates of convergence .
we also only considered one particular though quite convenient way of sieving the model , i.e. , one particular class of finite dimensional approximations .
there are many other possibilities and some of them might allow easier analysis and easier computation .
for instance , instead of truncating the matrix @xmath11 in a given basis , one could truncate in an arbitrary basis , so that the finite dimensional approximations would correspond to specifying @xmath13 arbitrary state vectors and a probability distribution over these `` pure states '' .
now the problem has become a missing data problem , where the `` full data '' would assign to each observation also the label of the pure state from which it came .
in the full data problem we need to reconstruct not a matrix but a set of vectors , together with an ordinary probability distribution over the set , so the `` full data '' problem is statistically speaking a much easier problem that the missing data problem .
we shall use a version of this to derive an expectation - maximization algorithm ( em ) for the practical implementation of the sml estimator , see section [ sec.experimental ] .
one could imagine that bayesian reconstruction methods could also exploit this structure .
the analogy with density estimation could suggest new statistical approaches here .
finally , it is most important to add to the estimated parameter , estimates of its accuracy .
this is absolutely vital for applications , but so far no valid approach is available . in section [ sec.problem ]
we introduce first , very briefly , the statistical problems we are concerned with .
we then give a short review of the basic notions of quantum mechanics which are needed in this paper .
concepts such as observables , states , measurements and quantum state tomography are explained by using finite dimensional complex matrices . at the end of the section
we show how to generalize this to the infinite dimensional case and describe the experimental set - up of quantum homodyne detection pointing out the relation with computerized tomography . in section [ sec.densitymatrixestimation ] we present results on consistency of density matrix estimators : projection estimator based on pattern functions , and sieve maximum likelihood estimator .
the last subsection extends the results of the previous ones to estimating the wigner function .
section [ sec.efficiency ] deals with the detection losses occurring in experiments due in part to the inefficiency of the detectors .
this adds a deconvolution problem on the top of our tomography estimation .
section [ sec.experimental ] shows experimental results .
we illustrate the behavior of the studied estimators and propose some practical tools for the implementation em algorithm .
the last section finishes with some concluding remarks to the whole paper and open problems .
the main purpose of the paper is to bring the attention of the statistical community to these problem ; thus , some proofs are just sketched .
they fully appear in a complementary paper , @xcite .
our statistical problem is to reconstruct the quantum state of light by using data obtained from measuring identical pulses of light through a technique called quantum homodyne detection ( qhd ) . in particular we will estimate the quantum state in two different representations or parameterizations : the density matrix and the wigner function .
the physics behind this statistical problem is presented in subsection [ sec.qmeasurements ] which serves as introduction to basic notions of quantum theory , and subsection [ sec.qhomodyne ] which describes the model of quantum homodyne detection from quantum optics .
the relations between the different parameters of the problem are summarized in the diagram at the end of subsection [ sec.qhomodyne ] followed by table [ tbl.states ] containing some examples of quantum states . for the reader
who is not interested in the quantum physics background we state the statistical problem in the next subsection and we will return to it in section [ sec.densitymatrixestimation ] . we observe @xmath17 , i.i.d .
random variables with values in @xmath18 $ ] and distribution @xmath19 depending on the unknown parameter @xmath11 which is an infinite dimensional matrix @xmath20_{j , k=0,\ldots,\infty}$ ] such that @xmath21 ( trace one ) and @xmath22 ( positive definite ) .
the probability density of @xmath19 is @xmath23 where the functions @xmath24 to be specified later , form a orthonormal basis of the space of complex square integrable functions on @xmath25 . because @xmath11 is positive definite and has trace @xmath26 , this is a probability density : real , nonnegative , integrates to @xmath26 . the data @xmath27 comes from independent qhd measurements on identically prepared pulses of light whose properties or state are completely encoded in the matrix @xmath11 called a density matrix .
we will consider the problem of estimating @xmath11 from a given sample .
previous attempts by physicists to estimate the density matrix @xmath11 have focused mainly on the estimation of the individual matrix elements without considering the accuracy of the estimated density matrix with respect to natural distances of the underlying parameter space . in section [ sec.densitymatrixestimation ]
we will present consistency results in the space of density matrices w.r.t . @xmath28 and @xmath29-norms using two different types of estimators , namely projection and sieve maximum likelihood estimators .
an alternative representation of the quantum state is through the wigner function @xmath30 whose estimation is close to a classical computer tomography problem namely , positron emission tomography ( pet ) , @xcite . in pet
one would like to estimate a probability density @xmath31 on @xmath32 from i.i.d .
observations @xmath33 , with probability density equal to the radon transform of @xmath31 : @xmath34(x,\phi)=\int_{-\infty}^\infty f(x\cos\phi+t\sin\phi , x\sin\phi - t\cos\phi)dt.\ ] ] although the wigner function is in general _ not _ positive , its radon transform is always a probability density , in fact @xmath35(x,\phi ) = p_\rho(x,\phi)$ ] .
as the wigner function @xmath36 is in one - to - one correspondence with the density matrix @xmath11 , our state reconstruction problem can be stated as to estimate the wigner function @xmath37 .
this is an ill posed inverse problem as seen from the formula for the inverse of the radon transform @xmath38 where @xmath39 makes sense only as a generalized function . to correct
this one usually makes a cut - off in the range of the above integral and gets a well behaved kernel function @xmath40 .
then the tomographic estimator of @xmath37 is the average sampled kernel @xmath41 for consistency one needs to let the ` bandwidth ' @xmath42 depend on the sample size @xmath14 and @xmath43 as @xmath44 at an appropriate rate . in this paper
we will not follow this approach , which will be treated separately in future work . instead
, we use a plug - in type estimator based on the property @xmath45 where @xmath46 s are known functions and we replace @xmath11 by its above mentioned estimators
. we shall prove consistency of the proposed estimators of the wigner function w. r. t. @xmath29 and supremum norms in the corresponding space .
this subsection serves as a short introduction to the basic notions of quantum mechanics which will be needed in this paper . for simplicity
we will deal first with finite dimensional quantum systems and leave the infinite dimensional case for the next subsection . for further details on quantum statistical inference
we refer to the review and the classic textbooks @xcite and @xcite . in classical mechanics the state of macroscopic systems like billiard balls , pendulums or stellar systems
is described by points on a manifold or `` phase space '' , each of the point s coordinates corresponding to an attribute which we can measure such as position and momentum .
therefore the functions on the phase space are called observables .
when there exists uncertainty about the exact point in the phase space , or we deal with a statistical ensemble , the state is modelled by a probability distribution on the phase space , and the observables become random variables .
quantum mechanics also deals with observables such as position and momentum of a particle , spin of an electron , number of photons in a cavity , but breaks from classical mechanics in that these are no longer represented by functions on a phase space but by hermitian matrices , that is , complex valued matrices which are invariant under transposition followed by complex conjugation .
for example , the components in different directions of the spin of an electron are certain @xmath47 complex hermitian matrices @xmath48 .
any @xmath49-dimensional complex hermitian matrix @xmath50 can be diagonalized by changing the standard basis of @xmath51 to another orthonormal basis @xmath52 such that @xmath53 for @xmath54 , with @xmath55 .
the vectors @xmath56 and numbers @xmath57 are called eigenvectors and respectively eigenvalues of @xmath50 . with respect to the new basis
we can write @xmath58 the physical interpretation of the eigenvalues is that when measuring the observable @xmath50 we obtain ( randomly ) one of the values @xmath57 according to a probability distribution depending on the state of the system before measurement and on the observable @xmath50 .
this probability measure is degenerate if and only if the system before measurement was prepared in a special state called an eigenstate of @xmath50 .
we represent such a state mathematically by the projection @xmath59 onto the one dimensional space generated by the vector @xmath56 in @xmath51 . given a probability distribution @xmath60 over the finite set @xmath61 , we describe a statistical ensemble in which a proportion @xmath62 of systems is prepared in the state @xmath59 by the convex combination @xmath63 .
the expected value of the random result @xmath64 when measuring the observable @xmath50 for this particular state is equal to @xmath65 which can be written shortly @xmath66 similarly , the probability distribution can be recovered as @xmath67 thanks to the orthogonality property @xmath68 .
now , let @xmath69 be a different observable and suppose that @xmath69 does not commute with @xmath50 , that is @xmath70 , then the two observables can not be diagonalized in the same basis , their eigenvectors are different .
consequently , states which are mixtures of eigenvectors of @xmath50 typically will not be mixtures of eigenvectors of @xmath69 and vice - versa .
this leads to an expanded formulation of the notion of state in quantum mechanics independent of any basis associated to a particular observable , and the recipe for calculating expectations and distributions of measurement results .
any preparation procedure results in an statistical ensemble , or state , which is described mathematically by a matrix @xmath11 with the following properties 1 .
@xmath22 ( positive definite matrix ) , 2 .
@xmath71 ( normalization ) . in physics @xmath72 is called a _ density matrix _ , and is for a quantum mechanical system an analogue of a probability density .
notice that the special state @xmath73 defined above is a particular case of density matrix , since it is a mixture of eigenstates of the observable @xmath50 .
the density matrices of dimension @xmath49 form a convex set @xmath74 , whose extremals are the _ pure _ or _ vector _ states , represented by orthogonal projections @xmath75 onto one dimensional spaces spanned by _ arbitrary _
vectors @xmath76 .
any state can be represented as a mixture of pure states which are not necessarily eigenstates of a particular observable . when measuring an observable , for example @xmath50 , of a quantum system prepared in the state @xmath11 we obtain a random result @xmath77 with probability distribution given by equation ( [ eq.quantumprobability ] ) , expectation as in equation ( [ eq.quantumexpectation ] ) , and characteristic function @xmath78 in order to avoid confusion we stress the important difference between @xmath50 which is a matrix and @xmath64 which is a real - valued random variable . more concretely ,
if we write @xmath11 in the basis of eigenvectors of @xmath50 then we obtain the map @xmath79 from states to probability distributions @xmath80 over results @xmath81 @xmath82 notice that @xmath83 is indeed a probability distribution as a consequence of the defining properties of states , and it does not contain information about the off - diagonal elements of @xmath11 , meaning that measuring only the observable @xmath50 is not enough to identify the unknown state . roughly speaking , as @xmath84 , one has to measure on many identical systems each one of a number of @xmath85 mutually non - commuting observables in order to have a one - to - one map between states and probability distributions of results .
the probing of identically prepared quantum systems from different ` angles ' in order to reconstruct their state is broadly named _
quantum state tomography _ in the physics literature .
let us suppose that we have at our disposal @xmath14 systems identically prepared in an unknown state @xmath86 , and for each of the systems we can measure one of the fixed observables @xmath87 .
we write the observables in diagonal form @xmath88 where @xmath89 eigenvalues and @xmath90 eigenstates .
we will perform a randomized experiment , i.e. for each system we will choose the observable to be measured by randomly selecting its index according to a probability distribution @xmath91 over @xmath92 .
the results of the measurement on the @xmath93 system are the pair @xmath94 where @xmath95 are i.i.d .
with probability distribution @xmath96 and @xmath97 is the result of measuring the observable @xmath98 whose conditional distribution is given by @xmath99 the statistical problem is now to estimate the parameter @xmath11 from the data @xmath100 . in the next subsection
we will describe quantum homodyne tomography as an analogue of this problem for infinite dimensional systems .
although correct and sufficient when describing certain quantum properties such as the spin of a particle , the model presented above needs to be enlarged in order to cope with quantum systems with ` continuous variables ' which will be central in our statistical problem .
this technical point can be summarized as follows : we replace @xmath51 by an infinite dimensional complex hilbert space @xmath101 , the hermitian matrices becoming _ selfadjoint operators _ acting on @xmath101 .
the spectral theorem tells us that selfadjoint operators can be ` diagonalized ' in the spirit of ( [ eq.diagonalization ] ) but the spectrum ( the set of ` eigenvalues ' ) can have a more complicated structure , for example it can be continuous as we will see below .
the density matrices are _ positive _ selfadjoint operators @xmath102 such that @xmath71 and can be regarded as infinite dimensional matrices with elements @xmath103 for a given orthonormal basis @xmath104 in @xmath101 .
the central example of a system with continuous variables in this paper is the quantum particle .
its basic observables position and momentum , are two unbounded selfadjoint operators @xmath105 and @xmath106 respectively , acting on @xmath107 , the space of square integrable complex valued functions on @xmath108 @xmath109 for @xmath110 arbitrary functions .
the operators satisfy heisenberg s _ commutation relations _ @xmath111 which implies that they can not be measured simultaneously .
the problem of ( separately ) measuring such observables has been elusive until ten years ago when pioneering experiments in quantum optics by @xcite , led to a powerful measurement technique called _ quantum homodyne detection_. this technique is the basis of a continuous analogue of the measurement scheme presented at the end of the previous subsection where @xmath85 observables were measured in the case of a @xmath49-dimensional quantum system .
the quantum system to be measured is a beam of light with a fixed frequency whose observables are the electric and magnetic field amplitudes which satisfy commutation relations identical to those characterizing the quantum particle , with which they will be identified from now on .
their linear combinations @xmath112 are called _ quadratures _ , and homodyne detection is about measuring the quadratures for _ all _ phases @xmath113 $ ] . the experimental setup shown in figure [ fig.qhtomography ] contains an additional laser called a local oscillator ( lo ) of high intensity @xmath114 and relative phase @xmath115 with respect to the beam in the unknown state @xmath11 .
the two beams are combined through a fifty - fifty beam splitter , and the two emerging beams are then measured by two photon detectors .
a simple quantum optics computation ( see @xcite ) shows that in the limit of big lo intensity the difference of the measurement results ( countings ) of the two detectors rescaled by the lo intensity @xmath116 has the probability distribution corresponding to the measurement of the quadrature @xmath117 .
the result @xmath64 takes values in @xmath108 and its probability distribution @xmath118 has a density @xmath119 and characteristic function ( see equation ( [ eq.generating.function ] ) ) @xmath120 the phase @xmath115 can be controlled by the experimenter by adjusting a parameter of the local oscillator .
we assume that he chooses it randomly uniformly distributed over the interval @xmath9 $ ] .
then the joint probability distribution @xmath19 for the pair consisting in measurement result and phase @xmath121 has density @xmath122 equal to @xmath123 with respect to the measure @xmath124 on @xmath125 $ ]
. an attractive feature of the homodyne detection scheme is the invertibility of the map @xmath126 that associates @xmath19 to @xmath11 , making it possible to asymptotically infer the unknown parameter @xmath11 from the i.i.d .
results @xmath127 of homodyne measurements on @xmath14 systems prepared in the state @xmath11 .
we will see now why this state estimation method is called _ quantum homodyne tomography _ by drawing a parallel with computerized tomography used in the hospitals . in quantum optics
it is common to represent the state of a quantum system by a certain function on @xmath128 called the _ wigner function _
@xmath129 which is much like a joint probability density for @xmath105 and @xmath106 , for instance its marginals are the probability densities for measuring @xmath105 and respectively @xmath106 .
the wigner function of the state @xmath11 is defined by demanding that its fourier transform @xmath130 with respect to _ both _ variables has the following property @xmath131(u , v)= \mathrm{tr}\big(\rho \exp(-iu\mathbf{q}-iv\mathbf{p})\big).\ ] ] we see from this equation that if @xmath105 and @xmath106 were commuting operators then @xmath132 would indeed be the joint probability density of outcomes of their measurement .
as the two observables can not be measured simultaneously , we can not speak of a joint density , in fact the wigner function need not be positive , but many interesting features of the quantum state can be visualized in this way .
let @xmath133 , then @xmath134(t)= \mathrm{tr}\big(\rho \exp(-it\mathbf{x}_\phi)\big)\ ] ] where the fourier transform @xmath135 in the last term is with respect to the first variable , keeping @xmath115 fixed .
the equations ( [ def.wigner ] ) and ( [ eq.fourier transform.tildew ] ) are well known in the theory of radon transform @xmath136 and imply that for each fixed @xmath115 , the probability density @xmath122 is the marginal of the wigner function with respect to the direction @xmath115 in the plane , @xmath137(x,\phi)=\int_{-\infty}^\infty w_{\rho}(x\cos\phi+y\sin\phi , x\sin\phi - y\cos\phi)dy,\ ] ] adding quantum homodyne tomography to a number of applications ranging from computerized tomography to astronomy and geophysics , @xcite . in computerized tomography
one reconstructs an image of the tissue distribution in a cross - section of the human body by recording events whereby pairs of positrons emitted by an injected radioactive substance hit detectors placed in a ring around the body after flying in opposite directions along an axis determined by an angle @xmath113 $ ] . in quantum homodyne tomography
the role of the unknown distribution is played by the wigner function which is in general not positive , but has a probability density @xmath119 as marginal along any direction @xmath115 .
the following diagram summarizes the relations between the various objects in our problem : + ( 10,30)@xmath11 ( 30,30)@xmath37 ( 65,30)@xmath138 ( 110,30)@xmath139 .
( 47,10)@xmath140 ( 11,30)(29,30 ) [ , ] ( 29,30)(11,30 ) [ , ] ( 31,30)(60,30)@xmath136 [ , ] ( 31,29)(45,11)@xmath141 [ , ] ( 60,29)(47,11)@xmath142 [ , ] ( 70,30)(95,30)experiment [ , ] finally in table [ tbl.states ] we give some examples of density matrices and their corresponding wigner function representations for different states .
the matrix elements @xmath143 are calculated with respect to the orthonormal base corresponding to the wave functions of @xmath144 photons states @xmath145 where @xmath146 are the hermite polynomials normalized such that @xmath147 .
a few graphical representations can be seen in figure [ fig.examples ] .
.density matrix and wigner function of some quantum states [ cols="^,^,^",options="header " , ] the vacuum is the pure state of zero photons , notice that in this case the distributions of @xmath105 and @xmath106 are gaussian .
the thermal state is a mixed state describing equilibrium at temperature @xmath148 , having gaussian wigner function with variance increasing with the temperature .
the coherent state is pure and characterizes the laser pulse .
the photon number is poisson distributed with an average of @xmath13 photons .
the squeezed states have gaussian wigner functions whose variances for the two directions are different but have a fixed product . the parameters @xmath13 and @xmath149 satisfy the condition @xmath150
, @xmath151 is a normalization constant , @xmath152 , and @xmath153 .
@xcite presented the density matrix analogue of formula ( [ eq.inverse.radon.transform ] ) of the wigner function as inverse radon transform of the probability density @xmath154 @xmath155 where @xmath156 is the generalized function given in equation ( [ eq.kernel ] ) whose argument is a selfadjoint operator @xmath157 .
the method has been further analyzed in @xcite , @xcite , see also @xcite .
we recall that in the case of the wigner function we needed to regularize the kernel @xmath156 by introducing a cut - off in the integral ( [ eq.kernel ] ) . for density matrices
the philosophy will be rather to project on a finite dimensional subspace of @xmath158 whose dimension @xmath13 will play the role of the cut - off .
in fact all the matrix elements of the density matrix @xmath11 with respect to the orthonormal basis @xmath159 defined in ( [ eq.psi_n ] ) , can be expressed as kernel integrals @xmath160 with @xmath161 bounded real valued functions which in the quantum tomography literature are called _
pattern functions_. the singularity of the kernel @xmath156 is reflected in the asymptotic behavior of @xmath162 as @xmath163 .
a first formula for @xmath162 was found in @xcite and uses laguerre polynomials .
this was followed by a more transparent one due to @xcite , @xmath164 for @xmath165 , where @xmath166 and @xmath167 represent the square integrable and respectively the unbounded solutions of the schrdinger equation , @xmath168\psi=\omega~\psi , \qquad \omega\in\r.\ ] ] figure [ fig.ptnfnc ] shows pattern functions for different values of @xmath144 and @xmath169 .
we notice that the oscillatory part is concentrated in an interval centered at zero whose length increase with @xmath144 and @xmath169 , the number of oscillations increases with @xmath144 and @xmath169 and the functions become more irregular as we move away from the diagonal . it can be shown that tails of the pattern function decay like @xmath170 .
more properties of the pattern function can be found in @xcite and @xcite .
equation ( [ eq.quantum_tomographic_rho_n,n+d ] ) suggests the _ unbiased estimator _
@xmath171 of @xmath11 , based on @xmath14 i.i.d .
observations of @xmath10 , whose matrix elements are : @xmath172 where @xmath173 , see @xcite , @xcite . by the strong law of large numbers the individual matrix elements of this estimator
converge to the matrix elements of the true parameter @xmath11 .
however the infinite matrix @xmath171 need not be positive , normalized , or even selfadjoint , thus it can not be interpreted as a quantum state .
these problems are similar to those encountered when trying to estimate an unknown probability density by using unbiased estimators for all its fourier coefficients .
the remedy is to estimate only a finite number of coefficients at any moment , obtaining a projection estimator onto the subspace generated by linear combinations of a finite subset of the basis vectors .
in our case we will project onto the space of matrices of dimension @xmath174 with respect to the basis @xmath159 , @xmath175 and @xmath176 for @xmath177 . in order to test the performance of our estimators we introduce the @xmath28 and @xmath29 distances on the space of density matrices .
let @xmath11 and @xmath178 be two density matrices with @xmath179 the diagonal form of their difference , and notice that some of the eigenvalues are positive and some negative such that their sum is zero due to the normalization of the density matrices .
we define the absolute value @xmath180 and the norms @xmath181^{1/2}= \left[\sum_{j , k= 0}^\infty|\rho_{k , j}-\tau_{k , j}|^2\right]^{1/2}.\end{aligned}\ ] ] let us consider now the mean integrated square error ( mise ) and split it into the bias and variance parts : @xmath182 by choosing @xmath183 as @xmath44 the bias @xmath184 converges to zero
. for the variance we have the upper bound @xmath185 the proof of the following lemma on the norms of the pattern functions can be found in @xcite .
there exist constants @xmath186 such that @xmath187 by applying the lemma to equation ( [ eq.variance.upper.bound ] ) we conclude that the estimator @xmath188 is consistent with respect to the @xmath29 distance if we choose @xmath189 as @xmath190 such that @xmath191 .
based on the property @xmath192 we can prove a similar result concerning @xmath28-consistency , see @xcite .
[ th.norm2 ] let @xmath189 be the dimension of the pattern function projection estimator .
if @xmath191 , then @xmath193 if @xmath194 then @xmath195 rates of consistency can be obtained by assuming that the state belongs to a given class for which upper bounds of the bias can be calculated and @xmath15 is chosen such as to balance bias and variance . this problem will be attacked in future work within the minimax framework . in section [ sec.experimental ]
we present a data - dependent way of selecting the dimension of the projection estimator based on the minimization of the empirical @xmath29-risk using a cross - validation technique .
we will consider now a maximum likelihood approach to the estimation of the state @xmath11 .
let us recall the terms of the problem : we are given a sequence @xmath196 of i.i.d .
random variables @xmath197 with values in @xmath125 $ ] and probability density @xmath154 depending on the parameter @xmath11 which is an infinite dimensional density matrix .
we would like to find an estimator of @xmath11 @xmath198 let @xmath11 and @xmath178 be density matrices and denote by @xmath199 the hellinger distance between the two probability distributions .
the following relations are well known @xmath200 an important property which is true for any measurement is the following inequality between the classical and quantum distances , cf .
@xcite , @xmath201 from ( [ eq.inequalities.hellinger.norm_1 ] ) and ( [ eq.norm_1.inequality ] ) we obtain @xmath202 for arbitrary states @xmath203 . as we have seen previously ,
the inverse map from @xmath19 to @xmath11 is unbounded thus we do not have an inequality in the opposite direction to the one above
. however we can prove the continuity of the inverse map by using a matrix analogue of the scheff s lemma from classical probability ( see @xcite ) stating that if a sequence of probability densities converges pointwise almost everywhere to a probability density , then they also converge in total variation norm
. the matrix scheff s lemma which can be found in @xcite says that if @xmath204 are density matrices ( positive and trace one ) and if the coefficients @xmath205 converge to @xmath206 as @xmath14 goes to infinity , for any fixed indices @xmath207 , then @xmath208 .
but by equation ( [ eq.quantum_tomographic_rho_n,n+d ] ) if the sequence @xmath209 converges to @xmath19 as @xmath44 with respect to the @xmath210-distance then @xmath205 converges to @xmath206 and thus @xmath208 , completing the proof of the continuity of the map from @xmath19 to @xmath11 .
in particular we have @xmath211 due to the inequality between the @xmath28 and @xmath29 norms @xmath212 the maximum likelihood estimator of @xmath11 is defined as @xmath213 where the maximum is taken over all density matrices in on the space @xmath158 .
however there exist density matrices @xmath178 such that the probability density @xmath214 takes arbitrarily high values at all the points @xmath27 . to see this
let us first remind the reader that any density matrix is a convex combination of `` pure states '' which are projections @xmath75 on one dimensional sub - spaces of @xmath158 generated by vectors @xmath215 which can be written as a fourier sum @xmath216 in the basis @xmath217 given in equation ( [ eq.psi_n ] ) , with @xmath218 . for any such state
the corresponding probability density is @xmath219 .
it is clear that there exists a one - to - one relation between @xmath215 and @xmath220 which preserves the @xmath29-norms , thus we can choose vectors @xmath221 such that @xmath222 and @xmath223 for all @xmath224 and arbitrary @xmath225 .
then the density matrix @xmath226 representing a statistical mixture of the pure states leads to the likelihood @xmath227 which can be arbitrarily high for any fixed @xmath14 .
this drawback can be corrected by using for example penalized maximum likelihood estimators or by restricting the state space to some subspace @xmath228 of density matrices such that for any amount of data the maximum of the likelihood over @xmath228 exists and @xmath229 is dense in the space of all density matrices @xmath230 with respect to some chosen distance function .
such a method is called sieve maximum likelihood and we refer to @xcite and for the general theory . the choice of the sieves @xmath228 should be tailored according to the problem one wants to solve , the class of states one is interested in , etc . here
we will use the same subspaces as for the projection estimator of the previous subsection , that is @xmath228 consists of those states with maximally @xmath231 photons described by density matrices over the subspace spanned by the basis vectors @xmath232 .
we will call @xmath233 the _ number states sieves _ and the dimension @xmath15 will be an increasing function of @xmath14 which will be fixed later so as to guarantee consistency .
@xmath234 notice that the dimension of the space @xmath228 is @xmath235 .
let now the estimator be @xmath236 where the maximum can be shown to exist for example by using compactness arguments .
we will denote the corresponding sieve in the space of probability densities by @xmath237 the theory of m - estimators @xcite tells us that the consistency of ml estimators depends essentially on the `` size '' of the parameter space , in our case the sieves @xmath228 or @xmath238 , which is measured by entropy numbers with respect to some distance , for example the @xmath28-norm on density matrices or the hellinger distance between probability distributions .
let @xmath239 be a class of density matrices .
let @xmath240 be the smallest value of @xmath241 for which there exist pairs of hermitian matrices ( not necessarily density matrices ) @xmath242 $ ] with @xmath243 such that @xmath244 for all @xmath169 , and such that for each @xmath245 there is a @xmath246 satisfying @xmath247 then @xmath248 is called _
@xmath249-entropy with bracketing _ of @xmath239 .
we note that this definition relies on the concept of positivity of matrices and the existence of the @xmath250-distance between states .
but the same notions exist for the space of integrable functions thus by replacing density matrices with probability densities and selfadjoint operators with functions we obtain the definition of the @xmath249-entropy with bracketing @xmath251 for some space of probability densities @xmath252 , see @xcite .
moreover by using equation ( [ eq.norm_1.inequality ] ) and the fact that the linear extension of the map from density matrices to probability densities sends a positive matrix to a positive function , we get that for any @xmath249-bracketing @xmath242 $ ] for @xmath228 , the corresponding functions @xmath253 $ ] form a @xmath249-bracketing for @xmath238 , i.e. they satisfy @xmath254 and for any @xmath255 there exists a @xmath256 such that @xmath257 . thus @xmath258 the following proposition gives an upper bound of the `` quantum '' bracketing entropy and in consequence for @xmath259 .
its proof can be found in @xcite and relies on choosing a maximal number of nonintersecting balls centered in @xmath228 having radius @xmath260 and then providing a pair of brackets for each ball .
let @xmath228 be the class of density matrices of dimension @xmath15 .
then @xmath261 for some constant @xmath262 independent of @xmath14 and @xmath249 .
by combining the previous inequalities with equation ( [ contractivity.t ] ) we get the following bound for the bracketing entropy of the class of square - root densities @xmath263 with respect to the @xmath29-distance @xmath264 we will concentrate now on the hellinger consistency of the sieve maximum likelihood estimator @xmath265 .
we will appeal to a theorem from @xcite , which is similar to other results in the literature on non - parametric @xmath266-estimation ( see for example ) .
there are two competing factors which contribute to the convergence of @xmath267 .
the first is related to the approximation properties of the sieves with respect to the whole parameter space .
such a distance from @xmath11 to the sieve @xmath228 can take different expressions , for example in terms of the @xmath268-distance between the corresponding probability measures @xmath269 where the @xmath268-distance between two probability distributions is given by @xmath270 notice that @xmath271 depends on @xmath14 through the growth rate of the sieve @xmath15 .
the second factor influencing the convergence of @xmath267 is the size of the sieves which is expressed by the bracketing entropy .
the non - parametric sieve maximum likelihood estimation theory shows that consistency holds if there exists a sequence @xmath272 such that the following _ entropy integral inequalities _ are satisfied for all @xmath14 @xmath273 where @xmath274 is some universal constant , @xcite . from ( [ eq.bracketing.entropy.hellinger ] )
we get @xmath275,\ ] ] which implies the following constraint for @xmath183 and @xmath272 , @xmath276 [ th.hellinger.consistency ] suppose that the state @xmath11 satisfies @xmath277 .
let @xmath278 be the sieve mle with @xmath15 and @xmath279 satisfying ( [ eq.entropy.integral.inequality ] ) , then @xmath280 _ proof .
_ details can be found in @xcite based on theorem 10.13 of @xcite . from the physical point of view , we are interested in the convergence of the state estimator @xmath278 with respect to the @xmath28 and @xmath29-norms on the space of density matrices
. clearly the rates of convergence for such estimators are slower than those of their corresponding probability densities .
as shown in the beginning of this subsection the map sending probability densities @xmath154 to density matrices @xmath11 is continuous , thus an estimator @xmath281 taking values in the space of density matrices @xmath230 is consistent in the @xmath28 or @xmath29-norms if and only if @xmath265 converges to @xmath19 almost surely with respect to the hellinger distance .
[ cor.consistency.mle.densitymatrix ] the hellinger consistency of @xmath265 is equivalent to the @xmath282-consistency of @xmath278 . in particular ,
if @xmath283 and the assumptions of theorem [ th.hellinger.consistency ] hold , then we have @xmath284 , a.s .. the wigner function plays an important role in quantum optics as an alternative way of representing quantum states and calculating an observable s expectation : for any observable @xmath50 there exists a function @xmath285 from @xmath128 to @xmath108 such that @xmath286 besides , physicists are interested in estimating the wigner function for the purpose of identifying features which can be easier visualized than read off from the density matrix , for example a `` non - classic '' state may be recognized by its patches of negative wigner function , while `` squeezing '' is manifest through the oval shape of the support of the wigner function , see table [ tbl.states ] and figure [ fig.examples ] . as described in subsection [ sec.qhomodyne ] the wigner function
should be seen formally as a joint density of the observables @xmath105 and @xmath106 which may take non - negative values , reflecting the fact that the two observables can not be measured simultaneously .
however the wigner function shares some common properties with probability densities , in particular their marginals @xmath287 and @xmath288 are probability densities on the line .
in fact this is true for the marginals in any direction which are nothing else then the densities @xmath122 . on the other hand there
exist probability densities which are not wigner functions and vice - versa , for example the latter can not be too `` peaked '' : @xmath289 as a corollary of this uniform boundedness we get @xmath290 for any density matrices @xmath11 and @xmath178 .
indeed we can write @xmath291 where @xmath292 and @xmath293 represent the positive and negative part of @xmath294
. then @xmath295 another important property is the fact that the linear span of the wigner functions is dense in @xmath296 , the space of real valued , square integrable functions on the plane , and there is an isometry ( up to a constant ) between the space of wigner functions and that of density matrices with respect to the @xmath29-distances @xmath297 in section [ sec.problem ] we have described the standard estimation method employed in computerized tomography which used a regularized kernel @xmath298 with bandwidth @xmath299 converging to zero as @xmath44 at an appropriate rate .
this type of estimators for the wigner function will be analyzed separately in future work in the minimax framework along the lines of @xcite .
the estimators which we propose in this subsection are of a different type , they are based on estimators for @xmath11 plugged into the following linearity equation @xmath300 where @xmath46 are known functions corresponding to the matrix with the entry @xmath301 equal to @xmath26 and all the rest equal to zero , see @xcite .
the isometry ( [ eq.isometry ] ) implies that the family @xmath302 forms an orthogonal basis of @xmath303 .
following the same idea as in the previous section we consider the projection estimator @xmath304 [ th.wigner.norm2 ] let @xmath15 be such that @xmath183 and @xmath305 then @xmath306 _ proof : _ apply isometry property and theorem [ th.norm2 ] .
similarly we can extend the sml estimator of the density matrix to the wigner function .
define the subspace @xmath307 with @xmath228 as in equation ( [ def.q(n ) ] ) , and define the corresponding sml estimator as @xmath308 where @xmath278 was defined in ( [ def.smle ] ) .
[ th.hellinger.consistency.wigner ] suppose that @xmath11 satisfies @xmath277 .
let @xmath309 be the sml estimator with @xmath15 and @xmath279 satisfying ( [ eq.entropy.integral.inequality ] ) and @xmath283 .
then we have @xmath310 almost surely . under the same conditions
@xmath311 almost surely .
_ proof : _ apply the inequalities ( [ eq.norm1.norminfty.inequality ] , [ eq.isometry ] ) and corollary [ cor.consistency.mle.densitymatrix ] .
the homodyne tomography measurement as presented up to now does not take into account various losses ( mode mismatching , failure of detectors ) in the detection process which modify the distribution of results in a real measurement compared with the idealized case .
fortunately , an analysis of such losses ( see @xcite ) shows that they can be quantified by a single _ efficiency _ coefficient @xmath312 and the change in the observations amounts replacing @xmath313 by the noisy observations @xmath314 with @xmath315 a sequence of i.i.d .
standard gaussians which are independent of all @xmath316 .
the problem is again to estimate the parameter @xmath11 from @xmath317 for @xmath318 .
the efficiency - corrected probability density is then the convolution @xmath319 ~dx.\ ] ] the physics of the detection process detailed in @xcite offers an alternative route from the state to the probability density of the observations @xmath320 . in a first step one
performs a _ bernoulli transformation _
@xmath321 on the state @xmath11 which is a quantum equivalent of the convolution with noise for probability densities , and obtains a new density matrix @xmath322 . to understand the bernoulli transformation
let us consider first the diagonal elements @xmath323 and @xmath324 which are both probability distributions over @xmath325 and represent the statistics of the number of photons in the two states .
let @xmath326 be the binomial distribution .
then @xmath327 which has a simple interpretation in terms of an `` absorption '' process by which each photon of the state @xmath11 goes independently through a filter and is allowed to pass with probability @xmath328 or absorbed with probability @xmath329 .
the formula of the bernoulli transformation for the whole matrix is @xmath330^{1/2}\rho_{j+p , k+p}.\ ] ] the second step is to perform the usual quantum tomography measurement with ideal detectors on the `` noisy '' state @xmath322 obtaining the results @xmath320 with density @xmath331 .
it is noteworty that the transformations @xmath321 form a semigroup , that is they can be composed as @xmath332 and the inverse of @xmath321 is simply obtained by replacing @xmath328 with @xmath333 in equation ( [ eq.bernoulli ] ) .
notice however that if @xmath334 the power series @xmath335 appearing in the inverse transformation diverges , thus we need to take special care in this range of parameters .
a third way to compute the inverse map from @xmath331 to @xmath11 is by using pattern functions depending on @xmath328 which incorporate the deconvolution map from @xmath331 to @xmath336 : @xmath337 such functions are analyzed in @xcite where it is argued that the method has a fundamental limitation for @xmath334 in which case the pattern functions are unbounded , while for @xmath338 numerical calculations show that their range grows exponentially fast with both indices @xmath339 .
the two estimation methods considered in section [ sec.densitymatrixestimation ] can be applied to the state estimation with noisy observations .
the projection estimator has the same form as in subsection [ sec.pfe ] with a similar analysis of the mean @xmath29-risk taking into account the norms of the new pattern functions @xmath340 .
the sieve maximum likelihood estimator follows the definition in subsection [ sec.mle ] and a consistency result can be formulated on the lines of corollary [ cor.consistency.mle.densitymatrix ] .
we expect however that the rates of convergence will be dramatically slower and we will leave this analysis for a separate work .
in this section we study the performance of the pattern function projection estimator and the sieve maximum likelihood estimator using simulated data . in table
[ tbl.states ] we showed some examples of density matrices and wigner functions of quantum states . in figure [
fig.examples ] , we display their corresponding graphical representation . for some of them the corresponding probability distribution can be expressed explicitly and it is possible to simulate data . in particular we shall simulate data from qht measurements on squeezed states with efficiency @xmath341 .
+ + + in order to implement the two estimators we need to compute the basis functions @xmath342 and the functions @xmath343 , which are solutions of schrdinger equation ( [ eq.schrodinger ] ) . for this
, we use an appropriate set of recurrent equations , see @xcite , ch .
pattern functions can then be calculated as @xmath344 for all @xmath165 , otherwise @xmath345 and then @xmath346 . on the practical side ,
finding the maximum of the likelihood function over a set of density matrices is a more complicated problem due to the positivity and trace one constraints which must be taken into account .
a solution was proposed in @xcite , where the restriction on positivity of a density matrix @xmath11 is satisfied by writing the cholevski decomposition @xmath347 where @xmath348 of _ upper triangular _ matrices of the same dimension as @xmath11 with complex coefficients above the diagonal and reals on the diagonal .
the normalization condition @xmath349 translates into @xmath350 which defines a ball in the space of upper triangular matrices with the @xmath29-distance .
we will denote by @xmath351 the set of such matrices having dimension @xmath15 .
the sieve maximum likelihood estimator is the solution of the following optimization problem with @xmath174 @xmath352 the numerical optimization was performed using a classical descendent method with constraints .
notice that we have an optimization problem on @xmath353 real variables . given the problem of high dimensionality and computational cost we propose an alternative method to the procedure mentioned above .
it exploits the mixing properties of our model .
any density matrix of dimension @xmath13 can be written as convex combination of @xmath13 _ pure _ states , i.e. @xmath354 , where @xmath355 , @xmath356 and @xmath357 is a one dimensional projection whose cholevski decomposition is of the form @xmath358 where @xmath359 is the row vector of dimension @xmath13 on which @xmath357 projects , and @xmath360 is the column vector of the complex conjugate of @xmath359 .
it should be noted that even though decomposition of our state in pure states is not unique this is not a problem given we are actually not interested in this representation but in the resulting convex combination , the state itself .
now we can state the problem as to find the maximizer of the loglikelihood @xmath361 where @xmath362 represents the m@xmath363 coordinate of @xmath359 .
we now maximize over all @xmath364 where @xmath365 we propose an em algorithm as an alternative method to the one presented in @xcite .
see , @xcite for an exposition on the formulation of the em algorithm for problems of ml estimation with mixtures of distributions .
the iteration procedure is given then in the following steps .
+ 1 ) compute the expectation of the conditional likelihood : @xmath366 2 ) maximize @xmath367 over all @xmath364 and obtain @xmath368 with components @xmath369 where @xmath370 as initial condition one could take very simple ad hoc states .
for example , take @xmath371 as the vector @xmath372 and @xmath373 and @xmath359 to be the null vector and @xmath374 for @xmath375 .
this corresponds to the _ one photon _ state .
another possible combination is to take @xmath376 and @xmath377 .
this corresponds to the state represented by a diagonal matrix , called a chaotic state .
another strategy is to consider a preliminary estimator , based on just few observations , diagonalize it and take @xmath359 equal to its eigenstates and @xmath378 the corresponding eigenvalues . in this way one
hopes to start the iteration from a state closer to the optimum one . in terms of speed
our simulations suggests to use the em version as dimension grows . establishing any objective comparison between direct optimization and em algorithm
has proven to be difficult given the dependency on initial conditions , and high dimensionality of the problem . in figure
[ fig.estimation.squeezed ] we show the result of estimating the squeezed state defined in table [ tbl.states ] using samples of size 1600 , for both pattern function and maximum likelihood estimators . at a first glance
one can see that the pattern function estimator result is rougher when compared to the maximum likelihood estimator .
this is due to the fact discussed in subsection [ sec.pfe ] that the variance of @xmath379 increases as a function of @xmath144 and @xmath169 as we move away from the diagonal .
the relation between quality of estimation and dimension of the truncated estimator is seen more clearly in figure [ fig.mlevspf.l2error ] where the @xmath29-errors of estimating the coherent state is shown for both estimators at different sample sizes .
the @xmath380-dot represents the point of minimum and thus , optimum dimension @xmath381 for each curve .
the curves presented there are the mean @xmath29-error estimated using 15 simulations for each sample size .
from there we can see that the optimum pfp estimator for the sample of size @xmath382 is the one corresponding to @xmath383 while the optimum sml would be obtained using the sieve of size @xmath384 .
let us first analyze the performance of pfp estimator .
notice that for @xmath385 the mean @xmath29-error increases quadratically with @xmath13 due to contribution from the variance term .
as @xmath14 increases the variance decreases like @xmath386 for a fixed dimension @xmath13 and consequently , the optimal dimension @xmath381 increases .
one can see that the minimum is attained rather sharply .
this suggests that , in order to get a good result a refined method of guessing the optimum dimension becomes necessary , eg .
bic , aic or cross - validation .
figure [ fig.cross-validation ] shows a cross - validation estimator for the @xmath29-error of the pfp estimator for three simulations ( continuous lines ) , each one based on 1600 observations with a squeezed state , and for comparison the expected @xmath29-error ( dashed line ) .
this represents only a first attempt to implement model selection procedures for this problem which should be investigated more thoroughly from the theoretical and practical point of view .
we pass now to the sml estimator . in figure [ fig.mlevspf.l2error ]
we see that it has smaller @xmath29-error the the pfp estimator at its optimum sieve dimension .
it is remarkable that the behavior of the @xmath29-error , for @xmath385 has a different behavior in this case , increasing much slower than the pfp estimator at the right side of its corresponding optimal dimension .
this suggests that sml estimators could have a lower risk if the optimum dimension is overestimated .
error for pattern function estimator and sieve maximum likelihood estimator and different sample sizes : @xmath387 .
last graphic represents the optimum @xmath29 error for different sample size , using a logarithmic scale on both axis.,width=529 ] the bottom right pane of figure [ fig.mlevspf.l2error ] shows the optimum value of the @xmath29-risk in terms of sample size .
both axis are represented in a logarithmic scale .
the observed linear pattern indicates that the @xmath29-risk decreases as @xmath388 .
the slope of both curves correspond to @xmath389 , showing an almost parametric rate which is not surprising given the smoothness of the example that we consider .
the value of @xmath178 for the pfp estimator is a bit smaller than for the sml estimator , confirming its worse performance .
notice also that the constant @xmath390 is bigger for pfp than for the sml estimator .
we expect that the contrast between the two estimators will be accentuated when @xmath391 .
finally , in figure [ fig.estimation.wigner ] we show the result of estimating the wigner function of the squeezed state using both estimators .
as explained in section [ sec.wigner ] the corresponding estimator can be obtained by plugging - in the density matrix estimator in equation [ eq.linearity.wigner.rho ] .
the density matrix estimators for the same state are represented in figure [ fig.estimation.squeezed ] .
in this paper we have proposed a pattern function projection estimator and a sieve maximum likelihood estimator for the density matrix of the quantum state and its wigner function .
we proved they are consistent for different norms in their corresponding spaces .
there are many open statistical questions related to quantum tomography and we would like to enumerate a few of them here . * _ cross - validation .
_ for both types of estimators , a data dependent method is needed for selecting the optimal sieve dimension .
we mention criteria such as unbiased cross - validation , hard thresholding or other types of minimum contrast estimators . * _ efficiency @xmath312 . _
a realistic detector has detection efficiency @xmath312 which introduces an additional noise in the homodyne data .
from the statistical point of view we deal with a gaussian deconvolution problem on top of the usual quantum tomography estimation . *
_ rates of convergence . _ going beyond consistency requires the selection of classes of states which are natural both from the physical , as well as statistical point of view .
one should study optimal and achieved rates of convergence for given classes . for @xmath312 the rates
are expected to be significantly lower than in the ideal case , so it becomes even more crucial to use optimal estimators . in applications ,
sometimes only the estimation of a functional of @xmath11 such as average number of photons or entropy may be needed .
this will require a separate analysis , cf .
@xcite . * _ kernel estimators for wigner function .
_ when estimating the wigner function it seems more natural to use a kernel estimator such as in @xcite and to combine this analysis with the deconvolution problem in the case noisy observations @xmath391 , @xcite .
* _ other quantum estimation problems .
_ the methods used here for quantum tomography can be applied in other problems of quantum estimation , such as for example the calibration of measurement devices or the estimation of transformation of states under the action of quantum mechanical devices .
smithey , d. t. , beck , m. , raymer , m. g. , and faridani , a. ( 1993 ) .
measurement of the wigner distribution and the density matrix of a light mode using optical homodyne tomography : application to squeezed states and the vacuum .
_ 70 _ , 12441247 . | the quantum state of a light beam can be represented as an infinite dimensional density matrix or equivalently as a density on the plane called the wigner function .
we describe quantum tomography as an inverse statistical problem in which the state is the unknown parameter and the data is given by results of measurements performed on identical quantum systems .
we present consistency results for pattern function projection estimators as well as for sieve maximum likelihood estimators for both the density matrix of the quantum state and its wigner function .
finally we illustrate via simulated data the performance of the estimators .
an em algorithm is proposed for practical implementation .
there remain many open problems , e.g. rates of convergence , adaptation , studying other estimators , etc . , and a main purpose of the paper is to bring these to the attention of the statistical community . |
this paper presents a series of numerical studies which investigate the use of shear in the production of chaos .
the phenomenon in question can be described roughly as follows : an external force is applied to a system with tame , nonchaotic dynamics .
if the forcing is strategically applied to interact with the shearing in the underlying dynamics , it can sometimes lead to the folding of phase space , which can in turn lead to positive lyapunov exponents for a large set of initial conditions .
this phenomenon , which we call _ shear - induced chaos _
, occurs in a wide variety of settings , including periodically - forced oscillators . for a topic as general as this ,
it is difficult to compile a reasonable set of references .
we have not attempted to do that , but mention that the first known observation of a form of this phenomenon was by van der pol and van der mark 80 years ago @xcite .
other references related to our work will be mentioned as we go along .
the starting point of the present work is a series of papers by wang and young @xcite . in these papers ,
the authors devised a method for proving the existence of strange attractors and applied their techniques to some natural settings ; two of their main examples are periodically - kicked oscillators and systems undergoing hopf bifurcations .
they identified a simple geometric mechanism to explain how the chaotic behavior comes about .
because of the perturbative nature of their analysis , however , the kicks in these results have to be followed by very long periods of relaxation .
in other words , the chaos in these attractors develop on a very slow time scale .
relevant parts of the results of wang and young are reviewed in sect .
[ wang - young ] .
the aim of the present paper is to study shear - induced chaos in situations not accessible by current analytic tools .
we believe that this phenomenon is widespread , meaning it occurs for large sets of parameters , and that it is robust , meaning it does not depend sensitively on the type of forcing or even background dynamics as long as certain geometric conditions are met .
we validate these ideas through a series of numerical studies in which suitable parameters are systematically identified following ideas from @xcite and @xcite .
four separate studies are described in sects .
[ study1] [ study4 ] .
the first three studies involve an oscillator driven by different types of forcing ( both deterministic and stochastic ) ; in these studies , the unforced system is a simple linear shear flow model . in the fourth study ,
the unforced dynamics are that of a coupled oscillator system described by a ( periodic or quasi - periodic ) flow on the @xmath0-torus .
the linear shear flow used in studies 13 has been studied independently in @xcite and @xcite .
it is the simplest system known to us that captures all the essential features of typical oscillator models relevant to shear - induced chaos .
moreover , these features appear in the system in a way that is easy to control , and the effects of varying each are easy to separate .
this facilitates the interpretation of our theoretical findings in more general settings in spite of the fact that numerical studies necessarily involve specific models .
we mention that our results on shear flows are potentially applicable to a setting not discussed here , namely that of the advection and mixing of passive scalar tracers in ( weakly compressible ) flows .
finally , we remark that this work exploits the interplay between deterministic and stochastic dynamics in the following way : the geometry in deterministic models are generally more clear - cut .
it enables us to extract more readily the relationship between quantities and to deduce the type of results these relationships may lead .
results for stochastic models , on the other hand , tend to be more _ provable _ than their counterparts in deterministic models , where competing scenarios lead to very delicate dependences on parameters . our numerical results on stochastic forcing in studies 24 point clearly to the possibility of ( rigorous ) theorems , some versions of which
, we hope , will be proved in the not too distant future .
in this section , we review some rigorous results of wang and young ( mainly @xcite , also @xcite ) and the geometric mechanism for producing chaos identified in the first two of these papers .
we will focus on the case of limit cycles , leaving the slightly more delicate case of supercritical hopf bifurcations to the reader . the material summarized in this section form the starting point for the numerical investigations in the present paper . consider a smooth flow @xmath1 on a finite dimensional riemannian manifold @xmath2 ( which can be @xmath3 ) , and let @xmath4 be a
_ hyperbolic limit cycle _ , _ i.e. _ @xmath4 is a periodic orbit of @xmath1 with the property that if we linearize the flow along @xmath4 , all of the eigenvalues associated with directions transverse to @xmath4 have strictly negative real parts .
the _ basin of attraction _ of @xmath4 , @xmath5 , is the set @xmath6 .
it is well known that hyperbolic limit cycles are robust , meaning small perturbations of the flow will not change its dynamical picture qualitatively .
a _ periodically - kicked oscillator _ is a system in which kicks " are applied at periodic time intervals to a flow @xmath1 with a hyperbolic limit cycle .
for now let us think of a kick " as a mapping @xmath7 .
if kicks are applied @xmath8 units of time apart , then the time evolution of the kicked system can be captured by iterating its time-@xmath8 map @xmath9 . if there is a neighborhood @xmath10 of @xmath4 such that @xmath11 , and the relaxation time is long enough that points in @xmath12 return to @xmath10 , _
i.e. _ , @xmath13 , then @xmath14 is an attractor for the periodically kicked system @xmath15 . in a sense
, @xmath16 is what becomes of the limit cycle @xmath4 when the oscillator is periodically kicked .
since hyperbolic limit cycles are robust , @xmath17 is a slightly perturbed copy of @xmath4 if the kicks are weak .
we call it an `` invariant circle . ''
stronger kicks may break " the invariant circle , leading to a more complicated invariant set . of interest in this paper
is when @xmath17 is a strange attractor , _
i.e. _ , when the dynamics in @xmath10 exhibit sustained , observable chaos .
two theorems are stated below .
theorem 1 is an abstract result , the purpose of which is to emphasize the generality of the phenomenon .
theorem 2 discusses a concrete situation intended to make transparent the relevance of certain quantities .
let leb@xmath18 denote the lebesgue measure of a set .
@xcite let @xmath1 be a @xmath19 flow with a hyperbolic limit cycle @xmath4 .
then there is an open set of kick maps @xmath20 with the following properties : for each @xmath21 , there is a set @xmath22 with leb@xmath23 such that for each @xmath24 , @xmath17 is a `` strange attractor '' of @xmath15 .
the term `` strange attractor '' in the theorem has a well - defined mathematical meaning , which we will discuss shortly . but first let us take note of the fact that this result applies to all systems with hyperbolic limit cycles , independent of dimension or other specifics of the defining equations .
second , we remark that the kicks in this theorem are very infrequent , _ i.e. _ @xmath25 , and that beyond a certain @xmath26 , the set @xmath27 is roughly periodic with the same period as the cycle @xmath4 .
the term `` strange attractor '' in theorem 1 is used as short - hand for an attractor with a package of well defined dynamical properties .
these properties were established for a class of rank - one attractors ( see @xcite for the 2-dimensional case ; a preprint for the @xmath28-dimensional case will appear shortly ) .
in @xcite , the authors identified a set of conditions that implies the existence of such attractors , and the verification of the conditions in @xcite in the context of theorem 1 is carried out in @xcite ( see also @xcite and @xcite for other applications of these ideas ) .
we refer the reader to the cited papers for more details , and mention only the following three characteristics implied by the term strange attractor " in this section .
* there is a set @xmath29 of full lebesgue measure in the basin of attraction of @xmath17 such that orbits starting from every @xmath30 have ( strictly ) positive lyapunov exponents .
* @xmath15 has an ergodic srb measure @xmath31 , and for every continuous observable @xmath32 , @xmath33 * the system @xmath34 is mixing ; in fact , it has exponential decay of correlations for hlder continuous observables . an important remark before leaving theorem 1 : notice that the existence of `` strange attractors '' is asserted for @xmath15 for only a positive measure set of @xmath8 , not for all sufficiently large @xmath8 .
this is more a reflection of reality than a weakness of the result .
for large enough @xmath8 in the complement of @xmath27 , the attractor is guaranteed to contain horseshoes , the presence of which will lead to some semblance of chaotic behavior . after a transient , however , typical orbits may ( or may not ) tend to a stable equilibrium . if they do , we say @xmath15 has _
transient chaos_. this is to be contrasted with properties ( 1)(3 ) above , which represent a much stronger form of chaos .
the next result has an obvious analog in @xmath28-dimensions ( see @xcite ) , but the 2-d version illustrates the point .
@xcite consider the system @xmath35 where @xmath36 are coordinates in the phase space , @xmath37 are constants , and @xmath38 is a nonconstant smooth function . if the quantity @xmath39 is sufficiently large ( how large depends on the forcing function @xmath40 ) , then there is a positive measure set @xmath41 such that for all @xmath24 , @xmath15 has a strange attractor in the sense above . [ shear ] here
, the term involving @xmath42 defines the kick , and @xmath43 .
we explain intuitively the significance of the quantity @xmath44 .
as noted earlier , to create a strange attractor , it is necessary to `` break '' the limit cycle .
the more strongly attractive @xmath4 is , the harder it is to break . from this
we see the advantage of having @xmath45 small . by the same token , a stronger forcing , _
i.e. _ , larger @xmath46 , helps .
the role of @xmath47 , the _ shear _ , is explained pictorially in fig .
[ shear ] : since the function @xmath40 is required to be nonconstant , let us assume the kick drives some points on the limit cycle @xmath4 up and some down , as shown .
the fact that @xmath47 is positive means that points with larger @xmath48-coordinates move faster in the @xmath49-direction . during the relaxation period ,
the `` bumps '' created by the kick are stretched as depicted . at the same time , the curve is attracted back to the limit cycle .
thus , the combination of kicks and relaxation provides a natural mechanism for repeated stretching and folding of the limit cycle .
observe that the larger the differential in speed in the @xmath49-direction , _ i.e. _ the larger @xmath47 , and the slower the return to @xmath4 , _
i.e. _ the smaller @xmath45 , the more favorable the conditions are for this stretch - and - fold mechanism . in eq .
( [ thm2 ] ) , the quantities @xmath45 , @xmath47 and @xmath46 appear naturally .
but what about in general limit cycles , where the direction of the kicks vary ?
what , for example , will play the role of @xmath47 , or what we called shear in eq .
( [ thm2 ] ) ?
the aim of this subsection is to shed light on the general geometric picture , and to explain how the dynamics of @xmath15 for large @xmath8 can be understood .
let @xmath4 be a hyperbolic limit cycle as in the beginning of sect .
1.1 . through each @xmath50
passes the _ strong stable manifold _ of @xmath51 , denoted @xmath52 @xcite . by definition , @xmath53 as @xmath54 ; the distance between @xmath55 and @xmath56 in fact decreases exponentially .
some basic properties of strong stable manifolds are : ( i ) @xmath52 is a codimension one submanifold transversal to @xmath4 and meets @xmath4 at exactly one point , namely @xmath51 ; ( ii ) @xmath57 , and in particular , if the period of @xmath4 is @xmath58 , then @xmath59 ; and ( iii ) the collection @xmath60 foliates the basin of attraction of @xmath4 , that is to say , they partition the basin into hypersurfaces .
-foliation . shown
are the kicked image of a segment @xmath61 and two of its subsequent images under @xmath62 . ]
[ wss ] we examine next the action of the kick map @xmath63 in relation to @xmath64-manifolds . fig .
[ wss ] is analogous to fig .
[ shear ] ; it shows the image of a segment @xmath61 of @xmath4 under @xmath9 . for illustration purposes ,
we assume @xmath61 is kicked upward with its end points held fixed , and assume @xmath65 for some @xmath66 ( otherwise the picture is shifted to another part of @xmath4 but is qualitatively similar ) .
since @xmath62 leaves each @xmath64-manifold invariant , we may imagine that during relaxation , the flow `` slides '' each point of the curve @xmath67 back toward @xmath4 along @xmath64-leaves . in the situation
depicted , the effect of the folding is evident . fig . [ wss ] gives considerable insight into what types of kicks are conducive to the formation of strange attractors .
kicks along @xmath64-leaves or in directions roughly parallel to the @xmath64-leaves will not produce strange attractors , nor will kicks that essentially carry one @xmath64-leaf to another .
what causes the stretching and folding is the _ variation _ in how far points @xmath68 are moved by @xmath63 as measured in the direction transverse to the @xmath64-leaves . without attempting to give a more precise characterization , we will refer to the type of chaos that results from the geometry above as _ shear - induced chaos_. we emphasize that the occurrence of shear - induced chaos relies on the interplay between the geometries of the kicks and the underlying dynamical structures .
returning to the concrete situation of theorem 2 , since eq .
( [ thm2 ] ) without the kick term is linear , it is easy to compute strong stable manifolds . in @xmath69-coordinates , they are lines with slope @xmath70 .
variations in kick distances here are guaranteed by the fact that @xmath40 is nonconstant . with
@xmath40 fixed , it is clear that the larger @xmath71 and @xmath46 , the greater these variations .
note that the use of the word kick amplitude " in the statement of theorem 2 is a little misleading , for it is not the amplitude of the kicks _ per se _ that leads to the production of chaos . when @xmath73 , _ i.e.
_ when kicks are very infrequent , the map @xmath15 sends a small tube @xmath74 around @xmath4 back into itself .
this is an example of what is called a _
rank - one map _ in @xcite . roughly speaking ,
a rank - one map is a smooth map whose derivative at each point is strongly contractive in all but one of the directions .
rank - one maps can be analyzed using perturbative methods if they have well - defined `` singular limits . '' in the context of limit cycles , these singular limits do exist ; they are a one - parameter family of maps @xmath75 obtained by letting @xmath76 in the following way : for each @xmath77 ( recall that @xmath78 ) , let @xmath79 equivalently , @xmath80 is the unique point @xmath81 such that @xmath82 .
notice that @xmath83 , where we identify @xmath4 with @xmath84 $ ] ( with the end points identified ) . for eq .
( [ thm2 ] ) , @xmath85 is easily computed to be @xmath86 where the right side should again be interpreted as mod 1 .
( in the setting of driven oscillators , singular limits are sometimes known as `` phase resetting curves '' ; they have found widespread use in _
e.g. _ mathematical biology @xcite . )
it is shown in @xcite that a great deal of information on the attractor @xmath17 of @xmath15 for @xmath73 can be recovered from these singular limit maps .
the results are summarized below .
these results hold generally , but as we step through the 3 cases below , it is instructive to keep in mind eq .
( [ thm2 ] ) and its singular limit ( [ sing ] ) , with @xmath44 increasing as we go along : * if @xmath85 is injective , _
i.e. _ , it is a circle diffeomorphism , the attractor @xmath17 for @xmath15 is an invariant circle .
this happens when the kicks are aimed in directions that are unproductive " ( see above ) , or when their effects are damped out quickly . in this case , the competing scenarios on @xmath17 are quasi - periodicity and `` sinks , '' _
i.e. _ the largest lyapunov exponent of @xmath15 is zero or negative . * when @xmath85 loses its injectivity , the invariant circle is `` broken '' . when that first happens , the expansion of the 1-d map @xmath85 is weak , and all but a finite number of trajectories tend to sinks .
this translates into a gradient type dynamics for @xmath15 . *
if @xmath85 is sufficiently expanding away from its critical points , @xmath17 contains horseshoes for all large @xmath8 . for an open set of these @xmath8 , the chaos is transient , while on a positive measure set
, @xmath15 has a strange attractor with the properties described in sect .
these are the two known competing scenarios .
( they may not account for all @xmath8 . ) since @xmath87 for large @xmath8 , both sets of parameters are roughly periodic .
+ the analyses in the works cited suggest that when horseshoes are first formed , the set of parameters with transient chaos is more dominant .
the stronger the expansion of @xmath85 , the larger the set of parameters with strange attractors . in the first case ,
the largest lyapunov exponent of @xmath15 may appear positive for some time ( which can be arbitrarily long ) before turning negative .
in the second case , it stays positive indefinitely .
in hyperbolic theory , there is , at the present time , a very large discrepancy between what is thought to be true and what can be proved .
maps that are dominated by stretch - and - fold behavior are generally thought to have positive lyapunov exponents
although this reasoning is also known to come with the following caveat : maps whose derivatives expand in certain directions tend to contract in other directions , and unless the expanding and contracting directions are well separated ( such as in anosov systems ) , the contractive directions can conspire to form sinks .
this is how the transient chaos described in sect .
1.2 comes about .
still , if the expansion is sufficiently strong , one would expect that positive lyapunov exponents are more likely to prevail even though for any one map the outcome can go either way .
_ proving _ results of this type is a different matter .
few rigorous results exist for systems for which one has no _ a priori _ knowledge of invariant cones , and invariant cones are unlikely in shear - induced chaos .
the rigorous results reviewed in the last two subsections have the following limitations : ( i ) they pertain to @xmath15 for only very large @xmath8 .
this is because the authors use a perturbative theory that leans heavily on the theory of 1-d maps .
no non - perturbative analytic tools are currently available .
( ii ) a larger than necessary amount of expansion is required of the singular limit maps @xmath85 in the proof of strange attractors .
this has to do with the difficulty in locating suitable parameters called misiurewicz points from which to perturb .
( this problem can be taken care of , however , by introducing more parameters . )
we point out that ( i ) and ( ii ) together exacerbate the problem : @xmath85 is more expanding when @xmath45 is small , but if @xmath88 is to be near its singular limit , then @xmath89 must be very small , _
i.e. _ @xmath90 must be very large . that brings us to the present paper , the purpose of which is to supply numerical evidence to support some of our conjectured ideas regarding situations beyond the reach of the rigorous work reviewed .
our ideas are based on the geometry outlined in sect .
[ sec1.2 ] , but are not limited to periodic kicks or to the folding of limit cycles .
our first model is the periodic kicking of a linear shear flow with a hyperbolic limit cycle .
the setting is as in theorem 2 with @xmath91 , _ i.e. _ , we consider @xmath92 where @xmath36 , @xmath93 $ ] with the two end points of @xmath84 $ ] identified . in the absence of kicks , _
i.e. _ , when @xmath94 , @xmath95 tends to the limit cycle @xmath96 for all @xmath97 .
as before , the attractor in the kicked system is denoted by @xmath17 .
the parameters of interest are : @xmath98 our aim here is to demonstrate that the set of parameters with chaotic behavior is considerably larger than what is guaranteed by the rigorous results reviewed in sect .
[ wang - young ] , and to gain some insight into this parameter set . by `` chaotic behavior , '' we refer in this section to the property that @xmath15 has a positive lyapunov exponent for orbits starting from a `` large '' set of initial conditions , _
i.e. _ a set of full or nearly full lebesgue measure in the basin of attraction of @xmath17 .
more precisely , we _ assume _ that such lyapunov exponents are well defined , and proceed to compute the largest one , which we call @xmath99 .
we begin with some considerations relevant to the search for parameters with @xmath100 : * it is prudent , in general , to ensure that orbits do not stray too far from @xmath4 .
this is because while the basin of attraction of @xmath4 in this model is the entire phase space , the basin is bounded in many other situations .
we therefore try to keep @xmath101 with relatively small @xmath102 .
this is guaranteed if @xmath46 is small enough that @xmath103 ; the bound is improved if , for example , no point gets kicked to maximum amplitude two consecutive iterates .
* let @xmath104 .
a simple computation gives @xmath105 \cdot ( 1 - e^{-\lambda t})\qquad\mbox{(mod 1)}\ , \\
y_t & = & e^{-\lambda t } [ y_0 + a \sin(2\pi\theta_0)]\ . \end{array}\ ] ] for @xmath102 relatively small , we expect the number @xmath106 to be a good indicator of chaotic behavior : if it is large enough , then @xmath15 folds the annulus @xmath107 with two turns and maps it into itself .
the larger this number , the larger the folds , meaning the more each of the monotonic parts of the image wraps around in the @xmath49-direction .
[ [ summary - of - findings . ] ] summary of findings .
+ + + + + + + + + + + + + + + + + + + + _ _ * with the choice of parameters guided by ( a ) and ( b ) above , we find that as soon as the folding described in ( b ) is definite , @xmath15 becomes `` possibly chaotic '' , meaning @xmath99 is seen numerically to oscillate ( wildly ) between positive and negative values as @xmath8 varies .
we interpret this to be due to competition between transient and sustained chaos ; see ( iii ) in sect .
1.2 . for larger @xmath44 ,
_ i.e. _ , as the stretching is stronger , and for @xmath8 beyond an initial range , this oscillation stops and @xmath99 becomes definitively positive for all the values of @xmath8 computed . * as for the range of parameters with chaotic dynamics , we find that @xmath108 occurs under fairly modest conditions , _
e.g. _ , for @xmath109 , we find @xmath108 starting from about @xmath110 , which is very far from the `` @xmath111 '' in rigorous proofs . also , while shear - induced chaos is often associated with weak damping , we find that the phenomenon occurs as well for larger @xmath45 , _
e.g. _ , for @xmath112 , provided its relation to the other parameters are favorable .
_ supporting numerical evidence .
_ figures [ fig : kick - cycle ] and [ fig : kick - cycle - too ] show the largest lyapunov exponent @xmath99 of @xmath15 versus the kick period @xmath8 .
( note that this is the expansion rate per kick period and is @xmath8 times the rate per unit time . ) in fig .
[ fig : kick - cycle ] , @xmath45 and @xmath46 are fixed , and @xmath47 is increased .
we purposefully start with too small a @xmath47 so that we may see clearly the gradual changes in @xmath99 .
the results are in excellent agreement with the description at the end of sect .
[ sec1.2 ] ( which pertains to regimes with very large @xmath8 ) , even though @xmath8 is not so large here : in the top picture , where @xmath44 is small , the plot confirms a competition between quasi - periodicity and sinks ; in the middle picture , we see first @xmath99 becoming increasingly negative , then transitions into a competition between transient and sustained chaos , with the latter dominating in the bottom picture . fig . [
fig : kick - cycle - too ] shows the same phenomena in reverse order , with @xmath47 and @xmath46 fixed and @xmath45 increasing .
notice that even for @xmath113 and @xmath46 leading to chaotic dynamics , @xmath99 is negative for small @xmath8 .
this is in agreement with the influence of the factor @xmath114 in eq .
( [ time - t map ] ) . as explained in ( a ) above ,
when @xmath90 is too small relative to @xmath46 , orbits stray farther from @xmath4 .
data points corresponding to parameters for which this happens are marked by open squares . for purposes of demonstrating the phenomena in question ,
there is nothing wrong with these data points , but as explained earlier , caution must be exercised with these parameters in systems where the basin of @xmath4 is smaller .
+ + + [ fig : kick - cycle ] + + + [ fig : kick - cycle - too ] _ simulation details . _
the numbers @xmath99 are computed by iterating the map in eq .
( [ time - t map ] ) and its jacobian , and tracking the rate of growth of a tangent vector .
we use @xmath115 iterates of @xmath15 in each run .
mindful of the delicate situation due to competition between transient and sustained chaos , and to lower the possibility of atypical initial conditions , we perform 10 runs for each choice of @xmath116 , using for each run an independent , random ( with uniform distribution ) initial condition @xmath117 $ ] . among the 10 values of
@xmath118 computed , we discard the largest and the smallest , and plot the maximum and minimum of the remaining 8 . as one
can see in figs .
[ fig : kick - cycle ] and [ fig : kick - cycle - too ] , the two estimates occasionally do not agree .
this may be because not all initial conditions in the system have identical lyapunov exponents , or it may be that the convergence to the true value of @xmath118 is sufficiently slow and more iterates are needed , _
i.e. _ there are long transients .
we consider next a variant of eq .
( [ model1 ] ) in which deterministic , periodic kicks are replaced by `` random kicks . '' here , random kicks refer to kicks at random times and with random amplitudes .
more precisely , we consider @xmath119 where the kick times @xmath120 are such that @xmath121 , @xmath122 are independent exponential random variables with mean @xmath8 , and the kick amplitudes @xmath123 are independent and uniformly distributed over the interval @xmath124 $ ] for some @xmath125 .
( we do not believe detailed properties of the laws of @xmath126 and @xmath127 have a significant impact on the phenomena being addressed . )
the analog here of the time-@xmath8 map in study 1 is the _ random map _
@xmath128 where @xmath126 and @xmath127 are random variables . by the standard theory of random maps , lyapunov exponents with respect to stationary measures
are well defined and are nonrandom , _
i.e. _ they do not depend on the sample path taken @xcite .
notice that if @xmath129 , the system ( [ poisson ] ) has a unique stationary measure which is absolutely continuous with respect to lebesgue measure on @xmath130 : starting from almost every @xmath131 , after one kick , the distribution acquires a density in the @xmath48-direction ; since vertical lines become slanted under @xmath1 due to @xmath129 , after a second kick the distribution acquires a ( two - dimensional ) density . in terms of overall trends , our assessment of the likelihood of chaotic behavior follows the analysis in study 1 and will not be repeated .
we identify the following two important differences : * _ smooth dependence on parameters .
_ due to the averaging effects of randomness , we expect lyapunov exponents to vary smoothly with parameter , without the wild oscillations in the deterministic case .
* _ effects of large deviations . _
a large number of kicks occurring in quick succession may have the following effects : * * they can cause some orbits to stray far away from @xmath132 .
this is guaranteed to happen , though infrequently , in the long run .
thus , it is reasonable to require only that a large fraction not all of the stationary measure ( or perhaps of the _ random attractors _
@xmath133 ) to lie in a prescribed neighborhood of @xmath4 . * * it appears possible , in principle , for a rapid burst of kicks to lead to chaotic behavior even in situations where the shear is mild and kick amplitudes are small . to picture this ,
imagine a sequence of kicks sending ( or maintaining ) a segment far from @xmath4 , allowing the shear to act on it for an uncharacteristically long time .
one can also think of such bursts as effectively setting @xmath45 to near @xmath134 temporarily , creating a very large @xmath135 . on the other hand ,
if @xmath47 is small , then other forces in the system may try to coax the system to form sinks between these infrequent events .
we do not have the means to assess which scenario will prevail .
[ [ summary - of - findings.-1 ] ] summary of findings .
+ + + + + + + + + + + + + + + + + + + + _ in terms of overall trends , the results are consistent with those in study 1 .
two differences are observed .
one is the rapid convergence of @xmath99 and their smooth dependence on parameters .
the other is that positive lyapunov exponents for @xmath136 are found both for smaller values of @xmath44 and for apparently very small @xmath8 ( which is impossible for periodic kicks ) , lending credence to the scenario described in ( b)(ii ) above . _
+ ( a ) _ increasing shear _
+ + ( b ) _ increasing damping _
+ lyapunov exponents for the linear shear flow with poisson kicks .
squares indicate the corresponding orbit spends more than 20% of the time in the region @xmath137 .
[ fig : poisson ] _ supporting numerical evidence . _ fig .
[ fig : poisson ] shows @xmath99 as a function of the mean kick interval @xmath8 . as in study 1
, we first show the effects of increasing @xmath47 and then the effects of increasing @xmath45 . without the oscillations seen previously , the present plots are straightforward to interpret . in case one
wonders how @xmath118 curves can switch from strictly - decreasing to strictly - increasing behavior , the middle panel of fig .
[ fig : poisson](b ) catches such a swtich `` in the act . ''
squares indicate that the orbit computed spends @xmath138 of its time outside of the region @xmath139 .
in this section , we investigate the effect of forcing by white noise .
the resulting systems are described by stochastic differential equations ( sdes ) .
we consider two ways to force the system : * study 3a : degenerate white noise applied in chosen direction : * @xmath140 * study 3b : isotropic white noise : * @xmath141 in study 3a , @xmath142 is standard 1-dimensional brownian motion ( meaning with variance @xmath143 ) . in study 3b , @xmath144 is a standard 2-d brownian motion , _
i.e. _ , they are independent standard 1-d brownian motions . for definiteness , we assume the stochastic terms are of it type .
notice that the two parameters @xmath46 and @xmath8 in studies 1 and 2 have been combined into one , namely @xmath145 , the coefficient of the brownian noise . by standard theory @xcite
, the solution process of an sde can be represented as a stochastic flow of diffemorphisms .
more precisely , if the coefficients of the sde are time - independent , then for any time step @xmath146 , the solution may be realized , sample path by sample path , as the composition of random diffeomorphisms @xmath147 , where the @xmath148 are chosen _ i.i.d .
_ with a law determined by the system ( the @xmath148 are time-@xmath149 flow - maps following this sample path ) .
this representation enables us to treat an sde as a _ random dynamical system _ and to use its lyapunov exponents as an indicator of chaotic behavior .
it is clear that system ( [ nondeg ] ) has a unique invariant density , which is the solution of the fokker - planck equation . even though the stochastic term in system ( [ deg ] ) is degenerate , for the same reasons discussed in study 3 , it too has a unique stationary measure , and this measure has a density .
the lyapunov exponents considered in this section are with respect to these stationary measures . before proceeding to an investigation of the two systems above
, we first comment on the case of purely additive noise , _
i.e. _ eq . ( [ nondeg ] ) without the @xmath150 factor in either brownian term . in this case
it is easy to see that all lyapunov exponents are @xmath151 , for the random maps are approximately time-@xmath149 maps of the unforced flow composed with random ( rigid ) translations .
such a system is clearly not chaotic . with regard to system ( [ deg ] ) , we believe that even though the quantitative estimates from study 1 no longer apply , a good part of the _ qualitative reasoning _ behind the arguments continues to be valid .
in particular , we conjecture that * trends , including qualitative dependences on @xmath47 and @xmath45 , are as in the previous two studies ; * the effects of large deviations noted for poisson kicks ( study 2 , item ( b ) ) are even more prominent here , given that the forcing now occurs continuously in time . as for system ( [ nondeg ] ) , we expect it to be less effective in producing chaos , _ i.e. _ more inclined to form sinks , than system ( [ deg ] ) .
this expectation is based on the following reasoning : suppose first that we force _ only _ in the @xmath49-direction , _ i.e. _ , suppose the @xmath152 term in ( [ nondeg ] ) is absent .
then the stochastic flow leaves invariant the circle @xmath153 , which is the limit cycle of the deterministic part of the system . a general theorem tells us that when a random dynamical system on a circle has an invariant density , its lyapunov exponent is always @xmath151 ; in this case , it is in fact strictly negative because of the inhomogeneity caused by the sine function @xcite .
thus the corresponding 2-d system has `` random sinks . ''
now let us put the @xmath48-component of the forcing back into the system .
we have seen from previous studies that forcing the @xmath48 direction alone may lead to chaotic behavior .
the tendency to form sinks due to forcing in the @xmath49-direction persists , however , and weakens the effect of the shear - induced stretching .
we now discuss the results of simulations performed to validate these ideas .
[ [ summary - of - findings.-2 ] ] summary of findings .
+ + + + + + + + + + + + + + + + + + + + _ _ * in the case of degenerate white noise , the qualitative dependence of @xmath99 on @xmath47 and @xmath45 are as expected , and the effects of large deviations are evident .
in particular , @xmath99 is positive for very small values of @xmath154 and @xmath145 provided @xmath155 is large .
this can not happen for periodic kicks ; we attribute it to the effect of large deviations .
* isotropic white noise is considerably less effective in producing chaos than forcing in the @xmath48-direction only , meaning it produces a smaller ( or more negative ) @xmath99 .
* in both cases , we discover the following approximate scaling : under the scaling transformations @xmath156 and @xmath157 , @xmath99 transforms approximately as @xmath158 . in the case of degenerate white noise , when both @xmath47 and @xmath159 are not too small ( _ e.g. _ , @xmath160 ) , this scaling gives excellent predictions of @xmath99 for the values computed .
we remark that ( iii ) does not follow by scaling time in the sde . indeed , scaling time by @xmath161 in eq .
( [ deg ] ) , we obtain @xmath162 thus the approximate scaling in ( iii ) asserts that the lyapunov exponent of system ( [ scaleddeg ] ) , equivalently @xmath161 times the @xmath99 for eq .
( [ deg ] ) , is roughly equal to that of the system obtained by changing the first equation in ( [ scaleddeg ] ) to @xmath163 . in other words , @xmath99 seems only to depend minimally on the frequency of the limit cycle in the unforced system .
_ supporting numerical evidence .
_ plots of @xmath99 as functions of @xmath145 are shown in figs .
[ fig : whitenoise ] [ fig : scaling ] .
+ [ fig : whitenoise ] [ fig : small sigma ] + ( a ) @xmath164 + + ( b ) @xmath165 [ fig : isotropic ] [ fig : scaling ] in fig .
[ fig : whitenoise ] , the forcing is degenerate , and for fixed @xmath47 , @xmath118 decreases with increasing damping as expected .
notice that compared to the two previous studies , a somewhat larger damping is required to maintain a good fraction of the attractor near @xmath4 .
[ fig : small sigma ] shows that @xmath99 is positive for values of @xmath47 and @xmath45 as small as @xmath166 and @xmath167 , and white noise amplitudes @xmath145 close to @xmath134 .
notice first that this is consistent with the scaling conjectured in ( iii ) above , and second that in the case of periodic kicks , comparable values of @xmath47 and @xmath45 would require a fairly substantial kick , not to mention long relaxation periods , before chaotic behavior can be produced .
we regard this as convincing evidence of the significant effects of large deviations in continuous - time forcing .
( it must be pointed out , however , that in our system , the basin of attraction is the entire phase space , and a great deal of stretching is created when @xmath168 is large .
that means system ( [ deg ] ) takes greater advantage of large deviations than can be expected ordinarily .
[ fig : isotropic](a ) shows @xmath99 in the isotropic case for the same parameters as in fig .
[ fig : whitenoise ] .
a comparison of the two sets of results confirms the conjectured tendency toward negative exponents when the forcing is isotropic .
[ fig : isotropic](b ) shows that this tendency can be overcome by increasing @xmath47 .
[ fig : scaling ] shows four sets of results , overlaid on one another , demonstrating the scaling discussed in item ( iii ) above .
fixing @xmath169 , we show the graphs of @xmath170 as functions of @xmath171 for four values of @xmath47 .
the top two curves ( corresponding to @xmath172 and @xmath173 ) coincide nearly perfectly .
similar approximate scalings , less exact , are observed for smaller values of @xmath155 , both when @xmath99 is positive and negative . _
simulation details .
_ we compute lyapunov exponents numerically by solving the corresponding variational equations ( using an euler solver with time steps of @xmath174 ) and tracking the growth rate of a tangent vector . to account for the impact of the realization of the forcing on the computed exponents , for each choice of @xmath175 we perform 12 runs in total , using 3 independent realizations of the forcing and , for each realization , 4 independent initial conditions ( again uniformly - distributed in @xmath176 $ ] ) . for almost all the parameter values
, the estimates agree to fairly high accuracy , so we simply average over initial conditions and plot the result .
[ [ related - results . ] ] related results .
+ + + + + + + + + + + + + + + + the asymptotic stability of dynamical systems driven by random forcing has been investigated by many authors using both numerical and analytic methods . particularly relevant to our study are results pertaining to the random forcing of oscillators ( such as duffing - van der pol oscillators ) and stochastic hopf bifurcations ; see e.g. @xcite .
most of the existing results are perturbative , _
i.e. _ , they treat regimes in which both the noise and the damping are very small .
positive lyapunov exponents are found under certain conditions .
we do not know at this point if the geometric ideas of this paper provide explanations for these results .
in this section , we will show that external forcing can lead to shear - induced chaos in a coupled phase oscillator system of the form the governing equations are @xmath177\ , \\
\dot{\theta}_2 & = \nu_2 + z(\theta_2 ) [ { a_{\rm fb}}g(\theta_1)].\nonumber\end{aligned}\ ] ] the state of the system is specified by two angles , @xmath178 , so that the phase space is the torus @xmath179 .
the constants @xmath180 and @xmath181 are the oscillators intrinsic frequencies ; we set @xmath182 and @xmath183 ( representing similar but not identical frequencies ) .
the constants @xmath184 and @xmath185 govern the strengths of the feedforward and feedback couplings .
the oscillators are pulse - coupled : the coupling is mediated by a bump function @xmath186 supported on @xmath187 $ ] and normalized so that @xmath188 .
the function @xmath189 , which we take to be @xmath190 $ ] , specifies the sensitivity of the oscillators to perturbations when in phase @xmath49 .
finally , we drive the system with an external forcing @xmath191 , which is applied to only the first oscillator .
this simple model arises from neuroscience @xcite and is examined in more detail in @xcite .
let @xmath1 denote the flow of the unforced system , _
i.e. _ , with @xmath192 .
flowlines are roughly northeasterly and are linear except in the strips @xmath193 and @xmath194 , where they are bent according to the prescribed values of @xmath184 and @xmath185 .
let @xmath195 denote the rotation number of the first return map of @xmath1 to the cross - section @xmath196 .
it is shown in @xcite that for @xmath197 , @xmath195 is monotonically increasing ( constant on extremely short intervals ) as one increases @xmath185 , until it reaches @xmath198 at @xmath199 , after which it remains constant on a large interval . at @xmath200 ,
a limit cycle emerges in which each oscillator completes one rotation per period ; we say the system is 1:1 phase - locked , or simply _ phase - locked_. in @xcite , it is shown numerically that forcing the system by white noise after the onset of phase - locking leads to @xmath100 .
the authors of @xcite further cite wang - young theory ( the material reviewed in sect .
[ wang - young ] ) as a geometric explanation for this phenomenon . in this section ,
we provide geometric and numerical evidence of shear - induced chaos both before and after the onset of phase - locking at @xmath201 . our results for @xmath202 support the assertions in @xcite .
for @xmath203 , they will show that _ limit cycles are not preconditions for shear - induced chaos_. we will show that in eq .
( [ eq:2osc ] ) , the mechanism for folding is already in place before the onset of phase - locking , where the system is quasi - periodic or has periodic orbits of very long periods ; the distinction between these two situations is immaterial since we are concerned primarily with finite - time dynamics . in the rest of this section , we will , for simplicity , refer to the regime prior to the onset of phase - locking as `` near - periodic . ''
the dynamical picture of kicks followed by a period of relaxation has a simpler , more clear - cut geometry than that of continuous , random forcing .
thus we use the former to demonstrate why one may expect chaotic behavior over the parameter ranges in question .
the kick map is denoted by @xmath63 as in section 1 .
_ folding in the periodic ( _ i.e. _ phase - locked ) regime .
_ we will use @xmath204 for illustration purposes ; similar behavior is observed over a range of @xmath185 from @xmath205 to @xmath206 .
note that the system is phase - locked for a considerably larger interval beyond @xmath207 , but the strength of attraction grows with increasing @xmath185 , and when the attraction becomes too strong , it is harder for folding to occur . ) in the phase - locked regime . _
blue : _ a lift of the limit cycle .
_ red : _ the image of the cycle after a single kick . _
green : _ strong - stable foliation . here , the parameters are @xmath182 , @xmath183 , @xmath197 , and @xmath204 . ] [ cycle+wss ] -images of @xmath61 , the part of the limit cycle between @xmath208 and @xmath209 . _
red curves : _ @xmath67 and its images .
_ purple dot : _ the point on @xmath61 which does not move under @xmath63 .
the parameters are the same as in fig .
[ cycle+wss].,title="fig : " ] + [ cycle snapshots ] fig .
[ cycle+wss ] shows the limit cycle @xmath4 ( _ blue curve _ ) of the unforced system at @xmath204 ; more precisely , it shows a `` lift '' of @xmath4 to @xmath210 , identifying the torus @xmath211 with @xmath212 .
also shown is the image @xmath213 of the cycle after a single kick ( _ red curve _ ) , where the kick map @xmath63 corresponds to @xmath214 with @xmath215 , _
i.e. _ @xmath63 is given by @xmath216 where @xmath217 is the solution of @xmath218 .
notice the special form of the kicks : @xmath63 acts horizontally , and does not move points on @xmath219 . in particular
, @xmath63 fixes a unique point @xmath220 on the cycle ; this point is , in fact , not affected by _ any _ kick of the form considered in eq .
( [ eq:2osc ] ) .
several segments of strong stable manifolds ( _ green curves _ ) of the unforced system are drawn . recall that if @xmath58 is the period of cycle and @xmath66 , then @xmath221 lies on the @xmath64-curve through @xmath222 and is pulled toward the cycle as @xmath28 increases ( see sect . [ sec1.2 ] ) . from the relation between the @xmath64-curves and the cycle
, we see that for @xmath223 , @xmath224 will lag behind @xmath95 during the relaxation period .
notice in particular that there are points on @xmath213 above the line @xmath225 that are pulled toward the part of @xmath4 below @xmath225 . since @xmath220 stays put , we deduce that some degree of folding will occur if the time interval between kicks is sufficiently long .
[ cycle snapshots ] illustrates how this folding happens through three snapshots .
we begin with a segment @xmath226 between @xmath208 and @xmath209 ( _ blue curve _ ) and its image after a single kick ( _ red curve _ ) .
both curves are then evolved forward in time and their images at @xmath227 and @xmath228 are shown .
the purple dot marks the point on @xmath61 which does not move when kicked .
notice that these pictures are shown in a _ moving frame _ to emphasize the geometry of @xmath229 relative to @xmath230 . ) in a near - periodic regime . _ blue curves : _ a segment @xmath61 of an orbit and its forward images @xmath230 at @xmath228 , @xmath231 .
_ red curves : _ @xmath67 and its forward images .
_ purple dot : _ the point on @xmath61 which does not move under @xmath63 .
the parameters are @xmath182 , @xmath183 , @xmath197 , and @xmath232.,title="fig : " ] + [ quasiper snapshots ] [ finite - time - wss ] _ folding in the near - periodic regime . _
[ quasiper snapshots ] shows snapshots of a similar kind for @xmath232 ; this value of @xmath185 puts the system in the near - periodic regime .
the snapshots begin with an ( arbitrary ) orbit segment @xmath61 and its image @xmath67 ; the location of @xmath61 is near that of the limit cycle in fig .
[ cycle+wss ] .
the kicked segment clearly folds ; indeed , the picture is qualitatively very similar to that of the limit cycle case .
note that at @xmath232 , the rotation number of the return map to @xmath233 is a little below 1 , so that @xmath229 has an overall , slow drift to the left when viewed in the fixed frame @xmath234 .
this slow , left - ward drift is not especially relevant in our moving frame ( which focuses on the movement of @xmath229 relative to that of @xmath230 ) . on successive laps around the torus , the orbit in question returns to the part of the torus shown in the figure , and the sequence of actions depicted in fig .
[ quasiper snapshots ] is repeated .
we regard this as geometric evidence of shear - induced chaos .
we have seen that in the phase - locked regime , the folding of the limit cycle ( when the time interval between kicks is sufficiently large ) can be deduced from the geometry of the strong stable foliation .
a natural question is : in the quasi - periodic regime , are there geometric clues in the unforced dynamics that will tell us whether the system is predisposed to chaotic behavior when forced ?
since folding occurs in finite time , we believe the answer lies partially in what we call _ finite - time stable manifolds _ , a picture of which is shown in fig .
[ finite - time - wss ] .
we first explain what these manifolds are before discussing what they can and can not tell us .
fix @xmath235 . at each @xmath236 , let @xmath237 be the most contracted direction of the linear map @xmath238 if it is uniquely defined , _ i.e. _ if @xmath239 is a unit tangent vector at @xmath240 in the direction @xmath237 , then latexmath:[$|d\phi_t(z)v| \leq @xmath240 .
a smooth curve is called a _
time-@xmath242 stable manifold _ if it is tangent to @xmath243 at all points ; these curves together form the _
time-@xmath242 stable foliation_. in general , time-@xmath242 stable manifolds are not necessarily defined everywhere ; they vary with @xmath242 , and may not stabilize as @xmath242 increases .
when `` real '' ( _ i.e. _ infinite - time ) stable manifolds exist , time-@xmath242 stable manifolds converge to them as @xmath244 .
the blue curve in fig . [ finite - time - wss ] is an orbit segment of @xmath1 . the angles between this segment and the time-@xmath231 stable manifolds ( _ green curves _
) reflect the presence of shear .
for example , if a kick sends points on the blue curve to the right , then within @xmath231 units of time most points on the kicked segment will lag behind their counterparts on the original orbit segment except for the point with @xmath219 at the time of the kick .
pinching certain points on an orbit segment while having the rest slide back potentially creates a scenario akin to that in fig .
2 ; see sect . [ sec1.2 ] .
one is also likely to find shear along the black curve in fig .
[ finite - time - wss ] , a second orbit segment of @xmath1 . whether or not the shear here is strong enough to cause the formation of folds in 5 units of time can not be determined from the foliation alone ; more detailed information such as contraction rates are needed . what fig .
[ finite - time - wss ] tells us are the mechanism and the shapes of the folds if they _
do _ form .
notice also that shearing occurs in opposite directions along the blue and black segments .
this brings us to a complication not present previously : each orbit of @xmath1 spends only a finite amount of time near , say , the blue curve before switching to the region near the black curve , and when it does so , it also switches the direction of shear .
finite - time stable foliations for system ( [ eq:2osc ] ) have also been computed for @xmath245 and a sample of @xmath246 ( not shown ) .
they are qualitatively similar to fig .
[ finite - time - wss ] , with most of the leaves running in a northeasterly direction . in summary , for @xmath242
not too large , time-@xmath242 stable foliations generally do not change quickly with @xmath242 or with system parameters .
they are good indicators of shear , but do not tell us if there is _ enough _ shear for folds to form . for the system defined by ( [ eq:2osc ] ) ,
given that the finite - time stable manifolds are nearly parallel to flowlines and the kick map acts unevenly with respect to this foliation , we conclude the presence of shear .
[ quasiper snapshots ] and similar figures for other @xmath185 ( not shown ) confirm that folding does indeed occur when the system is forced in the near - periodic regime . to provide quantitative evidence of shear - induced chaos in the situations discussed above , we compute @xmath99 .
recall that while periodic kicks followed by long relaxations provide a simple setting to visualize folding , it is not expected to give clean results for @xmath99 because of the competition between transient and sustained chaos ( see sect . [ sec1.2 ] ) .
continuous - time random forcing , on the other hand , produces numerical results that are much easier to interpret .
[ [ study-4a - stochastic - forcing . ] ] study 4a : stochastic forcing .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + we consider system ( [ eq:2osc ] ) with @xmath197 and @xmath247 $ ] .
the forcing is of the form @xmath248 where @xmath142 is standard brownian motion . [
[ study-4b - periodic - kicks . ] ] study 4b : periodic kicks .
+ + + + + + + + + + + + + + + + + + + + + + + + + the equation and parameters are as above , and the forcing is given by @xmath249 .
[ cols="^,^ " , ] [ osc lyaps periodic kicks ] [ [ summary - of - findings.-3 ] ] summary of findings .
+ + + + + + + + + + + + + + + + + + + + _ positive @xmath99 are found for stochastic forcing in the parameter interval studied , both before and after the onset of phase - locking at @xmath201 . for periodic kicks with large enough @xmath46 and @xmath8
, it appears that @xmath99 is positive for a fraction of the forcing periods tested , but the results are hard to interpret due to the competition between transient and sustained chaos .
_ _ supporting numerical evidence .
[ osc lyaps white noise ] shows some results for stochastic forcing .
for @xmath204 , negative lyapunov exponents are found for very small amplitudes of forcing , while slightly stronger forcing ( _ e.g. _ @xmath250 ) is needed before @xmath108 can be concluded with confidence .
in contrast , even fairly small values of forcing seem to lead to @xmath251 when @xmath232 , _ i.e. _ in the near - periodic regime .
this may be explained by the damping in the limit cycle case , especially for larger @xmath185 .
notice also that in this model large amplitudes of forcing do not lead to larger @xmath99 .
this is due to the fact that unlike the system in studies 13 , a very strong forcing merely presses most of the phase space against the circle @xmath219 , which is not very productive from the point of view of folding phase space .
[ osc lyaps periodic kicks ] shows plots of @xmath99 for periodic kicks . here
, roughly 40% of the kick periods @xmath8 for which lyapunov exponents were computed yield a positive exponent .
more generally , we find that @xmath251 for over 40% of kick intervals @xmath8 as @xmath46 varies over the range @xmath252 $ ] .
see simulation details in study 1 .
shear - induced chaos , by which we refer to the phenomenon of an external force interacting with the shearing in a system to produce stretches and folds , is found to occur for wide ranges of parameters in forced oscillators and quasi - periodic systems .
highlights of our results include : * for periodically kicked oscillators , positive lyapunov exponents are observed under quite modest impositions on the unforced system and on the relaxation time between kicks ( in contrast to existing rigorous results ) .
these regimes are , as expected , interspersed with those of transient chaos in parameter space . *
continuous - time stochastic forcing is shown to be equally effective in producing chaos .
the qualitative dependence on parameters is similar to that in deterministic forcing .
we find that suitably directed , degenerate white noise is considerably more effective than isotropic white noise ( and additive noise will not work ) .
we have also found evidence for an approximate scaling law relating @xmath118 to @xmath47 , @xmath45 , and @xmath145 .
other types of random forcing such as poisson kicks are also studied and found to produce chaos .
* the shear - induced stretching - and - folding mechanism can operate as well in quasi - periodic systems as it does in periodic systems , _
i.e. _ limit cycles are not a precondition for shear - induced chaos .
we demonstrate this through a pulse - coupled 2-oscillator system .
chaos is induced under both periodic and white noise forcing , and a geometric explanation in terms of finite - time stable manifolds is proposed .
the conclusions in ( i ) and ( ii ) above are based on systematic numerical studies of a linear shear flow model . as this model captures the essential features of typical oscillators
, we expect that our conclusions are valid for a wide range of other models . our numerical results , particularly those on stochastic forcing , point clearly to the possibility of a number of ( rigorous ) theorems . | guided by a geometric understanding developed in earlier works of wang and young , we carry out some numerical studies of shear - induced chaos . the settings considered include periodic kicking of limit cycles , random kicks at poisson times , and continuous - time driving by white noise .
the forcing of a quasi - periodic model describing two coupled oscillators is also investigated . in all cases ,
positive lyapunov exponents are found in suitable parameter ranges when the forcing is suitably directed . |
it is believed that at high temperatures and densities there should be phase transition(s ) between confined nuclear matter and the quark - gluon plasma ( qgp ) , where quarks and gluons are no longer confined inside hadron bags @xcite .
various theoretical studies have been devoted to tackle the possible change in properties of the strongly interacting matter , when the phase transition(s ) between hadronic and partonic phases takes place under the effect of an external magnetic field @xcite .
it is conjectured that the strongly interacting system ( hadronic or partonic ) can response to the external magnetic field with magnetization , @xmath5 , and magnetic susceptibility , @xmath6 @xcite .
both quantities characterize the magnetic properties of the system of interest .
thus , the effects of the external magnetic field on the chiral condensates should be reflected in the chiral phase - transition @xcite .
also , the effects on the deconfinement order - parameter ( polyakov - loop ) which includes the confinement - deconfinement phase - transition can be studied @xcite . in an external magnetic field
, the hadronic and partonic states are investigated in different models , such as the hadron resonance gas ( hrg ) model @xcite , and other effective models @xcite . the nambu - jeno - lasinio ( njl ) model @xcite , the chiral perturbation theory @xcite , the quark model @xcite and certain limits of qcd @xcite are also implemented
. furthermore , there are some studies devoted to the magnetic effects on the dynamical quark masses @xcite .
the chiral magnetic - effect was studied in context of the polyakov njl ( pnjl ) model @xcite .
recently , it was reported about lattice qcd calculations in an external magnetic field @xcite .
the polyakov linear @xmath0-model ( plsm ) was implemented to estimate the effects of the magnetic field on the system @xcite . in the present work ,
we add some restrictions to the quarks energy due to the existence of free charges in the plasma phase . to this end
, we apply the landau theory ( landau quantization ) @xcite , which quantizes of the cyclotron orbits of the charged particles in the magnetic fields .
we notice that this proposed configuration requires an additional temperature to drive the system through the chiral phase - transition .
accordingly , we find that the value of the chiral condensates increase with increasing the external magnetic field @xcite .
a few remarks are now in order . in many different calculations for the thermal behavior of the chiral condensates and the deconfinement order - parameter ( polyakov - loop ) using pnjl or njl @xcite ,
the external magnetic field was not constant .
also , the dependence of the critical temperatures of chiral and confinement phase - transitions on the magnetic field was analysed @xcite .
almost the same study was conducted in plsm @xcite .
all these studies lead to almost the same pattern , the critical temperature of chiral phase - transition increases with increasing the external magnetic field .
but , the critical temperature of the confinement phase - transition behaves , oppositely .
the latter behavior agrees - to some extend - with the lattice qcd calculations @xcite . in the present work ,
we study the effects of external magnetic field on the phase transition and deduce the phase - diagram curve using su(3 ) plsm @xcite . in light of this , we recall that the plsm is widely implemented in different frameworks and different purposes .
the lsm was introduced by gell - mann and levy in 1960 @xcite long time before qcd was known to be the theory of strong interaction .
many studies have been performed with lsm like @xmath7 lsm @xcite , @xmath7 lsm at finite temperature @xcite and @xmath8 lsm for @xmath9 , @xmath10 or even @xmath11 quark flavors @xcite . in order to obtain reliable results ,
polyakov - loop corrections have been added to lsm , in which information about the confining glue sector of the theory was included in form of polyakov - loop potential .
this potential is to be extracted from the pure yang - mills lattice simulations @xcite .
so far , many studies were devoted to investigating the phase diagram and the thermodynamics of plsm at different polyakov - loop forms with two @xcite and three quark flavors @xcite .
also , the magnetic field effect on the qcd phase - transition and other system properties are investigated using plsm @xcite .
the present paper is organized as follows . in section [ sec :
approaches ] , we introduce details about su(3 ) plsm under the effects of an external magnetic field .
section [ sec : results ] gives some features of the plsm in an external magnetic field , such as the quark condensates , polyakov loop , some thermal quantities , the phase - transition(s ) and scalar and pseudoscalar meson masses under the magnetic field effect . in section [ sec : conclusion ] , the final conclusions and outlook shall be presented .
the lagrangian of lsm with @xmath12 quark flavors and @xmath13 color degrees of freedom , where the quarks couple to the polyakov - loop dynamics , was introduced in ref .
@xcite , @xmath14 where the chiral part of the lagrangian @xmath15 has @xmath16 symmetry @xcite .
the lagrangian with @xmath12 consists of two parts .
the first part represents fermions , eq .
( [ lfermion ] ) with a flavor - blind yukawa coupling @xmath17 of the quarks .
the coupling between the effective gluon field and quarks , and between the magnetic field , @xmath18 , and the quarks is implemented through the covariant derivative @xcite @xmath19 where the summation @xmath20 runs over the three flavors , @xmath21 for @xmath22- , @xmath23- and @xmath24-quark , respectively , @xmath25 is the gell - man matrices .
the flavor - blind yukawa coupling , @xmath17 , should couple the quarks to the mesons @xcite .
the coupling of the quarks to the euclidean gauge field , @xmath26 , was discussed in ref @xcite . for the abelian gauge field , the influence of the external magnetic field , @xmath27 ,
@xcite is given by the covariant derivative @xcite , @xmath28 where @xmath29 and @xmath30 and @xmath31 is a matrix defined by the quark electric charges @xmath32 for up , down and strange quarks , respectively .
the interaction of charged pion @xmath33 with the magnetic field is included by @xmath34 with @xmath35 is the electric charge @xcite . the second part of chiral lagrangian stands for the the mesonic contribution , eq .
( [ lmeson ] ) , @xmath36 ^ 2 \nonumber\\ & - & \lambda_2 \mathrm{tr}(\phi^{\dag } \phi)^2+c[\mathrm{det}(\phi)+\mathrm{det}(\phi^{\dag } ) ] + \mathrm{tr}[h(\phi+\phi^{\dag } ) ] .
\label{lmeson}\end{aligned}\ ] ] in eq .
( [ lmeson ] ) , @xmath37 is a complex @xmath38 matrix , which depends on the @xmath39 and @xmath40 @xcite , where @xmath41 are the chiral spinors , @xmath39 are the scalar mesons and @xmath40 are the pseudoscalar mesons . the second term in eq .
( [ plsm ] ) , @xmath42 , represents the polyakov - loop effective potential @xcite , which is expressed by using the dynamics of the thermal expectation value of a color traced wilson loop in the temporal direction @xmath43 .
then , the polyakov - loop potential and its conjugate read @xmath44 and @xmath45 , respectively .
@xmath46 , which stands for the polyakov loop , can be represented by a matrix in the color space @xcite @xmath47,\label{loop}\end{aligned}\ ] ] where @xmath48 is the inverse temperature and @xmath49 is the polyakov gauge @xcite .
the polyakov loop matrix can be given as a diagonal representation @xcite . in the plsm lagrangian ,
( [ plsm ] ) , the coupling between the polyakov loop and the quarks is given by the covariant derivative of @xmath50 @xcite .
it is apparent that the plsm lagrangian is invariant under the chiral flavor - group .
this is similar to the original qcd lagrangian @xcite . in order to reproduce the thermodynamic behavior of the polyakov loop for pure gauge
, we use a temperature - dependent potential @xmath51 .
this should agree with the lattice qcd simulations and have @xmath52 center symmetry as that of the pure gauge qcd lagrangian @xcite . in case of vanishing chemical potential ,
then @xmath53 and the polyakov loop is considered as an order parameter for the deconfinement phase - transition @xcite . in the present work ,
we use @xmath51 , landau - ginzburg type potential , as a polynomial expansion in @xmath54 and @xmath55 @xcite @xmath56 where , @xmath54 and @xmath57 are introduced previously and @xmath58 , where constants are @xmath59 , @xmath60 , @xmath61 , @xmath62 , @xmath63 and @xmath64 . in eq .
( [ uloop ] ) , the vandermonde jacobian contribution , @xmath65 $ ] , was ignored due the small value of @xmath66 . in principle , the vandermonde term comes from the change of variables from vector potential to @xmath54 in the path integral and should guarantee a reasonable behavior of the mean field approximation @xcite , i.e. it was suggested to solve the problem that the normalized polyakov loop becomes greater than @xmath67 at very high temperatures . j[,^
* ] & = & , where @xmath68 is the vandermonde determinant , which is not explicitly space - time dependent .
the dimensionless parameter @xmath66 would be dependent on the temperature and the chemical potential .
therefore , @xmath66 should be estimated , phenomenologically . in order to reproduce the pure gauge qcd thermodynamics and the behavior of the polyakov loop as a function of temperature , we use the parameters listed out above in this section ( [ sec : approaches ] ) @xcite . in calculating the grand potential
, we use the mean field approximation @xcite , @xmath69 the purely mesonic potential is given as , @xmath70 where @xmath71 , @xmath72 , @xmath73 , @xmath74 , @xmath75 and @xmath76 are the model fixed parameters @xcite .
the quarks and antiquark contribution to the medium potential was introduced in ref @xcite and based on landau quantization and magnetic catalysis concepts , app . [ appnd:1 ] , we get @xmath77 \right .
\nonumber \\ & & \hspace*{8.5 mm } \left .
+ \ln \left [ 1 + 3\left(\phi^*+\phi e^{-\frac{(e_f + \mu_f)}{t}}\right)\ , e^{-\frac{(e_f + \mu_f)}{t}}+e^{-3\frac{(e_f + \mu_f)}{t}}\right ] \right\ } , \label{new - qqpotio}\end{aligned}\ ] ] it is worthwhile to highlight that the chemical potential used everywhere in the manuscript is the quark one , @xmath78 with @xmath79 being the quark flavor .
the different variables are elaborated in the app .
[ appnd:1 ] .
the potential at vanishing @xmath1 reads @xmath80 \right .
\nonumber \\ & & \hspace*{3.2 mm } \left .
+ \ln \left [ 1 + 3(\phi^*+\phi e^{-(e_f+\mu_f)/t})\ , e^{-(e_f+\mu_f)/t}+e^{-3 ( e_f+\mu)/t}\right ] \right\}. \label{thermalomg}\end{aligned}\ ] ] this is the system free of landau quantization .
the landau theory quantizes of the cyclotron orbits of charged particles in magnetic field . for small magnetic fields ,
the number of occupied landau levels ( ll ) is large and the quantization effects are washed out , while for large magnetic fields , the landau levels are less occupied and the chiral symmetry restoration occurs for smaller values of the chemical potential . according to eqs .
( [ new - qqpotio ] ) and ( [ thermalomg ] ) , eq .
( [ potential ] ) get an additional term , @xmath81 where @xmath82 represents the potential term at vanishing magnetic field , @xmath83 switches between the two systems ; one at vanishing and one at finite magnetic field .
we notice that the sum in eqs .
( [ uloop ] ) , ( [ new - qqpotio ] ) and ( [ upotio ] ) give the thermodynamic potential density as in eq .
( [ potential ] ) . by using the minimization condition , app .
[ appnd:2 ] , we can evaluate the parameters .
having the thermodynamic potential , eq .
( [ potential ] ) , we can determine all thermal quantities including the higher - order moments of particle multiplicity , and then mapping out the chiral phase - diagram @xcite .
the meson masses are defined by the second derivative with respect to the corresponding fields of the grand potential , eq .
( [ potential ] ) , evaluated at its minimum .
the results of the chiral condensates @xmath84 and @xmath85 , section [ subsec : condensates ] , the thermodynamic quantities , section [ subsec : thermo ] , the non - normalized and normalized higher - order moment of particle multiplicity , section [ sec : h - m ] and section [ sec : nh - m ] , respectively , the chiral phase - transition , section [ subsec : phase - t ] and finally the meson masses , section [ sec : masses ] , are introduced as follows .
the thermal evolution of the chiral condensates , @xmath84 and @xmath85 , and the polyakov order parameters , @xmath54 and @xmath57 is calculated from eq .
( [ potential ] ) at finite chemical potential and finite magnetic field using the minimization conditions given in eq .
( [ cond1 ] ) . the dependence on the four parameters , temperature @xmath3 , chemical potential @xmath86 , magnetic field @xmath18 and minimization parameter with respect to it the minimization condition shall be analysed . in left - hand panel ( a ) of fig .
[ fig : sig - with - t ] , the normalized chiral condensates , @xmath84 and @xmath85 , are given as function of temperature at vanishing chemical potential and different magnetic field values , @xmath87mev@xmath88 ( double - dotted curve ) , @xmath89mev@xmath88 ( solid curve ) and @xmath90mev@xmath88 ( dotted curve ) .
we notice that both condensates increase with increasing the magnetic field , @xmath4 .
this dependence seems to explain the increase in the chiral critical temperature @xmath2 with the magnetic field .
this - in turn - agrees with various studies using plsm and pnjl @xcite .
the condensates become moderated ( smoother ) with increasing magnetic field .
[ sigxy ] the right - hand panel ( b ) of fig .
[ fig : sig - with - t ] shows the chiral condensates , @xmath84 and @xmath85 , as function of temperature at constant magnetic field @xmath91mev@xmath88 , and finite chemical potentials , @xmath92mev ( solid curve ) , @xmath89mev ( dotted curve ) and @xmath93mev ( double - dotted curve ) .
both condensates decrease with increasing the chemical potentials .
this dependence gives a signature for the decreasing behavior of the chiral critical temperature @xmath2 with increasing the chemical potential , which obviously agrees with our previous calculations @xcite .
the condensates become rowdy ( sharper ) with increasing chemical potential . the left - hand panel ( a ) of fig . [
fig : fi - with - t ] shows the polyakov - loop field and it is conjugation , @xmath54 ( upper curves ) and @xmath57 ( lower curves ) , as function of temperature at a vanishing chemical potential and different magnetic field values , @xmath94gev@xmath88 ( double - dotted curve ) , @xmath95gev@xmath88 ( solid curve ) and @xmath96gev@xmath88 ( dotted curve ) .
both fields decrease with increasing the magnetic field .
this behavior explains the dependence of the confinement critical temperature on the magnetic field . at vanishing chemical potential , @xmath97 .
both polyakov - loop fields become smoother with increasing magnetic field .
the right - hand panel ( b ) draws the same as in left - hand panel but at a constant magnetic field @xmath98gev@xmath88 and different quark chemical potential values , @xmath99gev ( solid curve ) , @xmath95gev ( dashed curve ) and @xmath100gev ( double - dotted curve ) .
we find that @xmath54 increases with increasing the chemical potential values but @xmath57 decreases .
this behavior seems to agree with our previous calculations @xcite . at finite chemical potential , @xmath101 .
we conclude that the polyakov - loop fields , @xmath54 and @xmath57 , increase with @xmath3 , fig .
[ fig : fi - with - t ] . at vanishing @xmath86 , both @xmath54 and @xmath57 decrease with increasing @xmath4 . at finite @xmath86
, we find that @xmath54 increases , while @xmath57 decreases with @xmath4 .
in this section , we introduce some thermal quantities like energy density and trace anomaly . as we discussed in ref .
@xcite , the purely mesonic potential , eq .
( [ upotio ] ) gets infinity at very low temperature and entirely vanishes at high temperature . from this numerical estimation , we concluded that this part of potential is only effective at very low temperatures .
its dependence on the external magnetic field has been checked and was found that finite @xmath4 comes up with very tiny contribution to this potential part .
as the present study is performed at temperatures around the critical one , this potential part can be removed from the effective potentials given in eq .
( [ potential ] ) . in eq .
( [ upotio ] ) , the chiral condensates , @xmath0 s , are small at finite temperature , fig . [
fig : sig - with - t ] . therefore
, much smaller values are expected for their higher orders and multiplications .
opposite situation is likely at very small temperatures .
the energy density , @xmath102 , at finite quark chemical potential , @xmath78 , can be obtained as @xmath103 in section [ subsec : condensates ] , we have estimated the parameters , the two chiral condensates , @xmath84 and @xmath85 and the two order parameters of the polyakov - loop and it s conjugation , @xmath54 and @xmath57 , respectively .
thus , we can substitute all these into eq .
( [ e1 ] ) . the left - hand panel ( a ) of fig .
[ fig : pr ] presents the normalized energy density , @xmath102 , as function of temperature at vanishing chemical potential . in calculating the results , eqs .
( [ new - qqpotio ] ) and eq .
( [ thermalomg ] ) are implemented as given in eq .
( [ potentialnew ] ) .
the general temperature - dependence is not absent . also , we notice that @xmath102 is sensitive to the change in @xmath4 @xcite .
increasing @xmath4 seems to increase the critical temperature , at which the system undergoes phase transition .
as the chiral condensates become smoother with increasing @xmath4 , the thermodynamic quantities , such as energy density , behave accordingly , i.e. the phase transition becomes smoother as well .
the right - hand panel ( b ) shows @xmath102 as function of temperature at a constant magnetic field @xmath98gev@xmath88 and varying quark chemical potentials , @xmath99gev ( long - dashed curve ) , @xmath95gev ( dash - dotted curve ) and @xmath100gev ( double - dotted curve ) .
the solid curve represents the results in absence of an external magnetic field but at @xmath99gev .
we note that @xmath102 is not as sensitive to the change in @xmath86 @xcite as to the external magnetic field . despite the lack of chemical potential dependency , which can be understood due to the large magnetic field applied ,
it is believed to affect contrary to the chemical potential . to this indirect dependency of @xmath86 and @xmath1
, we shall devote a separate work . again
, it seems that increasing @xmath78 , decreasing @xmath2 . at finite quark chemical potential ,
the trace anomaly known as interaction measure reads @xmath104 in fig .
[ fig : tr ] , we notice that the normalized trace - anomaly under the effect of an external magnetic field becomes smaller than the corresponding quantity in absence of magnetic field @xcite at high temperature .
this can be explained due the restrictions added to the quark energy by the landau quantization through the magnetic field .
we find that increasing @xmath4 increases the critical temperature .
this behavior can be understood because of the dependence of the chiral condensates , fig .
[ fig : sig - with - t ] and the polyakov - loop potential , fig .
[ fig : fi - with - t ] on @xmath4 . in the left - hand panel ( a ) ,
the trace anomaly @xmath105 , is given as function of temperature @xmath3 at vanishing chemical potential but different values of the magnetic fields , vanishing @xcite ( solid curve ) , @xmath94gev@xmath88 ( long - dashed curve ) , @xmath98gev@xmath88 ( dash - dotted curve ) and @xmath106gev@xmath88 ( double - dotted curve ) .
we notice that the trace anomaly increases with @xmath3 until the chiral symmetry is restored . then , increasing @xmath3 reduces the normalized trace anomaly .
the peak represents the critical temperature @xmath2 corresponding to a certain magnetic field .
we find that @xmath2 increases with increasing @xmath4 .
the right - hand panel ( b ) of fig .
[ fig : tr ] shows the same as in the left - hand panel but at a constant magnetic field @xmath98gev@xmath88 and different chemical potentials , @xmath107gev ( solid curve ) , @xmath108gev ( long - dashed curve ) , @xmath89mev ( dash - dotted curve ) and @xmath93gev ( double - dotted curve ) .
the trace anomaly increases with @xmath3 until the chiral symmetry is fully restored .
the peaks are positioned at @xmath2 of the certain value for chemical potential .
here , we find that @xmath2 decreases with increasing @xmath86 .
the sensitivity to @xmath86 is not as strong as to @xmath4 .
this might be interpreted as the high magnetic field applied seems to contradict the effects of the chemical potential . in other words
, should the magnetic field adds energy to the system , the chemical potential requires energy in order to produce new particles .
we notice that the dependence on the quark chemical potential is more obvious that that shown in fig .
[ fig : pr ] . in app .
[ appnd:1 ] , we discuss the magnetic catalysis , eq .
( [ dr ] ) , and the so - called dimension reduction concepts , eq.([dr ] ) . due to the effects of the magnetic field
, the latter would mean modifying the sum over the three - dimensional momentum space to a one - dimensional one .
according to ref .
@xcite , the effect of this reduction reduces also the value of the quantity by almost two third from the expected value .
this would explain the difference between results at vanishing and that at finite @xmath4 , left - hand panels ( a ) of figs .
[ fig : pr ] and [ fig : tr ] , for instance . in the present work
, we distinguish between two types of systems . in the first one ,
the landau quantization should be implemented , i.e. taking into account the magnetic effects , while in the other system , the external magnetic field is not taken into consideration , i.e. no magnetic contribution to the thermal system .
the higher - order moment of the particle multiplicity is defined @xcite as m_i & = & , where the pressure @xmath109 is related to the partition function , which in tern is related to the potential , @xmath110 . in this section
, we introduce the first four non - normalized moments of the particle multiplicity calculated in plsm under the effects of an external magnetic field .
the thermal evolution is studied at a constant chemical potential but different magnetic fields and also at a constant magnetic field but different chemical potentials .
doing this , it is possible to map out the chiral phase - diagram , for which we determine the irregular behavior in the higher - order moments as function of @xmath3 and @xmath86 . here , we introduce the non - normalized higher - order moments of the particle multiplicity @xcite .
the left - hand panels ( a ) of figs .
[ fig : m1 ] , [ fig : m2 ] , [ fig : m3 ] and [ fig : m4 ] show the first four non - normalized moments of the quark distributions .
these quantities are given as function of temperature at a constant chemical potential @xmath99gev and different magnetic fields , @xmath111gev@xmath88 ( double - dotted curve ) , @xmath96gev@xmath88 ( dashed curve ) and @xmath112gev@xmath88 ( dotted curve ) .
we find that increasing temperature rapidly increases the four moments .
furthermore , the thermal dependence is obviously enhanced , when moving from lower to higher orders .
the values of the moment are increasing as we increase the magnetic field .
the fluctuation in the third- and fourth - order moments reflect the increase of the critical temperature @xmath2 with increasing the magnetic field .
the critical temperature can , for instance , be defined where the peaks are positioned .
the right - hand panels ( b ) of figs .
[ fig : m1 ] , [ fig : m2 ] , [ fig : m3 ] and [ fig : m4 ] , present the same as in the left - hand panels but at a constant magnetic field @xmath111gev@xmath88 and different chemical potentials , @xmath113gev ( double - dotted curve ) , @xmath108gev ( dashed curve ) and @xmath95gev ( dotted curve ) .
it is apparent that increasing temperature rapidly increases the four moments of quark number density .
furthermore , the thermal dependence is obviously enhanced , when moving from lower to higher orders . the values of the moment are increasing as we increase the chemical potential .
but the critical temperature @xmath2 decrease with @xmath86 .
the peaks are positioned at the critical temperature .
the statistical normalization of the higher - order moments requires a scaling of the non - normalized quantities , section [ sec : non - h - m ] , with respect to the standard deviation @xmath0 , which is related to the susceptibility @xmath114 or the fluctuations @xcite in the particle multiplicity .
it is conjectured that the dynamical phenomena could be indicated by large fluctuations in these dimensionless moments and therefore , the chiral phase - transition can be mapped out @xcite .
due to the sophisticated derivations , we restrict the discussion here to dimensionless higher - order moments @xcite . this can be done when the normalization is done with respect to the temperature or chemical potential .
the higher - order moments of the particle multiplicity normalized with respect to temperature are studied in dependence on the temperature at a constant chemical potential and different magnetic fields .
also they are studied at different chemical potentials and a constant magnetic field .
the corresponding expressions were deduced in ref .
@xcite . in left - hand panel ( a ) of figs .
[ fig : mn1 ] , [ fig : mn2 ] and [ fig : mn3 ] the first three normalized moments are given as function of temperature at a constant chemical potential @xmath99gev and different magnetic fields , @xmath111gev@xmath88 ( double - dotted curve ) , @xmath96gev@xmath88 ( dashed curve ) and @xmath112gev@xmath88 ( dotted curve ) .
we find that the values of the moments are increasing as the magnetic field increases .
the fluctuations in the normalized moments would define the dependence of the critical temperature @xmath2 on the magnetic field .
the right - hand panels ( b ) of figs .
[ fig : mn1 ] , [ fig : mn2 ] and [ fig : mn3 ] present the first three normalized moments as function of temperature but at a constant magnetic field @xmath111gev@xmath88 and different chemical potentials , @xmath113gev ( double - dotted curve ) , @xmath108gev ( dashed curve ) and @xmath95gev ( dotted curve ) .
we notice that the moments of quark multiplicity increase with the chemical potentials . that the peaks at corresponding critical temperatures can be used to map out the chiral phase - diagrams , @xmath3 vs. @xmath4 and @xmath3 vs. @xmath86 .
now we can study the effects of the magnetic field on the chiral phase - transition . in a previous work @xcite ,
we have introduced and summarized different methods to calculate the critical temperature and chemical potential , @xmath115 , by using the fluctuations in the normalized higher - order moments of the quark multiplicity or by using the order parameters .
the latter is implemented in the present work .
the plsm has two order - parameters .
the first one presents the chiral phase - transition .
this is related to strange and non - strange chiral condensates , @xmath84 and @xmath85 .
the second one gives hints for the confinement - deconfinement phase - transition , the polyakov - loop fields , @xmath54 and @xmath57 .
therefore , for the models having polyakov - loop potential , we can follow a procedure as follows .
we start with a constant value of the magnetic field . by using strange and non - strange chiral - condensates , a dimensionless quantity reflecting the difference between the non - strange and strange condensates @xmath116 as a function of temperature at fixed chemical potentials
will be implemented .
this procedure give _
one point _ in the @xmath3-@xmath86-chart , at which the chiral phase transition takes place . at the same chemical potential as in previous step ,
we deduce the other order - parameter related to the polyakov - loop fields as a function of temperature .
these calculations give another point ( _ in @xmath3 and @xmath86 chart _ ) , at which the deconfinement phase - transition takes place . by varying the chemical potential ,
we repeat these steps .
then , we find a region ( or point ) , in ( at ) which the two order - parameters , chiral and deconfinement , cross each other , i.e. equal each other .
it is assumed that such a point represents phase transition(s ) at the given chemical potential . in doing this
, we get a set of points in a two - dimensional chart , the qcd phase - diagram . in fig .
[ fig : p - d ] , we compare five chiral phase - diagrams , @xmath117 vs. @xmath118 , with each others at @xmath119 , @xmath120 , @xmath121 , @xmath122 and @xmath123gev@xmath88 from top to bottom .
@xmath117 are plotted against @xmath118 , where the two normalization quantities @xmath124gev and @xmath125gev were deduced from ref .
this should give an indication about the behavior of the critical temperature and the critical chemical potential of the system under the effect of the magnetic field .
apparently , we conclude that both critical temperature and critical chemical potential increase with increasing the magnetic field . the masses can be deduced from the second derivative of the grand potential , eq .
( [ potential ] ) with respect to the corresponding fields , evaluated at its minimum , which is estimated at vanishing expectation values of all scalar and pseudoscalar fields @xmath126 where @xmath127 and @xmath128 range from @xmath129 and @xmath130 and @xmath131 are scalar and pseudoscalar mesonic fields , respectively .
obviously , @xmath132 stands for scalar and pseudoscalar mesons .
the scalar meson masses @xcite @xmath133 where @xmath134 is the scalar mixing angle @xcite _ s & = & , with @xmath135 .
the expressions for @xmath136 and @xmath137 can be found in ref .
@xcite . on the tree level
, @xmath138 can be determined according to @xmath139 . in fig .
[ fig : sif_f0_mu ] , the scalar meson masses , @xmath140 from eq .
( [ ms1 ] ) , and @xmath141 from eq .
( [ ms2 ] ) are given as function of temperature at a constant magnetic field @xmath111gev@xmath88 and different chemical potentials @xmath107gev ( dotted curve ) , @xmath108gev ( dashed curve ) and @xmath95gev ( double - dotted curve ) .
we conclude that the scalar meson masses decrease as the temperature increases .
this remains until @xmath3 reaches the critical value .
then , the vacuum effect becomes dominant and rapidly increases with the temperature .
the effect of the chemical potential is very obvious .
the masses decrease with the increase in chemical potential .
this explains the phase diagram of temperatures and chemical potentials at a certain magnetic field .
the decrease of the critical temperature with increasing chemical potential is represented by the bottoms ( minima ) in thermal behavior of meson masses before switching on the vacuum effect . in fig .
[ fig : sif_f0 ] , the four scalar meson masses , @xmath140 from eq .
( [ ms1 ] ) , @xmath141 from eq .
( [ ms2 ] ) , @xmath142 from eq .
( [ ms3 ] ) and @xmath143 from eq .
( [ ms4 ] ) are given as function of temperature at two values of chemical potential , @xmath99gev left - hand panel ( a ) and @xmath144gev right - hand panel ( b ) and different magnetic fields , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath145 and @xmath146gev@xmath88 from top to bottom .
we notice that the scalar meson masses decrease as the temperature increases , until it reaches the critical temperature .
then , the vacuum effect gets dominant and apparently increases with the temperature .
the effect of magnetic field is very obvious .
the masses increase as the magnetic field increases .
this explains the phase diagram of temperatures and magnetic field at a certain chemical potential . in fig .
[ fig : sif_f0_norm ] , the normalized scalar meson masses , @xmath140 from eq .
( [ ms1 ] ) , @xmath141 from eq .
( [ ms2 ] ) , @xmath142 from eq .
( [ ms3 ] ) and @xmath143 from eq .
( [ ms4 ] ) are given as function of temperature at two values of chemical potential , @xmath99gev in left - hand panel ( a ) and @xmath144gev in right - hand panel ( b ) and different magnetic fields , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath145 and @xmath146gev@xmath88 from top to bottom .
the normalization is done due to the lowest matsubara frequencies , @xmath147 , app .
[ sec : matsub ] . at high temperatures ,
we notice that the masses of almost all meson states become temperature independent , i.e. constructing a kind of a universal bundle .
this would be seen as a signature for meson dissociation into quarks . in other words ,
the meson states undergo deconfimement phase - transition .
it is worthwhile to highlight that the various meson states likely have different critical temperatures . at low temperatures ,
the scalar meson masses normalized to the lowest matsubara frequency rapidly decrease as the temperature increases .
then , starting from the critical temperature , we find that the thermal dependence almost vanishes .
the magnetic field effect is clear , namely the meson masses increase with increasing magnetic field .
this characterizes @xmath3 vs. @xmath4 phase diagram . the pseudoscalar meson masses @xcite @xmath148 where @xmath149 is the pseudoscalar mixing angle @xcite _ p & = & , with @xmath150 .
the expressions for @xmath151 are given in eq .
( 11c ) in ref .
@xcite . in fig .
[ fig : eta_etad_mu ] , the pseudoscalar meson masses , @xmath152 from eq .
( [ mp1 ] ) , @xmath153 from eq .
( [ mp2 ] ) , @xmath154 from eq .
( [ mp3 ] ) and @xmath155 from eq .
( [ mp4 ] ) are given as function of temperature at a constant magnetic field @xmath111gev@xmath88 and different chemical potentials , @xmath107gev ( dotted curve ) , @xmath108gev ( dashed curve ) and @xmath95gev ( double - dotted curve ) .
it is obvious that the pseudoscalar meson masses remain constant at low temperature . at temperatures @xmath156 ,
the vacuum effect becomes dominant . accordingly .
the pseudoscalar meson masses increase with the temperature .
we shall notice that even contribution by the vacuum will be moderated through the normalization with respect to the lowest matsubara frequency . in fig .
[ fig : sif_f0_p ] , the four pseudoscalar meson masses , @xmath152 from eq .
( [ mp1 ] ) , @xmath153 from eq .
( [ mp2 ] ) , @xmath154 from eq .
( [ mp3 ] ) and @xmath155 from eq .
( [ mp4 ] ) are given as function of temperature at two constant chemical potentials , @xmath99gev in left - hand panel ( a ) and @xmath144gev in right - hand panel ( b ) and different magnetic fields , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath145 and @xmath146gev@xmath88 from top to bottom .
again , at low temperature , the masses remain temperature - independent . at @xmath157 ,
the vacuum effect is switched on .
accordingly , the masses increase rapidly with the temperature . in fig .
[ fig : sif_f0_p_norm ] , the four pseudoscalar meson masses , @xmath152 from eq .
( [ mp1 ] ) , @xmath153 from eq .
( [ mp2 ] ) , @xmath154 from eq .
( [ mp3 ] ) and @xmath155 from eq .
( [ mp4 ] ) normalized with respect to the lowest matsubara frequency are given as function of temperature at two constant chemical potentials , @xmath99gev in left - hand panel ( a ) and @xmath144gev in right - hand panel ( b ) and different magnetic fields , @xmath119 , @xmath120 , @xmath121 , @xmath122 , @xmath145 and @xmath146gev@xmath88 from top to bottom .
the normalization should result in temperature - independent mesonic states .
this would be seen as a signature for meson dissociation into quarks .
it is obvious that various critical temperatures can be assigned to various pseudoscalar meson states .
the normalized masses starts with high values reflecting confinement , especially at low temperatures .
then , they decrease as the temperature increases until the critical temperature , @xmath2 , which differs for different meson states . at higher temperatures ,
the dependence of meson masses on temperature is almost entirely removed .
.a comparison between pseudoscalar meson masses , @xmath158 , in mev and the corresponding results from pnjl @xcite .
both are compared with the experimental measurements , the particle data group ( pdg ) @xcite and the lattice qcd calculations @xcite . [ cols="^ " , ] + as introduced in ref .
@xcite , tab .
[ masscomp ] presents a comparison between pseudoscalar meson nonets in various effective models , like plsm ( present work ) and pnjl @xcite confronted to the particle data group @xcite and lattice qcd calculations @xcite .
the comparison for scalar states would be only partly possible .
some remarks are now in order .
the errors are deduced from the fitting for the parameters used in calculating the equation of states and other thermodynamics quantities .
the output results are very precise for some of the lightest hadron resonances described by the present model , plsm .
an extended comparison is given in ref .
the qcd phase - diagram at vanishing chemical potential and finite temperature subject to an external magnetic field gained prominence among high - energy physicists , for instance , our previous work @xcite was based on two concepts in order to explain the effects of external magnetic field on the qcd phase - diagram .
another study was done in the framework of nambu - jona - lasinio ( njl ) model and polyakov njl ( pnjl ) model @xcite .
the main idea is that the scalar coupling parameter is taken dependent on the magnetic field intensity .
thus , the scalar coupling parameter decreases with the magnetic field increase .
we also implemented the relation between the magnetic field and the scalar coupling parameter in order to fit for the lattice qcd results @xcite .
we conclude the increase in the magnetic field increases the critical temperature .
in the presence work , we use the polyakov linear @xmath0-model and assume that the external magnetic field adds some restrictions to the quarks due to the existence of free charges in the plasma phase .
in doing this , we apply landau theory ( landau quantization ) , which quantizes of the cyclotron orbits of charged particles in magnetic field .
first , we have calculated and then analysed the thermal evolution of the chiral condensates and the deconfinement order - parameters .
we notice that the landau quantization requires additional temperature to drive the system through the chiral phase - transition .
accordingly , we find that the value of the chiral condensates increase with increasing the external magnetic field . in the contrary to various previous studies , the effects of the external magnetic field are analysed , systematically . accordingly
, the dependence of the critical temperatures of chiral and confinement phase - transitions on the magnetic field could be characterized .
we deduced @xmath3-@xmath86 curves using su(3 ) plsm in external magnetic field .
furthermore , by using mean field approximation , we constructed the partition functions and then driven various thermodynamic quantities , like energy density and interaction rate ( trace anomaly ) .
their dependence on temperature and chemical potential recalls to highlight that the effects of external magnetic field on the chemical potential was disregarded in all calculations at finite chemical potential .
we have analysed the first four non - normalized higher - order moments of particle multiplicity .
the thermal evolution was studied at a constant chemical potential but different magnetic fields and also at a constant magnetic field but different chemical potentials .
doing this , the chiral phase - diagram can be mapped out .
we determined the irregular behavior as function of @xmath3 and @xmath86 .
we found that increasing temperature rapidly increases the four moments and the thermal dependence is obviously enhanced , when moving from lower to higher orders .
the values of the moment are increasing as we increase the chemical potential .
but the critical temperature @xmath2 decrease with @xmath86 .
the peaks are positioned at the critical temperature .
the higher - order moments normalized to temperature are studied at a constant chemical potential and different magnetic fields .
also they are studied at different chemical potentials and a constant magnetic field .
the statistical normalization requires scaling with respect to the standard deviation , @xmath0 , where @xmath0 is related to the susceptibility @xmath114 or the fluctuations . due to the sophisticated derivations ,
the discussion was limited to dimensionless higher - order moments .
this can be done when the normalization is done with respect to the temperature or chemical potential .
we find that the higher - order moments increase with the magnetic field .
we found that the moments increase with the chemical potentials . that the peaks at corresponding critical temperatures can be used to map out the chiral phase - diagrams , @xmath3 vs. @xmath4 and @xmath3 vs. @xmath86 .
the effects of the magnetic field on the chiral phase - transition have been evaluated .
there are different methods proposed to calculate the critical temperature and chemical potential , @xmath115 , through implementing fluctuations in the normalized higher - order moments or by the order parameters .
the latter was implemented in the present work .
it is obvious that plsm has two types of order - parameter .
the first one gives the chiral phase - transition and is related to strange and non - strange chiral condensates .
the second one gives hints for deconfinement phase - transition .
therefore , we can follow a procedure that at a constant magnetic field and by using strange and non - strange chiral - condensates , a dimensionless quantity would reflect the difference between the non - strange and strange condensates @xmath116 as a function of temperature at fixed chemical potentials , i.e. chiral phase - transition . at the same chemical potential
, we can deduce the other order - parameter related to the polyakov - loop fields as function in temperature .
both calculations give one point , at which the two order - parameters crossing each other .
it is assumed that such a point represents the transition point at the given chemical potential .
we repeat this at various chemical potentials and get a set of points in a two - dimensional chart , the qcd phase - diagram .
we have compared five qcd phase - diagrams , @xmath117 vs. @xmath118 , with each others at five different values of the magnetic field .
we found that both critical temperature and critical chemical potential increase with increasing the magnetic field .
the masses can be deduced from the second derivative of the grand potential with respect to the corresponding fields , evaluated at its minimum , which is estimated at vanishing expectation values of all scalar and pseudoscalar fields .
we have studied scalar and pseudoscalar meson masses as function of temperature at two different values of magnetic field and different chemical potentials .
we concluded that the meson masses decrease as the temperature increases .
this remains until @xmath3 reaches the critical value .
then , the vacuum effect becomes dominant and rapidly increases with the temperature .
the decrease of the critical temperature with increasing chemical potential is represented by the bottoms ( minima ) in thermal behavior of meson masses before switching on the vacuum effect . at low temperatures ,
the scalar meson masses normalized to the lowest matsubara frequency rapidly decrease as the temperature increases .
then , starting from the critical temperature , we find that the thermal dependence almost vanishes .
furthermore , the meson masses increase with increasing magnetic field .
this characterizes @xmath3 vs. @xmath4 phase diagram . at high temperatures ,
we notice that the masses of almost all meson states become temperature independent , i.e. constructing kind of a universal line .
this would be seen as a signature for meson dissociation into quarks . in other words ,
the meson states undergo deconfimement phase - transition .
it is worthwhile to highlight that the various meson states likely have different critical temperatures .
for simplicity , we assume that the direction of the magnetic field @xmath18 goes along @xmath159-direction . from the magnetic catalysis @xcite and by using landau quantization , we find that when the system is affected by a strong magnetic field , the quark dispersion relation will be modified to be quantized by landau quantum number , @xmath160 , and therefore the concept of dimensional reduction will be applied .
@xmath161 where @xmath0 is related to the spin quantum number and @xmath162 ( @xmath163 ) . here
, we replace @xmath164 by one quantum number @xmath165 , where @xmath166 is the lowest landau level ( lll ) and the maximum landau level ( mll ) was determined according to eq .
( [ mll ] ) @xcite , @xmath167 , where @xmath79 runs over @xmath22- , @xmath23- and @xmath24-quark mass ,
@xmath168 we apply another magnetic catalysis property @xcite , namely the dimensional reduction . as the name says ,
the dimensions will be reduced as @xmath169 . in this situation ,
the three - momentum integral will transformed into a one - momentum integral @xmath170 when @xmath171 represents the degenerate in the landau level , since for lll we have single degenerate and doublet for the upper landau levels , @xmath172 we use @xmath173 and @xmath174 for non - strange and strange quark mass , i.e. the masses of light quarks degenerate .
this is not the case for the electric charges . in section [ sec :
approaches ] , @xmath175 , @xmath176 and @xmath177 are elaborated .
we notice that the thermodynamic potential density as given in eq .
( [ potential ] ) , which has seven parameters @xmath178 and @xmath17 , two unknown condensates @xmath84 and @xmath85 and the order parameters for the deconfinement , @xmath54 and @xmath57 .
the six parameters @xmath179 and @xmath180 are fixed in the vacuum by six experimentally known quantities @xcite . in order to evaluate the unknown parameters @xmath84 , @xmath181 , @xmath54 and @xmath57 , we minimize the thermodynamic potential , eq .
( [ potential ] ) , with respect to @xmath84 , @xmath85 , @xmath54 and @xmath57 or @xmath182 and @xmath183 .
doing this , we obtain a set of four equations of motion , @xmath184 meaning that @xmath185 , @xmath186 , @xmath187 and @xmath188 are the global minimum .
in finite temperature field theory , the matsubara frequencies are a summation over the discrete imaginary frequency , @xmath189 , where @xmath190 is a rational function , @xmath191 for bosons and @xmath192 for fermions and @xmath193 is an integer playing the role of a quantum number . by using matsubara weighting function @xmath194 , which has simple poles exactly located at @xmath195 , then s _ & = & g(z ) h_(z ) dz , where @xmath196 stands for the statistic sign for bosons and fermions , respectively .
@xmath194 can be chosen depending on which half plane the convergence is to be controlled , @xmath197 where @xmath198 is the single - particle distribution function .
the mesonic masses are conjectured to have contributions from matsubara frequencies @xcite . furthermore , at high temperatures , @xmath156 , the behavior of the thermodynamic quantities , including the quark susceptibilities , the masses is affected by the interplay between the lowest matsubara frequency and the polyakov loop - correction @xcite .
we apply normalization for the different mesonic sectors with respect to lowest matsubara frequency @xcite in order to characterize the dissolving temperature of the mesonic bound states .
it is found that the different mesonic states have different dissolving temperatures .
this would mean that the different mesonic states have different @xmath2 s , at which the bound mesons begin to dissolve into quarks .
therefore , the normalized masses should not be different at @xmath199 .
to a large extend , their thermal and dense dependence should be removed , so that the remaining effects are defined by the free energy @xcite , i.e. the masses of _ free _ besons are defined by @xmath200 . that the masses of almost all mesonic states become independent on @xmath3 , i.e. constructing kind of a universal line , this would be seen as a signature for meson dissociation into quarks .
it is a deconfinement phase - transition , where the quarks behave almost freely .
in other words , the characteristic temperature should not be universal , as well .
so far , we conclude that the universal @xmath2 characterizing the qcd phase boundary is indeed an approximative average ( over various bound states ) .
this work is supported by the world laboratory for cosmology and particle physics , http://wlcapp.net/. at would like to thank abdel magied diab for the constructive discussion about the implementation of an external magnetic field in the sigma model ! g. s. bali , f. bruckmann , g. endrodi , z. fodor , s. d. katz and a. schafer , phys .
d * 86 * , 071502 ( 2012 ) .
p. v. buividovich , m. n. chernodub , e. v. luschevskaya and m. i. polikarpov , phys .
b * 682 * , 484 ( 2010 ) .
j. wambach , b .- j .
schaefer and m. wagner , acta phys .
. supp . * 3 * , 691 ( 2010 ) .
t. kahara and k. tuominen , phys .
d * 78 * , 034015 ( 2008 ) .
schaefer and m. wagner , prog .
62 , 381 ( 2009 ) .
h. mao , j. jin and m. huang , j. phys .
g*37 * , 035001 ( 2010 ) . c. ratti , m. a. thaler and w. weise , phys .
d * 73 * , 014019 ( 2006 ) .
k. fukushima , phys .
d * 77 * , 114028 ( 2008 ) .
b .- j . schaefer , j. m. pawlowski , and j. wambach , phys .
d * 76 * , 074023 ( 2007 ) .
k. dusling , c. ratti and i. zahed , phys .
d * 79 * , 034027 ( 2009 ) .
a. tawfik , soryushiron kenkyu * 114 * , b48-b50 ( 2006 ) .
simon hands , timothy j. hollowood and joyce c. myers , jhep * 1007 * , 086 ( 2010 ) .
j. beringer _
( particle data group ) , phys .
d * 86 * , 010001 ( 2012 ) .
t. xia , l. he and p. zhuang , phys .
d * 88 * , 056013 ( 2013 ) .
p. costa , m. c. ruivo , c. a. de sousa and y. l. kalinovsky , phys .
d * 70 * , 116013 ( 2004 ) ; + p. costa , m. c. ruivo , c. a. de sousa and y. l. kalinovsky , phys .
d * 71 * , 116002 ( 2005 ) ; + p. costa , m. c. ruivo , c. a. de sousa , h. hansen and w. m. alberico , phys .
d * 79 * , 116003 ( 2009 ) .
a. bazavov , _ et al .
_ ( hotqcd collaboration ) , phys .
d * 85 * , 054503 ( 2012 ) .
s. aoki , _ et al .
_ ( pacs - cs collaboration ) , phys . rev .
d * 81 * , 074503 ( 2010 ) .
a. tawfik and a. diab , _
polyakov su(3 ) extended linear @xmath0-model : sixteen mesonic states in chiral phase - structure _ , to appear in prc . | effects of an external magnetic field on various properties of the quantum chromodynamics ( qcd ) matter under extreme conditions of temperature and density ( chemical potential ) have been analysed . to this end
, we use su(3 ) polyakov linear @xmath0-model and assume that the external magnetic field ( @xmath1 ) adds some restrictions to the quarks energy due to the existence of free charges in the plasma phase .
in doing this , we apply the landau theory of quantization , which assumes that the cyclotron orbits of charged particles in magnetic field should be quantized .
this requires an additional temperature to drive the system through the chiral phase - transition .
accordingly , the dependence of the critical temperature of chiral and confinement phase - transitions on the magnetic field is characterized .
based on this , we have studied the thermal evolution of thermodynamic quantities ( energy density and trace anomaly ) and the first four higher - order moment of particle multiplicity .
having all these calculations , we have studied the effects of the magnetic field on the chiral phase - transition .
we found that both critical temperature @xmath2 and critical chemical potential increase with increasing the magnetic field , @xmath1 .
last but not least , the magnetic effects of the thermal evolution of four scalar and four pseudoscalar meson states are studied .
we concluded that the meson masses decrease as the temperature increases till @xmath2 .
then , the vacuum effect becomes dominant and rapidly increases with the temperature @xmath3 . at low @xmath3 ,
the scalar meson masses normalized to the lowest matsubara frequency rapidly decrease as @xmath3 increases .
then , starting from @xmath2 , we find that the thermal dependence almost vanishes .
furthermore , the meson masses increase with increasing magnetic field .
this gives characteristic phase diagram of @xmath3 vs. external magnetic field @xmath4 . at high @xmath3 , we find that the masses of almost all meson states become temperature independent .
it is worthwhile to highlight that the various meson states likely have different critical temperatures . |
in optical lattices @xcite , atoms are trapped in a periodic array of micron - sized potential wells .
atoms confined in these potential wells exhibit a rich dynamical behaviour which can be probed by a number of techniques , including probe absorption spectroscopy @xcite , heterodyne fluorescence spectroscopy @xcite , recoil - induced resonances @xcite , intensity correlation spectroscopy @xcite , and observation of the redistribution of photons between the lattice - inducing beams @xcite . using these techniques , the quantized nature of the motion of the atoms in the potential wells has been demonstrated , yielding information on the vibrational frequencies of the atoms , the motion of wavepackets , and on the decay of coherences between the bound oscillator states .
recently , a novel method has been suggested and experimentally demonstrated which effectively probes the coherences set up between different vibrational states of the trapped atoms @xcite .
this method has a number of advantages over other schemes , as we shall demonstrate .
most importantly , the possibility of selectively exciting one particular dynamical mode of the lattice allows one to fit theoretical predictions to experimental spectra easily and reliably .
+ in this paper we are predominantly concerned with resonances of the coherent transient spectra corresponding to stimulated raman transitions between adjacent vibrational levels .
the mechanisms responsible for the decay of the vibrational coherences and the resulting widths of the raman peaks are still a subject of debate , and a comprehensive study is as yet absent from the literature . in what follows , we present the results of our experiments conducted on a three - dimensional lattice of cesium atoms in the _ oscillating regime _
in which the oscillatory motion of the trapped atoms is only rarely interrupted by inelastic photon scattering events .
the converse case of optical lattices operating in the jumping regime has been treated by mennerat - robilliard and co - workers @xcite . in our experiments
the intensity , detuning , and geometry of the lattice were varied independently , and the resulting variations of the positions and widths of the vibrational raman peaks were measured and compared with theoretical models . + in another set of experiments , we loaded an optical lattice with atoms having an initial kinetic temperature higher than the steady - state temperature in the lattice and made time - resolved measurements of the vibrational frequency as the atoms came to equilibrium in the lattice .
as the distribution of atoms amongst the vibrational states evolved , the measured vibrational frequency was observed to change owing to the perturbations to the energy spacings of the bound levels caused by the anharmonicity of the potential .
+ this paper is organised as follows . in section [ setup ] the experimental setup
is described , and in section [ results ] we present the results of the experiments .
section [ theory ] contains a brief discussion of the various mechanisms involved in determining the frequency widths of the vibrational raman resonances observed in optical lattices and a comparison of the experimental data with theoretical models .
finally , section [ conclusion ] summarizes our findings .
in our experiment , we trap and cool cesium atoms in a standard six - beam magneto - optical trap ( mot ) operating on the @xmath0 transition @xcite . after a @xmath1 trapping period , the magnetic field gradient is switched off and the atoms are further cooled for a few milliseconds in an optical molasses to a temperature of around @xmath2 .
the molasses beams are then switched off and the lattice beams switched on .
the geometry of the lattice beams is the standard tetrahedron configuration @xcite with splitting half - angles @xmath3 and @xmath4 , where in our experiment @xmath4 is fixed and @xmath3 is variable ( see fig . [ fig : setup ] ) .
the lattice operates on the same transition as the mot , detuned to the low frequency side by @xmath5 to @xmath6 , where @xmath7 is the excited state lifetime ( @xmath8 for cesium ) .
the intensity of the lattice beams in our experiment lies between @xmath9 and @xmath10 .
in addition to the four lattice beams , a weak ( @xmath11 ) probe beam propagates either along the @xmath12- or the @xmath13-axis .
the detuning @xmath14 of the probe laser from the lattice laser can be controlled by an acousto - optic modulator ( aom ) connected to a tunable radio - frequency source .
+ coherent transient spectra are recorded as follows : after the atoms have equilibrated in the lattice for a few milliseconds , the probe beam is switched on for @xmath15 at detuning @xmath14 ( see also section [ theory ] ) .
after that time , at @xmath16 the detuning is increased by @xmath17 within less than @xmath18 .
after the increase in the detuning of the probe light , it no longer interacts with the atoms and serves as a local oscillator for the heterodyne detection of the light scattered by the atoms in the lattice .
the signal arising from the beating of the lattice light scattered by the atoms and the local oscillator is then recorded with a fast photodiode .
assuming that the coherently scattered light from the lattice decays exponentially with time at a rate @xmath19 , the observed signal ( expressed as a voltage @xmath20 ) is of the form @xmath21\nonumber \\
+ \frac{1}{2}(i_{probe}+i_{coh})\quad,\end{aligned}\ ] ] where @xmath22 and @xmath23 are the intensities of the probe and the coherently scattered light , respectively , and @xmath24 is the amount by which the probe frequency is suddenly switched at @xmath16 . a fast - fourier - transform ( fft ) is then performed on the signal to yield its frequency spectrum .
our spectra are shifted by subtracting the frequency jump @xmath24 , so that a rayleigh peak ( for which @xmath25 initially ) appears at zero frequency in our spectra .
+ in the case of the time - resolved equilibration experiment described in section [ equili ] , the lattice is loaded directly from the mot with a temperature of order @xmath26 , and the atoms are then allowed to equilibrate in the lattice for a variable length of time . after this period
the vibrational excitation of the trapped atoms is probed and read - out as described above .
some aspects of the dynamical behaviour of atoms in optical lattices are better understood than others .
the vibrational frequencies of the atoms , for instance , can be rather accurately predicted and modelled using a simple harmonic oscillator picture ( for potential wells that are deep enough for a tight - binding model to be valid ) @xcite .
discrepancies with experiments can be accounted for by the anharmonicity of the potential , which can be treated perturbatively @xcite .
+ on the other hand , the lifetimes of atoms in these oscillator states and of the coherences between different states , and hence the widths they give rise to in the coherent transient spectra , are not yet well understood .
several mechanisms have been proposed that could be responsible for the observed widths of these vibrational raman peaks @xcite . in particular , the photon scattering rate of atoms in a light field turns out to be much larger than the observed widths of the raman peaks , indicating that the decay of coherences between vibrational levels is strongly suppressed .
the mechanisms responsible for this suppression will be dealt with in more detail in section [ mechs ] .
+ in order to carry out a comprehensive study , we independently varied the detuning , intensity , and geometry ( characterized by the two splitting angles ) of the lattice and determined both the position and the width of the raman peaks in the coherent transient spectrum by fitting lorentzians to the experimentally obtained spectra .
typical experimental results are shown in fig .
[ fig : spectra ] .
the signal - to - noise ratio even of a single - shot measurement is good , so that little averaging is necessary .
the signal recorded by the photodiode is , in general , a sum of exponentially damped beat - notes between the light scattered by the trapped atoms and the probe light .
when the probe is initially tuned to the same frequency as the lattice beams , one sees a narrow rayleigh peak and two peaks resulting from off - resonantly excited vibrational coherences . when detuning the probe from the lattice beams by an amount equal to the separation between a raman peak and the rayleigh peak ( i.e. the vibrational frequency of the atoms ) , one can selectively excite the vibrational raman coherences alone and hence obtain a spectrum that only reflects one excitation mode of the lattice , in contrast to probe absorption spectra , where resonances associated with many modes are superimposed .
it is found that the resulting spectrum can be well fitted to a lorentzian , and in what follows we will refer to the width ( fwhm ) of such a lorentzian fit as the width of the raman peak .
a further advantage of the method of coherent transients is the fact that the preparation and the read - out of the vibrational coherence grating are separated in time .
the frequency - selectivity in the excitation phase therefore depends on the frequency bandwidth of the initial probe pulse , whilst the resolution of the spectrum only depends on the read - out time . by making the read - out time as short as possible whilst maintaining a reasonable resolution ,
time - dependent studies of the evolution of atoms in an optical lattice can be performed .
this is discussed in more detail in section [ equili ] . + as can be seen in fig .
[ fig : spectra ] , the rayleigh peaks are of much smaller width than the raman peaks .
these widths yield information on the rate of transfer of atoms between neighbouring wells and are discussed in detail in @xcite . in this paper , however , we shall concentrate our attention on the vibrational raman peaks . + for a tetrahedral lattice ( see fig . [
fig : setup ] ) in which the polarizations of the four lattice beams are perpendicular to the planes spanned by each pair of nearly co - propagating beams , one can derive simple expressions for the vibrational frequencies in a harmonic approximation of the potential close to a minimum @xcite : @xmath27 and @xmath28 where @xmath29 is the recoil energy with @xmath30 for an optical lattice employing cesium atoms and light tuned close to the d2 resonance line and @xmath31 is the potential well - depth given by @xmath32 ( @xmath33 is the intensity of a single beam , @xmath34 , and @xmath35 is the detuning of the lattice beams ) .
these theoretical predictions have been verified to yield the correct dependence on @xmath36 , @xmath33 , and @xmath37 , but the experimental results usually fall short of the theoretical values by 15 - 25 percent . in our experiment
, we varied the potential well - depth @xmath31 ( ranging from @xmath38 to @xmath39 ) both by varying the intensity ( between @xmath40 and @xmath41 ) and the detuning ( between @xmath42 and @xmath43 ) and measured the positions of the raman peaks of the corresponding coherent transient spectra along the @xmath12- and the @xmath13- directions .
the results of these measurements are summarized in fig .
[ fig : intdet ] ( a ) and confirm the dependence of the vibrational frequencies on @xmath31 as given by equations [ eqn : omxy ] and [ eqn : omz ] .
the splitting half - angles in this experiment were @xmath44 degrees and @xmath45 degrees , and the ratio of the slopes of fig .
[ fig : intdet ] ( a ) is @xmath46 , which is close to the theoretically predicted value of @xmath47 .
+ the results of measurements of the raman widths for the cases of the probe propagating along both the @xmath12- and the @xmath13-directions are shown in fig .
[ intdet](b ) where , for clarity , most of the error bars have been omitted @xcite . for the most part , the width of the raman peaks and the corresponding vibrational frequencies are proportional to each other , with @xmath48 . only for frequencies above @xmath49
do the raman widths deviate appreciably from this simple relationship . in order to investigate the dependence of the position and width of the raman peaks on the geometry of the lattice , the intensity and detuning of the lattice beams ( and hence @xmath31 )
were kept fixed fixed at @xmath41 and @xmath50 , respectively , and the splitting - angle @xmath3 was varied .
this resulted in a change in the lattice constants in the @xmath12- and @xmath13-directions . in our setup
, @xmath3 could be varied between @xmath51 and @xmath52 degrees , which according to equations [ eqn : omxy ] and [ eqn : omz ] should result in a large variation in @xmath53 , but only a small variation in @xmath54 .
figure [ fig : geoposw ] ( a ) shows the results of the measurements of @xmath55 as a function of @xmath56 .
as expected , there is a linear relationship , and again the experimentally measured frequencies lie about 20 percent below the theoretical values . +
the measurements of the raman widths as a function of geometry are summarized in fig .
[ fig : geoposw ] ( b ) , where @xmath57 is plotted against @xmath55 . as before , there is a linear relationship between the two , with @xmath58 . within the experimental errors
, this agrees with the empirical constant of proportionality found in section [ intdet ] .
in most of the work previously carried out on the excitation modes of atoms in optical lattices , the atoms were allowed to reach equilibrium in the potential wells before measurements were performed . to demonstrate that we can use the coherent transient technique for time - resolved , high - resolution spectroscopic studies
, we loaded our lattice directly from the mot , skipping the molasses phase .
the temperature of the atoms released from the mot was about @xmath26 .
the lattice was then switched on for a variable amount of time , after which either the temperature was measured ( using the time - of - flight method @xcite ) , or a coherent transient spectrum was recorded in about @xmath59 .
figure [ fig : equil ] shows the results of the corresponding temperature and vibrational frequency measurements .
the temperature of the atoms reaches its steady - state value governed by the lattice parameters in a few milliseconds @xcite . at the same time , the measured vibrational frequency of the atoms increases as the atoms are cooled and then approaches as steady - state value .
these results can alternatively be visualized by plotting the measured vibrational frequency against temperature ( fig .
[ fig : postemp ] ) .
evidently , the colder the atoms are , the higher is the observed vibrational frequency .
this can be understood in terms of the anharmonicity of the lattice potential , as will be explained in more detail in section [ anharmo ] .
briefly , the hotter the atoms are , the larger the population of higher - lying ( and more anharmonically shifted ) vibrational states in the potential wells will be , which results in a lower observed vibrational frequency .
as mentioned before , whereas the dependence of the vibrational frequencies on the lattice parameters is well understood and can be theoretically modelled , the widths of the raman peaks of the spectra have so far resisted a comprehensive explanation . in this section
, we shall summarize the models proposed so far and discuss them in the light of our findings . also , we give a brief overview of the results obtained by other groups and compare them to our own experimental results .
+ the first rate to consider when dealing with the lifetimes of vibrational levels in optical lattices and coherences between those levels is the photon scattering rate @xmath60 . after the first observation of distinct raman resonances in probe - absorption spectra
it was soon realized that the lifetimes could not be determined by @xmath61 since that would have meant that ( for most experimental parameters ) the resonances would not have been resolvable .
the experimentally observed widths , however , were more than an order of magnitude smaller than those predicted by the photon scattering rate .
et al . _
@xcite proposed the lamb - dicke effect @xcite arising from sub - wavelength localization of the atoms as an explanation for the unexpected narrowness of the resonances @xcite . according to this theory , the rate of inelastic scattering events in which the vibrational quantum number changes is reduced by the lamb - dicke factor @xmath62 in favour of elastic scattering events involving no change in vibrational level . for typical parameters ,
the lamb - dicke factor is of the order of @xmath63 or less , so that the inelastic scattering rate @xmath64 comes close to the experimentally observed widths . in experiments involving shallow potentials with small lamb - dicke parameters , however , kozuma _ et al . _ still found lifetimes that were much longer than predicted by the lamb - dicke model .
_ interpreted this as evidence for the existence of the transfer in inelastic scattering processes of coherence between different pairs of vibrational levels .
also , results of other experiments suggested that the broadening of the vibrational raman resonances was dominated by the anharmonicity of the lattice potential , which was further demonstrated through the observation of revivals in wavepacket oscillations . in the following
, we will show that our experiments support that interpretation . before discussing the experimental findings in detail ,
however , we shall give a brief outline of the effects of anharmonicity in optical lattices .
when deriving the expressions for the vibrational frequencies of atoms in optical lattices ( equations [ eqn : omxy ] and [ eqn : omz ] ) , the true light - shift potentials are often approximated by harmonic potentials characterized by the curvature of the lattice potentials close to a local minimum .
the effect of the anharmonicity of the potential can be treated perturbatively by considering the correction due to the fourth - order term in the expansion of the potential in first - order perturbation theory .
taking into account the geometrical factor @xmath56 characterizing the lattice constant in the @xmath12-direction , this leads to the vibrational frequency @xmath65 associated with a pair of adjacent vibrational levels characterized by @xmath66 and @xmath67 being reduced with respect to the harmonic frequency @xmath68 so that @xmath69 assuming that in an optical lattice the vibrational levels are populated according to a boltzmann distribution @xcite , and that the ratio @xmath70 is a constant ( as has been shown experimentally @xcite ) , the effect of the anharmonicity on the experimentally observed raman resonances can be estimated .
firstly , the observed vibrational frequency can be defined as an average over all the anharmonically shifted pairs of vibrational levels , weighted by their respective boltzmann factors , leading to the following expression @xcite : @xmath71 this indicates that all the vibrational frequencies should be reduced by a constant fraction , and that that fraction will be proportional to the temperature @xmath72 of the atoms in the lattice for a constant well - depth @xmath31 .
this satisfactorily accounts for our observations presented in section [ equili ] .
+ likewise , the broadening of the resonance due to the presence of several vibrational frequencies can be modelled by assuming that only levels with a boltzmann factor ( relative to that for @xmath73 ) larger than some small , arbitrary number @xmath74 contribute appreciably to the observed spectrum . the most shifted vibrational frequency , which is a measure of the frequency spread ,
is then given by @xmath75 this is proportional to the vibrational frequency @xmath68 for any value of @xmath74 , so that we would expect that the true linewidth , calculated from a properly weighted sum over all vibrational levels , also to be proportional to @xmath68 .
this is precisely what we observe in our experiments .
figure [ fig : geoex ] illustrates the dependence of the vibrational frequency and width of the raman peak on the geometrical factor @xmath56 . in a more quantitative treatment , we have performed numerical simulations of the coherent transient spectra taking into account the frequency spread by superposing the contributions of all possible pairs of vibrational levels .
the results exhibit the same characteristics as the qualitative models described above .
+ in principle , the anharmonicity of the potential should also manifest itself in an asymmetry of the raman peak .
one expects a slower drop - off of the raman peak on the low frequency side because the anharmonic shifts are all in the same direction ( see equation [ anhshifts ] ) .
this effect , however , will only be noticeable if the frequency spread due to anharmonicity is appreciable compared to the widths of the individual contributions , which are caused by the dissipative mechanisms described above . in our experiments
, we have not found any clear evidence of an asymmetry of the raman peaks . in fig .
[ fig : geoex ] , however , the lorentzian fit to the broader raman peak ( b ) has a larger @xmath76-value then the narrower peak ( a ) , which suggests that the former is distorted by anharmonicity .
one of the main objectives of our experiments was to determine the dominant mechanism involved in determining the width of the raman resonances in the coherent transient spectra .
in particular , we were interested in establishing whether the lamb - dicke model correctly predicted the dependence of the widths on all the experimental parameters . since our experiment was performed on a 3-d lattice , whereas the lamb - dicke treatment of courtois _ et al .
_ is based on calculations in one dimension , we have generalized that derivation to take into account the dependence of the lattice constant on the geometry of the lattice .
this leads to the appearance of an extra geometrical factor @xmath56 in the expression for the rate @xmath77 of inelastic raman processes leading to a change in the external state of the atom along the @xmath12-direction from an initial state with vibrational quantum number , @xmath66 , which then becomes @xmath78 this reproduces the observed linear dependence of @xmath79 on @xmath80 when the splitting - angle @xmath3 is varied ( see section [ geovar ] ) .
it also turns out that when the lattice intensity is varied , the expression for @xmath81 reduces to @xmath82 .
when @xmath31 is varied by changing the detuning , however , the relationship between @xmath80 and @xmath81 ( and hence @xmath79 ) is expected to be of the form @xmath83 ( for large detuning ) , which is not at all what we observe .
in fact , upon closer inspection the values of @xmath79 for identical @xmath80 but different detuning do differ slightly from each other , the value of @xmath79 for the larger detuning being smaller , as expected .
the difference , however , is very small and does not reflect the predicted @xmath84-dependence .
this is perhaps only to be expected because , for the case of the largest detunings used in these experiments , the proximity of the neighbouring @xmath85 transition means that our approximation of an isolated transition is of dubious validity .
nevertheless , we find that the observed dependence of @xmath79 on @xmath80 agrees very well with broadening due to anharmonicity , as derived above .
these observations suggest that the main broadening mechanism is , indeed , the anharmonicity of the potential , and the dissipative mechanisms based on inelastic photon scattering are strongly suppressed . in the harmonic limit ,
where the homogeneous width is small compared to the anharmonic frequency difference between successive vibrational levels , the suppression of line - broadening because of the transfer of vibrational coherence @xcite between pairs of levels in spontaneous raman processes is expected to be appreciable .
the lifetimes of the vibrational coherences are then determined by the cooling rate , rather than by the inelastic photon scattering rate , which in the oscillating regime is given by @xmath86 @xcite and , in our experiments varies between @xmath87 for small detuning to less than @xmath88 for the largest detuning employed . for large detuning ,
the associated width is given by @xmath89 .
conversely , in the anharmonic limit , it is the spatial localization that is expected to give rise to the suppression of inelastic raman processes . clearly , the interplay between the transfer of coherence , the lamb - dicke narrowing and the line broadening mechanisms is hard to model quantitatively for an intermediate case .
nevertheless , our results strongly suggest that the broadening associated with the anharmonicity generally dominates the widths of the raman peaks , with dissipative mechanisms contributing at a much reduced level .
we used the method of coherent transients in order to measure the vibrational frequencies of atoms in optical lattices and the widths associated with the corresponding raman peaks of the coherent transient spectra . varying the intensity , detuning , and geometry of the lattices beam
, we confirmed the validity of the theoretical predictions for the vibrational frequencies based on a harmonic oscillator model of the potential wells .
our measurements of the raman widths as a function of intensity , detuning , and geometry all yielded a linear dependence on the vibrational frequencies , which strongly suggests that anharmonicity is the main broadening mechanism .
this implies that the decay of the coherence between adjacent vibrational levels is suppressed , which we ascribe to the combined effects of the transfer of coherence between pairs of vibrational levels in inelastic raman processes and the suppression of the inelastic rate arising from the sub - optical wavelength spatial localization .
the anharmonicity was also found to be responsible for a reduction in the observed vibrational frequencies relative to the predictions of a harmonic model , which was tested by measuring the vibrational frequencies of ensembles of atoms with different temperatures in otherwise identical optical lattices .
we would like to thank d. lucas , g. birkl , and m. oberthaler for stimulating discussions and d. smith for help with the electronics .
this work was supported by the epsrc ( uk ) .
o.m . gratefully acknowledges support by the studienstiftung des deutschen volkes .
ourpaper current address : dipartimento di fisica , universita di pisa , via buonarroti 2 , i-56127 pisa , italy . for reviews of optical lattices ,
jessen and i. h. deutsch , adv . at .
* 37 * 95 ( 1996 ) ; d.r .
meacher , contemp.phys . * 39*(5 ) , 329 ( 1998 ) ; l. guidoni and p. verkerk , j. opt .
b * 1 * r23 ( 1999 ) . g. grynberg , b. lounis , p. verkerk , j .- y .
courtois , and c. salomon , phys.rev.lett .
* 70 * , 2249 ( 1993 ) p.s .
jessen , c. gerz , p.d .
lett , w.d .
phillips , s.l .
rolston , r.j.c .
spreeuw , and c.i .
westbrook , phys.rev.lett . *
69 * , 49 ( 1992 ) m. kozuma , y. imai , k. nakagawa , and m. ohtsu , phys.rev.a . *
52 * , r3421 ( 1995 ) c. jurczak , b.desruelle , k. sengstock , j .- y .
courtois , c.i .
westbrook , and a. aspect , phys.rev.lett . *
77 * , 1727 ( 1996 ) c.i .
westbrook , c. jurczak , g. birkl , b. desruelle , w.d .
phillips , and a. aspect , j.mod.opt . * 44 * , 1837 ( 1997 ) g. raithel , w.d .
phillips , and s.l .
rolston , phys .
* 81 * , 3615 ( 1998 ) c. trich , l. guidoni , p. verkerk , and g. grynberg , in osa tops on ultracold atoms and bec , 1996 , vol . 7 , ed .
k. burnett ( 1997 ) , 82 . c. trich , ph.d .
thesis , paris 1997 ( unpublished ) c. mennerat - robilliard , l. guidoni , j .- y . courtois and g.
grynberg , europhys . lett . * 38 * , 429 ( 1997 ) e.l .
raab , m. prentiss , a. cable , s. chu , and d.e .
pritchard , phys.rev.lett .
* 59 * , 2631 ( 1987 ) k.i .
petsas , a.b .
coates , and g. grynberg , phys.rev.a * 50 * , 5173 ( 1994 ) p. verkerk , d.r .
meacher , a.b .
coates , j.y .
courtois , s. guibal , b. lounis , c. salomon , and g. grynberg , europhys.lett .
* 26 * , 171 ( 1994 ) m. gatzke , g. birkl , p.s .
jessen , a. kastberg , s.l .
rolston , and w.d .
phillips , phys.rev.a . *
55 * , r3987 ( 1997 ) j .- y .
courtois and g. grynberg , phys.rev.a .
* 46 * , 7060 ( 1992 ) g. grynberg and c. trich , in proceedings of the international school of physics enrico fermi , course cxxxi ( 1995 ) , ed .
a. aspect , w. barletta , and r. bonifacio , ios press , amsterdam 1996 w.d .
phillips and c.i .
westbrook , phys.rev.lett . * 78 * , 2676 ( 1997 ) the errors on @xmath80 and @xmath79 stem from the uncertainty in the fitting of the lorentzian to the spectra .
the smaller the detuning , the greater the uncertainty due to the noise on the signal . the uncertainty in the intensity measurements was about 10 percent . c. salomon , j. dalibard , w.d .
phillips , a. clairon , and s. guellati , europhys . lett . * 12*,683 ( 1990 ) as in @xcite , we found a linear dependence of @xmath72 on the potential well - depth @xmath31 .
dicke , phys . rev . * 89 * , 472 ( 1953 ) this result is already contained implicitly in an earlier paper by castin _
et al . _
yvan castin , ph.d .
thesis , universit paris vi , 1992 ( unpublished ) c. cohen - tannoudji , j. dupont - roc , and g. grynberg , 1992 , _ atom - photon interactions _
( new york : wiley ) y. castin and j. dalibard , europhys .
* 14 * , 761 ( 1991 ) | we use a method based on optical coherent transients to study the vibrational coherence lifetimes of atoms trapped in the potential wells of a near - resonant optical lattice in the oscillating regime .
the dependence of the positions and widths of the vibrational raman resonances of the coherent transient spectra on the intensity , detuning , and geometry of the lattice beams is investigated and the results are compared with theoretical predictions . it is shown that the principal source of broadening of the vibrational raman resonances is the anharmonicity of the light - shift potential wells . we also perform time - resolved measurements of the vibrational excitation of atoms equilibrating in an optical lattice . |
the fractional quantum hall ( fqh ) states have long fascinated the condensed - matter community due to their remarkable transport properties and the exotic nature of their quasiparticle excitations .
it is in the context of fqh states that the notion of topological order in gapped two - dimensional states first arose .
@xcite recently there has been enhanced interest in fqh states with _ non - abelian statistics _
, @xcite due to the possibility of implementing quantum computation schemes topologically protected from decoherence .
@xcite the unusual features of fqh states have been notoriously difficult to characterize using traditional condensed - matter concepts such as local order parameters and @xmath1-point correlation functions . in a separate development , recent years have seen growing understanding that entanglement measures borrowed from the discipline of quantum information can be useful in probing global features of quantum many - particle states . @xcite it is thus natural to ask what features of fqh states can be characterized by entanglement measures . in a recent short report,@xcite
three of the present authors have shown that one such entanglement measure , the _
bipartite entanglement entropy _ , indeed elucidates the subtle correlations and topological order in the simplest fqh states , those in the so - called laughlin sequence . the bipartite entanglement entropy is defined by partitioning the system under question into two blocks @xmath2 and @xmath3 , and using the reduced density matrix of one part ( e.g. , @xmath4 obtained by tracing over @xmath3 degrees of freedom ) to calculate the von neumann entropy @xmath5 $ ] . in ref .
, numerical calculations of the entanglement entropy between two spatial regions allowed us to extract from the laughlin wavefunctions the so - called _ topological entanglement entropy _
, a concept introduced in ref . .
[ for brevity we write _ topological entropy _ where no confusion can arise . ] in addition , we provided results on the entanglement entropy between subsets of the particles making up the state .
we showed that such _ particle entanglements _ are bounded by expressions that manifest the exclusion statistics in the laughlin states . in this article , we present a systematic discussion of bipartite entanglement entropies for fqh states , elaborating on our results in ref . .
in addition to the abelian laughlin ( l ) states , we consider a series of non - abelian fqh states : the moore - read ( mr ) ( or pfaffian ) states .
@xcite in planar geometry , the respective wavefunctions are given by @xmath6 with @xmath7 denoting the antisymmetric pfaffian symbol .
we shall here consider these same states in spherical geometry , so as to eliminate boundary effects . for both series of states ,
we derive upper bounds @xmath8 for particle entanglement entropies .
a marked difference between the @xmath9 laughlin state and the @xmath10 moore - read states is that in the latter the leading correlations have a 3-body nature , whereas those in the laughlin states are 2-body effects .
this difference is nicely manifested in the leading terms of a @xmath11 expansion of the upper bounds @xmath8 , which are given by ( see section iii below for details ) @xmath12 & & m=2\ { \rm moore - read}\ { \rm state : } \nonumber \\ & & \qquad s_a^{\rm bound } = s^f_a - \frac{3}{4n^2 } n_a(n_a-1)(n_a-2 ) + \ldots\end{aligned}\ ] ] another marked difference between the abelian and non - abelian states is in the value for the topological entropy @xmath0 . comparing the laughlin and moore - read states at the same filling fraction @xmath13 we have @xmath14 the difference being due to the non - abelian nature of the moore - read states ( see section iv for details ) . in this article
we extract values for @xmath0 directly from wavefunctions for a limited number of particles in spherical geometry ( up to @xmath15 for the @xmath9 laughlin state and up to @xmath16 for the @xmath10 moore - read state ) , finding values that are consistent with the expected result .
these results illustrate how the entanglement entropy can be used in diagnosing the topological order for a fqh state that is only known in the form of wavefunctions for a limited number of particles , as is often the case in numerical studies .
in section [ sec_partition - choices ] we give a general discussion of possible bipartite entanglement measures in itinerant many body systems , carefully distinguishing between _ particle _ and _ spatial _ partitioning schemes . in section [ sec_pcle - partition ]
we present analytical and numerical results for particle entanglement in the laughlin and moore - read fqh states . in this , the eigenvalue distribution of the reduced density matrix @xmath17 plays a central role . in subsection [ sec_pcle - rdm - n - corrln - fn ]
we relate the eigenvalue distribution for @xmath18 to the two - particle correlation function @xmath19 . in section [ sec_spatial - n - gamma ]
we discuss spatial partitioning , paying particular attention to the numerical procedure followed in extracting the topological entropy @xmath0 . while most results in this article are for fermionic fqh states ( meaning @xmath20 odd in the laughlin sequence and @xmath20 even for the moore - read states ) , we briefly comment on bosonic states in subsection [ sec_bosonic ] .
the entanglement entropy , being a _ bipartite _ measure of entanglement , depends on the particular partitions being considered .
obviously , a many - particle system can be partitioned in many ways . rather than asking
which partition is the `` correct '' one , we find it more useful to ask what information one can extract from various kinds of partitioning .
accordingly , we have partitioned both the spatial degrees of freedom , and the particles themselves .
we find that both schemes are useful , for revealing distinct features of the many - particle state .
the kinds of partitioning one is able to study depend on the available degrees of freedom .
our calculations are all performed for fqh states in a spherical geometry .
@xcite in this representation the fermions are placed on a sphere containing a magnetic monopole .
the magnetic orbitals of the relevant landau level are then represented as angular momentum orbitals ; the total angular momentum is half the number of flux quanta , @xmath21 .
the @xmath22 orbitals are labeled either @xmath23 to @xmath24 or @xmath25 to @xmath26 . for @xmath27 particles at fractional filling @xmath13
, one finds the interesting fqh states for @xmath28 , where @xmath29 is a finite - size shift .
the laughlin states appear at @xmath30 while for the moore - read states @xmath31 .
the `` filling '' acquires the usual meaning @xmath32 only in the thermodynamic limit .
the orbitals are each localized around a `` circle of latitude '' on the sphere , with the @xmath23 orbital localized near one `` pole . '' since the fqh wavefunctions on a sphere
are obtained in terms of orbital occupancies , one can either partition orbitals or partition particles . because of the spatial arrangement of the orbitals , partitioning orbitals is in fact equivalent to partitioning spatial regions .
the difference between spatial and particle partitioning has not been stressed in the literature because the most common systems studied ( in the context of entanglement entropies in many - particle states ) are spin models , for which there is no such distinction . in the cases of itinerant particles where there is a difference , the common default scheme has been spatial partitioning .
in particular , conformal field theory results on entanglement scaling @xcite and the distinction between gapless and gapped states observed in entanglement scaling @xcite actually pertain to the blocking of _ space _ rather than the particles or spins themselves .
the definition of the topological entropy for two - dimensional topologically ordered states is also based on the entanglement entropy between spatial blocks .
@xcite in previous work , @xcite three of the present authors studied the entanglement entropy between subsets of particles making up a laughlin fqh state .
we presented upper bounds and gave an interpretation in terms of exclusion statistics . in this paper
we extend these results to the moore - read states , where the exclusion effects are more intricate .
we refer to refs . for other studies of particle entanglement properties . for orbital or spatial partitioning ,
we define block @xmath2 to be the first @xmath33 orbitals , extending spatially from one pole of the sphere out to some latitude . in the thermodynamic limit , this is equivalent to choosing a disk - shaped block @xmath2 within an infinite planar system . in this limit , since each orbital @xmath34 is associated with a wavefunction of the form @xmath35 in usual complex coordinate language , a disk with @xmath33 orbitals has radius @xmath36 .
the spatial arrangement of the orbitals constrains us to either disc or ring - shaped spatial regions as the @xmath2 partition .
since the orbital indexes do not give us access to the full two - dimensional degrees of freedom , it is not possible to experiment with the various kinds of topologically nontrivial partitions suggested by preskill and kitaev , @xcite levin and wen , @xcite and furukawa and misguich .
@xcite however , as we have reported previously , @xcite the spherical geometry is sufficient to probe the topological entropy of fqh wavefunctions . comparing particle and spatial entanglement entropies
, we remark that the effect of correlations is opposite between the two cases . for particle
partitioning , the maximal entropy is realized for uncorrelated fermions ; correlations tend to lower @xmath37 from the fermion bound @xmath38 .
spatial entanglement , on the other hand , is entirely due to correlations .
the point is illustrated by considering the @xmath39 laughlin state , where the fermions are uncorrelated . here
the particle entanglement entropy equals @xmath38 , while the spatial entanglement entropy vanishes .
in this section we provide close upper bounds to the entropy of entanglement between @xmath40 particles of the state and the remaining @xmath41 particles .
we also discuss the @xmath40-particle reduced density matrices @xmath17 that arise in this context . for fqh states on a sphere ,
the @xmath40-particle reduced density matrices @xmath17 commute with the total angular momentum operators @xmath42 and @xmath43 of the selected @xmath40 particles .
this implies that the eigenvalues of @xmath17 are organized in a multiplet structure of the corresponding @xmath44 algebra : an eigenvalue for total angular momentum @xmath45 will be @xmath46-fold degenerate . for @xmath47 fermions , each having angular momentum @xmath48 ,
the 2-particle states have total angular momenta @xmath49 , @xmath50 , @xmath51 , @xmath52 , for @xmath48 integer ( half - integer ) , giving a total number of @xmath53 states . a naive upper bound to the entanglement entropy
is thus @xmath54 .\ ] ] inspecting the explicit structure of the fermionic laughlin states with @xmath55 , one finds that the eigenvalues corresponding to 2-particle states with @xmath49 , @xmath50 , @xmath51 , @xmath56 all vanish .
the reason is that the correlations in the laughlin states are such that particles can not come too close together .
for example , if a first fermion occupies the @xmath23 orbital , localized near the north pole , the laughlin wavefunction has zero amplitude for finding a second fermion in orbitals @xmath57 , @xmath58 , @xmath51 , @xmath59 .
the highest possible value of the angular momentum of the two fermions combined is thus @xmath60 .
the remaining number of non - zero eigenvalues is @xmath61 , leading to an improved bound on the entropy @xmath62 @xmath63\ ] ] with @xmath64 as before . for @xmath65 , the multiplet structures are more complicated and we need to resort to a different method for finding a non - trivial upper bound to the particle entropy . in the next subsection
we give a general derivation for both the laughlin and the moore - read series of fermionic fqh states . for @xmath27 fermionic particles , @xmath40 particles in the @xmath2 block , and the total number of orbitals given by @xmath66 ,
fermionic statistics lead to an obvious upper limit @xmath67 to the entropy @xmath37 @xmath68 in the fqh states the correlations are such that the particles avoid each other and the entropy is further reduced . to obtain a handle on this , one may reason as follows .
the model fqh states in the laughlin and moore - read series can be characterized as zero - energy eigenstates of a hamiltonian penalizing pairs and/or triplets of particles coming to the same position . after tracing out the coordinates for the @xmath3 set ,
the dependence on those in the @xmath2 set is such that one still has a zero - energy eigenstate .
however , the number of orbitals available to the @xmath2 particles is larger than what is needed to make the model fqh state in the @xmath2 sector , and one instead has a certain number of quasi - holes on top of the @xmath2 set model state .
the total ground state degeneracy for this situation has been studied in the literature : see ref . for the laughlin and moore - read states and ref . for the read - rezayi and ardonne - schoutens series of non - abelian fqh states . for
the laughlin states the details are as follows .
the @xmath27-particle laughlin state is realized on a total of @xmath69 landau orbitals , corresponding to @xmath70 flux quanta .
the laughlin state for @xmath40 particles would need @xmath71 flux quanta ; we thus have an excess flux of @xmath72 . with the laughlin gauge argument this corresponds to the presence of @xmath73 quasi - holes over the groundstate .
according to ref .
each of the quasi - holes has a number of @xmath74 effective orbitals to choose from , with bosonic counting rules ( meaning that two or more quasi - holes can be in the same effective orbital ) .
this gives a number of quasi - hole states equal to @xmath75 leading to the following upper bound to the entropy @xmath37 @xmath76 we remark that this expression has a clear interpretation in terms of exclusion statistics : the counting factor in eq .
( [ laughlinbound ] ) gives the number of ways @xmath40 particles can be placed in @xmath69 orbitals , in such a way that a particle placed in a given orbital @xmath34 excludes particles from orbitals @xmath77 with @xmath78 . in a @xmath11 expansion
we find ( assuming @xmath79 ) @xmath80 & & \frac{1}{n } \frac{m-1}{m } n_a ( n_a-1 ) \nonumber \\[2 mm ] & & + \frac{1}{n^2 } \frac{m-1}{2 m^2 } n_a ( n_a-1 ) [ 2 m + ( n_a-1)(m+n_a-4 ) ] \nonumber \\[2 mm ] & & + \mathcal{o}(1/n^3 ) \label{laughlin1overn}\end{aligned}\ ] ] the particle entropy reaches a maximum for @xmath81 . for this case our eq .
( [ laughlinbound ] ) gives , in the limit of large @xmath27 , @xmath82/2 \ .\ ] ] this bound is sharper than a bound recently presented in ref .
, which gives a larger coefficient for the linear - in-@xmath27 behavior . for the fermionic moore - read states at @xmath13 , with @xmath83
, we can reason in a similar way , with now @xmath84 .
as for the laughlin states we have an excess flux of @xmath85 but now the number of quasi - holes is twice this number due to the fact that the fundamental quasi - holes correspond to half a flux quantum .
thus , @xmath86 .
we now take from ref .
the following result for the total quasi - hole degeneracy @xmath87 this gives us an upper bound @xmath88 as before .
putting @xmath10 , one easily checks that @xmath88 coincides with @xmath89 for @xmath47 . in a @xmath11 expansion ,
the leading deviation from @xmath38 is a 3-body term at order @xmath90 , @xmath91 this result nicely illustrates the fact that the leading correlations in the @xmath10 moore - read state have a 3-body character : the wave - function vanishes if at least three particles come to the same position .
for @xmath92 the leading correlations do have a 2-body character , as for the laughlin states , @xmath93 inspecting the particle entanglement at @xmath81 and for @xmath27 large , our bound implies that for the @xmath10 moore - read state @xmath94 this bound is reduced from the fermi bound @xmath95 , but it is larger than the bound for the @xmath10 ( bosonic ) laughlin state , which has asymptotic form @xmath96 .
this indicates that , at equal filling @xmath97 , the particles in a moore - read state are more entangled than those in a laughlin state .
the quasi - hole counting rules for the order-@xmath98 clustered spin - polarized ( read - rezayi ) and spin - singlet ( ardonne - schoutens ) states are all known in the literature . @xcite they can be used to generalize the upper bounds on particle entanglement entropy given in this subsection to these more intricate non - abelian fqh states .
we briefly comment on the case of bosonic fqh states .
the realization that a rapidly rotating bose gas may eventually enter a regime of bosonic quantum hall states motivates the theoretical study of the effects of bosonic statistics .
we consider bosonic laughlin states at filling fraction @xmath99 with @xmath83 .
the naive upper bound to the the entropy associated to placing @xmath40 bosons in @xmath69 orbitals is @xmath100 the expression for @xmath88 remains unchanged , giving the following leading correction in a @xmath11 expansion @xmath101 for a bosonic moore - read state , with filling fraction @xmath13 with @xmath102 ,
the leading @xmath11 correction becomes @xmath103 in the case @xmath39 the leading correlations have 3-body character , leading to the vanishing of the leading @xmath11 correction . in deriving the upper bound @xmath8 we relied on the fact that a certain number of eigenvalues of the reduced density matrix vanish .
the bounds would be exact if all non - zero eigenvalues were equal , but since they are not the bounds overestimate the actual values for the entropies .
[ fig_2-pcle_eigen ] plots the eigenvalues for the @xmath47-particle reduced density matrix for @xmath104 particles on a sphere in the @xmath9 laughlin state , for which the single particle angular momentum is @xmath105 .
the horizontal axis represents the degeneracy @xmath106 of the eigenvalues , in descending order .
the eigenvalue at @xmath107 , with degeneracy 47 , vanishes ; the non - zero eigenvalues show some scatter around an asymptotic value . due to this scatter
the entropy @xmath108 is somewhat lower than the upper bound @xmath109 .
( color online ) eigenvalues for the 2-particle reduced density matrix , plotted against their multiplicities , for @xmath104 particles in the @xmath9 laughlin state . ] an important difference between the @xmath9 laughlin and the @xmath10 moore - read states is the absence of vanishing eigenvalues for the 2-particle reduced density matrix . the eigenvalue distribution shown in fig .
[ fig_2-pcle_pf_eigen ] illustrates this point .
( color online ) eigenvalues for the 2-particle reduced density matrix , plotted against their multiplicities , for @xmath110 particles in the @xmath10 moore - read state . ] in the @xmath10 moore - read state , there are vanishing eigenvalues in the reduced density matrix of @xmath111 particles .
the number of nonzero eigenvalues predicted by eq .
( [ mrbound ] ) agrees with numerical results .
for example , for @xmath112 and @xmath15 particles there are 770 nonvanishing eigenvalues , in agreement with eq .
( [ mrbound ] ) . in figs .
[ fig_pcle_lghln_23 ] , [ fig_pcle_pf_23 ] we compare numerically computed particle entanglement entropies with the bounds derived above .
( color online ) entanglement entropy for @xmath47 and @xmath112 particles for the @xmath9 laughlin state .
dots are numerical exact values , the dotted line represents @xmath38 and the solid curve is the bound @xmath8 . ]
( color online ) entanglement entropy for @xmath47 and @xmath112 particles for the @xmath10 moore - read state .
dots are numerical exact values , the dotted line represents @xmath38 and the solid curve is the bound @xmath8 . ] it is interesting to consider in some detail the deviation between the bounds @xmath8 and the actual entropies computed numerically . as mentioned above , this deviation arises from the fact that the non - zero eigenvalues of the reduced density matrices are not all equal . to estimate the effect on @xmath113 $ ] of the spread in the non - zero eigenvalues , we do a rough modeling of the eigenvalue distribution ( fig.[fig_2-pcle_eigen ] ) of @xmath18 for the @xmath9 laughlin state . for this case
the number @xmath114 of non - zero eigenvalues is @xmath115 . if these nonzero eigenvalues were all equal ( to @xmath116 ) , the entanglement entropy would have the maximum value @xmath117 , which is the predicted upper bound ( [ laughlinbound ] ) .
we now take into account the deviations from @xmath116 , guided by fig .
[ fig_2-pcle_eigen ] , with the following toy distribution : we take @xmath118 out of @xmath114 of the eigenvalues to be equal to @xmath119 , with @xmath120 , while the rest of the eigenvalues are at value @xmath121 , @xmath122 , such that the sum of eigenvalues is unity . assuming @xmath123 leads to @xmath124 guided by the eigenvalue distributions in fig .
[ fig_2-pcle_eigen ] , we assume that @xmath125 is of order @xmath11 ; for concreteness we put @xmath126 equal to the multiplicity of the largest eigenvalue , which is @xmath127 . taking @xmath128 between 1.2 and 1.5 ( as observed for the largest available laughlin wavefunctions )
gives a @xmath11 correction in the entropy with coefficient in the range 0.03 0.14 .
fitting the difference @xmath129 to a form @xmath130 gives a coefficient @xmath131 .
the vanishing eigenvalues account for @xmath132 , see eq .
( [ laughlin1overn ] ) , and we see that the remaining difference @xmath133 is consistent with the @xmath11 correction due to the spread in the non - zero eigenvalues .
we made similar estimates for a the @xmath10 moore - read state with up to @xmath110 particles , where the eigenvalues are all non - zero and @xmath8 agrees with @xmath38 . in this case the deviation between data and bound show a @xmath11 dependence with a coefficient of about 0.14 .
these considerations are of some general interest , as they make the point that a @xmath11 expansion of particle entanglement entropies are indicative of correlations in a many - body state . in the concrete case studied here ,
the sizeable value of the leading @xmath11 correction in the laughlin state indicates strong 2-body correlations , while the small value for the moore - read state indicates the absence of such correlations .
since the @xmath40-particle reduced density matrices @xmath17 are obtained by integrating out all but @xmath40 of the particles , one expects these matrices to be related to the @xmath40-particle correlation functions . in this subsection
we study this relation for the case @xmath47 .
in particular , the eigenvalue distributions of @xmath18 in figs .
[ fig_2-pcle_eigen ] , [ fig_2-pcle_pf_eigen ] , although discrete , are reminiscent of the well - known two - particle correlation functions @xmath19 for laughlin and moore - read states .
we will show that the eigenvalue distributions are in fact very closely related to the correlation functions the eigenvalue distribution function is a kind of discretized version of @xmath19 .
the two - particle correlation function @xmath19 is conventionally defined as @xmath134 where @xmath27 is the number of particles and @xmath1 is a density , which is chosen such that @xmath135 .
we express the 2-particle reduced density matrix @xmath136 on a sphere in a basis of polar spherical coordinates .
because of the rotational symmetry it should be a function of the angular distance @xmath137 between the two particles . in appendix [ eigenvalues_correlations ] we show that @xmath138 can be written in the form @xmath139 in this expression , @xmath140 is the eigenvalue with multiplicity @xmath141 , corresponding to the total angular momentum of the two particles equal to @xmath34 .
the functions @xmath142 are explicitly given in eq .
( [ rtheta ] ) .
( color online ) two - particle correlation as a function of distance in units of the magnetic length , for the @xmath9 laughlin state with @xmath143 particles and for the @xmath10 moore - read state with @xmath110 . ] since the distance @xmath144 between two particles is simply equal to @xmath145 , with @xmath146 the radius of the sphere , the 2-particle correlation function @xmath19 is directly proportional to @xmath138 . through eq .
( [ rho2_theta ] ) it is expressed as a transfrom from @xmath34 space to @xmath137 space , with basis functions @xmath142 . in fig .
[ fig_g2_lghln_pf ] , we show some curves for @xmath138 ; they agree with known results . as illustrated in the inset to fig .
[ fig_g2_lghln_pf ] , the basis functions @xmath142 have a peak structure , with the position of the peak depending on the total angular momentum @xmath34 .
large values of @xmath34 correspond to small angular distances and vice versa .
this is easy to understand from the following classical picture .
when two particles with angular momenta @xmath48 have total angular momentum @xmath147 , the corresponding vectors @xmath148 and @xmath149 should point a the same direction .
the angular momentum vector points at the position of the particle on a sphere , therefore two particle should be close to each other . on the other hand , if total angular momentum is zero then the angular momentum vectors should point into opposite directions .
this means that particles are placed at the opposite sides of the sphere .
the fact that the @xmath142 are localized functions , peaked at @xmath137 values monotonically decreasing with @xmath34 , indicates that @xmath19 curve is simply a continuous form of the @xmath140 versus descending-@xmath150 curves of figs .
[ fig_2-pcle_eigen ] and [ fig_2-pcle_pf_eigen ] .
the similarity between the discrete @xmath140 and the continuous @xmath19 curves is not accidental . at small distances @xmath151 .
for a laughlin state the lowest value of @xmath152 is @xmath20 , thus @xmath153 .
this behavior is a direct consequence of the vanishing of eigenvalues with the largest multiplicities . for the @xmath10 moore - read state
the lowest value of @xmath152 is @xmath154 because there are no vanishing eigenvalues .
therefore at small distances @xmath155 . of course , our observations on the 2-particle correlations agree with known results ; our main point has been to stress the intimate relation with the eigenvalue distribution of the 2-particle reduced density matrices .
we now turn to dividing the landau level orbitals into two blocks and calculating the entropy of entanglement between them . in a previous publication , @xcite we used this scheme to extract the topological entropy of the laughlin state . here
, we mainly focus on the ( @xmath10 ) moore - read state . after reviewing the topological entropy @xmath0 and the total quantum dimension @xmath156 , especially in the context of the moore - read state ( [ subsec_gamma - review ] ) , we detail some issues with taking the thermodynamic limit necessary for extracting @xmath0 from numerical data ( [ subsec_extrapoln - issues ] ) , and then present our numerical results ( [ subsec_numerical - gamma ] ) . we also present observations on the spectral structure of the reduced density matrices ( [ subsec_orbital - rdm - spectrum ] ) . for spatial partitioning of many - particle states ,
the general rule ( `` area law '' ) is that the entanglement entropy scales as the size of the boundary between the @xmath2 and @xmath3 blocks . @xcite
subtle information about the nature of the many - particle state can be provided by the presence or absence of logarithmic corrections , values of coefficients , or subleading terms in this basic relationship . for topologically ordered states in two dimensions ,
the following theorem has been presented recently @xcite concerning the scaling of entanglement entropy between spatial partitions . if @xmath48 is the length of the boundary between the two blocks , the entanglement entropy scales as @xmath157 .
as usual the scaling law applies to situations where @xmath2 is large and the total system is infinite .
the subleading term @xmath0 is called the _
topological entanglement entropy_. the striking result of refs .
has been that it can be expressed as the logarithm of a quantity @xmath156 known as the _ total quantum dimension _ of the topological field theory describing the topological order of the state .
the total quantum dimension is given by @xmath158 where the @xmath159 s are the quantum dimensions of the individual sectors making up the topological field theory .
these quantum dimensions are set by fusion rules of the fundamental anyons in the field theory .
the topological field theory for a @xmath13 laughlin state has a fundamental anyon ( of fractional charge @xmath160 ) , which generates @xmath20 abelian sectors . the quantum dimension @xmath159 is unity in all sectors so that @xmath161 for the @xmath13 laughlin state . for @xmath9
this gives @xmath162 . for states with non - abelian quasiparticles ,
the situation is more interesting because some anyon sectors contribute @xmath163 .
details for some examples have been provided in refs . .
in particular , for the @xmath10 moore - read state , there are six sectors ( two each of quasiparticles denoted by @xmath164 , @xmath165 , @xmath166 ) which contribute @xmath167 , @xmath168 , @xmath169 , leading to @xmath170 and @xmath171 .
the non - abelian nature shows up in the fact that @xmath0 is larger than @xmath172 , six being the degeneracy of the @xmath10 moore - read state on the torus . + a compact general expression for the total quantum dimension for a read - rezayi state with order-@xmath98 clustering and at filling fraction @xmath173 is ( see also ref . ) @xmath174= \frac{\sqrt{(k+2)(km+2)}}{2 \sin(\pi /(k+2 ) ) } \ . \label{scriptdrr}\ ] ] it includes the laughlin states ( @xmath175 , @xmath176 ) and the moore - read states ( @xmath177 , @xmath178 ) as special cases . for a general spin - singlet non - abelian fqh state with order @xmath98 clustering and filling fraction @xmath179 ,
the result is @xmath180= \frac{(k+3)\sqrt{(2km+3)}}{16 \cos(\pi/(k+3 ) ) \sin^3(\pi /(k+3 ) ) } \ , \label{scriptdas}\ ] ] giving @xmath181 for the paired spin - singlet state ( @xmath177 , @xmath182 ) at @xmath183 .
in general , the @xmath184-dependence of these expressions for total quantum dimensions is linked to the ground state degeneracy in torus geometry .
denoting the latter by @xmath185 $ ] we have the relation @xmath186= \mathcal{d}[k,0 ] \sqrt{\frac{\#[k , m]}{\#[k,0 ] } } \ , \ ] ] the conformal field theories underlying the states at @xmath187 are of wess - zumino - witten type ( @xmath188 for the rr states and @xmath189 for the as series ) and the quantities @xmath190 $ ] can be expressed in the modular @xmath29-matrix for these wzw models .
first of all , we note that , since our degrees of freedom are ordered essentially one - dimensionally , we can not use one of the two - dimensional schemes proposed previously @xcite in which an appropriate addition / subtraction of the entanglement entropies of several regions cancels the boundary parts of the entropy ( @xmath191 ) leaving the subleading term @xmath0 . with the orbital degrees of freedom on a sphere , we can choose only regions corresponding to disks , concentric rings , and combinations thereof .
any combination of entropies of disk- and ring - like regions that cancels out the boundary terms also unfortunately cancels out the @xmath0 term .
we are thus led to using directly the scaling law , @xmath192 .
our choice of block @xmath2 as the first @xmath33 orbitals , extending spatially from one pole out to some latitude , corresponds to a disk - shaped block only in the thermodynamic limit .
the block area is proportional to the square of @xmath193 while its boundary is proportional to @xmath194 ; these are equivalent only in the same @xmath195 limit . one way to numerically access
the thermodynamic limit is to take the entanglement entropy of @xmath33 orbitals with the rest , for accessible wavefunctions of various sizes @xmath27 , and then take the @xmath195 limit .
the @xmath196 versus @xmath193 points thus obtained should then follow a linear curve at large @xmath33 , whose vertical intercept gives the topological entropy .
results following this procedure were provided for the @xmath197 laughlin state in our earlier paper ; @xcite here we will focus on the moore - read state .
the extrapolation of @xmath198 values to the thermodynamic limit is a tricky issue .
we therefore discuss the extrapolation in some detail here , providing some general results .
we are interested in the function @xmath199 , where @xmath200 .
we have access to @xmath201 at several integer values of @xmath27 , and would like to estimate @xmath202 . for each dataset that we have access to ( each @xmath33 ; both laughlin and moore - read ) , we note the following : the @xmath201 versus @xmath203 values form a _ monotonic _ curve and this curve gets flatter ( slope magnitude decreases ) with decreasing @xmath204 .
two examples can be seen in the inset to fig .
[ fig_extrapolatn ] . in other words ,
the first and second derivatives of the @xmath199 function have the same sign and neither derivative changes sign . motivated by the above observations , we provide the following result . assuming only that the signs of the first two derivatives of the @xmath199 function are the same and that the signs remain unchanged until @xmath205
, we have : 1 .
the value @xmath206 corresponding to the smallest value @xmath207 of the available @xmath203 is a strict lower ( upper ) bound for @xmath202 if the @xmath208 is negative ( positive ) .
the intercept found by connecting the @xmath199 corresponding to the smallest two @xmath203 values ( @xmath207 , @xmath209 ) , namely @xmath210 is a strict upper ( lower ) bound if @xmath208 is negative ( positive ) and @xmath211 is positive ( negative ) .
the limits @xmath212 and @xmath213 thus obtained give us conservative bounds for the required entanglement entropies in the thermodynamic limit , @xmath196 . to obtain a sharper extrapolation
, one can use various polynomial extrapolations and take the average , as done in our earlier work .
@xcite here , we improve the extrapolation by using the extrapolation algorithm of bulirsh and stoer ( bst algorithm ) , based on rational polynomial fraction approximations .
@xcite .[table_bst-1 ] extrapolation using the bst algorithm , using three different @xmath214 values for the same initial dataset ( first column ) , corresponding to @xmath215 values for @xmath216 12,14,16 , 18 . in the last case
, @xmath214 has been tuned to ensure that the sets obtained after first and second iterations converge to the same value , i.e. , the lowest elements of the second and third columns are the same . [
cols="<,<,<,<",options="header " , ] in tables [ table_mr_entropies ] and [ table_laughlin_entropies ] we list orbital - partitioning entanglement entropies calculated using the numerical wavefunctions . for each wavefunction
( each colmun ) , the entanglement entropies are only listed up to their maximum value , typeset in bold , because the values after this are determined by the symmetry @xmath217 . the extrapolation procedure of sec .
[ subsec_extrapoln - issues ] involves extrapolating each row of numbers to @xmath195 .
note that , in each column ( for a particular @xmath27 ) , the omitted part after the midpoint ( in bold ) is a _ decreasing _ function of @xmath33 and thus does not give useful information about the thermodynamic limit of @xmath218 . in the extrapolation ,
it is therefore important to avoid values from these parts of the table .
we therefore restrict ourselves to values which in tables [ table_mr_entropies ] and [ table_laughlin_entropies ] are to the right of ( i.e. , above ) the diagonal line through the midpoint numbers typeset in bold .
kitaev , annals phys . * 303 * , 2 ( 2003 ) ; m. freedman , m. larsen , and z. wang , commun .
227 , 605 ( 2002 ) ; n. e. bonesteel , l. hormozi , g. zikos , s.h .
simon , phys .
95 * , 140503 ( 2005 ) ; s. das sarma , m. freedman , c. nayak , phys .
lett . * 94 * , 166802 ( 2005 ) .
j. i. latorre , e. rico , and g. vidal , quantum information and computation , * 4 * , 48 ( 2004 ) ; v.e .
korepin , phys .
. lett . * 92 * , 096402 ( 2004 ) ; v. popkov and m. salerno , phys .
a * 71 * , 012301 ( 2005 ) .
m. srednicki , phys .
71 * , 666 ( 1993 ) ; m. m. wolf , f. verstraete , m. b. hastings , j. i. cirac , arxiv : 0704.3906 ; m. b. plenio , j. eisert , j. dreiig , m. cramer , phys .
lett . * 94 * , 060503 ( 2005 ) . | we present a detailed analysis of bipartite entanglement entropies in fractional quantum hall ( fqh ) states , considering both abelian ( laughlin ) and non - abelian ( moore - read ) states .
we derive upper bounds for the entanglement between two subsets of the particles making up the state .
we also consider the entanglement between spatial regions supporting a fqh state .
using the latter , we show how the so - called topological entanglement entropy @xmath0 of a fqh state can be extracted from wavefunctions for a limited number of particles . |
1 chugai , n. n. , chevalier , r. a. , kirshner , r. p. & challis , p. m. hubble space telescope spectrum of sn1987a at an age of 8 years : radioactive luminescence of cool gas .
_ astrophys .
j. _ * 483 * , 925940 ( 1997 ) .
borkowski , k. j. , _ et al . _ radioactive scandium in the youngest galactic supernova remnant g1.9 + 0.3 .
_ astrophys .
j. _ * 724 * , l161l165 ( 2010 ) .
sunyaev , r. a. _ et al .
_ discovery of hard x - ray emission from supernova 1987a .
_ nature _ * 330 * , 227229 ( 1987 ) .
churazov , e. _ et al .
_ positron annihilation spectrum from the galactic centre region observed by spi / integral revisited : annihilation in a cooling ism ? _ mon . not .
r. astron .
soc . _ * 411 * , 17271743 ( 2011 ) .
we thank r.a .
sunyaev for reading of the manuscript and for comments ; v.s .
imshennik , d.k .
nadezhin and n.n .
chugai for discussions about different aspects of the physics of the supernova1987a explosion ; r.a .
krivonos for information regarding the analysis of integral / ibis data ; and m. coe for allowing us access to his integral data of the lmc observations ( @xmath60 ms ) before the end of the proprietary period .
this work is based on data obtained through the russian and european integral science data centers , and was supported by grants rfbr-11 - 02 - 12285ofi - m-2011 and ras - p20 ` the origin , structure and evolution of objects of the universe ' .
s.a.g . was principal investigator of the proposal requesting integral time ( @xmath61 ms ) to observe snr1987a ; made a preliminary analysis of the ibis / sigri data and detected an excess at the snr1987a position ; and wrote a draft of the text .
was co - investigator of the proposal , and participated in the analysis of ibis / isgri data and the verification of the result .
s.s.t . carried out the analysis of spi data .
provided general support for this project and participated in its different aspects , and c.w . and a.a.l .
contributed substantially to the final text .
all authors discussed the results and their presentation .
reprints and permissions information is available at www.nature.com/reprints . the authors declare no competing financial interests .
reader are welcome to comment on the online version of the paper .
correspondence and requests for materials should be addressed to s.a.g .
( [email protected] ) . | it is assumed@xcite that the radioactive decay of @xmath0ti powers the infrared , optical and uv emission of supernova remnants after the complete decay of @xmath1co and @xmath2co ( the isotopes that dominated the energy balance during the first three to four years after the explosion ) until the beginning of active interaction of the ejecta with the surrounding matter .
simulations@xcite show that the initial mass of @xmath0ti synthesized in core - collapse supernovae is @xmath3 solar masses ( _ _ * m*__@xmath4 ) .
hard x - rays and @xmath5-rays from the decay of this @xmath0ti have been unambiguously observed from cassiopeia a only@xcite , leading to the suggestion that the values of the initial mass of @xmath6ti near the upper bound of the predictions occur only in exceptional cases@xcite . for the remnant of supernova 1987a@xcite , an upper limit to the initial mass of @xmath0ti of @xmath7 _ _
* m*__@xmath4 has been obtained from direct x - ray observations@xcite , and an estimate of @xmath8 _ _ * m*__@xmath4 has been made from infrared light curves and ultraviolet spectra by complex model - dependent computations@xcite . here
we report observations of hard x - rays from the remnant of supernova 1987a in the narrow band containing two direct - escape lines of @xmath0ti at 67.9 and 78.4 kev .
the measured line fluxes imply that this decay provided sufficient energy to power the remnant at late times .
we estimate that the initial mass of @xmath0ti was @xmath9 _ _ * m*__@xmath4 , which is near the upper bound of theoretical predictions . + space research institute , ras , profsoyuznaya 84/32 , 117997 moscow , russia .
max - planck - institutfrastrophysik , karl - schwarzschild - strasse1,d-85741garching , germany .
finca , university of turku , vtslntie 20 , fi-21500 piikki , finland .
astronomy division , department of physics , university of oulu , fi-90014 oulu , finland .
european space agency , estec , keplerlaan 1 , 2200 ag noordwijk , the netherlands . the only immediate way to determine the initial mass of @xmath10ti ( denoted @xmath11 ) in a supernova remnant is through its emission in the direct - escape lines at energies @xmath124.1 , 67.9 , 78.4 , 511 and 1,157 kev .
this isotope decays according to the chain @xmath10ti@xmath13sc@xmath13ca , with a characteristic time of @xmath14 years ( ref . ) .
photons in the three x - ray lines , namely 4.1 , 67.9 and 78.4 kev , are emitted with efficiencies ( average numbers of photons per decay ) @xmath15% , 87.7% and 94.7% during the first stage of the decay , and photons in the @xmath5-ray lines , namely 511 and 1,157 kev , are produced with @xmath16% and 99.9% during the second very quick ( @xmath17 h ) stage .
when the ejecta becomes transparent for hard - x - ray and @xmath5-ray lines ( typically at @xmath18 years ) , the corresponding fluxes can be estimated as @xmath19 where @xmath20 is the distance to the remnant and @xmath21 is the proton mass .
measurements@xcite of the 67.9- , 78.4- and 1,157-kev lines from casa have yielded the value @xmath22 : tentative detections of the 1,157-kev line from the vela junior remnant@xcite and the 4.1-kev line from the g1.9 + 0.3 remnant@xcite have been also reported , but they are more uncertain . assuming a value of @xmath11 similar to that in cas a , we can expect the photon fluxes in these lines from supernova remnant ( snr ) 1987a ( @xmath23 kpc ) about @xmath24 years after the explosion to be about @xmath25 photons @xmath26 s@xmath27 .
note that this remnant was the first for which the x - ray and @xmath5-ray emission from the decay of @xmath28co and @xmath29co was detected@xcite .
the international gamma - ray astrophysics laboratory ( integral)@xcite invested considerable observing time in trying to detect these lines from snr1987a . the first observation was carried out in 2003 soon after the launch of integral .
a number of important results have been obtained , including constraints on the luminosity of the stellar remnant@xcite , but the dedicated exposure ( @xmath30 ms ) was insufficient for the detection of the @xmath10ti lines .
the next set of observations was carried out in 201011 with a total exposure of @xmath31 ms .
figure1 presents the results of our analysis of all these data .
images of the snr1987a field obtained with the ibis / isgri hard - x - ray telescope on board integral are shown in three energy bands , 4865 , 6582 and 8299 kev , which encompass the @xmath10ti lines at 67.9 and 78.4 kev .
two known sources , the black - hole binary and the 50-ms crab - like pulsar psrb0540 - 69 , are seen in all images .
remarkably , there is an obvious excess at the position of snr1987a , but only in the energy band that includes the @xmath10ti lines .
the location of the excess coincides within @xmath32 with that of snr1987a a typical uncertainty for moderately bright sources ( the angular resolution of ibis is @xmath33 full - width at half - maximum ( fwhm ) ) .
the signal - to - noise ratio for this excess , @xmath34 ( related to the flux measured in both lines at once ) , corresponds to a rather small probability of detecting such an excess by chance , @xmath35 .
the probability of detecting a random excess with the same @xmath36 ratio at an arbitrary position in the whole @xmath37 ibis field of view is much higher ( it can be estimated as @xmath38 where @xmath39 is the number of uncorrelated spots within the ibis field of view ; a ` spot ' is the area of the sky image that can be occupied by a single source ) . in reality
, the probability of detecting the @xmath10ti lines by chance is even smaller than @xmath40 , because significant excesses were simultaneously detected at the position of snr1987a in the images accumulated in the narrow energy bands 62.773.2 and 73.283.7 kev .
each of these bands contains one of the @xmath10ti lines and they are nearly independent in terms of the energy resolution of ibis / isgri , which is 8% fwhm at @xmath41 kev .
the @xmath36 ratios measured for snr1987a in these images are respectively 3.0 and 3.1 , and the corresponding chance probabilities are @xmath42 and @xmath43 .
the combined probability of detecting the @xmath10ti emission by chance is then equal to @xmath44 ( @xmath45 ) .
the hard - x - ray spectrum obtained with ibis / isgri from snr1987a is presented in fig.2 .
we fitted it using two lines of gaussian shape with fixed centroid energies , equal widths and self - consistent ( according to the @xmath10ti emission efficiencies ) normalizations .
the flux measured in both lines at once , @xmath46 , corresponds to a value for initial amount of synthesized @xmath10ti of @xmath47 ( the uncertainty is purely statistical ) .
this estimate can be slightly decreased if we assume the presence of some underlying continuum in the snr1987a spectrum for example , in the form of a power law with a photon index of 2.1 ( a crab - like spectrum ) , which should describe the hard - x - ray emission well up to @xmath48 kev ( ref . ) .
its possible contribution at 1 kev , @xmath49 , coincides with the normalization of the non - thermal x - ray component in the snr1987a spectrum following from the xmm - newton observations in 2009@xcite . with such a component ( shown in fig.2 by a solid curve ) the flux in the lines decreases to @xmath50 , which may be translated to @xmath51 .
the spi @xmath5-ray spectrometer on board integral observed snr1987a simultaneously with ibis / isgri in the 511- and 1,157-kev lines of @xmath10ti ; however , the smaller effective area and presence of strong background features at 511 and @xmath52 kev preclude their confident detection .
the photon spectrum measured in the 1,1061,205-kev band is presented in fig.3 .
the solid curve shows the @xmath53 upper limit ( 90% confidence ) for the flux in the 1,157-kev gaussian line , namely @xmath54 s@xmath27 .
it can be converted to the limit for the amount of @xmath10ti , @xmath55 . for comparison ,
the dotted curve in fig.3 shows the gaussian line corresponding to the value of @xmath11 measured with ibis / isgri .
the detection of emission in the 67.9- and 78.4-kev lines allows us to measure directly the amount of radioactive @xmath10ti synthesized during the explosion , that is , @xmath56 .
this amount formally exceeds by a factor of 1.5 to 2 the estimates based on theoretical simulations of explosive nucleosynthesis in supernova 1987a and on the infrared and ultraviolet data ( although the discrepancy represents only one standard deviation ) .
we note that @xmath10ti may be produced either as a result of strong deviation from nuclear statistical equilibrium during the explosion ( so - called @xmath57-rich freeze - out)@xcite or as a result of incomplete si - burning in a thin shell surrounding the region of main energy release@xcite . in both cases ,
the simulations are not straightforward and many poorly known parameters ( for example , the mass of the collapsed core , the maximum temperature and density behind the shock wave , the neutron excess and the composition of the pre - supernova interior ) could affect the results . in this sense , the value of @xmath11 that we report here may be useful for studying explosive nucleosynthesis and physical conditions reached during the shock - wave passage in the ejecta .
we note finally that the shock interaction of the ejecta and the surrounding matter is responsible for the synchrotron emission observed at present from snr1987a in the radio- and soft - x - ray bands@xcite .
the reprocessing of this emission in the ejecta is contributing to the late - time optical - infrared emission of snr1987a@xcite .
our estimate of the power - law component in the spectrum of snr1987a may be translated to a @xmath58 upper limit for the luminosity of synchrotron emission of @xmath59 in the 2060 kev band .
this value can be also considered as the limit for the luminosity of the stellar remnant ( pulsar or black hole ) harboured inside the ejecta . _
received 23 november 2011 ; accepted 13 august 2012 ; published 18 october 2012 . _ + |
the first spatial oscillations of coronal loops were discovered in extreme ultraviolet ( 171 @xmath1 ) by aschwanden et al .
( 1999a ) . using a mhd wave theory developed by edwin & roberts ( 1983 ) , aschwanden et al .
( 1999b ) and nakariakov et al .
( 1999 ) interpreted them as the fundamental fast ( kink ) mode of coronal loops .
subsequently , wang et al .
( 2003 ) observed frequency spectrum of the standing slow modes of the loops .
since then theoreticians have been supplementing the observed data , oscillation periods , loop lengths , etc . , with plausible auxiliary parameters , such as plasma density , magnetic field strength and configuration , etc .
, with the aim of obtaining reasonably realistic models for the structure of the loops ( bennett et al . , 1998 ;
verwichte et al .
, 2004 ; daz et al . , 2002 , 2004 ;
van doorsselaere et al . , 2004 ; andries et al . ,
2005a , b , arregui et al .
, 2005 ; dymova & ruderman , 2005 ; erdlyi & fedun , 2006 ; daz & roberts , 2006 ; mcewan et al . , 2006
; donnelly et al .
, 2006 ; erdlyi & verth 2007 ; dymova & ruderman , 2005 ; dymova & ruderman , 2006 ; verth et al . ,
2007 doorsselaere et al . , 2007 ; erdlyi & verth , 2007 ; safari et al . , 2007 ;
karami & asvar 2007 ) . in their recent work , dymova & ruderman ( 2005 ) and safari et al .
( 2007 ) reduce the mhd wave equations to a single sturm - liouville equation for the z - component of the perturbation in the magnetic field . here , we use the formalism of safari et al . and show that , for an exponential plasma density stratification along the loop axis , the problem has a closed analytical solution . the loop model , equations of motion , and boundary conditions are presented in sec .
the closed solutions , including the dispersion relation , are treated in sec 3 .
concluding remarks are given in sec .
a coronal loop is approximated by a cylinder of length @xmath2 and radius @xmath3 .
loop ends are fixed at the photosphere .
loop curvature is neglected , on account of @xmath4 .
no initial flow is assumed inside the loop .
a uniform magnetic field along the axis pervades the loop , @xmath5 .
gas pressure , gravity and all dissipative and viscous forces are neglected .
the density is discontinuous on the lateral surface of the cylinder and varies exponentially along the axis .
thus , @xmath6 where @xmath7 is the density scale height parameter , and @xmath8 , @xmath9 are the interior and exterior footpoint densities , respectively .
the assumption of exponential density is in accord with the findings of aschwanden et al .
they conclude this from their stereoscopic analysis of 30 loop oscillations in euv .
restriction of @xmath10 to the interval @xmath11 $ ] is permissible on account of the symmetry of the loop configuration about its midpoint the linearized ideal mhd equations are @xmath12 where @xmath13 and @xmath14 are the eulerian perturbations in the velocity and magnetic fields , respectively .
an exponential @xmath15 and @xmath16 dependence , is assumed , @xmath17 $ ] . by straightforward calculations
one can express all components of @xmath18 and @xmath19 in terms of @xmath0 .
the latter , in turn , is obtained from the following second order pde ( see safari et al .
2007 ) , @xmath20 where @xmath21 , local alfv@xmath22n speed , has different values inside and outside of the loop .
equation ( [ dbz ] ) , admits of a separable solution @xmath23 , where @xmath24 satisfies bessel s equation and will not be further referred to here , and @xmath25 satisfies the following @xmath26 where @xmath27 and @xmath28 are dimensionless frequency and footpoint densities , respectively
. equation ( [ z ] ) is an eigenvalue problem weighted by @xmath29 .
changing the variable @xmath30 to @xmath31 reduces eq.(7 ) to a bessel equation with the following solutions @xmath32 where @xmath33 and @xmath34 are constants , and @xmath35 and @xmath36 are bessel functions of first and second kind , respectively .
the boundary conditions are @xmath37imposing the boundary condition at @xmath38 on eq .
( [ zsol ] ) gives @xmath39 substituting these coefficients in eq .
( [ zsol ] ) and imposing the boundary conditions at @xmath40 gives the dispersion relations @xmath41 equations ( [ dispo ] ) and ( [ dispe ] ) are similar to those of daz & roberts ( 2006 ) with @xmath42 in their analysis . in the remainder of this section
they are solved analytically for weakly stratified loops and numerically for arbitrary stratifications .
bessel functions can be expanded as @xmath43 where @xmath44 we insert eq .
( [ bess ] ) into eqs .
( [ dispo ] ) and ( [ dispe ] ) and reduce them for @xmath45 and find @xmath46 where @xmath47 , and @xmath48^{-\frac{1}{2}}$ ] is the fundamental kink frequency of homogenous loops .
expectedly , @xmath49 as @xmath50 .
the following ratios are noteworthy : @xmath51 the frequencies and the ratio of any two odd numbered frequencies begin decreasing linearly with @xmath52 .
the ratio of two even modes , however , begins decreasing quadratically with @xmath52 .
these features are also seen on the diagrams of fig [ fig3]@xmath53 .
observational verification of these points , however , has to await the availability of more extended and higher resolutions data .
presently only two frequencies in three loops are available ( verwitche et al . 2004 , van doorsselaere 2007 ) .
we use newton - raphson s numerical method to solve eqs .
( [ dispo ] ) and ( [ dispe ] ) for the eigenfrequencies . in the range , @xmath54 , the fundamental and three higher kink frequencies , @xmath55 , @xmath56 , and @xmath57 , and the ratios , @xmath58 , are computed .
the data are plotted in fig .
[ fig3 ] for two density contrasts , @xmath59 and 0.5 .
as the density contrast increases the frequencies shift down .
their ratios , however , remain unchanged .
the ratio @xmath60 begins with @xmath61 and decreases with increasing @xmath52 , in compliance with eqs .
( [ ratio ] ) .
andries et al . ( 2005b ) maintain that the frequency ratios could be used as a seismological tool to estimate the coronal density scale heights . from the trace data , verwichte et al .
( 2004 ) find the ratio @xmath62 to be 1.64 and 1.81 for loops designated by @xmath63 and @xmath64 in their records , respectively .
van doorsselaere , nakariakov , & verwichte ( 2007 ) revisited the same ratios from the observational data to be 1.58 and 1.82 for the same loops , respectively , and 1.795 for another loop in their current analysis . with the help of fig .
[ fig1 ] , we find the corresponding @xmath52 to be 4.98 and 1.9 , respectively . assuming typical loop lengths , @xmath65 mm , the density scale heights fall in the range of @xmath66 and @xmath67 mm , respectively .
these scale heights are slightly different from the findings of andries et al .
( 2005a , b ) and safari et al . ( 2007 ) for a sinusoidal density profile .
a noteworthy point is the effect of column mass on frequencies . in fig .
[ fig1 ] we assume a constant footpoint density contrast , @xmath68 , and vary @xmath52 . consequently , the total column mass of the loop changes . compared with variable density contrast but fixed column mass
, the mode profiles and the frequency ratios remain unchanged .
the frequencies themselves , however , behave slightly differently . for variable column mass models , the frequencies increase more sharply with @xmath52 .
the mode profiles , @xmath69 of eq .
( [ zsol ] ) , are shown in fig .
[ fig2 ] for @xmath70 , and 4 . for the unstratified case , @xmath71
, the profiles are sinusoidal . with increasing @xmath52 , they depart from the sine curves .
antinodes shift toward the footpoints .
stronger the stratification , the greater the shift is , in agreement with the findings of safari et al .
( 2007 ) and verth et al .
van doorsselaere et al .
( 2007 ) point out , the shift of the antinodes is potentially , a coronal seismological tool to estimate the density scale heights . in fig .
[ fig3 ] , we have plotted the antinode shift , @xmath72 , of the first overtone versus @xmath52 .
the shift @xmath73 , grows approximately linearly with @xmath52 .
our numerical result shows that , the shift in the antinode for different density contrasts , @xmath68 and @xmath74 , are the same . for typical loops , of lengths @xmath75 mm and density scale heights , @xmath76 and @xmath77 mm
, the antinode shift falls in the range @xmath78 mm and @xmath79 mm , respectively .
observation wise , the resolution of current solar satellite facilities , e.g. , trace , sdo , so , etc .
, seems adequate to detect such antinode shifts and estimate the density scale height of solar coronae .
verth et al .
( 2007 ) study semi circular loops of sinusoidal density profiles and find the antinode shifts @xmath80 .
our numerical results show that , for a given @xmath52 , the mode profiles are insensitive to changes in density contrast , @xmath81 .
the differences between the mode profiles of the stratified and unstratified cases , @xmath82 , @xmath83 and 4 are plotted in fig .
erdlyi & verth ( 2007 ) maintains that these differences in the mode profile are so small to be resolved by current euv instruments of trace mission .
another interesting point is negative scale heights , suggested by andries ( 2005a , b ) on the basis of the error bars in the observations of verwichte et al .
( 2004 ) . here
, the density at the apex is higher than at footpoints . unlike the positive scale scale height case : a ) @xmath84 decreases with increasing @xmath85 ,
[ fig5 ] ; b ) higher overtones , @xmath86 , @xmath87 , however , increase with @xmath85 , though at slower rate ; c ) the ratios @xmath60 increase with @xmath85 ( not presented in a diagram ) . as @xmath85 increases , the mode profiles and their node and antinodes move away from the footpoints and concentrate more and more in the inner regions of the loop .
suggested theoretical models of 3d coronal loops are , still , far from the realities . many complicating factors , such as variable cross sections , variable magnetic fields , non - zero @xmath88 plasmas , etc .
, are to be accounted for in a realistic study of both the equilibrium structure and the perturbed state of actual loops . here , we study the oscillations of loops with exponential density variations along the loop axis .
- : : analytical dispersion relations , eqs .
( [ dispo ] ) and ( [ dispe ] ) , and analytical mode profiles , eq .
( [ zsol ] ) and fig .
[ fig2 ] , are derived .
- : : for weak stratifications , the kink frequencies and the frequency ratios are found up to the second order in @xmath52 , eqs .
( [ weak ] ) and ( [ ratio ] ) .
- : : increasing the density contrast decreases the frequencies but their ratios and shape of the profiles remain unchanged . - : : models with variable total column mass , but constant footpoint densities , are investigated . compared with models of constant total mass ,
the frequencies increase more sharply with increasing @xmath52 . - : : the case of negative scale heights is investigated and results are compared with those of positive @xmath52 s . -
: : for @xmath89 , and for typical loop lengths , 100 - 400 mm , the density scale heights fall in the range of 20 - 210 mm , in agreement with andries et al .
( 2005a , b ) , safari et al . ( 2007 ) , mcewan et al . ( 2006 ) , and donnelly et al .
( 2006 ) . - : : based on our simple theoretical model and typical coronal conditions , the antinode shift of the first overtone mode profiles are in the range of @xmath90 mm .
they are in the range of the detectability of the resolution of the current observational instruments . | the influence of longitudinal structuring on the fast kink modes of coronal loops is investigated .
analytical dispersion relations and mode profiles are derived for the second - order ordinary differential equation governing the z- component of the perturbation in the magnetic field , @xmath0 .
all other components are given in terms of @xmath0 .
deviations from the frequencies and mode profiles of homogenous loops are given as functions of the density scale height .
the effects of the fixed and variable column masses , negative scale heights , and density contrasts inside and outside of the loops are studied .
the frequency ratios , mode profiles , and the antinode shifts from those of the sine profiles of the homogenous loops , are tools to estimate solar photospheric parameters . to this end
, we have expanded the relevant factors up to the second order in the stratification parameter . in particular
, we verify that the first overtone antinode shifts are in the mm range and are within the reach of the resolutions of the present day observations . |
the production of charm quarks in association with electroweak gauge bosons at hadron colliders can provide important information on strange and charm quark pdfs , complementary to that obtained by tagging charm quarks in the final state in deep inelastic scattering experiments @xcite .
in particular , the tevatron cdf and d0 experiments @xcite have measured the cross section for charm quarks produced in association with @xmath4 bosons , using muon tagging of the charm - quark jet . however
the accuracy of these measurements is limited to @xmath530% by low statistics .
the lhc is expected to provide a more precise measurement , and indeed the cms collaboration has recently performed a similar study @xcite of @xmath6 production , again using muons to tag the charm quark jet in the final state . at leading order ( lo ) , the feynman diagrams for charm production in association with a w boson are shown in fig .
the dominant contribution comes from strange quark
gluon scattering , as the corresponding down - quark contribution is strongly cabibbo suppressed .
production at leading order . ]
the cross section for @xmath2 production ( where @xmath7 denotes a tagged charm - quark jet ) is measured in ref .
@xcite ( cdf ) , while ref .
@xcite ( d0 ) introduces the ratio of charm jets to all jets , which is expected to suffer less from both experimental and theoretical uncertainties .
cms has performed a similar analysis @xcite using the 2010 lhc data set . also of interest
is charm production in association with @xmath8 bosons .
the leading - order process is simply @xmath9 , and so this process can be used to extract information on the charm quark pdf .
it is important to note that for both @xmath2 and @xmath10 production at hadron colliders , the strange and charm quarks are probed at much higher @xmath3 ( @xmath11 gev@xmath12 ) values than in the traditional determinations from deep inelastic scattering , i.e. @xmath13 and @xmath14 where typically @xmath15 gev@xmath12 ) . taken together , the measurements therefore also test dglap evolution for these quark flavours . in this letter
we study @xmath16-jet production in the context of the cms analysis @xcite , analysing the different quark contributions and comparing the predictions of various widely - used pdf sets .
we also study the corresponding cross - section ratio for @xmath10-jet production , which should be measurable with the 2011 lhc data set .
the two relevant cross - section ratios introduced by cms @xcite are : @xmath18 the advantage of using ratios is that many of the theoretical and experimental uncertainties cancel . in particular , the ratios are fairly insensitive to higher - order perturbative qcd corrections .
note that @xmath19 at the tevatron .
we calculate the cross sections at nlo pqcd using mcfm @xcite , applying the cms cuts @xcite to the final - state : @xmath20 gev , @xmath21 , @xmath22 gev , @xmath23 , @xmath24 , @xmath25 .
five different nlo pdf sets are used : ct10 @xcite , mstw2008 @xcite , nnpdf2.1 @xcite , gjr08 @xcite and abkm09 @xcite , as implemented in lhapdf @xcite .
the renormalisation and factorisation scales are set to @xmath26 , although the cross - section ratios are rather insensitive to this choice .
( we have also considered dynamical scales of the form @xmath27 , but the differences for the cross - section ratios are similar in magnitude to the pdf uncertainties . ) the results are summarised in table [ fractions2 ] where we also include the ratio : @xmath28 .comparison of results for the cross - section ratios defined in ( [ eq : wcratios ] ) and ( [ eq : wpmrat ] ) at nlo using different pdf sets , including 68%cl ( asymmetric , where available ) pdf errors . [ cols="^,^,^,^",options="header " , ] in principle @xmath29 provides direct information on the charm content of the proton , complementary to that obtained from dis experiments via @xmath30 .
( -240,-95 ) we note that the differences between the predictions of different pdf sets are much smaller than for the strange quark distributions , presumably because in all these global fits the charm distributions arise perturbatively from @xmath31 splitting , with the small-@xmath32 gluon well determined from the hera structure function data .
this can be seen in fig .
[ charm1 ] , which compares the ratio of charm quarks to all quarks for the three pdf sets .
when pdf errors are taken into account the three sets are very difficult to distinguish , and this is reflected in the similarity in the @xmath29 predictions in table [ zres ] .
the use of @xmath33 as a pdf discriminator will therefore require a very precise measurement .
we can also consider the ( charm ) charge asymmetry ratio : @xmath34 at first sight it might appear that @xmath35 is automatically equal to 1 , but this is only true if @xmath36 in the initial state .
this is indeed the case for all the pdf sets considered here , since as already noted the charm distributions are generated by charge symmetric @xmath37 splitting .
however this symmetry does not necessarily hold if we allow for an _ intrinsic _ charm component @xcite .
pdf studies incorporating intrinsic charm , see for example @xcite , suggest that it is probably a small effect compared to perturbatively generated charm , particularly at the small @xmath32 values relevant to the lhc . in any case ,
any such intrinsic charm analysis is highly model dependent , and beyond the scope of the present study .
we do note however that in the process @xmath38 , which dominates @xmath39jet production over most of the kinematic range at the lhc , the @xmath40 is more likely to be positively charged than negatively charged , and the @xmath41jet is therefore more likely to contain a @xmath42 ( say ) than a @xmath43 . hence there may well be a natural charge asymmetry in the misidentified charm - jet background .
finally we can also compare the ratios : @xmath44 the nlo predictions for mstw2008nlo are 0.045 and 0.082 respectively , for the selection cuts described above for @xmath4 and @xmath8 including leptonic decays .
we can relate the ratio of these ratios to the ratio of @xmath45 over @xmath46 which can be read from figs .
[ strerr ] and [ charm1 ] at the appropriate momentum fraction .
of course in practice it is more complicated , as multiple scales , momentum fractions and couplings are involved , but an estimate @xmath47 can be extracted for this ratio and this matches well with the 0.082/0.045 ratio above .
we have investigated charm production in association with @xmath4 and @xmath8 bosons at the lhc .
we have presented predictions relevant to the recent ( and ongoing ) cms analysis for @xmath4 bosons produced in association with a charm - quark jet .
use of a larger lhc data sample should lead to precise measurements of the ratios @xmath33 and @xmath48 which can in turn provide useful information on the strange content of the proton , and in particular the asymmetry between @xmath49 and @xmath50 at small @xmath32 and high @xmath3 .
we have also shown results for several differential distributions that can in principle be used to provide additional information on the @xmath32 dependence of the strange and anti - strange quark distributions .
we also propose a measurement of the corresponding ratio for z bosons , @xmath29 , which can be used as a measure of the charm content of the proton .
finally , we note that information on the strange quark content of the proton can also be obtained by comparing the _ total _ @xmath4 and @xmath8 cross sections at the lhc , see for example the recent atlas study reported in @xcite .
this exploits the fact that the dependence on the strange pdf is linear for the @xmath4 and quadratic for the @xmath8 : @xmath51 , @xmath52 .
the method is complementary to the charm - jet tagging method discussed in the present study , and it will be interesting to compare the results from the two analyses when more precise data become available .
useful discussions with isabel josa and robert thorne are acknowledged .
ev is grateful to the uk science and technology facilities council for financial support .
99 h. l. lai , p. m. nadolsky , j. pumplin , d. stump , w. k. tung and c. p. yuan , jhep * 0704 * ( 2007 ) 089 [ arxiv : hep - ph/0702268 ] .
t. aaltonen _ et al . _
[ cdf collaboration ] , phys .
lett . * 100 * , 091803 ( 2008 ) [ arxiv:0711.2901 [ hep - ex ] ] . v.
m. abazov _ et al . _
[ d0 collaboration ] , phys .
b * 666 * , 23 ( 2008 ) [ arxiv:0803.2259 [ hep - ex ] ] .
cms collaboration , cms - pas - ewk-11 - 013 .
j. m. campbell and r. k. ellis , http://mcfm.fnal.gov/. h. l. lai , m. guzzi , j. huston , z. li , p. m. nadolsky , j. pumplin and c. p. yuan , phys .
d * 82 * , 074024 ( 2010 ) [ arxiv:1007.2241 [ hep - ph ] ] .
a. d. martin , w. j. stirling , r. s. thorne and g. watt , eur .
j. c * 63 * , 189 ( 2009 ) [ arxiv:0901.0002 [ hep - ph ] ] .
r. d. ball _ et al .
_ [ the nnpdf collaboration ] , nucl .
b * 855 * , 153 ( 2012 ) [ arxiv:1107.2652 [ hep - ph ] ] .
m. gluck , p. jimenez - delgado and e. reya , eur .
j. c * 53 * , 355 ( 2008 ) [ arxiv:0709.0614 [ hep - ph ] ] .
s. alekhin , j. blumlein , s. klein and s. moch , phys .
d * 81 * , 014032 ( 2010 ) [ arxiv:0908.2766 [ hep - ph ] ] .
m. goncharov _ et al .
_ [ nutev collaboration ] , phys . rev .
d * 64 * ( 2001 ) 112006 [ arxiv : hep - ex/0102049 ] .
s. j. brodsky , p. hoyer , c. peterson and n. sakai , phys .
b * 93 * ( 1980 ) 451 .
j. pumplin , h. l. lai and w. k. tung , phys .
d * 75 * ( 2007 ) 054029 [ arxiv : hep - ph/0701220 ] .
g. aad _ et al . _ [ atlas collaboration ] , arxiv:1203.4051 [ hep - ex ] . | the production of charm quark jets in association with electroweak gauge bosons at the lhc can be used as a tool to constrain quark parton distribution functions ( pdfs ) .
motivated by recent measurements at the tevatron and lhc , we calculate cross sections for @xmath0 , comparing these to @xmath1 , for various pdf sets .
the cross - section differences can be understood in terms of the different underlying pdfs , with the strange quark distribution being particularly important for @xmath2 production .
we suggest measurements of appropriately defined ratios and comment on how these measurements at the lhc can be used to extract information on the strange and charm content of the proton at high @xmath3 scales . |
mass is the fundamental property that sets the evolutionary path of a star .
the masses of young stars are of particular interest in many astrophysical problems : they provide unique information about the star formation process ( e.g. , accretion histories , the initial mass function ; see * ? ? ? * ) and are thought to have a substantial influence on the evolution of their circumstellar material , and therefore the efficiency of the planet formation process ( e.g. , see * ? ? ?
unfortunately , measurements of pre - main sequence ( pre - ms ) star masses are difficult and accordingly rare . unlike their more evolved counterparts , the location of a young star in the hertzsprung - russell ( hr ) diagram does not provide a robust estimate of @xmath9 @xcite .
the theoretical models for stellar evolution at these early stages are plagued with uncertainties related to rotation @xcite , accretion @xcite , magnetic fields @xcite , atmosphere properties ( e.g. , convection , opacities ; * ? ? ?
* ) , and unknown initial conditions .
ultimately , a more nuanced understanding of star and planet formation requires pre - ms evolution models that are empirically calibrated with direct , independent , and accurate @xmath9 measurements .
the only _ direct _ methods available for measuring stellar masses are based on orbital dynamics . in sufficiently close pre - ms binary star systems ,
@xmath9 can be estimated from the stellar orbits using multi - epoch radial velocity ( rv ) measurements ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and/or long - term astrometric monitoring @xcite . for a double - lined spectroscopic binary ( sb2 ) , the rv method provides a robust estimate of @xmath10 for each component , where @xmath11 is the inclination angle of the orbit projected on the sky . for visual " binaries ,
the astrometric monitoring technique offers a constraint on the quantity @xmath12 , where @xmath13 is the sum of the stellar masses and @xmath14 is the distance to the binary . generally , the stellar mass estimates from these methods are inherently uncertain due to their strong dependences on the unknown values of @xmath11 or @xmath14 .
the \{@xmath9 , @xmath11 } degeneracy is broken for the special case of an eclipsing sb2 , but few pre - ms systems with such favorable orientations are known ( e.g. , * ? ? ? * ; * ? ? ? * and references therein ) . in a subset of ideal cases ,
the rv and astrometric techniques can be combined to alleviate the uncertainties related to \{@xmath11 , @xmath14 } and extract accurate @xmath9 values ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these standard methods are only applicable for binary stars with a narrow range of orbital separations .
alternatively , for any _ isolated _ young star with a circumstellar disk , @xmath9 can be determined from a single millimeter - wave interferometric observation of an optically thick emission line ( with a linear dependence on @xmath14 ) .
this latter technique relies on modeling the spatially and spectrally resolved keplerian rotation curve of the molecular gas disk that orbits the young star @xcite .
while this method has extraordinary value in its more general applicability , it has only been successfully employed for small samples .
attempts to expand its reach have been frustrated by molecular cloud contamination and observational limitations in resolution and sensitivity . moreover , a reconstruction of the disk velocity field necessarily involves fitting such data with a relatively complicated model of the disk structure @xcite .
given that added complexity , there is naturally some concern about the absolute accuracy of the @xmath9 estimates from this method ( e.g. , * ? ? ?
* ) , despite the impressive formal precision of the measurements ( @xmath72 - 3% ; e.g. , * ? ? ?
the young binary v4046 sgr provides a rare opportunity to benchmark the disk kinematics method for estimating @xmath9 against the more traditional rv technique for a sb2 system .
v4046 sgr is a nearly equal mass ( @xmath15 ) pair of solar - type pre - ms stars in a circular ( @xmath16 ) , non - eclipsing orbit with a 2.4 day period ( @xmath17r@xmath5 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the system is completely isolated from any known molecular clouds and has been kinematically associated with the @xmath78 - 20myr - old @xmath18 pic moving group ; a moving cluster analysis suggests it is relatively nearby , @xmath19pc @xcite . despite its advanced age ,
v4046 sgr hosts a large and massive circumbinary disk that exhibits a rich molecular emission line spectrum @xcite .
since the binary orbit is tight and circular , it has no dynamical impact on the disk structure outside a radius of @xmath70.1au ( e.g. , see * ? ? ?
* ) . with its central sb2 host , rare proximity to the sun , lack of molecular cloud contamination , and intrinsically bright , spatially extended line emission ,
the v4046 sgr disk is an ideal target to assess the accuracy of the disk kinematics technique for measuring @xmath9 . in this article
, we build on some initial work by @xcite and use high - quality spatially and spectrally resolved observations of the co @xmath1=2@xmath21 emission line to measure the velocity field of the v4046 sgr disk and extract the total mass of the close binary at its center .
our millimeter - wave observations with the submillimeter array ( sma ) and data calibration procedures are described in [ sec : observations ] .
a detailed overview of the modeling analysis is provided in [ sec : modeling ] .
the modeling results are presented and compared with the complementary rv analysis of the central sb2 by @xcite in [ sec : results ] .
these results are discussed in the context of the v4046 sgr system in particular , pre - ms evolution models more generally , and future prospects for @xmath9 estimates from the disk kinematics method in [ sec : discuss ] .
some key conclusions from this work are summarized in [ sec : summary ] .
the v4046 sgr circumbinary disk was observed at 225ghz ( 1.3 mm ) with the submillimeter array ( sma ; * ? ? ?
* ) on four occasions starting in 2009 , using each of the available antenna configurations : sub - compact ( baseline lengths of 9 - 25 m ; 2011 mar 18 ) , compact ( 16 - 70 m ; 2009 apr 25 ) , extended ( 28 - 226 m ; 2009 feb 23 ) , and very extended ( 68 - 509 m ; 2011 sep 4 ) .
the sma double sideband receivers were tuned to simultaneously observe the @xmath1=2@xmath21 transitions of @xmath0co , @xmath20co , and c@xmath21o at 230.538 , 220.399 , and 219.560ghz , respectively , and the adjacent dust continuum .
the correlator was configured to place those emission lines in separate 104mhz spectral chunks and sample them finely with 512 channels per chunk , corresponding to a native velocity resolution of @xmath70.25 km s@xmath22 .
the continuum was observed with a more coarse frequency sampling ( in 3.25mhz channels ) , with a total bandwidth of 1.6 and 3.6ghz in 2009 and 2011 , respectively .
the observations cycled between v4046 sgr and the quasars j1924 - 292 ( 15 away ) and j1733 - 130 ( 22 from v4046 sgr , 30 from j1924 - 292 ) with a total loop time of 10 - 20 minutes .
the bright quasars 3c 84 and 3c 454.3 were also observed as bandpass calibrators , along with uranus , callisto , and ceres for use in determining the absolute scaling of the amplitudes .
all of the data were collected in outstanding weather conditions for this observing frequency , with an atmospheric zenith optical depth of only @xmath70.05 ( corresponding to a precipitable water vapor level of @xmath71.0 mm ) .
each individual dataset was calibrated independently with the mir software package .
the bandpass response was corrected based on observations of bright quasars , and broadband continuum channels were generated from the central portions of all line - free spectral chunks .
the visibility amplitude scale was derived by bootstrapping the gain calibrator ( quasar ) flux densities from the observations of uranus , callisto , or ceres , with a systematic uncertainty estimated at 10 - 15% .
the antenna - based complex gain response of the system was determined with reference to j1924 - 292 , and the quality of the phase transfer was assessed using the observations of j1733 - 130 .
that comparison suggests only a small amount of seeing " ( @xmath701 ) was introduced by atmospheric phase noise ( or small baseline errors ) , consistent with the excellent observing conditions . after applying the appropriate ( and small ) phase shifts to account for the v4046 sgr proper motion ( @xmath23yr@xmath22 , @xmath24yr@xmath22 ; *
* ) and confirming that the continuum amplitudes from different array configurations were consistent on overlapping baseline lengths , the visibility datasets from each observation were combined .
the observations of the dust continuum and co isotopologue emission will be presented in a separate article ; the focus here will be solely on the @xmath0co @xmath1=2@xmath21 emission .
note that although the 2009 data were originally presented by @xcite , those data have been re - calibrated here for consistency ( and modest improvements ) .
the co visibilities were continuum - subtracted and truncated outside a projected baseline length of 200k@xmath25 to reduce the data volume and improve the signal - to - noise ratio .
they were then fourier inverted , deconvolved with the clean algorithm , and restored with a synthesized beam using the miriad package .
the naturally - weighted spectral images shown as channel maps in figure 1 were synthesized on a 0.4 km s@xmath22 smoothed velocity scale with a @xmath26 beam ( at a position angle of 24 ) .
the typical rms noise level in each channel is 70mjy beam@xmath22 .
there is co emission firmly detected ( @xmath273@xmath28 ) out to @xmath294.8 km s@xmath22 from the systemic velocity , estimated to be @xmath30 km s@xmath22 ( corresponding to @xmath31 km s@xmath22 in the heliocentric frame ) , with an integrated intensity of @xmath32jy km s@xmath22 and a peak flux density of @xmath33jy beam@xmath22 ( @xmath34k ; a peak s /
n = 21 ) , including the calibration uncertainties . figure [ fig : mom ] shows a map of the velocity - integrated co intensities ( 0@xmath35 moment ; _ contours _ ) overlaid on the intensity - weighted velocities ( 1@xmath36 moment ; _ colorscale _ ) , as well as a spatially integrated co spectrum .
these data exhibit a molecular gas disk with a clear rotation pattern , from east ( blueshifted ) to west ( redshifted ) , and suggest a modest inclination angle to the line of sight .
near the systemic velocity , the co emission subtends @xmath75 in radius ( @xmath7365au ; see * ? ? ?
the fundamental goal here is to derive a dynamical estimate of the central stellar mass based on the kinematic properties of the v4046 sgr gas disk . in order to extract @xmath9 from a measurement of the disk velocity field
, we need to construct a detailed physical model of the disk structure .. in this particular case , @xmath9 corresponds to the sum of the stellar masses in the v4046 sgr binary .
] we adopt a modeling formalism motivated by @xcite , which makes three basic assumptions about disk properties .
first , the disk material is assumed to be orbiting in keplerian rotation around a point mass , meaning the central stars are treated as one object that dominates the disk velocity field .
previous work has found little evidence to support ( simple parametric ) deviations from keplerian @xmath37-fields in disks ( e.g. , see * ? ? ?
* ) ; the corollary criterion that the disk mass is only a small fraction of the stellar mass ( @xmath38 ) can be confirmed _ a posteriori _ ( see 4 ) .
second , the disk is assumed to be geometrically thin at all radii . although @xcite argued that this may not be valid at very large radii ( @xmath27800au ) , the higher @xmath9 for the v4046 sgr binary ( which decreases the disk scale height ; see below ) and the limited extent of the co emission from its disk ( @xmath39400au ) suggest it is a reasonable approximation in this case . and third , in the context of the disk structure , we assume that the gas is vertically isothermal and in hydrostatic pressure equilibrium .
a more realistic model would include a temperature inversion ( e.g. , * ? ? ?
however , excluding that kind of added complexity does not diminish the accuracy of an @xmath9 estimate : the emission from a single co line is generated in a narrow vertical layer of the disk atmosphere , which has a roughly constant temperature @xcite . the model for the gas density distribution
is constructed in cylindrical coordinates ( @xmath40 , @xmath41 , @xmath42 ) , @xmath43,\ ] ] where @xmath44 is the radial surface density profile and @xmath45 is the pressure scale height at each radius .
we assume a parametric version of the former that is appropriate for an accretion disk with a static , power - law viscosity profile @xcite and is currently the basis for most dust - based disk density measurements ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , @xmath46,\ ] ] where @xmath47 is a density gradient , @xmath48 is a characteristic radius , and @xmath49 is a normalization equivalent to @xmath50 .
the scale heights are calculated with the explicit assumption of hydrostatic equilibrium and a constant vertical temperature profile , such that @xmath51 where @xmath52 is the sound speed , @xmath53 is the angular velocity , @xmath54 is the radial temperature profile , @xmath55 is the boltzmann constant , @xmath56 is the gravitational constant , @xmath57 is the mass of a hydrogen atom , and @xmath58 is the mean molecular weight of the gas .
we further assume a simple power - law behavior for the radial temperature profile , @xmath59 where @xmath60 is a temperature gradient and @xmath61 is the gas temperature at a radius of 10au .
the parametric descriptions in equations ( 1)-(4 ) completely characterize the temperature and density structure of a model gas disk .
the gas kinematics are described by a keplerian velocity field , @xmath62 meaning there is only rotation , and no net motion in the radial or vertical dimensions .
given a physical disk structure , we can construct a model of the co @xmath1=2@xmath21 emission with the added assumption that the energy levels relevant for this transition are populated according to local thermodynamic equilibrium ( lte ) .
although the lte approximation is not always valid in protoplanetary disks , @xcite demonstrated that it is an appropriate simplification for the low energy and high optical depths associated with this particular species and transition .
we model the emission line intensity distribution by integrating the radiative transfer equation along each sight line @xmath63 , so that @xmath64 where @xmath65 is the absorption coefficient , @xmath66 is the source function ( here equivalent to the planck function ) , and @xmath67 is the optical depth along the line - of - sight . the absorption coefficient is the sum of contributions from dust and gas . for the former ,
we assume @xmath68 , where the dust - to - gas mass ratio is @xmath69 and the grain opacity is @xmath70100ghz)@xmath71 g@xmath22 @xcite . in the particular case of the v4046 sgr disk ,
the contribution of dust to the absorption coefficient is effectively negligible compared to the gas ; the specific choices of @xmath72 and @xmath73 ( within reasonable limitations ) have no tangible effect on our results ( moreover , continuum emission is removed from the data before our modeling analysis ) .
the absorption coefficient of the co gas is calculated from the transition cross section weighted by the population of the lower energy level , @xmath74 , such that @xmath75 ( note that in this specific case , @xmath76 ) .
since we assume the line is thermally populated in lte , the level populations are determined by the local disk temperature via the boltzmann equation , @xmath77 where @xmath78 is the transition energy , @xmath79 is the statistical weight , @xmath80 is the partition function , and @xmath81 is the co fractional abundance ( assumed to be constant everywhere in the disk ) .
the details of the emission line model are encoded in the absorption cross section , @xmath82 where @xmath83 is the line profile function and the integrated cross section is @xmath84 where we have used the einstein relation in the last equality to express the einstein-@xmath85 coefficient in terms of the einstein-@xmath86 coefficient provided by the lamda molecular database @xcite .
the line profile function naturally determines the shape of the emission line , where a given frequency @xmath87 corresponds to an intrinsic doppler velocity @xmath88 relative to the line center , @xmath89 .
the gas in a given disk model has a projected , line - of - sight velocity field given by @xmath90 , where @xmath91 is the inclination of the disk relative to the observer ( such that @xmath92 is edge - on ; for the geometry , see * ? ? ?
the line profile shape at each frequency is determined by the difference between the doppler and line - of - sight velocities , @xmath93,\ ] ] and the effective line width , @xmath94 .
the latter is comprised of the quadrature sum of contributions from thermal and non - thermal broadening terms , @xmath95 where @xmath96 is the mean molecular weight of co and @xmath97 is the contribution from microturbulence .
we assume that the turbulent velocity width @xmath97 is constant throughout the disk .
the @xcite model formalism highlighted above is admittedly complex .
a single model is completely described by 14 parameters : four quantify the distribution of co densities \{@xmath81 , @xmath49 , @xmath98 , @xmath47 } , two characterize the temperature structure ( and therefore the vertical density structure ) \{@xmath61 , @xmath60 } , three play pivotal roles describing the disk kinematics \{@xmath9 , @xmath97 , @xmath89 } , and the remaining five relate the model to the observations
i.e. , convert from physical coordinates in the disk to the observed plane projected on the sky including the disk inclination and major axis position angle \{@xmath91 , pa@xmath99 } , distance \{@xmath14 } , and disk center \{@xmath100 , @xmath101}. some of these parameters can be reliably determined in a simple way , and then fixed to facilitate the data modeling problem ( with no tangible impact on the quality of the @xmath9 determination ) . in this case , the systemic lsr velocity ( effectively @xmath89 ) was measured directly from the sma data and set to @xmath102 km s@xmath22 .
the disk center coordinates were estimated from both the dust continuum and co emission morphology at the systemic velocity ( see fig .
[ fig : co21 ] ) and set to @xmath103 , @xmath104 ( j2000 ) , coincident with the composite v4046 sgr stellar position in the ucac3 catalog @xcite .
we adopted the distance inferred by @xcite from the kinematic parallax ( moving cluster ) technique , @xmath19pc ( but see 5 regarding alternative values ) .
some additional practical simplications can be made , since the focus here is on the velocity field and not the disk structure details .
because the line opacity is so much greater than the dust opacity , the normalization parameters @xmath81 and @xmath49 are not independent : we can only constrain their product . in practice ,
we adopt a joint co disk mass parameter , @xmath105 , for that purpose .
moreover , after extensive experimentation , we concluded that two more parameters should be fixed .
first , we found that the disk orientation was very well - determined ( within 1 ) at pa@xmath106 ( e of n ) , so that continued iteration on a more precise value was a waste of computational resources : fixing this parameter has negligible quantitative impact on the other model parameters or their uncertainties .
and second , we fixed the surface density gradient , @xmath107 .
the key parameters that set the density profile , \{@xmath47 , @xmath48 } are anti - correlated and strongly degenerate ; statistically indistinguishable model fits can be found over a large range of these parameters .
the degeneracy is remarkably narrow , and could easily be missed with standard minimization algorithms ( resulting in local @xmath108 minima with misleadingly tight constraints on the gradient ) .
fortunately , this degeneracy has minimal impact on the parameters most relevant for characterizing the disk velocity field : the precision and accuracy of a @xmath9 determination are not notably affected by the uncertainties in \{@xmath47 , @xmath48}. with that in mind , we have made the modeling process more tractable with a fixed @xmath47 . making use of those simplifications
, a synthetic co spectral datacube can be calculated by specifying 7 free parameters , \{@xmath109 , @xmath48 , @xmath61 , @xmath60 , @xmath97 , @xmath91 , @xmath9}. that model datacube is then resampled at the observed velocities and spatial frequencies and processed in the same way as the data ( see 2 ) to produce a set of synthetic spectral visibilities .
those model visibilities are evaluated with respect to the data by computing a composite @xmath108 value , summed over the real and imaginary components in 25 spectral channels .
the best - fit set of model parameters and their associated uncertainties were determined with a monte carlo markov chain ( mcmc ) technique , using the affine invariant ensemble sampler developed by ( * ? ? ?
* see also foreman - mackey et al .
2012 ) , using a jump probability @xmath110 . to our knowledge , this is the first application of these more sophisticated mcmc methods for parameter estimation in this particular context : previous studies have relied on downhill simplex routines ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , sometimes with clever modifications to address asymmetric uncertainties @xcite .
although comparatively the mcmc technique is computationally expensive , this bayesian treatment has the distinct advantage of providing the posterior probability distributions for each parameter in the complex , multi - dimensional parameter - space of the underlying disk model .
the best - fit model derived from the sma data is compiled in table [ tab : params ] , which lists the mode ( peak ) of the posterior probability distribution for each parameter along with its uncertainty , based on the range of values that encompass 68% of the distribution area ( i.e. , 1@xmath28 uncertainties for a gaussian distribution ) .
figure [ fig : bestfit ] makes a direct comparison between the data and the best - fit model in the spectral image plane , demonstrating the fit quality with only low - level ( statistically insignificant ) residuals present in the channel maps .
the best - fit model has a reduced @xmath111 .
a more detailed view of the multi - dimensional results of the modeling analysis is presented in figure [ fig : posterior ] , where we have taken the 30,000 mcmc samples computed and marginalized over parameter subsets to display one and two - parameter posterior probability distributions .
the marginalized distributions for individual parameters are single - peaked , while the paired distributions highlight internal degeneracies in the model .
for example , there are tight correlations between normalizations and sizes ( e.g. , \{@xmath109 , @xmath48 } , to conserve the integrated line intensity ; this would be modified were @xmath47 a free parameter ) or gradients ( e.g. , \{@xmath61 , @xmath60 } , to maintain an appropriate temperature in the outer disk ; * ? ? ?
* ; * ? ? ?
* ) , as well as more subtle associations amongst parameters related to line broadening ( amongst @xmath48 , @xmath61 , and @xmath97 ; * ? ? ?
* ; * ? ? ?
the degeneracy of most interest here is the \{@xmath9 , @xmath91 } anti - correlation , which is a natural consequence of reproducing the observed line - of - sight velocity pattern , @xmath112 ( i.e. , @xmath113 ) .
the best - fit model parameters that describe the v4046 sgr disk structure are typical for young protoplanetary disks .
the adopted ( fixed ) surface density gradient ( @xmath107 ) and characteristic radius ( @xmath114au ) lie near the median values for a survey of disk structures in the ophiuchus region @xcite ; the temperature profile ( @xmath115k , @xmath116 ; and therefore vertical density distribution ) and turbulent linewidth ( @xmath117 km s@xmath22 ) are comparable to those inferred from co emission in other t tauri disks ( e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
if we assume the co mass fraction found in dark clouds ( @xmath118 ) is also applicable in the disk , the molecular density normalization parameter @xmath119 implies a modest gas mass , @xmath120m@xmath5 , validating _ a posteriori _ our assumption that self - gravity is negligible ( @xmath121 ) . in any case , our simple treatment of the disk structure is really only used as a means to an end : the focus here is to place a firm constraint on the key parameter that determines the behavior of the disk velocity field the central stellar mass , @xmath9 .
we infer a total stellar mass of @xmath4m@xmath5 for the v4046 sgr spectroscopic binary , a formally precise constraint with a relative uncertainty of 3 - 5% .
figure [ fig : posterior ] demonstrates that this estimate of @xmath9 is essentially independent of the disk structure parameters in the model .
the sole relevant degeneracy is with the disk inclination , which we infer to be @xmath6 ( a relative precision of @xmath72 - 5% ) . within the quoted uncertainties , these \{@xmath9 , @xmath91 } values
are consistent with the previous determinations by @xcite , who used a simple @xmath108 grid search and an altogether different underlying disk structure model .
these \{@xmath9 , @xmath91 } measurements based on the circumstellar gas disk kinematics can be directly compared with the joint ( degenerate ) constraints on \{@xmath9 , @xmath11 } imposed by rv measurements of the spectroscopic binary itself .
although various groups have studied this sb2 system @xcite , by far the most robust constraints on the stellar parameters come from the long - term , extensive , and high resolution optical spectroscopic monitoring campaign conducted by @xcite . in that work ,
the rv data are found to be best explained with a binary that has a total mass @xmath122m@xmath5 .
the inset in figure [ fig : posterior ] confirms that our disk kinematics constraints and the rv constraints by @xcite coincide in the stellar mass@xmath2inclination plane : the two _ completely independent _ methods find the same results well within their formal 1@xmath28 uncertainties .
adopting our best estimate of @xmath9 and propagating the relevant uncertainties , the rv constraints suggest that @xmath123 .
therefore , we quantitatively confirm the finding of @xcite : the v4046 sgr spectroscopic binary and its associated , large - scale circumbinary disk are co - planar within @xmath8 across @xmath74 orders of magnitude in radial scale .
we have presented spatially and spectrally resolved submillimeter array observations of the @xmath0co @xmath1=2@xmath21 line emission from the isolated , nearby , and gas - rich circumstellar disk that orbits the close , pre - main sequence double - lined spectroscopic binary v4046 sgr . adopting a simple parametric model for the structure of a flared keplerian disk
, we employ a monte carlo markov chain technique to infer the properties of the disk velocity field from these data , and thereby provide a robust statistical estimate for the total mass of the central binary .
we find that these co line data are best reproduced for a binary mass @xmath4m@xmath5 and a disk viewing angle inclined by @xmath6 from face - on .
those values are in excellent agreement with the completely independent inferences of @xcite , made from their extensive radial velocity monitoring campaign of the spectroscopic binary itself .
the orbital planes of the binary ( @xmath124au ) and its associated circumbinary disk ( on radial scales out to @xmath7400au ) are aligned within @xmath70.1@xmath291 .
these results demonstrate that , despite its complexity , the disk - based kinematic method for estimating the masses of young stars is both _ precise _ ( at the level of a few percent ) and _ accurate _ in an absolute sense , as verified here by an entirely independent dynamical constraint .
the disk - based dynamical estimate of @xmath9 can be combined with the rv constraints of @xcite and other information from the literature to assess the predictions of pre - ms stellar evolution models .
coupling our @xmath9 estimate with the stellar mass ratio ( @xmath60 ) and mass function ( @xmath10 ) determined by @xcite , we can derive the masses of the individual stellar components in the v4046 sgr binary , @xmath125 and @xmath126m@xmath5 , and their orbital inclination , @xmath123 .
assuming that the stellar rotation axes are aligned with the binary orbital axis ( a spin - orbit alignment presumably induced by tides in this close , circularized system ; e.g. , * ? ? ?
? * ; * ? ? ?
* ) , the radial and _ rotational _ velocities measured for each stellar component by @xcite can be used to determine the stellar radii , @xmath127 and @xmath128r@xmath5 .
@xcite inferred spectral types of k5 and k7 in this system , corresponding to effective temperatures of @xmath129 and @xmath130k for the standard conversion advocated by @xcite ; an ambiguity of one spectral subclass has been assumed to estimate the temperature uncertainties .
those measurements imply that the individual stellar luminosities are @xmath131 and @xmath132l@xmath5 ( a total luminosity of @xmath133l@xmath5 ) .
figure [ fig : spectrum ] shows the composite optical / near - infrared spectrum expected from the v4046 sgr binary , given the effective temperatures , stellar radii , and masses derived above from the combination of rv data and our disk - based measurements ( _ black curve _ , with uncertainty in _ shaded gray _ ) .
this spectral energy distribution ( sed ) prediction was generated from an interpolated grid of @xcite model spectra , assuming the appropriate stellar gravities ( @xmath134 in both cases ) , negligible interstellar reddening ( @xmath135 ) , and our adopted @xmath19pc .
representative broadband photometric data are marked as the datapoints in figure [ fig : spectrum ] , constructed from the weighted mean magnitudes of @xcite , @xcite , @xcite , and the 2mass @xcite and denis @xcite surveys : uncertainties were taken as the quadrature sum of the standard deviation of the weighted mean and the maximal deviation from the weighted mean , in an effort to more faithfully represent potential uncertainties due to variability .
a quick examination of figure [ fig : spectrum ] demonstrates that the spectral morphology of the data and model prediction are a good match , although the normalizations appear discrepant .
scaling the ( best - fit ) predicted spectrum down by @xmath730@xmath136% provides a good match to the observed spectrum ; a composite luminosity of @xmath137l@xmath5 is more appropriate ( _ red curve _ ) .
therefore , the predicted and observed composite binary luminosities are marginally ( @xmath71.4@xmath28 ) different .
although formally the conflict in these luminosity values is not statistically significant , it is still interesting to consider some potential paths for a better reconciliation of all the data .
perhaps the most straightforward means of doing so is to modify the assumed distance to the system : taken at face value , a @xmath730% increase to @xmath138pc implies an inherently more luminous pair of stars with a spectrum that would be in good agreement with the observations .
however , that same re - normalization impacts the total stellar mass inferred from the co data , with a linear scaling to @xmath139m@xmath5 that introduces a substantial ( @xmath72@xmath28 ) discrepancy in the \{@xmath9 , @xmath140}-plane between the disk - kinematics and rv techniques for estimating stellar parameters .
although it hardly seems worthwhile to trade one marginal discrepancy for another ( which is technically less marginal ) , there is no _ a priori _ reason that the orbital planes of the disk and binary need to be aligned with the precision inferred in 4 .
unfortunately , little guidance is provided in the form of an uncertainty on the kinematic parallax measurement of @xcite . in any case
, discrepancies in effective temperatures and luminosities are not necessarily uncommon for active young stars @xcite , and particularly for those in close binary systems ( e.g. , * ? ? ?
ultimately , dwelling on a marginal luminosity discrepancy is not well justified , especially given the independent distance estimate from the @xcite study . in the following ,
we adopt the stellar parameters inferred from the joint constraints of the co disk spatio - kinematics and the optical rv studies of @xcite assuming @xmath19pc , but use individual stellar luminosities scaled down by 30% to best match the observed optical and near - infrared spectrum : @xmath141l@xmath5 and @xmath142l@xmath5 ( note that each component is slightly less luminous than reported by * ? ? ?
with those results , we can make comparisons with the predictions of pre - ms evolution models to estimate the ages ( @xmath143 ) and masses of the individual stars in the v4046 sgr binary . to accomplish that
, we follow the bayesian methodology of @xcite , which employs a finely - interpolated grid of pre - ms models in the hr diagram to estimate the probability distributions of stellar mass and age given the measured values ( and uncertainties ) of effective temperatures and luminosities ( see also * ? ? ?
if we define the observables as \{@xmath144 , @xmath145 } = \{@xmath146 , @xmath147 } with associated uncertainties \{@xmath148 , @xmath149 } , and the pre - ms model predictions \{@xmath150 , @xmath151 , @xmath152 , @xmath151 } , we can write the likelihood function as a multivariate gaussian , @xmath153\right\}.\ ] ] the best - fit \{@xmath9 , @xmath143 } are then directly determined from the \{@xmath154 , @xmath155 } that maximize the likelihood , with uncertainties that can be calculated from the shape of the likelihood distribution .
this procedure was conducted for four different pre - ms evolution models , assuming solar composition , a fractional deuterium abundance of @xmath156 , and a convective mixing parameter @xmath157 - 2.0 : @xcite , @xcite , @xcite , and @xcite .
the results of these comparisons are presented in table [ tab : hr ] .
figure [ fig : hr ] shows the v4046 sgr stellar properties that were inferred from each of the four reference sets of pre - ms evolutionary model tracks in the hr diagram .
all of the evolutionary models make predictions for the v4046 sgr stellar masses that are remarkably consistent with each other and the dynamical masses inferred in 4 .
the stellar masses that formally maximize the likelihood in equation 12 are systematically @xmath72 - 10% below the best - fit dynamical masses , an under - prediction typical of pre - ms models regardless of the dynamical method used to estimate @xmath9 @xcite . however , those modest discrepancies are not statistically significant , given the uncertainties on \{@xmath158 , @xmath54 } and the dynamical masses .
an examination of figure [ fig : hr ] demonstrates that the v4046 sgr stars are nearly evolved off the hayashi track , having presumably developed radiative cores as expected for their solar - like masses . the clear implication of their location in the hr diagram is that the v4046 sgr binary is comparatively _ old _ for a pre - ms system , as has been previously reported by @xcite and @xcite .
the models used here suggest a large range of ages are plausible , from @xmath75 - 30myr , and confirm that the binary components are coeval . if we include a gaussian prior representing the dynamical masses into the bayesian analysis of the hr diagrams described above ( labeled as ` + dyn ' in table [ tab : hr ] ) , we can infer a smaller range of acceptable dynamical " ages from each set of pre - ms models .
the different pre - ms model predictions with these dynamical priors are shown together , and averaged ( weighted by the posterior probability for each model ; * ? ? ?
* ) , in figure [ fig : ages ] : the averaged results suggest ages of @xmath159 and @xmath160myr for v4046 sgr a and b , respectively . the corresponding coeval age is 13@xmath161myr , in good agreement with the age constraints from the putative far - flung companion(s ) v4046 sgr c[d ] ( at a separation of @xmath712,350au , or 28 on the sky ; * ? ? ?
@xcite suggested that the observed space motion of v4046 sgr is consistent with a high probability of membership in the @xmath18 pic moving group , a widespread and local ( @xmath162pc ) stellar association with an age range estimated to be @xmath78 - 20myr ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the kinematic parallax calculated by torres et al .
( @xmath3pc ) and component ages estimated here for the v4046 sgr binary are certainly supportive of that conclusion .
if that is the case , it is worth noting the uniqueness of the v4046 sgr circumstellar environment : there are no other @xmath18 pic moving group members that are known to host such a massive , long - lived , gas - rich accretion disk . regardless of its original association , the proximity and advanced age of v4046 sgr make for a remarkable case study in both the long - term evolution of protoplanetary disk structure and the fundamental properties of pre - ms binary stars . although v4046 sgr is a particularly striking example of the methodology behind disk - based dynamical estimates of stellar masses , we anticipate that these techniques will find considerably more use in the near future as the atacama large millimeter array ( alma ) project is completed . even with vastly improved data quality , this simplified model to extract the gas disk rotation curve from such interferometric observations ( see 3 ) remains complex
. however , we have demonstrated that the method is robust , accurate , and precise , by using independent dynamical constraints from the v4046 sgr spectroscopic binary radial velocity monitoring results . in practice , these constraints and others like them ( notably for uz tau e ; see * ? ? ?
* ) effectively serve as a check on the absolute calibration of the modeling technique described in 3 .
the consistency between the independent dynamical constraints on the v4046 sgr stellar masses validates the application of this procedure to dynamically weigh " single , isolated pre - ms stars based on their gas disk kinematics .
along with an investment in accurate stellar luminosity and temperature measurements , alma observations of the molecular gas kinematics in circumstellar disks will usher in a new era of precision in the fundamental astrophysical properties of young stars .
we have presented sensitive , high - resolution sma observations of the @xmath0co @xmath1=2@xmath21 line emission from the massive , gas - rich disk orbiting the double - lined spectroscopic binary star v4046 sgr . using simple radiation transfer calculations for a disk structure model with a keplerian velocity field
, we fit the observed spectral visibilities using a stochastic model optimization technique that simultaneously infers model parameter values and their uncertanties .
our specific focus has been on the key model parameters that describe the disk velocity field , with a goal of placing a firm , dynamical constraint on the mass of the central binary .
the key conclusions of our analysis are : 1 . from modeling the co line emitted by the circumbinary disk
, we infer that the total stellar mass of the v4046 sgr binary is @xmath163 , assuming the kinematic parallax distance of 73pc estimated by @xcite .
that measurement is in excellent agreement with the independent dynamical constraints imposed from the analysis of an optical spectroscopic monitoring campaign of the radial velocity variations from the binary itself @xcite .
the mutual consistency of these distinct dynamical constraints on the stellar masses verifies that the mass determination based on the velocity field of the disk gas is accurate in an absolute sense , as well as remarkably precise ( with a 3 - 5% formal uncertainty on @xmath9 ) .
that same combination of constraints from millimeter and optical spectroscopic measurements confirm that the orbital planes of the stars and their accompanying circumbinary disk are co - planar , with inferred inclination angles ( @xmath6 and @xmath164 ) that differ by @xmath165 over roughly four decades in radius , from @xmath70.04 to 400au .
4 . the inferred component masses of the v4046 sgr binary are in uniformly good agreement with a variety of pre - ms evolution model predictions .
when combined with our dynamical constraints , those same models confirm the coevality of the binary components , and suggest an average age for the system of @xmath166myr .
therefore , v4046 sgr hosts one of the oldest and nearest gas - rich primordial accretion disks currently known .
we are very grateful to joel kastner , guillermo torres and zachory berta for some insightful discussions and suggestions , as well as to germano quast for being kind enough to provide additional information on their optical observations of the v4046 sgr binary .
the sma is a joint project between the smithsonian astrophysical observatory and the academia sinica institute of astronomy and astrophysics and is funded by the smithsonian institution and the academia sinica . ccr @xmath109&@xmath167 $ ] & @xmath168 + @xmath48 & [ au ] & @xmath169 + @xmath60 & & @xmath170 + @xmath61 & [ k ] & @xmath171 + @xmath97 & [ km s@xmath22 ] & @xmath173 + @xmath174 & [ @xmath175 & @xmath176 + @xmath177 & [ m@xmath178 & @xmath179 lcc|ccccc|cc dm98 & @xmath180 & @xmath181 & @xmath182 & @xmath183 & & @xmath184 & @xmath185 & @xmath186 & @xmath187 + bcah98 & @xmath188 & @xmath189 & @xmath190 & @xmath191 & & @xmath192 & @xmath193 & @xmath194 & @xmath195 + sdf00 & @xmath196 & @xmath197 & @xmath198 & @xmath199 & & @xmath200 & @xmath201 & @xmath202 & @xmath203 + tpd11 & @xmath204 & @xmath205 & @xmath206 & @xmath207 & & @xmath208 & @xmath209 & @xmath210 & @xmath211 + @xmath212mean@xmath213 & @xmath214 & @xmath215 & @xmath216 & @xmath217 & & @xmath218 & @xmath219 & @xmath220 & @xmath221 alibert , y. , mordasini , c. , & benz , w. 2011 , , 526 , 63 andrews , s. m. , & williams , j. p. 2007
, , 659 , 705 andrews , s. m. , wilner , d. j. , hughes , a. m. , qi , c. , & dullemond , c. p. 2009 , , 700 , 1502 andrews , s. m. , wilner , d. j. , hughes , a. m. , qi , c. , & dullemond , c. p. 2010 , , 723 , 1241 artymowicz , p. , & lubow , s. h. 1994 , , 421 , 651 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 1998 , , 337 , 403 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 2002 , , 382 , 563 baraffe , i. , chabrier , g. , & gallardo , j. 2009 , , 702 , l27 baraffe , i. , & chabrier , g. 2010 , , 521 , 44 bastian , n. , covey , k. r. , & meyer , m. r. 2010 , , 48 , 339 beckwith , s. v. w. , sargent , a. i. , chini , r. s. , & gsten , r. 1990 , , 99 , 924 beckwith , s. v. w. , & sargent , a. i. 1993 , , 402 , 280 boden , a. f. , sargent , a. i. , akeson , r. l. , et al .
2005 , , 635 , 442 boden , a. f. , torres , g. , sargent , a. i. , et al .
2007 , , 670 , 1214 boden , a. f. , akeson , r. l. , sargent , a. i. , et al .
2009 , , 696 , l111 boden , a. f. , torres , g. , duchne , g. , et al .
2012 , , 747 , 17 byrne , p. b. 1986 , ir .
j. , 17 , 294 dalessio , p. , canto , j. , calvet , n. , & lizano , s. 1998 , , 500 , 411 dantona , f. , & mazzitelli , i. 1998 , in asp conf .
134 , brown dwarfs and extrasolar planets , ed .
r. rebolo , e. l. martin , & m. r. z. osorio ( san francisco , ca : asp ) , 442 dantona , f. , ventura , p. , & mazzitelli , i. 2000 , , 543 , l77 dartois , e. , dutrey , a. , & guilloteau , s. 2003 , , 399 , 773 donati , j .- f .
, et al . 2011 , , 417 , 1747 duchne , g. , beust , h. , adjali , f. , konopacky , q. m. , & ghez , a. m. 2006 , , 457 , l9 dutrey , a. , guilloteau , s. , & simon , m. 1994 , , 286 , 149 dutrey , a. , guilloteau , s. , prato , l. , simon , m. , duvert , g. , schuster , k. , & mnard , f. 1998 , , 338 , l63 epchtein , n. , et al . 1997 ,
messenger , 87 , 27 foreman - mackey , d. , hogg , d. w. , lang , d. , & goodman , j. 2012 , arxiv:1202.3665 gennaro , m. , prada moroni , p. g. , & tognelli , e. 2012 , , 420 , 986 goodman , j. , & weare , j. 2010 , comm .
math . comp .
, 5 , 65 gomz maqueo chew , y. , et al .
2012 , , 745 , 58 guilloteau , s. , & dutrey , a. 1998 , , 339 , 467 hartmann , l. , calvet , n. , gullbring , e. , & dalessio , p. 1998
, , 495 , 385 hillenbrand , l. a. , & white , r. j. 2004 , , 604 , 741 ho , p. t. p. , moran , j. m. , & lo , k. y. 2004 , , 616 , l1 hoeting , j. a. , madigan , d. , raftery , a. e. , & volinsky , c. t. 1999 , statist .
, 14 , 4 hughes , a. m. , wilner , d. j. , qi , c. , & hogerheijde , m. r. 2008 , , 678 , 1119 hughes , a. m. , wilner , d. j. , andrews , s. m. , qi , c. , & hogerheijde , m. r. 2011 , , 727 , 85 hut , p. 1981
, , 99 , 126 hutchinson , m. g. , evans , a. , winkler , h. , & spencer jones , j. 1990 , , 234 , 230 isella , a. , testi , l. , natta , a. , neri , r. , wilner , d. , & qi , c. 2007 , , 469 , 213 jensen , e. l. n. , & mathieu , r. d. 1997 , , 114 , 301 jrgensen , b. r. , & lindegren , l. 2005 , , 436 , 127 kastner , j. h. , zuckerman , b. , hily - blant , p. , & forveille , t. 2008 , , 492 , 469 kastner , j. h. , sacco , g. g. , montez , r. , et al . 2011 , , 740 , l17 koerner , d. w. , sargent , a. i. , & beckwith , s. v. w. 1993 , icarus , 106 , 2 lejeune , t. , cuisinier , f. , & buser , r. 1997 , , 125 , 229 lynden - bell , d. , & pringle , j. e. 1974 , , 168 , 603 mathieu , r. d. , walter , f. a. , & myers , p. c. 1989 , , 98 , 987 mathieu , r. d. , adams , f. c. , & latham , d. w. 1991 , , 101 , 2184 mathieu , r. d. , et al .
1997 , , 113 , 1841 melo , c. h. f. , covino , e. , alcal , j. m. , & torres , g. 2001 , , 378 , 898 mendes , l. t. s. , dantona , f. , & mazzitelli , i. 1999 , , 341 , 174 morales - caldern , m. , stauffer , j. r. , stassun , k. g. , et al .
2012 , , 753 , 149 mundy , l. g. , et al .
1996 , , 464 , l169 berg , k. i. , et al .
2011 , , 734 , 98 pavlyuchenkov , y. , semenov , d. , henning , t. , guilloteau , s. , pitu , v. , launhardt , r. , & dutrey , a. 2007 , , 669 , 1262 pitu , v. , dutrey , a. , & guilloteau , s. 2007 , , 467 , 163 prato , l. , simon , m. , mazeh , t. , zucker , s. , & mclean , i. s. 2002 , , 579 , l99 quast , g. r. , torres , c. a. o. , de la reza , r. , da silva , l. , & mayor , m. 2000 , iau symposium , 200 , 28p rodriguez , d. r. , kastner , j. h. , wilner , d. , & qi , c. 2010 , , 720 , 1684 schaefer , g. h. , simon , m. , nelan , e. , & holfeltz , s. t. 2003 , , 126 , 1971 schaefer , g. h. , simon , m. , beck , t. l. , nelan , e. , & prato , l. 2006 , , 132 , 2618 schaefer , g. h. , simon , m. , prato , l. , & barman , t. 2008 , , 135 , 1659 schmidt - kaler , t. 1982 , in landlt bornstein , group vi , vol .
hellwege ( berlin : springer ) , 454 schier , f. l. , van der tak , f. f. s. , van dishoeck , e. f. , & black , j. h. 2005 , , 432 , 369 simon , m. , dutrey , a. , & guilloteau , s. 2000 , , 545 , 1034 siess , l. , forestini , m. , & dougados , c. 1997 , , 326 , 1001 siess , l. , & livio , m. 1997 , , 490 , 785 siess , l. , dufour , e. , & forestini , m. 2000 , , 358 , 593 skrutskie , m. f. , et al .
2006 , , 131 , 1163 stassun , k. g. , mathieu , r. d. , vaz , l. p. r. , stroud , n. , & vrba , f. j. 2004 , , 151 , 357 stassun , k. g. , kratter , k. m. , scholz , a. , & dupuy , t. j. 2012 , , in press ( arxiv:1206.4930 ) steffen , a. t. , et al .
2001 , , 122 , 997 stempels , h. c. , & gahm , g. f. 2004 , , 421 , 1159 stempels , h. c. , 2012 , in preparation strassmeier , k. g. , hall , d. s. , fekel , f. c. , & scheck , m. 1993 , , 100 , 173 tamazian , v. s. , docobo , j. a. , white , r. j. , & woitas , j. 2002 , , 578 , 925 tognelli , e. , prada moroni , p. g. , & deglinnocenti , s. 2011 , , 533 , a109 torres , c. a. o. , quast , g. r. , da silva , l. , de la reza , r. , melo , c. h. f. , & sterzik , m. 2006 , , 460 , 695 torres , c. a. o. , quast , g. r. , melo , c. h. f. , & sterzik , m. f. 2008 , handbook of star forming regions , volume ii , 757 zacharias , n. , et al .
2010 , , 139 , 2184 zahn , j .- p .
1977 , , 57 , 383 zuckerman , b. , song , i. , bessell , m. s. , & webb , r. a. 2001 , , 562 , l87 zuckerman , b. , & song , i. 2004 , , 42 , 685 | we present sensitive , arcsecond - resolution submillimeter array observations of the @xmath0co @xmath1=2@xmath21 line emission from the circumstellar disk orbiting the double - lined spectroscopic binary star v4046 sgr .
based on a simple model of the disk structure , we use a novel monte carlo markov chain technique to extract the keplerian velocity field of the disk from these data and estimate the total mass of the central binary . assuming the distance inferred from kinematic parallax measurements in the literature ( @xmath3pc ) , we determine a total stellar mass @xmath4m@xmath5 and a disk inclination @xmath6 from face - on .
these measurements are in excellent agreement with independent dynamical constraints made from multi - epoch monitoring of the stellar radial velocities , confirming the absolute accuracy of this precise ( @xmath7few percent uncertainties ) disk - based method for estimating stellar masses and reaffirming previous assertions that the disk and binary orbital planes are well aligned ( with @xmath8 ) . using these results as a reference
, we demonstrate that various pre - main sequence evolution models make consistent and accurate predictions for the masses of the individual components of the binary , and uniformly imply an advanced age of @xmath75 - 30myr . taken together
, these results verify that v4046 sgr is one of the precious few nearby and relatively evolved pre - main sequence systems that still hosts a gas - rich accretion disk . |
the layered structure of cuprates makes these materials good candidates for observing zeeman response to a magnetic field @xmath2 directed parallel to the cu - o planes @xcite .
moreover , the @xmath3 symmetry of the order parameter ( hereafter @xmath0-wave ) leads in principle to substantial differences with respect to the zeeman response of isotropic @xmath1-wave superconductors .
for example , at zero temperature , the tunneling conductance @xmath4 of a @xmath0-wave superconductor - insulator - metal junction is nonzero for finite voltages @xmath5 provided @xmath6 @xcite , in sharp contrast to ordinary isotropic @xmath1-wave junctions for which @xmath4 is zero for @xmath7 , where @xmath8 is the energy gap and @xmath9 is the electron charge @xcite . on the other hand , the phase diagrams of pure @xmath1-wave and @xmath0-wave superconductors in the presence of a zeeman magnetic field have similar qualitative behaviours .
for example , for both symmetries of the order parameter , a first order phase transition to the normal state is found at low temperatures and for sufficiently strong magnetic fields @xcite .
however there are quantitative differences .
for example , at @xmath10 , the critical field is @xmath11 for @xmath1-wave @xcite and @xmath12 for @xmath0-wave @xcite , where @xmath13 is the zero temperature order parameter without magnetic field . so far ,
systematic theoretical studies of the zeeman response of anisotropic superconductors have been focused on the clean limit of @xmath0-wave bcs formulation .
a more realistic situation would require the inclusion of impurity effects , since these are known to have important effects on both thermodynamic and spectral quantities @xcite .
moreover , in addition to the disorder potential , the quasiparticles are also spin - orbit coupled to the impurities , so that the zeeman response is affected by spin - mixing processes . an additional source for spin - orbit effects could be provided by the electric fields in the vicinity of the conducting cu - o layers and the charge reservoirs interfaces .
although , in the past years , the effect of spin - orbit coupling has been largely studied for isotropic @xmath1-wave superconductors @xcite , the corresponding situation for @xmath0-wave superconductors ( or other anisotropic symmetries ) is still unknown .
however , it is expected that the spin - orbit effects on the zeeman response of @xmath0-wave superconductors differ from those of @xmath1-wave superconductors in a qualitative way . in fact , already at zero magnetic field , the spin - orbit scattering is pair - breaking and reduces both the critical temperature @xmath14 and the order parameter @xcite . as a consequence , for @xmath15 , the pair - breaking effects of both the external magnetic field and the spin - orbit coupling add together .
this situation must be contrasted with the @xmath1-wave case , where the spin - orbit potential is not pair - breaking and competes with the zeeman response reducing the pair - breaking effect of the magnetic field @xcite . in this paper , the effects of both impurity and spin - orbit scattering potentials are studied within a self - consistent born approximation for @xmath0-wave superconductors .
both thermodynamic and spectral properties are investigated and compared with those of @xmath1-wave superconductors .
let us consider a two - dimensional system with electrons ( holes ) moving in the @xmath16-@xmath17 plane under the effect of an external magnetic field @xmath2 directed along the plane .
in this situation , the coupling of the orbital motion of the charge carriers to the magnetic field is vanishingly small . in the following , no particular pairing mechanism is assumed and the condensate will be described within the bcs formalism . in this framework ,
the hamiltonian is : @xmath18 where @xmath19 and @xmath20 is the bohr magneton . for a @xmath21 symmetry of the gap ,
@xmath22 is parametrized as follows : @xmath23 where @xmath24 is the polar angle in the @xmath25-@xmath26 plane . in eq.([hami0 ] ) , @xmath27 and @xmath28 refer to the spin direction along and opposite to the direction of @xmath2 and it is assumed that @xmath2 is directed along the @xmath16 direction , so that @xmath29 . the interaction hamiltonian describing the coupling to the impurities located randomly at @xmath30 is given below : @xmath31\cdot \mbox{\boldmath $ \sigma$}_{\alpha\beta } ) c^{\dagger}_{{\bf k}\alpha}c_{{\bf k}'\beta},\end{aligned}\ ] ] where @xmath32 and @xmath33 refer to the non - magnetic and spin - orbit coupling to the impurities , respectively ( @xmath34 is the fermi momentum ) . from the elliott relation @xcite ,
the impurity and spin - orbit potentials are roughly given by @xmath35 , where @xmath36 is the shift of the g - factor which , for cuprates , is of order @xmath37 . here
, however , @xmath32 and @xmath33 will be treated as independent variables .
note that , since the momenta @xmath38 and @xmath39 are defined in the @xmath16-@xmath17 plane , the spin - momentum dependence of the spin - orbit interaction simplifies to : @xmath40\cdot \mbox{\boldmath $ \sigma$}_{\alpha\beta}= [ { \bf k}\times{\bf k}']\cdot { \bf \hat{z}}\,\sigma^z_{\alpha\beta}.\ ] ] since the spins have been quantized along the @xmath16-axis , the spin - orbit coupling leads to scattering events always accompanied by spin - flip transitions .
the following analysis is simplified by introducing the usual four - components field operators @xcite : @xmath41;\,\ , \psi^{\dagger}_{{\bf k}}=\left[c^{\dagger}_{{\bf k}\uparrow } , c^{\dagger}_{-{\bf k}\downarrow},c_{{\bf k}\uparrow } , c_{-{\bf k}\downarrow}\right].\ ] ] from eqs.([hami0],[hamiimp ] ) it is possible to evaluate the equation of motion of the field operator @xmath42 in the imaginary time @xmath43 : @xmath44\cdot { \bf \hat{z}}\,\tau_1\right]\psi_{{\bf k}'},\end{aligned}\ ] ] where the products @xmath45 are @xmath46 matrices acting on the field operators ( [ four ] ) .
they are constructed by treating the pauli matrices @xmath47 as elements of the pauli matrices @xmath48 as shown in the example below : @xmath49 equation ( [ motion ] ) permits to evaluate the equation of motion of the generalized green s function defined as : @xmath50 where @xmath51 is the @xmath43-order operator .
it is straightforward to obtain from eqs.([motion],[green0 ] ) the equation satisfied by the generalized green s function in the matsubara frequencies @xmath52 : @xmath53 where @xmath54\cdot { \bf \hat{z}}\,\tau_1 , \ ] ] and @xmath55^{-1 } , \ ] ] is the green s function in the absence of impurities . the average over all the impurity configurations of eq.([green1 ] )
leads to the averaged green s function @xmath56 which satisfies the following dyson equation @xcite : @xmath57 where @xmath58 is the electron self - energy resulting from the average procedure . in this paper , @xmath32 and @xmath33
are assumed to be sufficiently weak to justify a self - consistent born approximation for the self - energy @xmath58 .
because of the momentum dependence of the spin - orbit interaction , the feynmann diagrams describing the born approximation do not involve impurity spin - orbit mixed terms and @xmath58 is given by the diagrams showed in fig .
[ bornfig1 ] .
therefore , by using eq.([pot ] ) , the self - consistent born approximation for @xmath58 reads : @xmath59 where @xmath60 is the concentration of impurities .
equations ( [ green3 ] ) and ( [ self ] ) must be solved self - consistently and the solution can be written in terms of the following renormalized green s function @xcite : @xmath61 where for brevity the momentum and frequency dependence of the tilded quantities has been omitted .
the renormalized quantities can be calculated by substituting eq.([green4 ] ) into eqs.([green3],[self ] ) .
if there is particle - hole symmetry , the quasiparticle dispersion remains unaffected by the presence of impurities , _
i.e. _ , @xmath62 and @xmath63 . for the other quantities
it is useful to introduce the variables @xmath64 and @xmath65 defined by : @xmath66 in this way the self - consistent equations become : @xmath67 @xmath68 the summations over momenta are transformed in integrations over energy according to the usual procedure : @xmath69 where @xmath70 is the electronic density of states per spin state at the fermi level .
performing the integration over the energy @xmath71 , equations ( [ omega1 ] ) and ( [ delta1 ] ) reduce to : @xmath72^{1/2 } } + \frac{1}{\tau_{so}}\int\frac{d\phi'}{2\pi } \frac{s(\phi,\phi')\tilde{\omega}_{\mp } } { [ \tilde{\omega}_{\mp}^2+\tilde{\delta}_{\mp}(\phi')^2]^{1/2}},\ ] ] @xmath73^{1/2 } } + \frac{1}{\tau_{so}}\int\frac{d\phi'}{2\pi } \frac{s(\phi,\phi')\tilde{\delta}_{\mp}(\phi ' ) } { [ \tilde{\omega}_{\mp}^2+\tilde{\delta}_{\mp}(\phi')^2]^{1/2 } } , \ ] ] where @xmath74 and @xmath75 are the scattering rates for the non - magnetic and spin - orbit impurities , respectively .
they are given by : @xmath76 in eqs.([omega2],[delta2 ] ) , the function @xmath77 stems from the angular dependence of the spin - orbit factor @xmath78 by defining @xmath24 and @xmath79 as the polar angles of the versors @xmath80 and @xmath81 , respectively . in explicit form
, the function @xmath77 is given by : @xmath82 the presence of such angular function leads to important differences between non - magnetic and spin - orbit impurity effects also for zero magnetic field .
in fact , non - magnetic impurities does not renormalize the gap function when this has @xmath0-wave symmetry @xcite whereas the spin - orbit interaction , by means of the angular function @xmath77 , provides a finite renormalization .
this can be readily seen by realizing that if @xmath83 is of the form given by eq.([gap1 ] ) , then a consistent solution of eq.([delta2 ] ) is provided by setting @xmath84 , where @xmath65 is the solution of : @xmath85^{1/2}},\ ] ] and , in the same way , eq.([omega2 ] ) becomes : @xmath86^{1/2 } } + \frac{1}{\tau_{so}}\int\frac{d\phi}{2\pi}\sin(\phi)^2 \frac{\tilde{\omega}_{\mp } } { [ \tilde{\omega}_{\mp}^2+\tilde{\delta}_{\mp}^2\cos(2\phi)^2]^{1/2}}. \nonumber \\\ ] ] in obtaining eqs.([delta3],[omega3 ] ) , it has been used the identity : @xmath87= \int\frac{d\phi}{2\pi}\sin(\phi)^2 f[-\cos(2\phi)],\ ] ] where @xmath88 $ ] is a general function of @xmath89 .
as expected , the scalar impurity scattering contribution has disappeared from the gap renormalization ( [ delta3 ] ) . on the contrary
, the spin - orbit interaction modifies the gap function because of the presence of the angular function ( [ esse ] ) .
moreover , equations ( [ delta3 ] ) and ( [ omega3 ] ) are renormalized in a different way by @xmath33 so that , even at zero magnetic field , the spin - orbit interaction contributes to the thermodynamic and spectral properties of @xmath0-wave superconductors .
in fact , all the measurable quantities can be expressed in terms of @xmath90 @xcite which from eqs.([delta3],[omega3 ] ) satisfies the following equation : @xmath91^{1/2 } } + \frac{1}{\delta\tau_{so}}\int\frac{d\phi}{2\pi}\sin(\phi)^2 \frac{\tilde{u}_{\mp}-\tilde{u}_{\pm}\cos(2\phi ) } { [ \cos(2\phi)^2+\tilde{u}_{\mp}^2]^{1/2}}.\ ] ] the above equation should be compared with the corresponding expression for the two - dimensional isotropic @xmath1-wave case which , by setting @xmath92 in eqs.([omega2],[delta2 ] ) , is found to be @xcite : @xmath93^{1/2}},\ ] ] where the contribution of the impurity scattering has vanished because of anderson s theorem .
when @xmath94 , equation ( [ swave ] ) reduces to @xmath95 and does not depend on the spin - orbit scattering rate while eq.([u1 ] ) still depends on @xmath43 and @xmath96 .
in fact , in a @xmath0-wave superconductor , both the non - magnetic impurity and the spin - orbit scatterings are pair breaking and they tend to suppress superconductivity @xcite .
when @xmath6 , it is therefore expected that the zeeman response of a @xmath0-wave superconductor differs qualitatively from that of a @xmath1-wave condensate .
equation ( [ u1 ] ) permits to obtain all the informations needed to calculate the phase diagram of a dirt @xmath0-wave superconductor in a zeeman magnetic field .
let us start by considering the self - consistent equation for the order parameter @xmath8 : @xmath97= \lambda \pi t\sum_n \int\frac{d\phi}{2\pi } \mbox{re}\frac{\cos(2\phi)^2 } { [ \cos(2\phi)^2+\tilde{u}_+^2]^{1/2}},\ ] ] where @xmath98 is the pairing interaction and @xmath99 .
the summation over the frequencies is implicitly assumed to be restricted by a cut - off energy .
however , both the cut - off frequency and the pairing interaction can be absorbed in the definition of the critical temperature @xmath100 for a pure superconductor ( @xmath101 , @xmath102 ) without external magnetic field . in this way
the gap equation can be rewritten as : @xmath103^{1/2 } } , -\frac{1}{2\omega_n}\right\}\ ] ] in the hypothesis that the transition to the normal state is of the second order ( see below ) , the critical temperature @xmath14 is obtained from eq.([gap2 ] ) by setting @xmath104 and it is given by : @xmath105,\ ] ] where @xmath106 and @xmath107^{1/2}$ ] and @xmath108 is the di - gamma function . when @xmath109 equation ( [ tc1 ] ) reduces to : @xmath110 which coincides with the result obtained in ref.@xcite in the weak scattering limit @xcite .
equation ( [ tc2 ] ) shows that , even at zero magnetic field , the spin - orbit scattering contributes together with the non - magnetic impurity scattering to the suppression of @xmath14 . for large enough values of the external magnetic field ,
the transition to the normal state becomes of first order @xcite .
this situation is studied by evaluating the difference of the free energy between the superconducting and the normal states @xmath111 .
if , by rising the temperature and/or the magnetic field , @xmath112 changes sign while @xmath8 remains finite , then the system undergoes a first order phase transition to the normal state with critical field @xmath113 and @xmath14 determined by @xmath114 @xcite . by following ref.@xcite ,
@xmath112 is obtained by : @xmath115 and by using eqs.([u1],[gap2 ] ) one readily finds : @xmath116^{1/2}-2u_+-\frac{\cos(2\phi)^2 } { [ \cos(2\phi)^2+u_+^2]^{1/2}}\right\}.\ ] ] the numerical solution of equations ( [ tc1 ] ) and ( [ free2 ] ) are shown in fig .
[ bornfig2 ] for the pure limit and , for comparison , the @xmath0-wave solution is plotted together with the @xmath1-wave one . in the phase diagram ,
the solid and dashed lines are solutions of equations ( [ tc1 ] ) and ( [ free2 ] ) , respectively . for both @xmath0-wave and @xmath1-wave ,
the transition to the normal state is of second order for @xmath117 @xcite while for lower temperatures the transition , marked by dashed lines , becomes of first order . for @xmath118
the solid lines represent the supercooling field @xcite . as already said in the introduction , at zero temperature the first order transition to the normal state
is obtained by critical fields @xmath11 @xcite for @xmath1-wave and @xmath12 for @xmath0-wave @xcite . in this paper ,
the fulde - ferrel larkin - ovchinnikov state @xcite which appears at low temperatures has not been considered since disorder tends to restore the zero momentum pairing @xcite . for the pure @xmath0-wave case
, the reader can find the phase diagram including the non - zero momentum pairing state in ref.@xcite .
although for the pure limit the phase diagrams of the zeeman response of @xmath1-wave and @xmath0-wave superconductors are qualitatively similar , they drastically differ when the coupling to the non - magnetic and spin - orbit impurity scatterings is switched on . in figs .
[ bornfig3 ] and [ bornfig4 ] , the phase diagrams for @xmath1-wave and @xmath0-wave superconductors are plotted for finite values of @xmath74 and @xmath119 . in both figures ,
the impurity scattering parameter @xmath120 is set equal to @xmath37 , while the spin - orbit scattering parameter @xmath121 assumes four different values : @xmath122 . in the @xmath1-wave case ,
3 , the phase diagram is insensitive to @xmath123 , while finite values of @xmath124 enhance the critical field for all temperatures .
moreover , the temperature interval of first - order phase transition ( dashed lines ) decreases as @xmath124 increases and for @xmath125 the transition becomes continuous for all temperatures @xcite .
this behavior is due to the spin - mixing effect of the spin - orbit interaction which lowers the zeeman response and consequently the depairing effect of the magnetic field . on the other hand , in the @xmath0-wave case shown in fig .
4 , the spin - orbit scattering is pair breaking and for @xmath126 the critical field is lowered .
this situation can be understood by realizing that finite values of @xmath124 lead to a weakening of the superconducting state @xcite so that , with respect to the @xmath127 case , lower values of @xmath128 are needed to suppress completely superconductivity .
another striking feature is that , due to the nodes of the @xmath0-wave order parameter , the @xmath124 dependence of the phase diagram is much stronger than for the @xmath1-wave case .
in fact , already for @xmath129 there is not signature for first - order transition whereas for an @xmath1-wave superconductor the first - order transition disappears only for @xmath130 , _
i. e. _ , a difference of one order of magnitude .
it is important to stress that the remarkable difference between the @xmath0-wave and @xmath1-wave phase diagrams has been obtained in the born limit of non - magnetic and spin - orbit impurity scatterings .
however it is well known that in high-@xmath14 superconductors the effect of disorder is best described by the strongly resonant limit of the impurity potential so that the born approximation may result inadequate . in practice
, one should formulate the zeeman response by employing the t - matrix approximation for both the non - magnetic and the spin - orbit potentials .
such a calculation has already been reported in ref.@xcite for zero external magnetic field .
the generalization for @xmath15 is currently under investigation and the results reported here may provide a useful tool for testing the more general t - matrix solution .
having described the effect of @xmath131 and @xmath124 on the phase diagram , it is interesting to investigate also how the spectral properties are modified . to this end , equation ( [ u1 ] ) must be analytically continued to the real axis by setting @xmath132 and @xmath133 . in this way
, the quasiparticle density of states ( dos ) per spin direction in units of the normal state dos @xmath70 can be calculated by the following expression : @xmath134^{1/2}}.\ ] ] in fig .
[ bornfig5 ] it is reported the quasiparticle dos for @xmath135 , @xmath136 and different values of the spin - orbit parameter @xmath124 . for clarity , the curves with @xmath137 have been vertically shifted by @xmath138 , @xmath139 and @xmath140 with respect to those with @xmath127 . for @xmath127 ,
the two dos per spin state , @xmath141 ( dashed lines ) and @xmath142 ( solid lines ) , show a clear zeeman splitting and for @xmath143 the total dos @xmath144 is different from zero as expected for a @xmath0-wave superconductor . for @xmath126
the total dos at @xmath143 is enhanced at the expenses of the coherence peaks which show a decrease of spectral weight . moreover , at @xmath145 , @xmath142 develops a structure ( marked by the arrows ) which becomes a peak at @xmath129 .
such a structure is even more visible in fig .
[ bornfig6 ] where @xmath146 is plotted for @xmath136 , @xmath147 and different values of the external magnetic field .
the origin of this peak can be understood by the following reasoning . at the fermi wave - vector , and for a pure superconductor ,
the quasiparticle energies for spin up and down are @xmath148 and therefore , depending on the values of @xmath128 and @xmath8 , two quasiparticles with different spin orientation and angles @xmath24 can have equal energies .
for example , for @xmath149 and @xmath150 , the two energies @xmath151 and @xmath152 are equal to @xmath153 if @xmath154 .
since the spin - orbit potential connects quasiparticle states with different spin orientation but equal energies , the two states @xmath151 and @xmath152 are coupled by the spin - orbit interaction and a enhanced signal should be expected at @xmath145 .
note in fact that in fig .
[ bornfig6 ] the low energy peak is more pronounced for @xmath155 where , since @xmath156 , the condition @xmath154 is nearly fulfilled .
in conclusion , it has been shown within the born approximation that the presence of impurity and spin - orbit scattering centres strongly affects the zeeman response of a @xmath0-wave superconductor .
both the phase diagram and the quasiparticle density of states show features qualitatively different from those of a @xmath1-wave superconductor .
in fact , by increasing the value of the spin - orbit scattering parameter @xmath121 the critical field @xmath157 is strongly lowered whereas in a @xmath1-wave superconductor @xmath157 increases .
moreover , the influence of @xmath124 on the superconducting state is much stronger for the @xmath0-wave symmetry . concerning the spectral properties , the zeeman splitted density of states of a @xmath0-wave superconductor shows interesting features which are missing in a @xmath1-wave superconductor . in fact , for sufficiently large values of @xmath124 and/or @xmath128 a resonant peak develops at energies close to @xmath158 .
the origin of this feature is given by the anisotropy of the order parameter and the spin - flip transitions due to the spin - orbit scattering .
an important open question concerns with the possibility of going beyond the born approximation and employing a t - matrix approach for the zeeman response in the presence of impurity and spin - orbit scattering centres in order to test the solidity of the results here presented . in ref.@xcite
a different notation has been used in which the impurity potential is parametrized by @xmath159 and @xmath160 and the spin - orbit interaction is parametrized as @xmath161 , where @xmath36 is the shift of the electronic @xmath162-factor .
therefore , by using eq.([scatte ] ) , @xmath163 and @xmath164 . | the effects of impurity and spin - orbit scattering potentials can strongly affect the zeeman response of a @xmath0-wave superconductor . here , both the phase diagram and the quasiparticle density of states are calculated within the born approximation and it is found that the spin - orbit interaction influences in a qualitatively different way the zeeman response of @xmath0-wave and @xmath1-wave superconductors .
= 10000 + + pacs : 74.20.fg ; 71.70.ej ; 74.62.dh + _ keywords _ : @xmath0-wave superconductivity ; spin - orbit coupling ; zeeman splitting . |
this paper applies the string method@xcite to the phenomenon of hydrophobic collapse .
it is shown that the method can describe complex dynamics in large atomistic systems , ones for which other currently available rare event methods would seem intractable .
furthermore , the string method is used to demonstrate that atomistic dynamics can be usefully projected onto that of a coarse - grained field .
the specific application of the string method considered herein finds results that are consistent with the mechanism of hydrophobic collapse put forward by ten wolde and chandler.@xcite the hydrophobic effect - or the tendency of oil and water to separate on mesoscopic lengthscales - has long been recognized as an important driving force in molecular assembly.@xcite it stabilizes the formation of micelles , protein tertiary structure , multi - protein assemblies , and cellular membranes.@xcite recent theoretical developments have helped establish a quantitative understanding of the thermodynamics of hydrophobicity,@xcite but the dynamics of hydrophobic collapse remain poorly understood because it couples a large range of length- and timescales .
relevant processes include the atomic - scale motions of individual water molecules , collective solvent density fluctuations , and the nanometer - scale movements of the hydrophobic solutes .
bridging these dynamical hierarchies - and addressing the problem of complex dynamics in large systems - is a fundamental challenge for computational methods .
we address this challenge using the string method in collective variables.@xcite we consider the collapse of a chain composed of twelve spherical hydrophobes in an explicit solvent of approximately 34,000 water molecules .
the system is studied by coarse - graining the water molecule positions onto a set of 129,000 collective variables that represent the solvent density field and then using the string method in these variables to compute the minimum free energy path ( mfep ) for the hydrophobic collapse of the chain .
conventional molecular dynamics ( md ) simulations are subsequently performed to confirm that this coarse - grained description adequately describes the mechanism of hydrophobic collapse .
ten wolde and chandler@xcite have previously reported simulations of a non - atomistic model for the hydrated chain considered herein .
it was found that the key step in the collapse dynamics is a collective solvent density fluctuation that is nucleated at the hydrophobic surface of the chain .
however , it was not clear whether this proposed mechanism captured the essential features of hydrophobic collapse or whether it was an artifact of the model .
atomistic simulations were needed to resolve the issue .
previous efforts to characterize the mechanism of hydrophobic collapse using atomistic computer simulations neither confirm nor disprove the mechanism proposed by ten wolde and chandler . in recent work , for example , molecular dynamics ( md ) simulations showed that dewetting accompanies the collapse of hydrophobes in water.@xcite however , the rate - limiting step - and thus the mechanism for hydrophobic collapse - was not characterized .
specifically , in ref . , md trajectories were initiated at various separation distances for a pair of melettin protein dimers .
observation of the collapse dynamics was observed only when the initial configuration was on the `` near '' side of the free energy barrier , but the actual nature of that barrier - and the dynamics of crossing it - were not studied . in ref .
, the thermodynamics and solvation of a hydrophobic chain was studied as a function of its radius of gyration , but again , the dynamics of collapse was not characterized .
the results presented here are the first atomistic simulations to explicitly confirm the mechanism of hydrophobic collapse put forward by ten wolde and chandler.@xcite in particular , we show that the rate - limiting step in hydrophobic collapse coincides with a collective solvent motion and that performing the rate - limiting step involves performing work almost exclusively in the solvent coordinates .
we further show that the solvation free energy along the mfep can be decomposed into small- and large - lengthscale contributions .
this analysis demonstrates that atomistic solvent energetics can be quantitatively modeled using a grid - based solvent density field , and it suggests that the rate - limiting step for hydrophobic collapse coincides with lengthscale - dependent hydrophobic dewetting .
combined , the results presented in this study provide ( 1 ) evidence that the string method is a powerful tool for atomistically simulating complex dynamics in large systems , ( 2 ) a proof of principle that atomistic solvent dynamics can be usefully projected onto that of a coarse - grained field , and ( 3 ) an atomistic demonstration that dewetting is key to the dynamics of hydrophobic collapse .
we consider the atomistic version of the hydrated hydrophobic chain studied by ten wolde and chandler.@xcite the chain is composed of twelve spherical hydrophobes , each of diameter @xmath0 and mass @xmath1 amu .
these are ideal hydrophobes , as they exert purely repulsive interactions upon the oxygen atoms of the water molecules .
consecutive hydrophobes in the unbranched chain interact via harmonic bonds , and the chain is made semi - rigid by a potential energy term that penalizes its curvature .
the chain is hydrated with approximately @xmath2 rigid water molecules in an orthorhombic simulation box with periodic boundary conditions .
all simulations were performed at @xmath3 k. full details of the system are provided in the supporting information sec .
a. in describing hydrophobic collapse , it is necessary to choose an appropriate thermodynamic ensemble for the simulations .
nanometer - scale fluctuations in solvent density are suspected to play a key role in these dynamics .
use of the nvt ensemble ( with a 1 g/@xmath4 density of water ) might suppress these density fluctuations and bias the calculated mechanism .
we avoid this problem with a simple technique that is based on the fact that under ambient conditions , liquid water is very close to phase coexistence .
indeed , it is this proximity that leads to the possibility of large lengthscale hydrophobicity.@xcite by placing a fixed number of water molecules at @xmath3 k in a volume that corresponds to an average density of less than 1 g/@xmath4 , we obtain a fraction of the system at the density of water vapor and the majority at a density of bulk water .
since we are not concerned with solvent fluctuations on macroscopic lengthscales , the difference between simulating bulk water at its own vapor pressure compared to atmospheric pressure is completely negligible .
this strategy has been previously employed to study the dewetting transition between solvophobic surfaces.@xcite to ensure that the liquid - vapor interface remains both flat and well - distanced from the location of the chain , we repel particles from a thin layer at the top edge of the simulation box , as is discussed in supporting information sec .
a. throughout this study , we simulate the hydrated chain using atomistic md , but we employ the string method using collective variables that describe the solvent density field .
a coarse - graining algorithm is developed to connect these atomistic and collective variable representations of the solvent .
following ten wolde and chandler,@xcite the simulation box is partitioned into a three - dimensional lattice ( 48@xmath548@xmath556 ) of cubic cells with sidelength @xmath6 .
we label the cells with the vector @xmath7 , where each @xmath8 takes on integer values bounded as follows : @xmath9 , @xmath10 , and @xmath11 . on this grid ,
the molecular density , @xmath12 , is coarse grained into the field @xmath13 , where @xmath14 here , the integral extends over the volume of the system , @xmath15 denotes the unit vector in the @xmath16 cartesian direction , and the coarse graining function , @xmath17 , must be normalized , i.e. , @xmath18 .
the particular function that we have chosen to use is @xmath19\nonumber\end{aligned}\ ] ] where @xmath20 @xmath21 , and @xmath22 is unity when @xmath23 is in the @xmath24 interval and zero otherwise .
since @xmath17 is normalized , @xmath25 is the total number of water molecules . in effect
, this choice spreads the atomistic density field @xmath12 over the lengthscale @xmath26 and bins it into a grid of lenthscale @xmath27 in such a way as to preserve normalization .
while we have found this choice of coarse graining function to be convenient , others are possible .
( a ) the coarse - graining procedure is schematically shown to project the atomistic solvent density onto a discrete grid .
( b ) the same procedure is shown for an instantaneous solvent configuration of the actual system .
grid cells containing less solvent density than @xmath28 are colored white ; the remaining cells are left transparent against the blue background .
the hydrophobic chain is shown in red .
( c ) the potential in eq .
[ shmear4 ] is used to restrain atomistic solvent density to the reference distribution at the far left . with larger @xmath29 ,
reported numerically in units of @xmath30 , the average atomic solvent density reproduces the reference distribution in detail ( see text for notation ) . , title="fig:",width=359 ] + figs .
[ shmear]a and b illustrate the coarse - graining procedure . in part
a , the solvent is schematically shown before and after coarse - graining . in part
b , the same mapping is shown for the actual system considered herein .
cells are shaded white when their solvent occupation , @xmath13 , is less than half of its bulk average value of @xmath31 molecules .
small local density fluctuations are seen throughout the simulation box , as is expected for an instantaneous solvent configuration .
in addition to visualizing solvent density , the coarse - graining algorithm is useful for controlling the solvent density in md simulations .
for example , if it is desired that a particular cell @xmath32 exhibit a solvent occupation @xmath33 , the following potential energy term can be used to derive the appropriate forces on the atoms in the simulation , @xmath34 where @xmath29 is an appropriate force constant .
this technique is illustrated in fig .
[ shmear]c .
the left - most panel shows a density distribution that is exceedingly unlikely for a simulation of ambient liquid water .
panels to the right show the average solvent density distribution from md simulations that are restrained to this unlikely reference distribution with increasing @xmath29 .
force constant values greater than @xmath30 , in which the simulation incurs an energetic penalty of at least @xmath35 for placing the average bulk density in a cell that is restrained to be empty , ensure that the reference distribution is recovered in detail .
we use the string method in collective variables to calculate the minimum free energy path ( mfep ) for the collapse of the hydrated chain.@xcite to explain , we introduce some notation .
let @xmath36 be the position vector of length @xmath37 for the atomistic representation of the entire system , where @xmath38 is the position vector for the atoms in the chain , and @xmath39 is the position vector for the atoms in the water molecules .
similarly , let @xmath40 be the vector of length @xmath41 for the collective variable representation of the system , where the elements of @xmath42 are defined in eq .
[ cgdensity ] .
the mfep , @xmath43 , is parameterized by the string coordinate @xmath44 , where @xmath45 corresponds to the collapsed chain and @xmath46 corresponds to the extended chain .
it obeys the condition @xmath47 where @xmath48 is the free energy surface defined in the collective variables , and @xmath49\\ & & \times\left<\sum_{k=1}^n m_k^{-1 } \frac{\partial z_i({\bf x})}{\partial x_k } \frac{\partial z_j({\bf x})}{\partial x_k } \prod_{i=1}^{\cal n}\delta(z_i - z_i({\bf x}))\right>.\nonumber\end{aligned}\ ] ] here , angle brackets indicate equilibrium expectation values , @xmath50 is the reciprocal temperature , and @xmath51 is the mass of the atom corresponding to coordinate @xmath52 . if the employed collective variables are adequate to describe the mechanism of the reaction ( here , the hydrophobic collapse ) , then it can be shown that the mfep is the path of maximum likelihood for reactive md trajectories that are monitored in the collective variables.@xcite in the current application , we shall check the adequacy of the collective variables _ a posteriori _ by running md trajectories that are initiated from the presumed rate - limiting step along the mfep ( i.e. the configuration of maximum free energy ) and verifying that these trajectories lead with approximately equal probability to either the collapsed or extended configurations of the chain ( see section [ qsection ] ) .
the string method yields the mfep by evolving a parameterized curve ( i.e. a string ) according to the dynamics@xcite @xmath53 where the term @xmath54 enforces the constraint that the string remain parameterized by normalized arc - length .
the endpoints of the string evolve by steepest decent on the free energy surface , @xmath55 for @xmath56 and @xmath57 .
these artificial dynamics of the string yield the mfep , which satisfies eq .
[ mfepcondition ] . in practice , the string is discretized using @xmath58 configurations of the system in the collective variable representation .
the dynamics in eqs .
[ steepest1 ] and [ steepest2 ] are then accomplished in a three - step cycle where ( i ) the endpoint configurations of the string are evolved according to eq .
[ steepest2 ] and the rest of the configurations are evolved according to the first term in eq .
[ steepest1 ] , ( ii ) the string is ( optionally ) smoothed , and ( iii ) the string is reparameterized to maintain equidistance of the configurations in the discretization . step ( i )
requires evaluation of the mean force elements @xmath59 and the tensor elements @xmath60 at each configuration .
these terms are obtained using restrained atomistic md simulations of the sort illustrated for the solvent degrees of freedom in fig .
[ shmear]c .
the details of the string calculation are provided in supporting information sec .
fig . [ feprofile ] shows the mfep for the hydrophobic collapse of the hydrated chain . it is obtained using the string method in the collective variables for the chain atom positions and the grid - based solvent density field .
the string is discretized using @xmath61 configurations of the system , and it is evolved using the steepest descent dynamics in eqs .
[ steepest1 ] and [ steepest2 ] to yield the mfep that satisfies eq .
[ mfepcondition ] .
the free energy profile is obtained by integrating the projection of the mean force along the mfep , using @xmath62 upon discretization , @xmath63 is the configuration number that indexes the mfep .
the resolution of @xmath64 in fig .
[ feprofile ] could be improved by employing a larger @xmath58 , but at larger computational cost . to the extent that the collective variables employed in this study adequately describe the mechanism of hydrophobic collapse ,
the variable @xmath65 parameterizes the reaction coordinate for the collapse dynamics ; this assumption is checked below with the use of straightforward md simulations .
the statistical errors in the free energy profile between consecutive configurations are approximately the size of the plotted circles , and the small features in the profile at configurations 27 and 31 are due to noise in the convergence of the string calculation .
the lower panel of fig .
[ feprofile ] presents configurations along the mfep in the region of the free energy barrier . as in fig .
[ shmear ] , lattice cells with less than half of the bulk solvent occupation number fade to white .
the minimum free energy path obtained using the string method .
above , the free energy profile exhibits a single peak at configuration 22 .
below , the configurations of the path in the vicinity of the free energy peak are shown with configuration numbers indicated in white text .
, title="fig:",width=359 ] + the free energy profile in fig .
[ feprofile ] is dominated by a single barrier at configuration 22 , where a liquid - vapor interface is formed at a bend in the hydrophobic chain .
the sharply curved chain geometry presents an extended hydrophobic surface to molecules located in the crook of the bend , an environment that is analogous to that experienced by water trapped between hydrophobic plates and known to stabilize large - lengthscale solvent density fluctuations.@xcite the barrier in the calculated free energy profile clearly coincides with a collective motion in the solvent variables . a simple theory can be constructed to understand the contributions to the free energy profile in fig .
[ feprofile ] and to test the validity of coarse - graining solvent interactions .
noting that our choice of collective variables for the string calculation neglects the configurational entropy of the chain , the free energy profile @xmath66 can be decomposed into the configurational potential energy of the chain @xmath67 and the free energy of hydrating the chain @xmath68 , @xmath69 the term @xmath70 is easily evaluated from the chain potential energy term , yielding the components of the free energy profile shown in fig . [ theory]a .
a simple theory describes the solvation free energy of irregular hydrophobic solute geometries . *
( a ) * the calculated free energy profile @xmath71 is decomposed into contributions from the chain configurational potential energy @xmath72 and the free energy of solvation @xmath73 . * ( b ) * the volume of the hydrated chain is parsed into large - lengthscale ( @xmath74 ) and small - lengthscale ( @xmath75 ) components . * ( c ) * the surface area of the chain is similarly parsed . *
( d ) * the dotted line indicates the solvation free energy obtained using a relation ( eq . [ fsolvtheory ] ) that is based on the volume - scaling of the small - lengthscale solvation free energy and the surface area - scaling of the large - lengthscale solvation free energy .
the solvation free energy obtained from simulation ( solid line ) is included for comparison.,title="fig:",width=272 ] + we focus our attention on @xmath73 . the solvation free energy for hydrophobes of idealized geometry is well understood . for small - lengthscale hydrophobes (
@xmath76 1 nm ) , this energy scales linearly with the solute volume , whereas for large - lengthscale hydrophobes , it scales linearly with the solute surface area.@xcite but theoretical prediction of the solvation free energy for more complicated solute geometries is not necessarily trivial . for example , depending on its configuration , some parts of the chain considered here might be solvated like a large - lengthscale hydrophobe , whereas other parts might behave like a small - lengthscale hydrophobe .
we have developed a technique for parsing a general hydrophobic solute into components that belong to either the small - lengthscale or large - lengthscale regimes .
we define the solvent - depleted volume as the continuous set of lattice cells that are , on average , occupied by less than 50% of the average bulk solvent value ( this set includes both the volume that is excluded by the hard - sphere - like interactions between the solute and solvent interactions , as well as any additional volume in the vicinity of the solute that is not substantially occupied by the solvent ) .
we also introduce a probe volume that is a cubic @xmath77 set of cells , which is approximately the size at which the water solvation structure changes from the small - lengthscale to the large - lengthscale regime .
the probe volume is used to determine whether a given section of the solvent - depleted volume is in the small- or large - lengthscale regime .
this is done by moving the probe volume throughout the simulation box and determining whether it can be fit into different portions of the solvent - depleted volume .
if , at a given position , the vast majority of this probe volume ( specifically 59 out of 64 cells ) fits within the solvent - depleted volume , then the cells in that portion of the solvent - depleted volume are included in the large - lengthscale component , @xmath74 .
portions of the solvent - depleted volume that do not meet this criterion for any position of the probe volume are included in the small - lengthscale component , @xmath75 .
the volumes @xmath74 and @xmath75 , which are plotted in fig .
[ theory]b , primarily include contributions from the collapsed portions and extended portions of the chain , respectively . the total volume @xmath78 is obtained by counting the total number of cells in the solvent - depleted volume , such that @xmath79 . the total surface area , @xmath80 , and its large - lengthscale component , @xmath81 , were obtained by counting the number of external faces on the cells that comprise @xmath78 and @xmath74 , respectively .
the surface area of the small - lengthscale component was obtained using @xmath82 to account for the fact that we are calculating the area of a smooth surface by projecting it onto a cubic lattice , each surface area term was also multiplied by a @xmath83 , a correction factor that is exact for an infinitely large sphere .
the surface area components are plotted in fig .
[ theory]c .
we use the parsed components of the solute volume and surface area to estimate the solvation free energy for the chain along the minimum free energy path with the relationship @xmath84 the coefficients @xmath85 and @xmath86 are , respectively , the prefactors for the linear scaling of the solvation free energy in the small - lengthscale and large - lengthscale regimes , as obtained by calculations on a spherical hydrophobic solute.@xcite their values are taken to be @xmath87 mj/(m@xmath88 ) and @xmath89 mj / m@xmath88 . the results of this calculation are presented in fig .
[ theory]d .
the dotted line indicates our theoretical estimate of the solvation free energy using eq .
[ fsolvtheory ] , and the solid line indicates the data from the atomistic simulations ( from fig .
[ theory]a ) .
the agreement is very good , and as is shown in supporting information sec .
c , it is better than can be obtained from solvation free energy estimates that exclusively consider either the solute volume or the solute surface area .
this result demonstrates that atomistic solvent energetics can be quantitatively modeled using a grid - based solvent density field .
furthermore , we note in fig .
[ theory]b and c that the collapsing chain first develops large - lengthscale components in the vicinity of the peak in the free energy profile , which suggests that this peak coincides with the crossover from small- to large - lengthscale solvation ( i.e. dewetting ) .
this observation is consistent with the collective solvent density motion that is observed near the free energy barrier in fig .
[ feprofile ] .
the various assumptions that are employed in our implementation of the string method , including our choice of collective variables and the corresponding neglect of the chain configurational entropy , raise the possibility that the barrier in the free energy profile in fig .
[ feprofile ] does not correspond to the dynamical bottleneck for hydrophobic collapse .
this is a general concern in trying to relate free energy calculations to dynamical quantities , such as the reaction mechanism and the reaction rate . to confirm that the calculated free energy profile is dynamically relevant
, we evaluate the committor function at several configurations along the mfep .
the committor function reports the relative probability that md trajectories that are initialized from a particular collective variable configuration first proceed to the extended configurations of the chain , as opposed to the collapsed configurations . for initial collective variable configurations that coincide with the true dynamical bottleneck ,
the committor function assumes a value of exactly @xmath90 .
we first evaluate the committor function at configuration 22 , which corresponds to the peak in the calculated free energy profile in fig .
[ feprofile ] .
the committor function is obtained by performing straightforward md trajectories from initial conditions that are consistent with this collective variable configuration , and then tallying the fraction of those trajectories whose endpoints are `` extended , '' as opposed to `` collapsed '' .
the initial conditions for these trajectories are obtained from a 20 ps md trajectory that is restrained to the collective variables for configuration 22 ; the atomistic coordinates of the restrained trajectory are recorded every picosecond . from each set of atomistic coordinates ,
an unrestrained trajectory was run for 150 ps forwards and backwards in time with the initial atomistic velocity vector drawn from the maxwell - boltzmann distribution at @xmath3 k. to decide whether a given unrestrained trajectory terminates in either an extended or collapsed configuration of the chain , we employ an order parameter based on the number of chain atoms that are within @xmath91 of another chain atom to which it is not directly bonded .
if the number of non - bonded contacts exceeds three , the chain is considered to be in a collapsed configuration ; otherwise it is considered extended .
this order parameter need only distinguish between collapsed and extended configurations of the chain ; it need not be ( and in fact is not ) a good reaction coordinate . fig .
[ mdtraj ] illustrates representative forward and backwards unrestrained md trajectories .
snapshots from unrestrained md trajectories that are initiated from the calculated free energy barrier at configuration 22 .
the instantaneous atomistic configurations are visualized in the collective variable representation , using the same technique that was introduced in figs .
[ shmear ] a and b.,title="fig:",width=302 ] + the committor function at configuration 22 is @xmath92 , suggesting that the calculated free energy barrier is slightly biased towards the basin of stability for the collapse chain .
however , this deviation from the ideal value of @xmath90 is only marginally statistically significant , and we emphasize that the committor function is exponentially sensitive in the region of the dynamical bottleneck . to illustrate this point , we repeat the evaluation of the committor function at configuration 20 , which is seen in fig .
[ feprofile ] to be near the barrier peak but slightly closer to the collapsed configurations of the chain , as well as at configuration 24 , which is slightly closer to the extended configurations .
we find the committor function at configuration 20 to be @xmath93 and the value at configuration 24 to be @xmath94
. only very minor shifts along the mfep dramatically change the value of the committor function .
given the extreme sensitivity of the commitor function near the dynamical bottleneck , the fact that configuration 22 gives rise to a significant fraction of trajectories that proceed to both the collapsed and extended basins is very significant .
furthermore , the direction in which the committor function changes as a function of the configuration number is as is expected in the vicinity of the dynamical bottleneck .
we conclude that the barrier in the calculated free energy profile correctly characterizes the true free energy barrier - thus identifying the rate - limiting step for the collapse dynamics .
the committor function calculations , which are based on straightforward md simulations , indicate that our choice of collective variables provides a reasonable description of the mechanism of hydrophobic collapse .
this result yields atomistic support for the strategy of coarse - graining solvent dynamics . using ( i )
that it is straightforward@xcite to obtain the stochastic dynamics in the collective variables for which the most likely reaction pathway is the same as the mechanism obtained using the string method and ( ii ) that the mechanism for hydrophobic collapse obtained in the employed collective variables agrees with the mechanism obtained using atomistic dynamics , we obtain a proof of principle that atomistic solvent dynamics can be usefully projected onto a coarse - grained field .
the coarse - grained dynamics obtained via this argument are@xcite @xmath95 where @xmath96 is a white noise satisfying @xmath97 and @xmath98 is an artificial time that is scaled by the friction coefficient @xmath99 .
the computational feasibility of directly integrating these coarse - grained dynamics , however , hinges on the cost of calculating the mean force elements @xmath100 and the tensor elements @xmath60 and @xmath101 , since these terms are required at every coarse - grained timestep .
it is thus encouraging that the free energy surface in the collective variables can be quantitatively modeled in lieu of atomistic simulations , by using eqs .
[ decompose ] and [ fsolvtheory ] .
similar approximations for the tensor elements might also be possible . using the calculated mfep and committor function values
, we have established that the rate - limiting step for the hydrophobic collapse of the hydrated chain coincides with a collective solvent motion . using a simple analysis of the solvation free energy , we find that this collective solvent motion is consistent with lengthscale - dependent dewetting
. however , it remains to be shown whether the rate - limiting step involves performing work in the solvent , or in the chain , degrees of freedom .
this is an important distinction .
if the latter case is true , then dewetting merely accompanies hydrophobic collapse as a spectator .
but if the former case is true , then dewetting _ is _ the rate - limiting step to hydrophobic collapse . to address this issue
, we again decompose the free energy profile , this time into contributions from work performed along the solvent and the chain collective variables . the definition of the free energy profile in eq . [ feprofeq1 ]
can be written more explicitly as @xmath102 where @xmath103 is the vector of mean forces acting on the chain atom positions at configuration @xmath104 along the mfep , and @xmath105 is the corresponding mean force on the solvent collective variable @xmath13 .
the full free energy profile is obtained by setting the volume @xmath106 equal to the entire simulation box . but to understand the role of solvent in hydrophobic collapse , it is informative to calculate the free energy profile using various smaller solvent volumes .
the bottom curve in fig . [ febox ] is obtained from eq .
[ feprofeq ] by letting @xmath106 be the empty set , thus eliminating the second term .
the middle curve is obtained by letting @xmath106 be the set of @xmath107 lattice cells at the middle of the simulation box , as is indicated in the corresponding image , and the top curve is obtained by letting @xmath106 be the middle set of @xmath108 lattice cells .
consideration of larger sets of lattice cells does not further alter the free energy profile , as is shown in supporting information sec .
d. the solvent variables for regions of space that are distant from the collapsing chain remain constant along the path and do not contribute to the free energy profile .
[ ! tbp ] the free energy profile from fig .
[ feprofile ] is separated into contributions from the chain and solvent coordinates . in the bottom curve , only the first term in eq .
[ feprofeq ] , which corresponds to work performed in the chain coordinates , is included in the free energy profile . in the middle and top curves ,
solvent contributions are also included .
the solvent volume included for each curve is indicated with a wire box .
inclusion of larger solvent volumes does not substantially change the profile.,title="fig : " ] + the bottom curve in fig .
[ febox ] only shows the work performed on the chain atom positions during hydrophobic collapse .
remarkably , it lacks almost any free energy barrier , indicating that essentially no work is performed in the chain degrees of freedom in crossing the dynamical bottleneck .
instead , we see that the barrier emerges only upon inclusion of the work performed in the solvent collective variables . performing
the hydrophobic collapse involves traversing a free energy barrier that exists only in the solvent coordinates .
this figure shows that the dewetting transition not only accompanies the rate - limiting step for hydrophobic collapse , but that it is the rate - limiting step .
is supported by doe grants che-0345280 and de - fg03 - 87er13793 .
e . acknowledges support from the miller institute at uc berkeley , onr grant n00014 - 04 - 1 - 0565 , and nsf grants dms02 - 09959 and dms02 - 39625 .
d.c acknowledges support from nsf grant che-0543158 .
the authors also thank giovanni ciccotti , mauro ferrario , ray kapral , and luca maragliano for useful discussions and nersc for computing resources .
99 e w , ren w , vanden - eijnden e ( 2002 ) _ phys rev b _ 66:52301 .
e w , ren w , vanden - eijnden e ( 2005 ) _ j phys chem b _ 109:6688 - 6693 .
maragliano l , fischer a , vanden - eijnden e , ciccotti g ( 2006 ) _ j chem phys _ 125:024106 .
ten wolde pr , chandler d ( 2002 ) _ proc natl acad sci usa _ 99:6539 - 6543 .
kauzmann w ( 1959 ) _ adv prot chem _ 14:1 - 63 .
safron sa ( 1994 ) _ statistical thermodynamics of surfaces , interfaces and membranes _ , ( addison - wesley , reading ) , pp 1 - 95 .
tanford c ( 1978 ) _ science _ 200:1012 - 1018 .
chandler d ( 2005 ) _ nature _ 437:640 - 647 .
hummer g , garde s , garcia ae , pratt lr ( 2000 ) _ chem phys _ 258:349 - 370 .
huang x , margulis cj , berne bj ( 2003 ) _ proc natl acad sci usa _ 100:11953 - 11958 .
huang q , ding s , hua c - y , yang h - c , chen c - l ( 2004 ) _ j chem phys _ 121:1969 - 1977 .
liu p , huang x , zhou r , berne bj ( 2005 ) _ nature _ 437:159 - 162 .
athawale mv , goel g , ghosh t , truskett tm , garde s ( 2007 ) _ proc natl acad sci usa _ 104:733 - 738 .
bolhuis pg , chandler d ( 2000 ) _ j chem phys _ 113:8154 - 8160 .
huang x , zhou r , berne bj ( 2005 ) _ j phys chem b _ 109:3546 - 3552 .
e w , ren w , vanden - eijnden e , submitted .
lum k , chandler d , weeks jd ( 1999 ) _ j phys chem b _
103:4570 - 4577 .
luzar a , leung k ( 2000 ) _ j chem phys _ 113:5836 - 5844 .
leung k , luzar a ( 2000 ) _ j chem phys _ 113:5845 - 5852 .
huang dm , geissler pl , chandler d ( 2001 ) _ j phys chem b _
105:6704 - 6709 .
we consider a hydrophobic chain in explicit solvent .
the unbranched chain is composed of @xmath109 spherical atoms of diameter @xmath0 and mass @xmath110 amu .
the interactions between the chain atoms were treated with the purely repulsive wca potential ( @xmath111 kj / mol ) , which provides a slightly softened hard - sphere interaction that is convenient for molecular dynamics ( md ) simulations .
neighboring atoms are linked by harmonic bonds , using @xmath112 where @xmath113 is the distance between chain atoms at positions @xmath114 and @xmath115 , @xmath116 , and @xmath117 kj / mol / @xmath88 . to add some rigidity to the chain ,
a harmonic potential is included to penalize its curvature , @xmath118 where @xmath119 is the angle between neighboring bond vectors @xmath120 and@xmath121 , and @xmath122 kj / mol / radian@xmath88 .
the chain was solvated with @xmath123 water molecules in an orthorhombic simulation cell with periodic boundary conditions and dimensions of @xmath124 @xmath125 @xmath126 .
interactions among water molecules are described with the spc / e rigid water potential,@xcite and the chain atoms interact with the oxygen atoms in the water molecules via wca repulsions@xcite ( @xmath127 @xmath128 kj / mol ) .
all md simulations were performed at @xmath3 k using the nos - hoover thermostat and a timestep of @xmath129 fs.@xcite constant pressure conditions were maintained in the simulations using the liquid - vapor coexistence technique described in the text . to ensure that the liquid - vapor interface remains flat , as opposed to collapsing under the surface tension of the liquid , the solvent density restraint potential in eq .
[ shmear4 ] was applied to the top two layers of lattice cells in the simulation box with parameters @xmath130 and @xmath131 kj / mol / molecule@xmath88 .
all md calculations were performed using a modified version of the dl_poly_3 molecular simulation package.@xcite the string method @xciteis a technique for calculating the committor function for a dynamical process .
the constant - value contours of the committor function are approximated by a collection of hyperplanes that are perpendicular to a given path ( the string ) .
this approximation can be justified within the framework of transition path theory ( tpt ) , @xcite which yields a variational criterion for the string such that the perpendicular hyperplanes optimally approximate the isocommittor surfaces .
the string method in collective variables characterizes the committor function projected onto the space of collective variables . provided that the collective variables adequately describe the reaction and that the reactive trajectories projected onto the space of collective variables remain confined to a relativelly narrow tube , the approximation is not only optimal but also accurate . in this case
, it can be shown that the string coincides with the minimum free energy path ( mfep ) in the space of collective variables , which is also the path of maximum likelihood for the recation monitored in these variables .
unlike other free - energy mapping techniques , the string method focuses only on a linear subset of the space of collective variables .
it thus scales independently of the dimensionality of the full free energy surface .
the method does not require performance of long , reactive md trajectories . instead , a double - ended approach is employed in which the path is updated using short , restrained md simulations .
a complete discussion of the implementation of the string method in collective variables can be found in ref . .
we represent the string in the variables of the chain atom positions and the solvent cell densities , as is explained in the primary text .
data needed for the calculation of the minimum free energy path were obtained from restrained md simulations , @xmath132 and @xmath133 where @xmath134 the string is converged to the mfep using eqs .
[ steepest1 ] and [ steepest2 ] . between timesteps of these dynamics ,
the string is smoothed and reparameterized .
the smoothing procedure is performed using @xmath135 where @xmath136 and @xmath137 indicate the @xmath138 configuration of the smoothed and unsmoothed string , respectively .
the parameter @xmath139 was chosen to be sufficiently small ( @xmath140 , as is discussed in ref . )
that the smoothing procedure does not effect the accuracy of the calculated mfep .
reparameterization of the string was performed using the linear interpolation scheme described in ref . .
the string was initialized from configurations of an md trajectory in which the hydrated chain was artificially extended from a collapsed configuration with the aid of a greatly magnified force constant in the chain potential energy term @xmath141 . in the first stage of the string calculation ,
the string was discretized using @xmath142 configurations and evolved using large timesteps and weak collective variable restraints .
specifically , we employed chain atom restraint force constants of @xmath129 kj / mol / @xmath88 , solvent restraints of @xmath143 kj / mol / molecule@xmath88 , and a steepest descent timestep of @xmath3 fs . restrained md trajectories of @xmath143 ps were performed during this stage .
this first stage of the string calculation included nine steepest descent timesteps . in a second stage of the string calculation ,
the number of images used to represent the path was increased to @xmath144 , the chain atom restraints were increased to @xmath145 kj / mol / @xmath88 , the solvent restraints were increased to @xmath144 kj / mol / molecule@xmath88 , the steepest descent timestep was decreased to @xmath144 fs , and the restrained md trajectory time was increased to @xmath142 ps . the second stage of the string calculation was terminated after six optimization steps , at which point reasonable convergence , as determined by monitoring changes in the path and its corresponding free energy profile , was obtained . throughout both stages of the string calculation , the path endpoint image corresponding to the extended configuration was ( after local relaxation ) held fixed , and the other endpoint corresponding to the collapsed chain was allowed to relax on the free energy surface according to eq .
[ steepest2 ] .
the final , converged string bears little resemblance to the initial guess .
the string method is a local , rather than global , optimization scheme .
different minimized free energy paths might have been obtained from string calculations started with different initial paths .
however , the simple topology of the chain , as well as the unrestrained md simulations presented in the paper , suggest that the calculated mfep is a reasonable characterization of the reaction mechanism for the hydrophobic collapse of the hydrated chain .
all computations were performed in parallel using 2.2 ghz amd operton processors .
each step in the first of the stage of the string calculation required @xmath146 cpu hours .
each step in the second stage required @xmath147 cpu hours .
each evaluation of the committor function described in sec .
[ qsection ] required @xmath148 cpu hours . ( supporting information ) alternative models for the solvation free energy . the profile obtained from simulation is shown as a solid line , and that obtained using eq .
[ fsolvtheory ] is shown as a dotted line . in panel
a , the dot - dashed line indicates the solvation free energy profile obtained from consideration of only the chain surface area , using eq .
[ saonly ] . in panel b ,
the dot - dashed line indicates the solvation free energy profile obtained from consideration of only the chain volume , using eq .
[ vonly ] . , title="fig:",width=321 ]
+ even with the aide of a one - parameter fit , the solvation free energy profiles estimated on the basis of only solvent - depleted surface area or only solvent - depleted volume do not match the accuracy obtained using eq .
[ fsolvtheory ] in the text . in fig .
[ otherfits]a , we see the profile ( dot - dashed line ) obtained by fitting @xmath149 in the expression @xmath150 the height of the shoulder at configurations 14 - 18 and the barrier peak region do not reproduce the simulated results ( solid line ) as well as eq . [ fsolvtheory ] ( dotted line ) . as is seen fig .
[ otherfits]b , the solvation free energy profile ( dot - dashed ) that is obtained using the expression @xmath151 is less accurate than that obtained using either eqs .
[ fsolvtheory ] or [ saonly ] .
[ [ convergence - of - the - free - energy - profile - with - increasing - solvent - box - sizes ] ] convergence of the free energy profile with increasing solvent box sizes ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ we show that the free energy profile calculated using eq .
[ feprofeq ] converges with respect to the solvent contribution .
the curves in fig .
[ multi ] correspond to free energy profiles obtained with @xmath106 equal to the middle @xmath152 lattice cells of the simulation box .
the profile changes dramatically with @xmath153 until @xmath106 fully encompasses the region of the bending chain , where the dewetting transition occurs . for @xmath154 ,
no major changes are seen in the profile , since the added contributions are from solvent cells that are distant from the collapsing chain .
the small changes that are found with large @xmath153 are due to statistical noise .
this noise increases with larger @xmath153 since the number of included solvent cells increases as @xmath155 .
99 berendsen hjc , grigera jr , straatsma tp ( 1987 ) _ j chem phys _ 91:6269 - 6271 .
chandler d , weeks jd ( 1970 ) _ phys rev lett _ 25:149 - 152 .
weeks jd , chandler d , andersen hc ( 1971 ) _ j chem phys _ 54:5237 - 5247 .
weeks jd , chandler d , andersen hc ( 1971 ) _ j chem phys _ 55:5422 - 5423 .
andersen hc , weeks jd , chandler d ( 1971 ) _ phys rev a _
4:1597 - 1607 .
hoover wg ( 1985 ) _ phys rev a _
31:1695 - 1697 .
todorov i , smith w ( 2004 ) _ phil trans r soc lond a _ 362:1835 - 1852 .
e w , ren w , vanden - eijnden e ( 2002 ) _ phys rev b _ 66:52301 .
e w , ren w , vanden - eijnden e ( 2005 ) _ j phys chem b _ 109:6688 - 6693 .
maragliano l , fischer a , vanden - eijnden e , ciccotti g ( 2006 ) _ j chem phys _ 125:024106 .
vanden - eijnden e ( 2006 ) in _ computer simulations in condensed matter : from materials to chemical biology - vol . 2
_ , eds ferrario m , ciccoti g , binder k ( springer , lnp ) , 703:439 - 478 . | using computer simulations of over 100,000 atoms , the mechanism for the hydrophobic collapse of an idealized hydrated chain is obtained .
this is done by coarse - graining the atomistic water molecule positions over 129,000 collective variables that represent the water density field and then using the string method in these variables to compute the minimum free energy pathway ( mfep ) for the collapsing chain .
the dynamical relevance of the mfep ( i.e. its coincidence with the mechanism of collapse ) is validated _ a posteriori _ using conventional molecular dynamics trajectories .
analysis of the mfep provides atomistic confirmation for the mechanism of hydrophobic collapse proposed by ten wolde and chandler .
in particular , it is shown that lengthscale - dependent hydrophobic dewetting is the rate - limiting step in the hydrophobic collapse of the considered chain . |
damped ly@xmath0 ( dla ) systems are complexes of neutral gas detectable in the spectra of background quasars as hi absorption lines with large column densities , n(hi)@xmath9 10@xmath10 atoms @xmath11 ( wolfe et al .
the damped ly@xmath0 absorptions are found at all redshifts @xmath12 and are associated with metal lines of low ionization and often high ionized species as well .
although the exact nature of the dla systems has yet to be clarified , there is consensus on the fact that they are the progenitors of present - day galaxies . damped
galaxies might be massive rotating disks which will evolve to the present - day spiral galaxies ( wolfe et al .
1986 , 1995a , prochaska & wolfe 1997 ) or less massive objects progenitors of dwarf galaxies ( tyson 1988 , pettini et al .
1990 , meyer and york 1992 , steidel et al .
understanding which interpretation is the more appropriate has important implications for any theory of galaxy formation and evolution . imaging in the field of the background quasars suggests that more than a single population of galaxies give rise to the dlas , even though the results are sometimes ambiguous , especially at high redshift ( briggs et al .
1989 , steidel & hamilton 1992 ) . at @xmath13 1
the candidate associations include gas - rich galaxies with a variety of morfological types ( le brun et al .
1997 ; lanzetta et al .
very recently rao & turnshek ( 1998 ) discovered two dla absorbers at very low redshift ( z=0.091 , z=0.221 ) and by means of imaging observations they exclude that they are luminous spiral galaxies .
dla galaxies have low metallicities , typically between 1% and 10% of solar , a signature of the enrichment of the first stellar generations .
the precision obtained in abundance studies of dlas is remarkable and comparable to that attained in the interstellar studies of our own galaxy , thus offering important clues for understanding the nature of these objects .
element abundance ratios are tracers of galactic chemical evolution ( see e.g. lauroesch et al . 1996 , matteucci , molaro & vladilo , 1997 ) , however the results obtained for dla galaxies are subject to some debate .
the typical overabundance of @xmath0-elements compared to iron - peak elements ( @xmath0/fe ) found in galactic metal - poor stars is observed in dlas by means of the si / fe ratios , which is interpreted as a signature of the first stages of chemical evolution of a spiral galaxy like the milky way ( wolfe et al .
1995b , lu et al . 1996 ) . on the other hand , the @xmath0/fe ratios in dlas
are observed close to solar when a dust - free diagnostic such as s / zn is used ( molaro , centurin & vladilo 1998 ; hereafter mcv ) , or when the observed abundances of refractory elements ( as fe and si ) are corrected by dust depletion ( vladilo 1998 ) .
nitrogen is an important element when attempting to understand the chemical history of dla galaxies .
the nucleosynthesis of n , which takes place mainly in low and intermediate mass stars , is quite complex and can be used to probe conditions typical of early stages of galactic chemical evolution .
although the dominant processes which lead to the enrichment of n are still not fully understood it seems clear that secondary synthesis of n which occur in stars of all masses plays an important role at high metallicities ( vila - costas & edmunds 1993 ) . in this case secondary origin of
n refers to its production in the cno cycle from seed c and o nuclei created in earlier generation of stars . however , at low metallicity levels , nitrogen observations in halo stars ( carbon et al .
1987 , tomkin & lambert 1984 ) and in hii regions of dwarf irregular galaxies ( thuan , izotov & lipovetsky 1995 ; kobulnicky & skillman 1996 ) require a significant primary component in the synthesis of this element . primary production of n is obtained in intermediate mass stars when freshly synthesized c in the helium - burning shell penetrates into the hydrogen - burning shell of the same star where it is converted into n during the asymptotic giant branch phase of their evolution ( renzini & voli , 1981 ) .
even though a general consensus is lacking , it has been suggested that primary n might be also produced
esentially by the same mechanism in massive stars of low metallicities ( woosley & weaver , 1995 ) .
this production is strongly dependent on the assumed treatment of convection in stellar interiors but it seems necessary in order to explain the high n abundances relative to o observed in some dwarf irregular galaxies and some dlas ( matteucci , molaro & vladilo , 1997 and references therein ) .
a further advantage of n is that it is not depleted onto dust in the galactic interstellar medium ( savage & sembach 1996 ; mayer , cardelli & sofia 1997 ) ; therefore n abundances measured in dla systems and in extragalactic hii regions are expected to reflect the real ( gas plus dust ) n abundances .
the first attempt to detect nitrogen in dlas was made by pettini , lipman & hunstead ( 1995 ) who derived a stringent upper limit ( [ n / h]@xmath14 3.15 ) for the absorber at @xmath15=2.279 towards qso 23481444 .
the upper limit on the [ n / o ] ratio in this system ( @xmath14 1.24 ) implied the lowest ratio ever found either in stars or galaxies at low metallicity , in line with a secondary origin of nitrogen .
pettini and colleagues argued that this result is consistent with a delayed delivery of n created mainly in intermediate mass stars compared to o enrichment produced in short - lived massive stars ( edmunds & pagel 1978 ; garnett 1990 ) .
nitrogen was first detected in the damped absorbers at z=1.776 towards mc 1331 + 170 ( green et al .
1995 ; kulkarni et al . 1996 ) and at z=3.390 towards qso 0000 - 2620 ( vladilo et al .
1995 ; molaro et al .
1996 ) . in both cases
the reported n / o abundance ratio is much higher than the upper limit derived by pettini , lipman & hunstead ( 1995 ) .
these results gave the first indication of an intrinsic dispersion of nitrogen abundances in dlas , as pointed out by vladilo et al .
( 1996 ) and molaro et al .
the high nitrogen abundance ratios are not easy to understand and non conventional chemical evolution models of the type used for dwarf irregular galaxies have been proposed by matteucci , molaro & vladilo ( 1997 ) . in a subsequent work , lu et al .
( 1996 ) found a stringent upper limit in the dla at @xmath15=2.994 towards qso 1946 + 7658 , in line with the result reported by pettini and colleagues .
( 1996 ) interpreted this low n / o value as an evidence of the first stages of chemical evolution of a spiral galaxy like the milky way .
more recently , four new nitrogen abundances were obtained by lu , sargent & barlow ( 1998 ) offering further evidence for an intrinsic dispersion of the nitrogen abundances in dlas . from the analysis of the [ n / si ] ratios in dla galaxies
the latter argue that the observed dispersion can be expected considering the time delay between oxygen and nitrogen delivery . in this paper
we describe three new ni abundance determinations in the dla systems at @xmath15 = 2.309 towards qso 0100 + 1300 , at @xmath15 = 2.827 towards qso 1425 + 6039 , and at @xmath15 = 3.025 towards qso 0347@xmath23819 .
the detections are based on the six ni lines of the multiplets at 1134 and 1200 .
a preliminary report for the dla system towards qso 0347@xmath23819 was given in vladilo et al .
in section 2 and 3 we present our observations and column density measurements which also include sii , siii , oi , and feii determinations .
abundance measurements for all the dlas with nitrogen detections are compiled from our data and from the literature , and are presented in section 4 .
element - to - element ratios in particular n / zn , n / fe , and n / s are discussed in section 5 in connection to the origin of n in the framework of chemical evolution models .
the implications for our understanding of the nature of dla galaxies are summarised in section 6 .
as part of a search for nitrogen absorptions in dlas we obtained optical spectra of qso 0100 + 1300 ( v=16.6 ) , qso 1425 + 6039 ( v=15.8 ) , and qso 03473819 ( v=17.3 ) .
the spectra of the first two qsos were obtained with the utrecht echelle spectrograph ( walker & diego 1985 ) at the nasmyth focus of the 4.2 m william herschel telescope of the observatorio del roque de los muchachos on la palma island .
the spectra of qso 03473819 were obtained with the caspec spectrograph at the cassegrain focus of the eso 3.6 m telescope at la silla , chile ( pasquini & dodorico 1989 ) . in both
spectrographs an echelle grating of 31.6 grooves / mm and a 1024x1024 tektronik ccd with square pixels of 24 @xmath16 m in size were used .
the ccd was binned at a step of 2 pixels along the dispersion and the slit width was set at 2.1 `` ( caspec ) and 2.2 '' ( ues ) in order to project the slit onto 2 binned pixels of the detector .
the slit width matched the seeing which was around 2 arcsecs during the observations .
the full width at half maximum of the instrumental profile , @xmath17 , was measured from the emission lines of the thorium - argon lamp spectra recorded frequently during the observations .
the resulting resolving power r= @xmath18/@xmath17 was r @xmath19 27000 ( ues spectra ) and 19500 ( caspec spectra ) , corresponding to a velocity resolution of @xmath20v @xmath19 11.5 km s@xmath21 and 15.4 km s@xmath21 , respectively .
the journal of the observations is given in table 1 .
the spectra taken on different nights were reduced separately and then averaged using weights according to the square of their s / n ratios .
cosmic ray removal , sky subtraction , optimal order extraction and wavelength calibration were performed using the echelle routines implemented within the midas software package developed at eso .
typical internal errors in the wavelength calibrations are of @xmath19 0.01 - 0.02 .
the wavelength scale of the spectra was corrected to vacuum heliocentric scale . finally , each order of the spectra was normalised using a spline to connect smoothly the regions free from absorption features
. the signal - to - noise ratios of the final spectra estimated from the @xmath22 scatter of the continuum near the absorptions under study are typically in the range between 10 and 20 . for the two dlas at higher redshift ,
the oi 1302 , oi 1355 and siii 1304transitions fall in a part of the spectrum where the signal - to - noise ratios can be as high as 45 thanks to the increase of the ccd quantum efficiency towards the red .
column densities have been derived by best fitting theoretical voigt profiles to the observed absorption lines via @xmath23 minimization .
this step was performed using the routines fitlyman ( fontana & ballester 1995 ) included in the midas package . before the fit
, the theoretical profiles were convolved with the instrumental point spread function determined from the analysis of the emission lines of the arcs .
portions of the profiles contaminated by intervening ly @xmath0 absorbers were excluded from the fit .
the fitlyman routines determine column densities , broadening parameters ( @xmath24 values ) and redshifts of the absorption components , as well as the fit errors for each of these quantities .
the laboratory wavelengths of the transitions investigated are listed in table 2 together with the oscillator strenghts adopted for computation of theoretical profiles .
when several unsaturated transitions are detected for a given element , the column density is estimated by applying the fit procedure both to the individual transitions and to the full set of available lines .
no differences were found between the mean column density resulting from individual fits and the column density obtained from the simultaneous fit of all the available lines . in these cases
we adopted the dispersion from individual measurements as conservative column density errors .
nitrogen column densities were determined from the six transitions of the two ni multiplets at @xmath181134 and @xmath181200 ( see table 2 ) .
the ni lines occur in the ly @xmath0 forest , but the identification of the absorptions is quite reliable because it is very unlikely that each one of the six transitions is blended with a ly @xmath0 interloper .
the oscillator strengths span about one order of magnitude offering a large dynamical range for the measurement of the column density .
the detection of the faintest transitions of the @xmath18 1134multiplet is especially important to get rid of saturation effects .
these transitions are not saturated and therefore uncertainties in @xmath24(ni ) do not affect the derived ni column densities .
the fits of nitrogen lines for the three dla systems under investigation are shown in fig . 1 , 2 and 3 and the resulting column densities
are reported in table 3 . in the same table
we list the rest of extant determinations of ni column densities or upper limits taken from the literature . in total
there are nine ni detections available : four from our project , one from kulkarni et al .
( 1996 ) , four from the recent measurements reported by lu , sargent & barlow ( 1998 ) .
in addition there are eight upper limits ( pettini , lipman & hunstead 1995 ; lu , sargent & barlow 1998 ) . in order to understand the nucleosynthetic origin of nitrogen
it is necessary to compare its abundance with that of other elements .
in addition to the nitrogen lines , we therefore searched for transitions of important species such as oi , siii , sii , and feii which fall in the wavelength range of our spectra ( see table 2 ) .
column densities and @xmath24 values of these elements for the dlas towards qso 03473819 and qso 1425@xmath256039 are reported in table 4 and for the system towards qso 0100 + 1300 are given in mcv . for saturated transitions ,
lower limits to the column density were estimated from the equivalent width obtained from the best fit to the absorption .
for undetected transitions , upper limits to the column density were derived from 3@xmath26 upper limits of the equivalent width . in both cases
the conversion from equivalent widths to column densities was performed in the optically thin limit ( linear part of the curve of growth ) .
oxygen column densities in dlas are generally uncertain by more than one order of magnitude owing to the saturation of the oi 1302 line which is present even at low metallicities . in order to constrain the oi column density
we fitted the oi 1302 line by adopting the @xmath24 value obtained from the analysis of the ni multiplets , which is expected to trace the same neutral gas as oi .
the @xmath24 value of both atoms should in fact be very similar : @xmath27 = 0.94 @xmath28 for pure thermal broadening and @xmath27 = @xmath28 for pure turbulence .
the other accessible oi transition at 1356 is extremely weak ( f@xmath29 = 1.248 x 10@xmath30 ) and is not detected in the dla systems under study ( see e.g. fig . 4 and fig .
the derived upper limits are given in table 4 .
measurements of sii column densities are derived from the sii triplet at 1254 which is generally unsaturated .
this triplet , however , can be contaminated by the lyman @xmath0 forest and we have been able to measure the sulphur column density only for two of the three dla systems under study . as far as iron
is concerned , there are six feii transitions available , listed in table 2 . for each of the three dla galaxies studied here , there are one or two feii unsaturated lines , and free from blends , which have allowed us to derive reliable iron column densities .
silicon transitions in the available spectral ranges are usually heavily saturated hampering attempts to derive accurate siii column densities .
more details on the measurements performed in individual systems are given in the next sub - sections .
the ni column density of this system was measured via the simultaneous fit of the two lines at shorter wavelength of the ni 1134 multiplet .
as one can see in fig .
1 , the reddest transition of this multiplet is in fact contaminated by a ly @xmath0 interloper .
the ni 1200 multiplet presents strong contamination and is highly saturated .
the column density that we derive log n(ni ) = 14.69(@xmath60.07 ) is at variance with the one found by lu , sargent & barlow ( 1998 ) , log n(ni)= 15.29 @xmath6 0.35 ( see table 3 ) .
these authors obtained this result by means of keck hires data , but details on the measurement are deferred to a subsequent paper . the large error bar reported by lu , sargent &
barlow ( 1998 ) suggests that n(ni ) was obtained from the 1200 multiplet .
in fact , the 1134 multiplet falls below the spectral range usually covered by hires observations ( see table 1 in lu et al .
we stress that our measure relies instead on the analysis of the unsaturated ni absorptions at @xmath31 1134.1 and 1134.4 .
the main absorption component of the system towards qso 0100 + 1300 falls at @xmath1 = 2.30907 in our spectra . a secondary component at @xmath32 @xmath3 35 km s@xmath21 has been also detected in the most intense absorption lines of this system ( wolfe et al .
1994 , mcv ) .
analysis of the most intense ni lines does not yield information on this component , owing to ly@xmath0 contamination .
however , close inspection of our spectrum redwards of the 1134.41 line shows a portion free from ly@xmath0 contamination without traces of absorption exceeding the noise level . therefore the @xmath32 @xmath3 35 km s@xmath21 component if present in ni at all must be negligible compared to the main one .
wolfe et al . ( 1994 ) in their higher resolution ( @xmath33=8 km
s@xmath21 ) spectra at redder wavelengths ( 4427 - 6900 ) revealed an asymmetric profile of the main component suggesting the presence of an additional component at @xmath32 @xmath3 8 km s@xmath21 .
the effect of this double structure of the main component on the column density of ni is negligible .
in fact , using a two - cloud model the ni fit gives a total column density within 0.01 dex of the one - cloud model .
abundances of other elements like sii , feii , mgii , mnii , pii ( from our data ) and znii ( from wolfe et al .
1994 ) in this system are discussed in mcv .
the intense siii 1304 and oi 1302 absorptions outside the lyman @xmath0 forest show two components ; the stronger lying at @xmath1= 3.02476 and the weaker at @xmath1= 3.02562 ( see fig .
the column density of the secondary component is less than 10% of that of the main one for both species . for ni , feii and sii ,
only the main component is clearly detected . in these cases
we used a single component fit and adopted the resulting column density as the total column density .
the ni multiplets are clearly identified , but some contamination from lyman @xmath0 absorption is visible , particularly in the vicinity of some lines of the 1134 multiplet . to minimize contamination we measured the column density by fitting the central parts of the line profiles .
the weakest line of the 1134 multiplet noisy and apparently contaminated was excluded from the fit .
results are shown in fig . 2 and in table 3 .
the column density of siii was estimated from the analysis of the 1304 line , which falls outside the lyman @xmath0 forest and has the lowest oscillator strength among the available siii transitions .
this fit gives log n(siii ) = 15.07 @xmath6 0.43 and @xmath24 = 21 @xmath6 5 km s@xmath21 .
an attempt to fit also simultaneously the 1193transition , which appears to be unblended , yields instead log n(siii ) = 15.50 @xmath6 0.69 and @xmath24 = 17 @xmath6 4 km s@xmath21 .
silicon abundance must obviously be treated with some caution . the analysis of ni and siii yields consistent values of the broadening parameter , although the value @xmath24= 17.9 @xmath6 1.1 km s@xmath21 derived from the ni multiplets is more precise .
we adopted this value , with a 3@xmath26 error ( @xmath6 3.3 km s@xmath21 ) , to fit species for which the @xmath24 value is undetermined .
there is only one transition of the sii triplet which is not contaminated by ly @xmath0 interlopers , namely the one at 1260 .
the identification is supported by the perfect coincidence of the redshift and by the narrowness of the absorption profile .
we obtain log n(sii ) = 14.77 @xmath6 0.06 where the error is the fit error at a fixed value of @xmath24 .
an additional @xmath6 0.06 error results when the above mentioned range of possible @xmath24 values is taken into account .
the sii column density is fairly independent of the adopted value of the broadening parameter .
many feii transitions of this system are potentially present in the spectral range of our spectra ( see fig .
however , all these lines lie in the lyman @xmath0 forest and many of them are expected to be close to the noise level .
we clearly detect the 1144 line and a hint of absorption in correspondence with other feii transitions .
our analysis is mostly based on the 1144 line , but the synthetic spectra were superposed to all the other lines for a consistency check . by excluding a few pixels of the violet wing of the 1144 absorption which we suspect
is affected by ly @xmath0 contamination we obtain log n(feii ) = 14.35 @xmath6 0.1 at @xmath24= 17.9 km s@xmath21 .
the 1096 line gives a firm upper limit at log n(feii ) @xmath34 14.45 , clearly indicating that the stronger 1144 line can not be significantly saturated . by fitting the full absorption profile of the 1144 line without adopting the above quoted value of @xmath24
, we obtain log n(feii ) = 14.45 at @xmath24 = 30 km s@xmath21 .
we estimated the oi column density from the 1302 line . by constraining the range of broadening values between @xmath24= 14.6 km s@xmath21and @xmath24= 21.2 km s@xmath21
we derive log @xmath35(oi ) = 16.97 @xmath6 0.39 and log @xmath35(oi ) = 15.61 @xmath6 0.19 , respectively .
the central value for the oxygen column density given in table 4 corresponds to the adoption of @xmath24= 17.9 km s@xmath21 , our best estimate of the ni broadening parameter .
the ni column densities of this system were measured via the simultaneous fit of two lines of the ni 1134 multiplet .
the reddest transition of this multiplet is in fact contaminated by a ly @xmath0 interloper ( see fig .
the ni 1200 multiplet is clearly saturated and presents contamination in its weakest line . from the analysis of the 1200 multiplet alone lu et al .
( 1996 ) derived a lower limit ( log n(ni ) @xmath36 14.62 ) which is consistent with the result ( log n(ni ) = [email protected] provided here ( cfr .
table 3 ) .
the ni absorptions in our spectra occur at @xmath37 = 2.82680 .
( 1996 ) found an absorber sub - complex at @xmath37 = 2.83058 ( @xmath38 300 km s@xmath21 ) with an hi column density at 5% that of the main component .
the shift in radial velocity of this secondary component prevents contamination of the 1134 multiplet from which we derived the ni column density . in fig .
5 we show the spectral regions of the siii , sii , feii and oi transitions .
three out of the four siii transitions available are severely blended with other absorptions .
only the saturated siii 1304 line could be fitted , providing a lower limit to the si abundance as explained in section 3 .
all three sii lines present contamination by the ly@xmath0 forest making it impossible to estimate , or even to derive a reliable upper limit to the sulphur column density in this dla . only the 1143 and 1144 transitions were used to derive the feii column density since the other lines are severely blended ( see fig .
the @xmath24 value derived from the iron lines ( 25.2 @xmath6 1.8 km s@xmath21 ) is in good agreement with that obtained from the nitrogen lines ( 23.0 @xmath6 1.8 km s@xmath21 ) .
the iron column density that we infer
log n(feii)= 14.45 @xmath6 0.06 matches very well the result reported by lu et al .
( 1996 ) for this system , namely log n(feii)= 14.48 @xmath6 0.04 .
the oi 1302 transition is heavily saturated and blended . in this case
we simply estimate an upper limit to the oi column density from the undetected 1356 transition .
table 5 lists measurements of the n , s , si , o , fe , and zn abundances taken from our search and from the literature for all the dlas with available nitrogen abundance determinations . the elemental abundances are scaled to the solar reference value by using the standard definition [ x / h ] = log ( x / h)@xmath39 log ( x / h)@xmath40 . the column density ratio ( x / h)@xmath39 is obtained by using the hi column densities given in table 3 .
the solar system values adopted for all the elements discussed here are the meteoritic abundances reported by anders & grevesse ( 1989 ) , with the exception of n and o for which we adopt the most recent solar phostospheric values given by grevesse & noels ( 1993 ) ( see table 5 ) .
the zn abundances are included in table 5 because zn is a good indicator of the system s metallicity since it is practically unaffected by the presence of dust ( pettini et al .
1994 , 1997 ) .
the conversion from column densities to abundances requires several logical steps which are briefly discussed in the following sub - sections .
dla systems have large column densities
n(hi ) @xmath9 2 x 10@xmath41 @xmath11 and are optically thick to interstellar or intergalactic ionizing photons with h@xmath42 @xmath9 13.6 ev in the absorber rest frame .
detailed ionization calculations indicate that species with ionization potential i.p . @xmath9
13.6 ev , such as ni , oi , sii , siii , feii , and znii , are the dominant ionization states and can be used to infer directly the elemental abundance ( fan & tytler 1994 ; lu et al .
we assume that ionization corrections are negligible for these species .
if the galaxy associated to the dla system hosts an intervening hii region , the ionized gas may produce an extra contribution to the column densities of species with i.p .
@xmath9 13.6 ev .
this effect should not alter the abundances of species with ip very close to 13.6 ev such as oi and ni ( since these species will be ionized in an intervening hii region ) but could increase the apparent abundances of other species with higher ionization potentials , such as siii , sii , feii , and znii .
the net result would be an artificial decrease of the [ n / fe ] and [ n / s ] ratios and an increase of the apparent metallicities [ fe / h ] and [ s / h ] which are discussed in section 5 .
the effect should only be important when the intervening hii gas has the same radial velocity and comparable column density of the hi region .
ionized regions with n(hii ) @xmath14 2 x 10@xmath43 @xmath11 would not affect the abundance analysis in any case .
the intervening hii regions will have in general a different kinematics from the hi gas and this will lead to differences in the radial velocity profiles of neutrals ( originated only in hi gas ) and singly ionized species ( originated both in hi and hii gas )
. therefore careful analysis of the absorption profiles of metal lines should help disentangle the contribution of the ionized gas , if at all present . from the analysis of our data
we do not find systematic differences between the @xmath24 values derived from neutral nitrogen and those derived from singly ionized species .
inspection of the highest quality spectra available ( see e.g lu et al .
1996 ) does not reveal systematic differences between the profiles of neutral and low ionization species suggesting that contributions from hii regions are generally negligible .
a fraction of the elements in dlas are expected to be in the form of dust grains thus reducing the gas phase abundance which is precisely what we are measuring .
the evidence for dust in dlas is manyfold .
one is the differential reddening of quasars with and without foreground damped absorption ( pei , fold & betchold 1991 ) .
another piece of evidence comes from the apparent enhancement of the zn to cr abundance ratio in dla systems ( pettini et al .
1994 , 1997 ) .
based on the expectation that zn tracks cr , pettini and colleagues interpret the observed [ zn / cr ] enhancement as a signature of cromium depletion onto dust grains .
vladilo ( 1998 ) calculated dust - to - gas ratios in dlas with fe , cr , and zn measurements and found quantitative agreement with the dust - to - gas ratios based on the reddening determinations by pei , fold & betchold ( 1991 ) .
the detection of the 2175 emission bump in high redshift mgii absorbers ( malhotra 1997 ) and the realization that type ii snae can produce dust ( danziger et al . 1991 ) also suggest that dust is present in the early stages of galactic evolution .
depletions in dla absorbers are expected to be lower than in the galactic ism since the dust - to - gas ratios in dlas are between @xmath19 2% and @xmath19 25% ( vladilo 1998 ) that of the galactic ones .
however , care must be taken when converting column densities into abundances while considering elements which are known to be depleted from gas to dust .
there are different pieces of evidence which indicate that nitrogen is not depleted in the interstellar medium . from the theoretical point of view
, the large activation energy of n@xmath44 prevents involvement of nitrogen in the gas - phase reactions which lead to dust formation in stellar atmospheres ( gail & sedlmayr 1986 ) , and hence interstellar accretion of n onto dust grains is not expected . from the observational point of view , two recent works confirm that interstellar n is not depleted into dust grains . on the one side , interstellar hst observations of the ni @xmath181160 doublet ( meyer , cardelli & sofia 1997 ) show that the gas - phase abundance of nitrogen does not decrease in sight lines with higher h@xmath44 fraction which are indicative of self - shielding environments more hospitable to grains . on the other , iso observations of the n - h stretch spectral region at 2.96 @xmath16 m limit the solid - state of n abundance to log ( n / h)@xmath45 @xmath46 , corresponding to a n depletion lower than 0.01 dex , for the line of sight to iv cyg no . 12 which is known to be heavily reddened ( whittet et al . 1997 ) . among the elements considered here , also o , s and zn are almost unaffected by depletion in galactic interstellar clouds , their typical deficiency with respect to solar values being in the range [ @xmath47 , [ 0.00 ] , and [ @xmath48 dex , respectively ( meyer et al .
1998 , savage & sembach 1996 , roth & blades 1995 ) .
by contrast , refractory elements such as si and fe are known to be depleted by 1 or 2 dex in the galactic ism . in particular , a fraction of iron as high as about 94@xmath49 is locked in dust grains in the warm gas of the galactic disk , and a fraction even greater in cold clouds ( savage & sembach 1996 ) .
these elements are expected to be depleted also in dlas and in section 5 we discuss how to estimate the amount of depletion .
the measurement of dust depletion in the galactic ism relies on the adoption of a suitable reference level for the elemental abundances .
the solar abundance pattern is usually adopted as a standard , but there are indications that the pattern observed in nearby b - type stars might be more appropriate ( see savage & sembach 1996 ) .
metal abundances are typically @xmath50 dex lower in b - type stars than in the solar system and some authors prefer to use 2/3 of the solar system values as a standard for `` cosmic '' abundances ( see savage & sembach 1996 , mathis 1996 ) .
adoption of this abundance reference would yield dust depletions consistently lower in absolute values . the average interstellar abundance of nitrogen , @xmath14log(n / h)@xmath9@xmath51 = [email protected] ( meyer , cardelli & sofia 1997 ) , is slightly deficient with respect to the solar value log ( n / h)@xmath52=4.03 @xmath60.07 ( grevesse & noels 1993 ) , yielding an underabundance @xmath14[n / h]@xmath9@xmath51 = [email protected] .
also nearby b stars show a similar underabundance , with @xmath14log(n / h)@xmath9@xmath53 = 4.17@xmath54 ( venn 1995 ) .
these results indicate non depletion of n into interstellar grains if the b - type star reference is adopted , or depletion of @xmath55 dex if the solar pattern is adopted .
meyer et al . (
1997 ) pointed out that limitations in deriving [ n / h ] in the galactic ism no longer lie in the measurement itself , but in the accuracy of the oscillator strenghts and in the solar photospheric determinations .
due to these uncertainties the authors claimed that it is still premature to rule out an interstellar n abundance either equal to , or 2/3 of solar .
we adopt here the solar reference pattern , and for nitrogen the solar photospheric value of grevesse & noels ( 1993 ) quoted above .
this is not in sharp contrast with the fact that we assume nitrogen to be undepleted in dlas because the expected absolute value for galactic depletion @xmath56 dex would become negligible at the low dust - to - gas ratios typical of dla absorbers .
consequences of adopting the b stars as standard for the abundances are considered in the next section .
the absolute nitrogen abundances in dla absorbers shown in table 5 span more than two orders of magnitude , with 3.7 dex @xmath14 [ n / h ] @xmath57 dex . from the point of view of chemical evolution models ,
relative abundances are more reliable than absolute abundances because they generally depend only on the elemental nucleosynthesis and stellar lifetimes , whereas absolute abundances are affected by specific assumptions underlying the models . here
we discuss the abundance of nitrogen relative to the iron - peak and to @xmath0 elements , which are the main products of type ia snae and type ii snae , respectively .
we compare measurements of [ n / fe ] and [ n/@xmath0 ] ratios in dlas and in other astrophysical sites of similarly low metallicity , namely galactic halo stars , blue compact galaxies , and dwarf irregular galaxies .
these abundance ratios are then compared with predictions of galactic chemical evolution models .
the nitrogen abundances relative to iron measured in dla systems are plotted versus [ fe / h ] in fig .
the filled symbols represent our data while open symbols denote measurements taken from the literature .
the [ n / fe ] values are deficient compared to the solar ratio and do not exhibit a trend with metallicity .
the measurements show a relatively high amount of dispersion , with one value as low as [ n / fe ] = 1.3 ( @xmath15=2.84 towards qso 1946 + 7658 ) and several high values around [ n / fe ] = 0.3 .
there is also a hint of a possible higher dispersion at lower metallicities . the dotted lines in fig .
6 indicate the range of measurements in galactic halo dwarfs with @xmath58 [ fe / h ] @xmath59 ( carbon et al .
the solid line is the regression found by these authors with a scatter of @xmath600.19 dex .
the large scatter in the stellar measurements reflects the difficulties in deriving n abundances from nh bands .
the non dependence of [ n / fe ] with stellar metallicity has been considered as evidence for a primary component of nitrogen ( pagel 1985 , carbon et al .
1987 ) . the dot symbols in fig .
6 represent the measurements in blue compact galaxies ( bcg ) taken from the sample of thuan , izotov & lipovetsky ( 1995 ) .
bcgs abundances take up about the same region of metal - poor stars . however , most of them lie below the linear regression of stellar data and closer to dla galaxies . at the lowest metallicities
there are no bcgs to compare with dlas . at a first glance
there is general consistency between dla , bcg , and halo star measurements , with the exception of a few cases in which the [ n / fe ] are lower in dlas .
however , the interpretation of the observed [ n / fe ] ratios must take into account that iron is one of the most refractory elements , while nitrogen is undepleted , as we discussed in section 4.3 . before drawing any conclusions from the comparison of the [ n / fe ] ratios
it is hence necessary to take into account the presence of dust .
the [ n / fe ] values corrected for dust depletions are shown in fig .
7 where we use the same symbols for dlas as in fig . 6 .
the net effect of the correction is a diagonal shift of the points at different lengths for each object which results from the increase in fe abundance at invariant n abundance .
the iron depletions were estimated with two methods .
for the dlas which have both fe and zn abundance measurements we estimated the amount of depletion by assuming that zinc tracks closely iron , [ zn / fe ] @xmath19 0 , as is indeed observed in galactic stars with metallicities @xmath61 [ fe / h ] @xmath62 ( sneden , gratton , & crocker , 1991 ) and in the large magellanic cloud stars ( russel & dopita 1992 ) , and by assuming that dust in dlas is of the same type as the galactic one . with both assumptions we estimated the fe and zn depletions starting from the galactic interstellar values @xmath63(fe ) = @xmath64 dex and @xmath63(zn ) = @xmath65 dex ( savage & sembach , 1996 , roth & blades 1995 ) ) .
this approach is similar to that employed by pettini et al .
( 1994 , 1997 ) , with the difference that we take into account not only iron depletion , but also that of zinc .
the two dlas with both fe and zn measurements available are indicated in fig . 7 with a filled triangle ( system towards qso 0100 + 1300 ) and a star ( system towards qso 1331 + 1700 ) . for the remaining dlas without zn but with fe measurements available we estimated the correction for dust depletion from eq .
( 17 ) by vladilo ( 1998 ) by adopting the average dust - to - metals ratio in dlas @xmath66 given there . the resulting [ n / fe]@xmath7 values are indicated in fig . 7 with a 45 degree error bar which corresponds to the above quoted dispersion in the dust - to - metals ratio .
the left and up arrows shown in fig .
7 for the absorber towards qso 00002620 ( filled square ) were derived from the stringent zn upper limit available for the system ( see table 5 ) , i.e. by taking [ fe / h]@xmath7 @xmath19[zn / h]@xmath39 and [ n / fe]@xmath7 @xmath19 [
n / zn]@xmath39 , respectively . since the data points are now corrected for the presence of dust , the spread observed in fig .
7 should reflect intrinsic differences in the relative nitrogen abundance . particularly relevant
is the dla at @xmath15 = 3.390 towards qso 00002620 , ( filled square ) for which [ n / fe ] remains significantly higher than in the other two dlas with zinc available .
this spread is specific to n and has no equivalent in other abundance ratios measured in dlas , which are remarkably similar both at face values ( see e.g. fig .
23 by lu et al . 1996 ) and after correction for dust depletion ( vladilo 1998 ) . as can be seen in fig .
7 most of the [ n / fe]@xmath7 measurements in dlas fall well below those of the metal - poor halo stars .
the slight underabundance of the measurements taken at face value shown in fig .
6 is in fact amplified after correction for dust depletion .
this conclusion still holds if we use the b star abundances , instead of the solar ones , as a reference for the `` cosmic '' abundances since , in this case , the [ n / fe ] ratio remains unchanged and the data points in fig .
7 shift along the x - axis by only + 0.18 dex . in blue compact galaxies , the [ n / fe ] determinations
are based on observations of emission lines from hii regions where dust is also known to be present but is not accounted for ( garnett et al .
if the presence of dust is considered , the bcgs data points of fig .
6 would shift in the same direction as the dla points in fig .
therefore , we can not exclude that [ n / fe ] measurements in bcgs are consistent with dla determinations , as they were before the correction for the presence of dust .
after correction for dust depletion the majority of the dla data are in line with a pure secondary origin of nitrogen , which in fig .
7 is sketched as a dashed - line passing through the solar point .
this is particularly clear at least in two cases : namely the dlas towards qso 0100 + 1300 ( triangle ) and towards qso 1331 + 1700 ( star ) .
this behaviour is quite different from that of halo metal - poor stars which show a constant [ n / o ] with metallicity , consistent with a primary origin of nitrogen .
the difference between the nitrogen abundances of galactic metal - poor stars and dlas suggests that the chemical enrichment history of these dlas has been somewhat different from that experienced by our own galaxy .
however , there is at least one remarkable exception for qso 0000 - 2620 where the nitrogen remains particularly high , at the level observed in galactic halo stars .
the [ n/@xmath0 ] ratio is an important tool when unravelling the nucleosynthetic history of nitrogen .
the abundances for the @xmath0 elements s , si , and o are reported in table 5 .
unfortunately it is quite difficult to obtain accurate oxygen abundances in dlas , as we have discussed in section 3 .
this problem has often been circumvented by taking s or si as a proxy for o , under the assumption that the [ @xmath0/o ] ratio does not evolve with metallicity .
s and si are @xmath0-elements produced in the same massive stars that produce o , and therefore the ratios between these elements are predicted to vary little , if at all , with time ( timmes woosley & weaver 1995 ) .
however , some caution must be taken when using silicon as a proxy of oxygen . from the theoretical point of view , type i snae produce more silicon than oxygen ( nomoto et al .
1984 ; thielemann et al .
the observations indicate that si traces o in galactic stars ( kilian - montenbruck , gheren & nissen 1994 ; venn 1995 ) , but is slightly underabundant relative to o in extragalactic hii regions ( garnett et al . 1995 ) .
this underabundance is interpreted as due to si depletion onto dust grains .
in addition to the uncertainty due to silicon depletion , we remark that si lines are often saturated even at the low column densities typical of dlas . for these reasons
we prefer to adopt sulphur as a tracer of the @xmath0-elements .
sulphur is found to track closely oxygen in galactic stars ( see references in lauroesch et al .
1996 ) , in dwarf galaxies ( garnett 1989 , skillman & kennicutt 1993 , skillman et al . 1994 ) and in bcgs ( thuan et al . 1995 ) in the entire range of metallicities explored .
sulphur is not depleted and , when detected , its lines are unsaturated . hence , we consider it safe to assume that [ n / o ] @xmath19 [ n / s ] in dlas . in fig .
8 we compare [ n / o ] measurements in dla systems , in galactic stars , and in dwarf irregular galaxies with predictions emerged from chemical evolution models . unless stated otherwise , dla measurements have been derived from sulphur by assuming that [ o / s ] @xmath67 0 .
larger sized symbols represent dla measurements , and filled symbols highlight the determinations obtained by our group .
the dashed - line represents the mean trend of the [ n / o ] ratio in galactic stars of all metallicities ( tomkin & lambert 1984 ) .
the dots represent the dwarf galaxies compiled by van zee et al .
( 1996 ) , van zee l. , haynes p.m. & salzer j.j .
( 1997 ) and by kobulnicky & skillman ( 1996 ) which include the bcgs of thuan et al .
( 1995 ) .
the solid lines are theoretical predictions of [ n / o ] from a chemical evolution model for the solar neighbourhood under different assumptions for the nucleosynthesis of nitrogen , described in detail by matteucci , molaro & vladilo ( 1997 ) .
the curve labelled `` s '' assumes purely secondary nitrogen produced in stars of all masses .
the `` p '' curve assumes some amount of primary production in intermediate mass stars according to renzini & voli ( 1981 ) and with a convective scale length @xmath0=1.5 .
the `` pmassive '' curve with a plateau assumes primary n both in intermediate mass stars and in massive stars , where in all the other tracks it is secondary .
as can be seen in the figure , the [ n/@xmath0 ] ratios in dlas are highly scattered and the majority of the measurements have no precedents , at comparable level of metallicity , in galactic stars . instead , intrinsic dispersion is observed in dwarf galaxies , with a range of [ n / o ] values similar to those found in dla galaxies . since both n and s
depletions are expected to be negligible , the new plot provides further evidence in addition to fig .
7 that the dispersion of n abundances is real and not due to different amounts of dust among the dlas .
the detailed study of individual cases shows peculiar behaviours with respect to the predictions of chemical evolution models .
the [ n / s ] ratios in the dlas towards qso 0100 + 1300 ( triangle ) , qso 1331 + 1700 ( star ) , qso 03473819 ( octagon ) and the upper limit towards qso 2348 - 1444 ( empty square with arrow ) are in between the primary and secondary nitrogen evolutionary tracks .
the [ n / s ] value in the dla towards qso 0930 + 2858 ( empty square in the middle of fig .
8) is close to the curves which consider a primary component of n either in intermediate and massive stars .
the value for the dla towards qso 2343 + 1230 ( empty square at the top of the figure ) falls well above the area limited by the evolutionary tracks with secondary plus primary production of nitrogen .
this is probably also true for the absorber towards qso 0000 - 2620 . in this case
we show in fig .
8 both the lower limit derived from the conservative sulphur upper limit ( lu , sargent & barlow 1998 ; filled square with arrows ) and the value from the direct oxygen measurement ( molaro et al .
1996 ; filled square at the left top of the figure ) .
this last determination was obtained by constraining the @xmath24 value of the saturated oxygen lines from the broadening obtained from the nitrogen lines ( see discussion in section 3 ) . even considering the uncertainty of the oxygen measurement
it is clear that nitrogen has an intrinsically high abundance in this dla .
it is important to note that for specific dla systems the [ n/@xmath0 ] and [ n / fe]@xmath7 values are mutually consistent .
the same points which show a secondary behaviour in [ n/@xmath0 ] also show a secondary behaviour in [ n / fe]@xmath7 .
we stress that the [ n / fe ] and [ n / s ] results would not lead to consistent results if iron depletion were not taken into account .
the different values of n / o observed at a given o / h metallicity in dlas may be attributed to a delay between the delivery of o and n into the ism when the star formation proceeds in bursts ( edmunds & pagel 1978 ) .
this delay has also been suggested by garnett ( 1990 ) to explain the n / o scatter in dwarfs irregulars . according to this picture , galaxies which have experienced a recent episode of star formation show low n / o ratios due to the quick production of o in massive stars and the prompt return to the ism which is over after a few 10@xmath68 years . on the other hand , in galaxies that have been quiescent for a long period , nitrogen
would be returned to the ism after a few 10@xmath69 years mainly as a product of intermediate mass stars . during the long quiescence
the n / o ratio will increase at about constant o / h . in this model , different ages or temporal gaps from recent bursts of star formation
are responsible for the dispersion of the n / o ratios , with low values implying a recent burst and high values pointing to a long quiescent period .
skillman ( 1998 ) discussed two cases of dwarf irregular galaxies with values of [ n / o ] around 0.7 for which there is clear evidence of recent star formation , in agreement with the interpretation of a prompt o enrichment of the ism .
however , some problems arise in this scenario if dla abundances of other elements are considered .
if the delayed model also applies to dlas , then it seems reasonable to expect an overabundance of the @xmath0-elements relative to iron - peak elements when low values of [ n / o ] are observed , i.e. in the phase temporally near the starburst .
in fact in this phase the elemental production is dominated by type ii snae which are rich in @xmath0-elements .
the few cases in which a measurement of the [ @xmath0/fe ] ratio and of nitrogen abundances are available do not conform with this prediction .
for instance , as discussed in mcv , the dla galaxies towards qso 0100 + 1300 and qso 1331 + 1700 have [ s / zn ] @xmath19 0 even if they show the lowest [ n/@xmath0 ] values . in the absorber towards qso 1331 + 1700
it has been suggested that an intervening hii region could lead to an artificially low [ n / s ] ratio ( see section 4.2 ) , since [ n / o ] @xmath9 [ n / s ] in this system ( kulkarni et al .
for the system towards qso 0100 + 1300 it is improbable that the [ n / s ] ratio is artificially lowered due to an intercepted hii region since independent fits of the ni and sii lines give consistent @xmath24 values ( see mcv ) , while differences between the profiles of the two species would be expected if contribution from hii regions were present .
the low [ n / s ] value in this dla absorber is therefore difficult to understand in the framework of the n - delayed model since the expected overabundance of the [ @xmath0/fe ] ratio is not appreciable . also the highest [ n/@xmath0 ] ratios observed in the dlas in fig .
8 are difficult to reconcile with the delayed nitrogen release .
in fact in this model , n / o ratios are expected to lie in the region limited by the tracks of primary and secondary nitrogen production ( vila - costas & edmunds , 1993 ) .
as we mentioned before , the system towards qso 2343 + 1230 , and perhaps also the one towards qso 0000 - 2620 , lie well above the track of primary plus secondary production instead .
matteucci , molaro & vladilo ( 1997 ) showed that the highest n / o values observed in dla galaxies can only be reproduced by chemical evolution models which assume primary production of n in massive stars , together with selective galactic winds which carry away , preferentially , the products of type ii snae explosions , such as oxygen and other @xmath0 elements .
we conclude that the time delay model requires other _ ingredients _
( for example selective galactic winds among others ) in order to explain some of the n / o ratios observed in dlas , once they are integrated with the other abundances observed in these galaxies .
new abundance determinations are needed in order to understand how common are the very low and extremely high ratios in dla galaxies .
we have presented three new measurements of nitrogen abundances in dla galaxies , increasing by 50% the number of determinations available in the literature ( not including upper limits ) .
the abundances of nitrogen have been discussed in connection with the iron - peak elements ( iron and zinc ) and @xmath0-elements ( represented by sulphur ) for the nine absorbers with available n data .
the effect of dust depletion on the abundance determinations has been taken into account .
we have compared the nitrogen measurements in dla galaxies with measurements in other astrophysical sites of low metallicity and with predictions from chemical evolution models .
the main results of the present work can be summarised as follows . *
the n / s and n / fe abundance ratios show a scatter of one order of magnitude or more , at variance with the remarkably low scatter observed for other elemental ratios measured in dla galaxies .
* when the presence of dust is considered to correct the iron abundances , a substantial fraction of [ n / fe ] values are lower than those in galactic halo stars .
a similar conclusion holds for the [ n/@xmath0 ] ratios obtained from the [ n / s ] measurements which are expected to be unaffected by dust depletion .
these results suggest that the chemical evolution of dla galaxies has been somewhat different from that experienced by the milky way . *
the [ n / fe ] ratios in bcgs are similar to those measured in dla galaxies and this similarity may remain also after correction for the presence of dust , provided this correction is also appropriate for bcgs .
the n / o ratios in dwarf galaxies have similar values and exhibit a similar spread to those observed in dla galaxies .
* when compared with theoretical evolution models , it is not possible to explain the observations in dla galaxies by invoking a unique production mechanism .
some cases are consistent with a secondary behaviour of nitrogen , while others require primary production .
this may suggest that dla absorbers include galaxies with different chemical histories .
most of the scatter of the nitrogen abundances can also be interpreted in the framework of the delayed model for nitrogen production ( edmunds & pagel , 1978 ; garnett 1990 ) .
however in this scenario dla galaxies with low [ n/@xmath0 ] values , which should be temporally close to the starburst , are expected to show an enhancement of @xmath0-elements which is instead not observed , making necessary the addition of new ingredients to the model in order to explain the observed abundance ratios . * in one or possibly two dla galaxies there is evidence for an extremely high n / o ratio , well above the values measured in galactic halo stars or in extragalactic hii regions .
the few dla systems with n / o ratios higher than those in any other astrophysical site can not be accommodated in the framework of the delayed nitrogen model : in this case differential galactic winds plus primary n production in massive stars are required as ingredients of the chemical evolution model ( matteucci et al . | nitrogen lines of the ni 1134 and 1200 multiplets in the damped ly@xmath0 ( dla ) galaxies at @xmath1 = 2.309 , 2.827 and 3.025 toward the qsos 0100 + 1300 , 1425 + 6039 and 0347@xmath23819 respectively , have been detected by means of high resolution spectra ( r @xmath3 2 @xmath4 10@xmath5 ) obtained with 4-m class telescopes at eso ( la silla , chile ) and orm ( la palma , spain ) .
the two ni multiplets offer a considerable range in oscillator strengths and the possibility of disentangling ly@xmath0 interlopers .
the derived nitrogen abundances for the three damped systems are [ n / h]= [email protected] , [email protected] , [email protected] , respectively .
the behaviour of nitrogen relative to iron - peak and @xmath0-elements has been investigated by considering all the extant ni determinations for a total of 9 dla galaxies .
we have estimated the fraction of iron locked into dust grains to convert the observed [ n / fe ] ratios into overall ( dust plus gas ) relative abundances , [ n / fe]@xmath7 .
the ratios [ n/@xmath0 ] have been mostly determined by using sulphur as a tracer of @xmath0-elements which is unaffected by dust .
the [ n / fe ] and [ n/@xmath0 ] ratios show high dispersions , of one order of magnitude or more , which have no equivalent in other element - to - element ratios in dlas .
the lowest values of the [ n / fe]@xmath7 and [ n/@xmath0 ] ratios are at variance with the values measured in galactic halo stars of similar metallicity , suggesting that part of the dla galaxies do not follow the chemical evolution of the milky way .
the dla nitrogen abundances and their dispersion show some similarities with those observed in dwarf galaxies .
comparison with chemical evolution models show that the lowest [ n / fe]@xmath7 and [ n/@xmath0 ] dla values are close to what would be expected for a pure secondary origin of nitrogen , while higher values are mostly consistent with a primary component .
the behaviour of nitrogen abundance ratios can be ascribed , in general , to the delayed release of nitrogen in the course of evolution .
however it is difficult to conciliate this interpretation with the lowest [ n/@xmath0 ] values measured , since an expected enhancement of @xmath0-elements respect to the iron - peak elements is not observed simultaneously in these dla galaxies . in two cases ,
relatively high [ n/@xmath0 ] values are observed which require also a more complex chemical evolution to be explained . |
the cosmic clock ticks logarithmically .
the universe s past can be divided into three well defined epochs : the quantum era ( @xmath1 to @xmath2 ) ; the quark era ( @xmath2 to @xmath3 ) ; and the hot big - bang era ( @xmath3 to @xmath4 ) .
earlier than the quantum era is the planck era ( @xmath5sec ) ; it holds the key to understanding the birth of the universe , but requires a quantum theory of gravity to proceed .
just recently the universe has entered a new era , of undetermined duration , where dark energy and its repulsive gravity are dominating the dynamics of the universe ( more later ) .
the third era derives its name from the very successful cosmological model that describes it ; the important events include the synthesis of the light elements , the last scattering of radiation , the formation of large - scale structure and the onset of accelerated expansion .
the earlier two eras are still largely terra incognita , though we have some tantalizing ideas : the transition from quark / gluon plasma to hadronic matter , the electroweak phase transition , and the birth of particle dark matter during the quark era ; and inflation and the origin of the matter
antimatter asymmetry during the quantum era ( see figure ) .
there is no doubt that cosmology is in the midst of the most exciting period ever .
progress in understanding the origin and evolution of the universe is proceeding at a stunning pace .
i mention a few of the highlights of the past decade : cobe discovery of the small density inhomogeneities ( @xmath6 ) that seeded large - scale structure , the mapping of cmb anisotropy on scales down to 0.1 degree and the determination of the shape of the universe ( flat ! ) , identification of the epoch of galaxy formation , determination of the hubble parameter to a precision of 10% , and the discovery of cosmic accelerated expansion .
there is much more to come .
these recent discoveries and those still to be made will test the framework of the standard cosmology and begin to open the quark and quantum eras .
we are already on the way to establishing a new standard cosmology : a flat universe comprised of 2/3rds dark energy and 1/3rd dark matter that is accelerating today and whose seeds for structure came from quantum fluctuations stretched to astrophysical size during inflation .
much remains to be done to put the new cosmology on the same firm footing as the hot big - bang model ; much of cosmology over the next two decades will be devoted to this ( see e.g. , turner , 2001a ) .
the richness and redundancy of the measurements to be made will also allow a number of consistency tests of the big - bang framework and its theoretical foundation , general relativity .
many of these tests involve cosmic timescales , and that is the subject of my talk .
the product of the present age and hubble constant is a powerful consistency test .
the standard hot big bang has not always cleared this hurdle , cf .
the late 1940s when the hubble age of about 2 billion years fell short of the age of the solar system by a factor of two ( see e.g. , kragh , 1996 ) or the more recent scares when some measurements of @xmath8 and @xmath9 drifted upward . recognizing the importance of age consistency , sandage has devoted much of his career to measuring the hubble constant @xmath8 and independent indicators of the age of the universe ( and pioneered many of the methods ) ; hence the title of this section .
the nature of the test has changed over sandage s career ; the importance has not . until recently one
would have written @xmath7 in terms of one parameter , @xmath10 , the fraction of critical density in matter , also assumed to be the total fraction of critical density contributed by all forms of matter and energy ( @xmath11 ) .
the discovery of accelerated expansion increased the number of parameters to two , @xmath10 and @xmath12 ( fraction of critical density in a cosmological constant ; @xmath13 ) .
the realization that accelerated expansion is here to stay , while the cosmological constant may not , added another parameter , @xmath14 ( the equation - of - state of the dark energy ) ( see below ) . the determination from cmb anisotropy measurements that
the universe is flat , effectively reduced the number of parameters to two , @xmath10 and @xmath15 : @xmath16 for the first two cases , there are well known analytic formulae for @xmath17 ; for the others there are not .
there is hope , that in the next ten years @xmath10 will be pinned down by a combination of cmb , cluster and large - scale structure measurements , reducing the number of parameters to one again . at the moment ,
my analysis of this data gives , @xmath18 ( turner , 2001b ) , and i believe there will be significant improvement over the next decade .
the function @xmath17 that accounts for the effect of slowing / speeding up on the age of the universe is given by ( for a coasting universe @xmath19 ) : @xmath20^{1/2}\end{aligned}\ ] ] where the small contribution from the cmb and relativistic neutrinos ( @xmath21 ) has been neglected and @xmath15 has been assumed to be constant . to illustrate the status of this test , for @xmath8 and @xmath9 i will use the values reported at this meeting , @xmath22 ( freedman , 2001 ) and @xmath23gyr ( chaboyer , 2001 ) .
this leads to @xmath24 taking @xmath25 and @xmath26 , this is consistent with @xmath27 ( at @xmath28 ) the new cosmology passes the sandage consistency test with flying colors ! while not a tight constraint on @xmath15 , it does provides more evidence for dark energy ( i.e. , a smooth component with large , negative pressure ) : if the dark energy were pressureless ( @xmath29 ) , then @xmath30 , which is clearly inconsistent with current data . at present
, the sandage test has a precision of about 15% ; what kind of improvement could one hope for over the next decade . together ,
physics - based measures of @xmath8 ( especially the cmb ) and distance scale measures might well pin down the hubble constant to a few percent .
however , it is difficult to imagine independent measures of the age achieving similar accuracy . one irreducible uncertainty for all methods , is the time to formation ( of stars , globular clusters , white dwarfs , etc ) . reducing this uncertainty to 0.5gyr
would still leave a 5% uncertainty in @xmath7 .
as i will discuss , other tests of age consistency , at much earlier times , are likely to be significantly more precise .
the expansion rate at any epoch is related to the age of the universe .
for example : during a matter - dominated epoch , the scale factor @xmath31 and @xmath32 ; and during a radiation - dominated epoch @xmath33 and @xmath34 .
the expansion rate is more accessible than the age at early times .
the expansion rate determines a key observable : the comoving distance to an object a redshift @xmath35 , @xmath36 where a flat universe has been assumed .
the cosmological volume element , luminosity distance and angular - diameter distance are all directly related to @xmath37 @xmath38 a number of probes of the expansion history from @xmath39 to @xmath40 have been discussed recently .
they include gathering a large sample of type ia supernovae ( snap ) and counting halos or clusters of galaxies . from these `` experiments , '' individually or taken together
, one could imagine mapping out @xmath41 back to redshift @xmath40 ( see e.g. , huterer & turner , 2000 ; or tegmark , 2001 ) .
note that the volume element and luminosity distance taken together , can in principle directly yield @xmath41 .
loeb ( 1998 ) has gone one step further , suggesting ultra - precise redshift measurements of the lyman - alpha forest over a decade or more could be used to determine @xmath41 directly ( ! ) .
the idea is to measure the tiny time variation of the redshifts of thousands of lyman - alpha clouds , @xmath42 \delta t \nonumber \\ & \sim & 0.2\,(\delta t /10\,{\rm
yrs})\,{\rm m\,sec^{-1}}\qquad { \rm for\ } z=3\end{aligned}\ ] ] where @xmath43 is the time interval of the observations .
whether or not this bold idea can be carried out remains to be seen , but it certainly is exciting to think about .
as is now very familiar , the power spectrum ( in multipole space ) of cmb anisotropy arises due to acoustic oscillations in the baryon photon fluid around the time of last scattering ( see e.g. , hu et al , 1997 ) . at this time
, the baryons ( and electrons ) are falling into the dark - matter potential wells ; still coupled to photons ( through thomson scattering off electrons ) , the infall is resisted by photon pressure and oscillations ensue . at maximum compression
the photons are heated and at maximum rarefaction they are cooled .
the cmb is a snapshot of the universe at 400,000 yrs ; regions caught at maximum compression ( rarefaction ) lead to hot ( cold ) spots on the microwave sky . while the physics is most easily explained in real space , the signal is best seen in multipole space , as a series of peaks and valleys .
the power at multipole @xmath44 is largely due to @xmath45-modes satisfying : @xmath46 the condition for maxima in the power spectrum is : @xmath47 , where @xmath48 ; the odd peaks are compression peaks and the even peaks are rarefaction peaks , @xmath49 is the physical oscillation frequency at the time of last scattering , and @xmath50yrs is the age of the universe then .
thus , in a flat universe , the harmonic peaks occur at multipoles @xmath51 while the above formulae are approximate , they capture the physics .
in particular , that the positions of the peaks depends upon the age of the universe at last scattering .
there will be sufficient redundancy in the information encoded in the 3000 or so multipole amplitudes that will be measured to not only determine cosmological parameters , but also to check the consistency of the standard relationship between age and energy density .
in particular , an analysis by lopez et al ( 1999 ) has projected that the planck mission will be able to peg the neutrino contribution to the energy density of the universe at last scattering to about 1% .
the expansion rate is related to the energy density , @xmath52 , and neutrinos contribute about 1/5th of the total energy density at the time of last scattering .
a quick estimate from lopez et al ( 1999 ) indicates that ultimately , cmb anisotropy will provide an age consistency test at about 0.1% precision , 400,000 yrs after the beginning . the limitations of this test
should be noted however . actually determining the age of the universe at last scattering , whose redshift is readily determined by thermodynamic considerations ( @xmath53 ) , is pegged to the present hubble constant , @xmath54 , and
thus can be no more accurate than the age of the universe itself .
the cmb consistency above actually probes the relationship between expansion rate and the energy density of the universe , a test of the big - bang framework and general relativity .
big - bang nucleosynthesis ( bbn ) is a cosmological experiment of great importance . in essence
, it is a quenched nuclear reactor whose by - products are the primeval mix from which the first generation stars were born .
the expansion rate of the universe controls the quench rate : @xmath55 .
the yields of d , @xmath56he , @xmath57he and @xmath58li , the primary products of bbn , are sensitive to the quench rate , and thus to the expansion rate @xmath59 where the coefficient of proportionality varies from about @xmath60 for @xmath58li to about @xmath61 for @xmath57he
. measurements of the primeval deuterium abundance ( see e.g. , omeara et al , 2001 ) together with predictions for its bbn yield pin down the baryon density , @xmath62 ( burles et al , 2001a ) . using this value , the other light - element abundances can be predicted , e.g. , @xmath63 . of the four light elements ,
the primeval @xmath57he abundance is known most accurately , @xmath64 ( though there is still much debate about systematic error ; see e.g. , burles et al , 2001b , or olive et al , 2000 ) .
the sensitivity of the @xmath57he abundance to the expansion rate is : @xmath65 .
the current agreement of prediction with observation translates to a 2% test of the consistency of the big - bang prediction for the expansion rate around 1 sec , or about 5 times better than the sandage test ! with improved measurements of the baryon density , both from bbn and cmb anisotropy ( see below ) , one might hope to see an improvement of a factor of five or so in the next decade . finally , as carroll & kaplinghat ( 2001 )
have recently emphasized , bbn can be discussed without reference to the standard cosmological model .
the yields can be analyzed completely in terms of the quench rate , @xmath66 , and nuclear input data with only the assumption of the robertson
walker line element ( and not the friedmann equations which relate the quench rate to the temperature ) .
bbn not only offers a window on the very early universe , but an almost model - independent timescale test .
together , the cmb and bbn offer a remarkable test of the consistency of the standard cosmology and general relativity .
as mentioned in the previous section , measurements of the primeval deuterium abundance together with precise theoretical predictions can be used to infer the baryon density : @xmath67 since the first measurements of the primeval deuterium abundance in 1998 , bbn has provided the best determination of the baryon density ( see e.g. , schramm & turner , 1998 ) .
the cmb provides an independent measure of the baryon density based upon the physics of gravity - driven acoustic oscillations 400,000 yrs after bbn .
specifically , it is the ratio of the heights of the odd to even acoustic peaks that is sensitive to the baryon density ( and insensitive to other cosmic parameters ; see e.g. , hu et al , 1997 ) .
the ratio of the heights of the first and second peaks is @xmath68 where the peak heights are measured in microkelvin@xmath69 .
the boomerang and maxima experiments were the first cmb experiments to probe the first and second peaks ( and to show that the universe is flat ; de bernardis et al , 2000 ; hanany et al , 2000 ) ; based upon their results a value for the baryon density was inferred ( jaffe et al , 2001 ) @xmath70 while this baryon density is about @xmath71 higher than the bbn value , it confirmed a key prediction of bbn a baryon density far below that of the best estimates of the total matter density ( @xmath72 vs. @xmath73 )
. only a couple of months after this meeting , carlstrom s dasi experiment at the south pole announced even more accurate and independent measurements of the first three acoustic peaks ( pryke et al , 2001 ) and arrived at a slightly lower baryon density : @xmath74 which is bang on the bbn value ! boomerang analyzed more data and refined their beam map and pointing model and arrived at an identical value ( netterfield et al , 2001 ) . the agreement between these two numbers
is stunning .
using the fact that ( d / h)@xmath75 and the sensitivity of d / h to the expansion rate , @xmath76 it follows that @xmath77 thus , the agreement of the bbn and cmb baryon densities checks the consistency of the standard expansion rate ( at about 1sec ) to a precision of about 15% . eventually , both determinations of the baryon density should achieve about 1% or so accuracy , and a factor of ten in the precision of this test might be expected .
the evidence for particle dark matter has only gotten stronger : firmer evidence for @xmath78 ( discussed above ) ; the many successes of the cdm scenario of structure formation ( and no viable model for structure formation without particle dark matter ) ; the detection of acoustic peaks as predicted by inflation and cdm ( netterfield et al , 2001 ; pryke et al , 2001 ) ; and growing circumstantial evidence for supersymmetry .
the most promising particle candidate is the lightest supersymmetric particle , which in most models is a neutralino of mass 100 to 300 gev ( see e.g. , jungman et al , 1996 ) .
relic neutralinos remain numerous today because of the incompleteness of neutralino annihilations in the early universe . at temperatures when @xmath79 , neutralinos and anti - neutralinos are present in numbers comparable to that of photons ; as the temperature drops neutralinos must annihilate to maintain thermal abundance ( a factor @xmath80 less than that of photons ) .
eventually , annihilations can not keep pace with the quench rate ( the annihilation rate per neutralino , @xmath81 , falls rapidly as the neutralino abundance decreases exponentially ) and the neutralino abundance `` freezes out . ''
freeze out occurs at a temperature , @xmath82 , corresponding to a time of around @xmath83sec .
the mass density contributed by relic neutralinos is given by @xmath84 while an approximation , this formula captures the essence of the neutralino production process ( see e.g. , kolb & turner , 1990 ) . here is the future timescale test : @xmath85 will be measured by cmb anisotropy experiments to percent level precision .
the properties of the neutralino can be measured at an accelerator lab ( next linear collider ? ) to 10% precision ( brhlik et al , 2001 ) . from this , the expansion rate at @xmath86 can be inferred and compared to the standard formula .
( to be more precise , the measured properties of the neutralino and the boltzmann equation in the expanding universe can be used to predict the relic mass density and compared with the value inferred from cmb measurements . )
if the neutralino is indeed the dark - matter particle ( or another particle that can be produced in the lab ) , we can look forward to a 10% consistency test of the expansion rate at a time of less one microsecond !
if current ideas about particle physics are correct , then , during its earliest moments , the universe should have gone through a series of phase transitions associated with symmetry breaking ( e.g. , qcd , electroweak , grand unification , compactification ? ) . during a first - order phase transition ,
the universe gets `` shaken up '' as bubbles of the new phase expand and collide .
this can lead to the production of prodigious amounts of gravitational radiation , resulting in @xmath87 today .
laboratory - based knowledge of particle physics ( e.g. , for the electroweak phase transition , the mass of the higgs boson ) can be used to predict with precision the spectrum and amount of gravitational radiation ( kosowsky , turner & watkins , 1992 ) ; both depend directly upon the expansion rate at the temperature at which the phase transition takes place . if the stochastic background of gravitational waves from a phase transition can be detected and the particle physics independently probed in the laboratory a cosmological timescale test can be carried out at very early times ( e.g. , for the electroweak phase transition , @xmath88sec ) .
inflation also produces gravitational waves , by a different mechanism
desitter - space produced quantum fluctuations in the space - time metric .
the stochastic background of gravity waves produced by inflation have wavelengths from 1 km to the beyond the size of the present horizon .
gravity waves are one of the three key predictions of inflation ( together with a flat universe and a nearly scale - invariant spectrum of adiabatic , gaussian density perturbations ; see e.g. , turner , 1997a ) .
detection of gravitational waves , either by their imprint on cmb anisotropy and/or polarization or directly by a gravity - wave detector ( ligo ?
, lisa ? ) , would not only confirm a key prediction of inflation , but would also reveal the timescale for inflation : the level of gravitational radiation produced by inflation is directly related to the expansion rate during inflation : @xmath89 measuring the dimensionless metric strain @xmath90 gives the hubble parameter during inflation in units of the planck energy ( @xmath91gev ) . further ,
if the spectrum of gravitational radiation can be probed , then there is a consistency test of inflation ( and cosmology ) : @xmath92 , where @xmath93 ( @xmath94 ) is the contribution of gravity waves ( density perturbations ) to the quadrupole cmb anisotropy , and @xmath95 is the spectral index of the inflation - produced gravitational waves ( turner , 1997a , b ) . as a practical matter
, gravity waves from inflation can probably only be detected if @xmath96gev , corresponding to a timescale of @xmath97sec . said another way , _ if _ gravity waves from inflation are detected we will be probing the universe at a time at least as early as @xmath98sec .
for decades cosmologists have believed that geometry ( or equivalently @xmath99 ) and the fate of the universe were linked .
the shape of the universe has been determined through measurements of cmb anisotropy , @xmath100 ( pryke et al , 2001 ; netterfield et al , 2001 ; hanany et al , 2001 ) , but we are further away than ever from determining the destiny of the universe .
this is because 2/3rds of the critical density is in dark energy rather than matter , and the connection between geometry and destiny only applies when the universe is comprised of matter ( or more precisely , stress - energy with @xmath101 ; see krauss & turner , 1999 ) .
the discovery of accelerated expansion through type ia supernovae distance measurements ( see tonry , 2001 ) was surprising , but can easily be accommodated within the framework of general relativity and the hot big - bang cosmology : in general relativity , the source of gravity is @xmath102 , so that a fluid that is very elastic ( negative pressure comparable in magnitude to energy density ) has repulsive gravity .
the simplest example is the quantum vacuum ( mathematically equivalent to a cosmological constant ) , for which @xmath103 .
unfortunately , all estimates of the energy of the quantum vacuum exceed by at least 55 orders - of - magnitude what is required to explain the acceleration of the universe , suggesting to many that `` nothing weighs nothing '' and that something else with large negative pressure is causing the accelerated expansion ( e.g. , a rolling scalar field , aka as mini inflation or quintessence , or a frustrated network of cosmic defects ; see turner , 2000 or carroll , 2001 )
. borrowing from zwicky , i have coined the term `` dark energy '' to describe this stuff , which is clearly nonluminous and more `` energy like '' than `` matter like '' ( since @xmath104 ) .
it seems to be very smoothly distributed and its primary effect is on the expansion of the universe . the first handle we will have in determining its nature is measuring its equation - of - state @xmath105 through cosmological observations ( see e.g. , huterer & turner , 2000 ) . until we figure out the nature of the dark energy , the ultimate timescale question is on hold . accepting that the universe is flat ( or at least that our bubble is ; see guth , 2001 ) ,
the possibilities for destiny are wide open : continued accelerated expansion ( and the almost complete `` red out '' of the extragalactic sky in 150 billion years ) if the dark energy is vacuum energy ; eternal slowing if the dark energy dissipates and matter takes over the dynamics ; or even recollapse if the dark energy dissipates revealing a small , negative cosmological constant ( krauss & turner , 1999 ) .
cosmology is in the midst of a golden age ( see e.g. , turner , 2001a ) .
as this meeting has illustrated , other areas of astronomy have not been left behind either .
astronomy in general is in the midst of the most exciting period of discovery ever .
the enormous variety and range of timescales in astrophysics makes the subject rich .
cosmology provides an especially good illustration , because the cosmological clock ticks logarithmically .
consistency checks on the cosmological clock have and will continue to play an important role in validating the standard cosmological model . while the consistency of the present age and expansion rate ( sandage test ) is important and will improve in accuracy , from its present 15% to perhaps 5% , there are many other age consistency tests in cosmology , whose precision may well approach a few tenths of a percent ( e.g. , bbn and cmb ) , and extend to times as early as @xmath98sec . | age consistency for the universe today has been an important cosmological test .
even more powerful consistency tests at times as early as @xmath0sec lie ahead in the precision era of cosmology .
i outline tests based upon cosmic microwave background ( cmb ) anisotropy , big - bang nucleosynthesis ( bbn ) , particle dark matter , phase transitions , and inflation .
the ultimate cosmic timescale
the fate of the universe will be in doubt until the mystery of the dark energy is unraveled . |
schwinger pair production by temporally or spatially inhomogeneous electric fields has been a theoretically and experimentally interesting and challenging issue .
the minkowski or dirac vacuum under the influence of a strong electric field becomes unstable and emits electron - positron or charged particle - antiparticle pairs .
the possibility of directly measuring electron - positron pairs by strong laser sources such as extreme light infrastructure ( eli ) has recently boosted intensive researches on pair production by pulsed electric fields @xcite ( for review and references , see also refs.@xcite ) .
in contrast to vacuum polarization , pair production has been studied for various electromagnetic configurations , for which many analytical approximation schemes have been elaborated @xcite .
the keldysh approach @xcite ( review and references , see ref .
@xcite ) , the worldline instanton method @xcite and the phase - integral or wkb method @xcite , to list a few , have been widely used as analytical approximation schemes .
the two typical methods employed to numerically compute the pair - production rate in time - dependent electric fields are the kinetic approach @xcite and the wigner formalism @xcite . the quantum vlasov equation has been often used for numerical calculation of pair production with or without back - reaction included @xcite .
the main purpose of this paper is to employ the evolution operator formalism for numerical computation of pair production by pulsed electric fields of various configurations in scalar qed .
the stratagem is to express the hamiltonian in terms of the creation and annihilation operators of the minkowski vacuum and then find the evolution operator satisfying the time - dependent schrdinger equation .
in fact , the hamiltonian for a spinless charged boson in a time - dependent electric field in the momentum space consists of infinite number of oscillators with time - dependent frequencies and each oscillator can have the evolution operator represented in @xmath0 algebra .
we then choose the number operator , the one - pair creation operator and the one - pair annihilation operator in the minkowski vacuum as the generators for the evolution operator @xmath1 in quantum optics the evolution operator for a time - dependent oscillator has been studied in a different representation of @xmath0 @xcite . the oscillator representation or the creation and
annihilation operator representation has also been used to derive the quantum vlasov equation for pair production @xcite .
the difference of this paper from other earlier works is that we find the evolution operator by directly solving the time - dependent schrdinger equation , then find the exact quantum state evolved from the minkowski vacuum , and finally compute the number of pairs during the whole evolution in pulsed electric fields .
the evolution operator represented by the creation and annihilation operators has an additional advantage of expressing the exact quantum state as the squeezed vacuum of multi - pairs of particle and antiparticle under the influence of the pulsed electric fields .
further it provides a good measure for counting the number of minkowski particle and antiparticle pairs . to study pair production from nontrivial configurations of electromagnetic field
, we shall consider only pulsed electric fields , simplified model fields , which act for a finite duration but is uniform over the space .
thus the minkowski vacuum excites into multi - particle and antiparticle state during the interaction of the electric field and finally settles down in some nontrivial vacuum state and , in an exceptional case , returns to the minkowski vacuum .
the pulsed electric fields to be studied are classified into the mono - polarity type and the di - polarity type . in the first class we consider ( i ) @xmath2 ,
( ii ) @xmath3 , ( iii ) @xmath4 , and in the second class we also consider ( iv ) @xmath5 , ( v ) @xmath6 , ( vi ) @xmath7 . the electric fields ( i)-(vi ) provides a good arena to discuss some fundamental questions , not to mention possible applications to ultra - strong lasers .
the electric fields ( i)-(iii ) of mono - polarity type necessarily lead to non - zero gauge potential values , in which the adiabatic basis differs from the minkowski one .
then a question may be raised whether pairs produced by electric pulses are the minkowski ones or the adiabatic ones @xcite . and the oscillation of the number of pairs after the completion of electric pulses may be another interesting question since the kinetic approach using the adiabatic basis predicts non - oscillating pair production while the nonadiabatic method predicts oscillating pair production for general electric fields except for the solitonic gauge field @xcite .
the evolution operator provides asymptotic solutions in the remote future for all pulsed electric fields including ( i)-(vi ) , according to which the number of pairs converges to a momentum - dependent constant when the gauge potentials vanish or oscillates around a momentum - dependent time - averaged constant when the gauge potentials have nonzero constant .
another interesting issue is the structure of the longitudinal momentum spectrum of pairs discovered by hebenstreit et al in a sinusoidal field with gaussian envelope @xcite .
dumlu and dunne explained the substructure of the spectrum in the inverse square gauge potential by the stokes phenomenon , in which more than two pairs of complex turning points for the fourier mode equation contribute either in phase or out of phase to the wkb instanton action @xcite .
the analytical approximation schemes such as the keldysh approach or the wkb or phase - integral method without stokes phenomenon can not explain the substructure of the longitudinal momentum distribution . the evolution operator formalism is efficient enough to compute the longitudinal momentum distribution of produced pairs in the pulsed electric fields ( i)-(vi ) and confirms the substructure of the spectrum for the critical field strength and compton time scale .
the spectra for ( i ) , ( ii ) , ( iv ) and ( v ) exhibit rich structures but those for ( iii ) and ( vi ) have relatively simple structure . bunching of spectrum in the field ( iii ) is observed around one positive and one negative momentum with the same magnitude .
further the polarity of the electric field ( iv ) leads to negative momentum dominance or positive momentum dominance in the spectrum .
the organization of this paper is as follows . in sec .
ii , we introduce the evolution operator in algebraic form for scalar qed in a pulsed electric field and express the vacuum polarization in terms of the complex parameters for the evolution operator . and
we express the number of pairs by the imaginary part of the parameter for the number operator . in sec .
iii , by solving the time - dependent schrdinger equation , we derive a set of first order differential equations for three complex parameters for the evolution operator .
and we find the asymptotic solutions for the pulsed electric fields whose gauge potentials approach either zero or nonzero constant after the completion of the interaction . in sec .
iv we give an intuitive interpretation of pair production by pulsed electric fields from the analogy with quantum mechanics . in sec .
v , we numerically investigate pair production by several configurations of electric fields , some of which have recently been studied in literature . in sec .
vi , we discuss the physical implications of the results and conclude the paper .
in scalar qed the fourier - decomposed hamiltonian for a spinless charged boson with mass @xmath8 in an electric field along a fixed direction @xcite @xmath9 , \nonumber\\ \omega_k^2 ( t ) & = & ( k_{\parallel } - qa_{\parallel } ( t))^2 + { \bf k}_{\perp}^2 + m^2 , \label{f - ham}\end{aligned}\ ] ] may have an oscillator representation @xmath10 where @xmath11 and @xmath12 here we have used the oscillator representation in the minkowski vacuum @xmath13 where @xmath14 and @xmath15 are the particle and antiparticle operators . and these operators constitute the @xmath0 algebra @xmath16 = \pm 2 \hat{j}^{(\pm)}_k , \quad [ \hat{j}^{(+)}_k , \hat{j}^{(-)}_k ] = \hat{n}_k.\end{aligned}\ ] ] in this paper we shall focus on pulsed electric fields that act effectively for a finite duration . before the onset of a pulsed electric field , we may choose a gauge @xmath17 so that the in - vacuum is the minkowski vacuum .
the constant electric field with @xmath18 will not be considered in this paper since no gauge can be chosen to make the initial vacuum the minkowski one , for which the in- and out - vacua are defined as asymptotic states .
the evolution of the minkowski vacuum follows the time - dependent schrdinger equation @xmath19 the evolution operator in time - ordered integral @xmath20 does not provide useful information about the quantum state unless @xmath21 since the hamiltonian is then entangled such that @xmath22 \neq 0 $ ] for @xmath23 .
the particle and antiparticle operators evolve as @xmath24 the time - dependent vacuum state is given by @xmath25 where @xmath26 denotes the minkowski vacuum .
the number of the minkowski particle carried by the time - dependent vacuum , which is equal to the number of time - dependent particle carried by the minkowski vacuum , is @xmath27 the same is true for the antiparticle production .
further , the vacuum persistence is related to the mean number @xcite @xmath28\end{aligned}\ ] ] on the other hand , the @xmath0 algebra may lead to the evolution operator of the form @xmath29 here the complex parameters @xmath30 , @xmath31 and @xmath32 will be determined from the evolution equation ( [ ev eq ] ) .
it should be mentioned that eq.([ev op ] ) differs from eq.(31 ) of ref .
@xcite for a time - dependent oscillator , which would read @xmath33 another form of the evolution operator is realized by inverting the bogoliubov transformation ( [ bog tran ] ) in ref.@xcite .
the representation ( [ ev eq ] ) is particularly useful for pair production since it expresses the out - vacuum as a squeezed vacuum of the minkowski vacuum @xmath34 where @xmath35 and @xmath36 denote the @xmath37 particles with momentum @xmath38 and @xmath37 antiparticles with momentum @xmath39 , respectively .
thus pair production conserves charge and momentum .
it further leads to the vacuum polarization @xmath40,\end{aligned}\ ] ] and the vacuum persistence @xmath41,\end{aligned}\ ] ] where @xmath42 denotes the imaginary part of @xmath31 .
therefore we can find the number of produced pairs from the imaginary part of the parameter @xmath43 during the interaction of the electric field as well as after the completion of the interaction @xmath44 master equations for evolution operator --------------------------------------- the time - dependent schrdinger equation ( [ ev eq ] ) together with @xmath0 algebra leads to the differential equations for the complex parameters^{(n)}$ ] , where @xmath45^{(0 ) } = \hat{m}$ ] and @xmath45^{(n ) } = [ \hat{n } , [ \hat{n } , \hat{m}]^{(n-1)}]$ ] . ]
@xmath46 the differential equations for the real and imaginary parts of complex parameters can be grouped into the set that determines pair production @xmath47 and into another set that are relevant for the vacuum polarization @xmath48 to get the initial data for the parameters , we employ the baker - campbell - hausdorff formula to write the evolution in a single exponential operator @xmath49,\end{aligned}\ ] ] where the dots denote polynomials of @xmath30 , @xmath31 and @xmath32 higher than third order . before the onset of the electric field @xmath50
, the evolution operator takes the form @xmath51 therefore the initial data are given by @xmath52 for pulsed electric fields such that @xmath53 and @xmath54 for large @xmath55 , we find the asymptotic solutions . in the first case of @xmath56
so that @xmath57 and @xmath58 , the asymptotic solutions are @xmath59 here three integration constants are @xmath60 and the remaining two are identified with those at @xmath61 .
this implies that the number of pairs per unit volume and per unit time approaches constant and the squeezing parameter rotates as @xmath62 . in the second case of more general @xmath63 and @xmath64 ,
the asymptotic solutions are given by @xmath65 , \label{asym sol2}\end{aligned}\ ] ] where @xmath66 , @xmath67 and @xmath68 are integration constants and @xmath69 .
further integrating eq.([asym sol2 ] ) @xmath70 we find the number of produced pairs @xmath71 the integration constants of the asymptotic solutions ( [ asym sol1 ] ) or ( [ asym sol2 ] ) should be determined by solving the evolution equations ( [ gamma - i])-([xi - i ] ) or may be found from other analytical methods .
the general feature of pair production can be understood from the analogy with quantum mechanical scattering problem .
we write the mode equation for the scalar field as @xmath72 where the kinematic momentum is @xmath73 the mode equation ( [ qm eq ] ) in the domain of time can be interpreted as a nonrelativistic particle with mass @xmath74 and energy @xmath75 under the potential @xmath76 .
we assume that the pulsed electric field has the gauge potential @xmath77 but @xmath78 .
depending on the profile of @xmath79 during the interaction of the electric field pulse , the potential becomes a barrier or well and the asymptotic value @xmath80 may be the same as @xmath81 or higher ( the green line ) or lower ( the sky blue line ) ( see fig .
1 ) . the incident wave from the future is partially reflected by the well or barrier back into the future and partially transmitted into the past @xmath82 where @xmath83 first , in the limit of weak field @xmath84 or large momentum such that @xmath85 eq.([qm eq ] ) is the scattering by a high energy particle over a shallow well or barrier .
thus the reflection coefficient @xmath86 is suppressed for all @xmath87 .
second , in the limit of strong field @xmath88 or small momentum such that @xmath89 eq.([qm eq ] ) is the scattering over a relatively deep well or high barrier .
the kinetic energy for small @xmath87 is smaller than for large @xmath87 , which means that the reflection coefficient for small @xmath87 is larger than that for large @xmath87 . ; the blue line denotes a potential , @xmath90 , where @xmath91 ; the green or sky blue line denote a potential , @xmath92 , where @xmath77 but @xmath93.,width=264,height=188 ] the above quantum mechanical interpretation may have a direct interpretation of pair production when @xmath94 so that @xmath95 .
the positive frequency solution in the right hand side of eq.([scat sol ] ) splits into a positive solution and a negative solution with the same frequency in the left hand side .
thus @xmath96 is the mean number of produced pairs by the electric field pulse .
however , when @xmath93 and @xmath97 , we may write the solution in the future by another basis with the initial frequency @xmath98 using the wronskian @xmath99 = i , \quad { \rm wr } [ v_k , v_k^ * ] = i,\end{aligned}\ ] ] we find the relation between two set of coefficients @xmath100 + \beta_k { \rm wr } [ v_k^ * , u_k^ * ] \bigr ) , \nonumber\\ \nu_k & = & i \bigl ( \alpha_k { \rm wr } [ v_k , u_k ] + \beta_k { \rm wr } [ v_k^ * , u_k ] \bigr).\end{aligned}\ ] ] thus the number of produced pairs measured with respect to the minkowski vacuum @xmath101 oscillates with the frequency @xmath102 unless @xmath95 , which confirms the the result ( [ asym num2 ] ) from the asymptotic solution . note that @xmath103 counted with respect to the minkowski vacuum is the same as @xmath104 counted with respect to the adiabatic vacuum only when @xmath95 .
( for discussion on pair production in the adiabatic basis in qed , see ref.@xcite . )
a few remarks are in order .
first , the kinetic approach @xmath105 applied to the pulsed electric field results in the asymptotic solutions ( [ scat sol ] ) with the coefficients @xmath106 where @xmath107 is an arbitrary time beyond which @xmath108 .
thus the number of pairs from the kinetic approach is the same from eq.([scat sol ] ) , that is , @xmath109 .
second , some gauge potential may involve an odd function @xmath110 such that for a fixed constant @xmath111 @xmath112 and @xmath113 for @xmath114 is the negative of @xmath115 for @xmath116 .
this implies that the scattering equation for @xmath117 is the backward scattering for @xmath118 in time and has the same reflection and transmission coefficients .
thus the number of pairs is symmetric in the longitudinal momentum around @xmath119 in the remote future .
for the purpose of numerical calculations , we use the compton unit @xmath120 so that the time is measured in compton time , the field strength in the critical strength and the energy in the rest mass energy of the particle : @xmath121 we restrict to the zero - transverse momentum , which is equivalent to replacing @xmath122 by @xmath123 in all the calculations below . in this unit system @xmath124 in eqs.([gamma - i])-([xi - i ] ) read @xmath125 further we shall consider only the strong field regime since pair production is prominent for an electric field near or above the critical strength and dynamical characteristics is manifest at the compton time scale .
the sauter electric field is a frequently used model in qed which is uniform in space but nontrivial in time .
the green function and the asymptotic number of pairs of the adiabatic particle and antiparticle have been known @xcite .
the gauge potential is chosen @xmath127 such that the initial state is the minkowski vacuum .
the duration of the field is effectively characterized by @xmath128 .
we shift the peak of electric field by @xmath129 only for the numerical purpose . and @xmath130 is plotted as a function of time and longitudinal momentum . ] [ left panel ] and the real and imaginary parts of the function @xmath131 [ right panel ] are plotted as a function of time for @xmath132 ( red ) , @xmath133 ( blue ) , @xmath134 ( green ) , and @xmath135 ( magenta ) , where the real part is in the light color and the imaginary part is in the dark color .
, title="fig : " ] is plotted as a function of momentum in the range of @xmath136 [ left panel ] and magnified in the range of @xmath137 [ right panel].,title="fig : " ] at the critical strength @xmath138 for a compton time scale pulse @xmath130 , the number of pairs is order of one and exhibits both the temporal behavior and the substructure of the longitudinal momentum spectrum as shown in fig.[tanh - f1 ] .
the pair production for small momentum in the compton unit begins to oscillate around the time of interaction as shown in detail in fig.[tanh - f2 ] , in which the gauge potential at @xmath139 is the half of the asymptotic value in the remote future .
the pair production increases soon after the electric field is turned on , reaches the maximum within a few compton times and then oscillates with large amplitude for small @xmath87 and with small amplitude for large @xmath87 .
the small amplitude oscillation for large @xmath87 is not shown in fig.[tanh - f1 ] and in the left panel of fig.[tanh - f2 ] due to a different order of magnitude . however , the asymptotic solutions ( [ asym sol2 ] ) predict oscillations for all @xmath87 since @xmath93 , which is numerically confirmed .
the time - averaged number of pairs monotonically decreases as the momentum increases .
the suppression of pair production for large momentum is expected from the scattering picture in sec .
iv , according to which the particle has a large kinetic energy compared to the potential well or barrier and thus has a small reflection coefficient , implying small pair production .
the oscillatory and temporal behavior of the squeezed vacuum ( [ sq vac ] ) can understood from @xmath30 in the right panel of fig.[tanh - f2 ] , in which it oscillates with large amplitude for small momentum while it oscillates with small amplitude for large momentum , according to the asymptotic solutions ( [ asym sol2 ] ) .
the longitudinal momentum spectrum of pairs shows a substructure , which could not be seen by the number of adiabatic pairs @xcite . since the kinematic momentum @xmath115 for @xmath140 and @xmath141 is antisymmetric in time , the longitudinal momentum spectrum of number of pairs is symmetric around @xmath142 according to sec.[scat pic ] , which is numerically confirmed in fig.[tanh - f3 ] .
pairs are minimally produced for @xmath142 , for which @xmath143 and @xmath144 .
the pair production is suppressed for large momentum as expected . and @xmath130 is plotted as a function of time and longitudinal momentum . ]
[ left panel ] and the real and imaginary parts of the function @xmath131 [ right panel ] are plotted as a function of time for @xmath132 ( red ) , @xmath133 ( blue ) , @xmath134 ( green ) , and @xmath135 ( magenta ) , where the real part is in the light color and the imaginary part is in the dark color.,title="fig : " ] is plotted in the range of @xmath136 [ left panel ] and magnified in the range of @xmath137 [ right panel].,title="fig : " ] the gauge potential given by the error - function @xmath146 leads to the gaussian electric field @xmath147 the center of the gaussian field is shifted for the numerical purpose .
the gaussian electric field decays more rapidly and thus produces relatively smaller pairs than the sauter electric field in sec.[saut ] . for the numerical work we set @xmath138 and @xmath130 as for the sauter electric field .
the number of pairs as function of time and longitudinal momentum in fig.[erf - f1 ] exhibits a similar structure as that of the sauter field in fig.[tanh - f1 ] .
it has a structure for small momentum but it is suppressed for large momentum .
the similarity of the temporal behavior can be seen by comparing the left panel of fig.[erf - f2 ] with fig.[tanh - f2 ] .
the longitudinal momentum spectrum of pairs in fig.[erf - f3 ] also has a similar pattern as fig.[tanh - f3 ] .
as the error - function is an odd function @xmath148 , it is symmetric around @xmath149 according to sec.[scat pic ] , which is numerically confirmed in fig.[erf - f3 ] .
note that @xmath149 is the channel for minimal pair production for small momentum .
the suppression of pair production for large momentum is also shown in fig.[erf - f3 ] . ) is plotted in red color for @xmath138 , @xmath130 and @xmath151 [ left panel ] and @xmath152 [ right panel ] .
the curve in blue color in each panel is @xmath153 for the same value of @xmath154.,title="fig : " ] ) is plotted in red color for @xmath138 , @xmath130 and @xmath151 [ left panel ] and @xmath152 [ right panel ] .
the curve in blue color in each panel is @xmath153 for the same value of @xmath154.,title="fig : " ] and @xmath130 is plotted as a function of time and longitudinal momentum . ] for @xmath151 [ left panel ] and @xmath152 [ right panel].,title="fig : " ] finally , we consider a gauge potential @xmath155 which leads to the oscillating gaussian electric field @xmath156 as shown in fig.[gcos - f1 ] the first term is dominant in the region @xmath157 and oscillates with an gaussian envelope @xmath158 and beyond that region the second term has a linearly growing factor but both terms are exponentially suppressed . for the numerical work we set @xmath138 , @xmath159 and @xmath151 or @xmath152 .
it is surprising that the number of pairs in fig.[gcos - f2 ] is bunched around @xmath160 and @xmath134 . the longitudinal momentum spectrum in fig.[gcos - f3 ]
shows bunching with momentum separation of @xmath161 and @xmath162 for @xmath151 and @xmath152 , respectively . and the rapidly oscillating factor suppresses pair production , though the peak intensity in fig.[gcos - f1 ] is almost comparable to the sauter electric field in sec.[saut ] and the gaussian electric field in sec.[gaus ] .
the numerical study also shows a similar pattern in pair production and bunching of pairs in the gauge potential @xmath163
in this section we study how the polarity of the electric field influences pair production , in particular , when the electric field changes the polarity during the interaction .
it is interesting to understand how the produced pairs behave when another electric pulse of opposite polarity acts .
do they annihilate each other partially or completely ? or are there more pairs produced by the second pulse ? in order to answer some of these questions we consider three model fields . as the first model we consider the di - polarity sauter electric field , two sauter electric fields acting with a time lag and in opposite directions .
the model gauge field is given by @xmath165 though we can not exactly solve qed problem in this gauge potential , we may understand the characteristic feature of pair production by each of sauter electric fields when they are separated by a sufficient time gap as in quantum mechanics . and @xmath130 is plotted as a function of time and longitudinal momentum . ] in the di - polarity sauter electric field is plotted as a function of time for @xmath133 ( red ) , @xmath166 ( blue ) , @xmath134 ( green ) , @xmath142 ( magenta ) , @xmath167 ( gold ) , and @xmath160 ( skyblue ) [ left panel ] and the longitudinal momentum spectrum is plotted at @xmath168 [ right panel].,title="fig : " ] in @xmath169 is plotted as a function of time for @xmath133 ( red ) , @xmath166 ( blue ) , @xmath134 ( green ) , @xmath142 ( magenta ) , @xmath167 ( gold ) , and @xmath160 ( skyblue ) [ left panel ] and the longitudinal momentum spectrum is plotted at @xmath168 [ right panel].,title="fig : " ] for the numerical work we set @xmath138 and @xmath130 so that two sauter electric pulses are effectively separated . at the onset of the interaction
the peak of number of pairs in fig.[disaut - f1 ] is almost the same as the single sauter electric field in fig.[tanh - f1 ] , though it exhibits more structure .
however , the main difference appears after the completion of the interaction .
the di - polarity sauter electric field returns to the minkowski vacuum while the sauter electric field has a constant residual gauge and @xmath170 .
the asymptotic solutions ( [ asym sol1 ] ) predict that the number of pairs for the di - polarity sauter field approaches to a constant , as shown in figs.[disaut - f1 ] and [ disaut - f2 ] , while eq.([asym num2 ] ) predicts that the number of pairs for the sauter field oscillates with a constant amplitude around the time - averaged value , as shown in figs.[tanh - f1 ] and [ tanh - f2 ] .
the gauge potential in between two peaks of the di - polarity sauter field has approximately the shape of square potential barrier , in which pair production oscillates according to the asymptotic solution ( [ asym num2 ] ) .
no pair production for @xmath132 is the numerical coincidence that @xmath171 in the compton unit is equivalent to five wavelengths of a particle in ten compton length width of the square barrier in the scattering picture [ scat pic ] and has the zero reflection coefficient due to resonance .
the effect of the polarity of the electric field can be seen in the right panel of fig.[disaut - f2 ] and fig.[disaut - f3 ] , in which the longitudinal momentum spectrum shows the mirror symmetry when the polarity of the electric field changes .
there are residual pairs with negative momentum ( fig.[disaut - f2 ] ) when the first sauter field acts in the positive direction and then the second one in the negative direction , while more pairs with positive momentum survive ( fig.[disaut - f3 ] ) when the first sauter field acts in the negative direction and then the second one in the positive direction .
the number of pairs in time also shows the polarity effect as shown in the left panel of fig.[disaut - f2 ] and fig.[disaut - f3 ] .
[ left panel ] and the real and imaginary part of the function @xmath131 [ right panel ] are plotted as a function of time for @xmath132 ( red ) , @xmath133 ( blue ) , @xmath173 ( green ) , and @xmath134 ( magenta).,title="fig : " ] is plotted as a function of parallel momentum in the range of @xmath136 [ left panel ] and the detailed plot @xmath174 [ right panel].,title="fig : " ] the inverse square potential shares the same property of vanishing in the remote past and future and changing polarity as the di - polarity sauter electric field in sec.[di - saut ] .
the structure of the longitudinal momentum spectrum of pairs in the inverse square potential was discovered and explained as a consequence of stokes phenomenon by dumlu and dunne @xcite . for the numerical work we set @xmath138 and @xmath130 .
the peak of the number of pairs in fig.[inv - f1 ] is relatively small compared with the sauter or gaussian electric field almost with the same peak intensity .
also it is small even compared with the di - polarity sauter electric field .
after the completion of interaction the number of pairs oscillates with small amplitude as shown in fig.[inv - f2 ] .
the structure of the longitudinal momentum spectrum is shown in fig.[inv - f3 ] .
the relatively simple looking spectrum has also a fine substructure as shown in the right panel of fig.[inv - f3 ] .
the result of this section can not be directly compared with that in ref .
@xcite , in which the structure was found for a subcritical strength and longer time scale than ours in the compton scale .
the solitonic gauge potential has a special energy condition @xmath176 where @xmath37 is a natural number and @xmath177 is the minkowski energy of particle in the remote past .
the one - soliton @xmath178 and multi - soliton @xmath179 gauge fields correspond to soliton solutions for the inverse scattering of the korteweg - de vries ( kdv ) equation @xcite .
the solitonic gauge potential is interesting in understanding the peculiarity of schwinger mechanism by pulsed electric fields . among the gauge potentials that vanish in the remote past and future ,
the solitonic gauge potential is unique in that the number of pairs produced by the corresponding electric field exponentially increases from and then decreases to zero @xcite .
in fact , any gauge potential which has the reflectionless scattering for the field equation ( [ qm eq ] ) has zero number of pairs in the remote future and may belong to the solitonic gauge potential .
is plotted for for the solitonic gauge potential with @xmath180 [ left panel ] and for non - solitonic gauge potential with @xmath181 [ right panel].,title="fig : " ] for the numerical purpose , we consider the electric field with the center shifted @xmath182 and set @xmath159 but select a natural number and non - natural number for the strength @xmath183 .
the left panel of fig.[soliton - f ] is the number of pairs as a function of time for the solitonic gauge potentials @xmath184 and @xmath161 .
the number of pairs increases and then decreases in symmetric way around the center @xmath139 of the gauge potential and the number of local maxima is the same as the soliton number @xcite . on the other hand , for a non - natural number , for instance , @xmath185 and @xmath186 , the number of produced pairs increases from zero but decreases to a constant value .
this temporal behavior of the number of pairs is expected from the asymptotic solutions ( [ asym sol1 ] ) for @xmath187 regardless of @xmath183 .
it should be recollected that the inverse square potential has also finite number of pairs in the remote future in sec.[inv pot ] , in contrast to the solitonic gauge field .
we have employed the evolution operator formalism to numerically study pair production in scalar qed by pulsed electric fields .
the pulsed electric field is nontrivial in that it acts for a finite period of time and has the inhomogeneity of time .
the hamiltonian for a spinless charged boson in time - dependent electric field is an infinite sum of oscillators with time - dependent frequencies . after expressing the oscillator hamiltonian in terms of the creation and annihilation operators of particle and antiparticle in the minkowski vacuum ,
the evolution operator ( [ ev op ] ) in @xmath0 algebra is factorized into the pair annihilation part , the number part and the pair creation part .
the advantage of this factorization is that the exact quantum state ( [ sq vac ] ) from the minkowski vacuum under the influence of a pulsed electric field is the squeezed vacuum of particle and antiparticle with a complex phase factor .
the sum of all complex time - dependent phase factors for the number operator determines the vacuum polarization and the vacuum persistence , which in turn is related to the number of pairs produced by the pulsed electric field .
now the time - dependent schrdinger equation is equivalent to a set of first order differential equations for the complex parameters for the evolution operator .
the evolution of the minkowski vacuum is governed by three parameters in eqs.([gamma - i])-([xi - i ] ) , and thus the set of first order differential equations can implement numerical works for pair production in various configurations of the electric fields . remarkably the differential equations have the asymptotic solutions ( [ asym sol1 ] ) when the gauge potential vanishes and another solutions ( [ asym sol2 ] ) when the gauge potential approaches a constant value .
these asymptotic solutions put a strong constraint on the behavior of number of pairs such that the number of pairs per unit volume and per unit time is a constant for the zero - gauge potential in the remote future but it oscillates around the time - averaged value for a non - zero gauge potential .
for numerical works we have selected two classes of electric fields or gauge potentials . in the first class
the electric field does not change the polarity of the field .
therefore , the gauge potential , as the negative of the time integral of electric field , should have a non - zero value in the remote future .
we have considered ( i ) sauter electric field , ( ii ) gaussian electric field and ( iii ) oscillating gaussian gauge potential as the first class of mono - polarity electric field . in the second class
the electric field changes the polarity and has the gauge potential which vanishes in the remote future .
we have considered in the second class ( iv ) di - polarity sauter electric field , two sauter fields separated by a time gap , ( v ) inverse square gauge potential and ( vi ) solitonic gauge potential , all of which have the di - polarity of field .
we have computed the number of pairs as a function of time and longitudinal momentum for the pulsed electric fields of ( i)-(vi ) with the compton scale . in the first class ( i)-(iii ) , the number of pairs increases when the field acts and then oscillates around the time - averaged value as predicted by the asymptotic solutions . in general the number of pairs is suppressed for large longitudinal momentum as expected from the field equation in sec.[scat pic ] .
the structure of longitudinal momentum spectrum is found for small momentum for ( i)-(iii ) .
hebenstreit et al found the substructure of the spectrum in a sinusoidal electric field with gaussian envelope @xcite while the sinusoidal gauge potential is considered in this paper .
also , the structure of the momentum spectrum has been found for the class ( v ) by dumlu and dunne @xcite .
the number of pairs as a function time or a function of momentum shows similarity between the sauter electric field and the gaussian electric field .
however , the momentum spectrum for the oscillating gaussian electric field reorganizes and bunches around one positive and one negative momenta with the same magnitude and the separation in the momentum space equals to the angular frequency of the oscillating electric field .
we do not have a simple physical explanation for this bunching effect due to the oscillating field with a gaussian envelope . in the second class of di - polarity field ( iv)-(vi )
in which the gauge potential vanishes in the remote future , the number of pairs increases and then decreases to a constant rate and has relatively simpler structure of the longitudinal momentum than mono - polarity electric fields .
still the longitudinal momentum spectrum for the di - polarity sauter electric field exhibits a surprising feature of negative or positive momentum dominance of produced pairs depending on whether the positively directed pulse or the negatively directed pulse acts first , which is then followed by the oppositely directed second pulse .
further the solitonic gauge potential produces pairs in a symmetric way in time and returns back to the minkowski vacuum without any residual pairs . in this paper
we have confined our study to scalar qed in pulsed electric fields .
however , there are a few issues related to but not treated in this paper .
first , the evolution operator formalism can also apply to spinor qed , since the spin-1/2 fermions have @xmath188 algebra isomorphic to @xmath0 and may have an evolution operator similar to the spinless boson .
another issue is the vacuum polarization , the real part of the in- and the out - vacua scattering amplitude . to get a finite effective action one should properly regularize the sum over the momentum of the real part of the scattering amplitude .
still another issue is the back reaction of produced pairs .
the strong field investigated in this paper produces roughly one pair per unit compton volume , which could generate an induced electric field screening the external field and cause another mechanism for the oscillation of produced pairs @xcite .
finally , it would be extremely useful and promising for eli experimentation to have a similar evolution operator formalism for both spatially and temporally localized electric fields .
all these issues go beyond the scope of this paper and will be addressed in future publication .
k. and h. w. l. would like to thank professor remo ruffini for the warm hospitality at icranet where this paper was completed .
the work of s. p.
k. was supported in part by basic science research program through the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology ( 22012r1a1b3002852 ) .
the work of h. w. l was supported in part by the international research and development program of the national research foundation of korea ( nrf ) funded by the ministry of education , science and technology ( mest ) of korea(k21003002081 - 12b1200 - 00310 ) .
the work of r. r. was supported in part by icranet .
the visit to icranet by s. p. k. and h. w. l. was supported in part by icra .
s. s. bulanov , v. d. mur , n. b. narozhny , j. nees , and v. s. popov , `` multiple colliding electromagnetic pulses : a way to lower the threshold of e+e- pair production from vacuum , '' phys .
lett . * 104 * , 220404 ( 2010 ) [ arxiv:1003.2623 ] .
d. b. blaschke , a. v. prozorkevich , g. ropke , c. d. roberts , s. m. schmidt , d. s. shkirmanov , and s. a. smolyansky , `` dynamical schwinger effect and high - intensity lasers .
realising nonperturbative qed , '' eur . phys .
j. d * 55 * , 341 ( 2009 ) [ arxiv:0811.3570 ] .
f. hebenstreit , r. alkofer , g. v. dunne , and h. gies , `` momentum signatures for schwinger pair production in short laser pulses with sub - cycle structure , '' phys . rev
. lett .
* 102 * , 150404 ( 2009 ) [ arxiv:0901.2631 ] .
f. hebenstreit , r. alkofer , and h. gies , `` schwinger pair production in space and time - dependent electric fields : relating the wigner formalism to quantum kinetic theory , '' phys . rev .
d * 82 * , 105026 ( 2010 ) [ arxiv:1007.1099 ] .
s. a. smolyansky , g. ropke , s. m. schmidt , d. blaschke , v. d. toneev , and a. v. prozorkevich , `` dynamical derivation of a quantum kinetic equation for particle production in the schwinger mechanism , '' [ hep - ph/9712377 ] .
s. m. schmidt , d. blaschke , g. ropke , s. a. smolyansky , a. v. prozorkevich , and v. d. toneev , `` a quantum kinetic equation for particle production in the schwinger mechanism , '' int .
j. mod .
e * 7 * , 709 ( 1998 ) [ hep - ph/9809227 ] .
v. n. pervushin , v. v. skokov , a. v. reichel , s. a. smolyansky , and a. v. prozorkevich , `` the kinetic description of vacuum particle creation in the oscillator representation , '' int .
j. mod .
phys.a * 20 * , 5689 ( 2005 ) [ hep - th/0307200 ] .
a. m. fedotov , e. g. gelfer , k. y. korolev , and s. a. smolyansky , `` on the kinetic equation approach to pair production by time - dependent electric field , '' phys .
d * 83 * , 025011 ( 2011 ) [ arxiv:1008.2098 ] . | we numerically investigate the temporal behavior and the structure of longitudinal momentum spectrum and the field polarity effect on pair production in pulsed electric fields in scalar quantum electrodynamics ( qed ) . using the evolution operator expressed in terms of the particle and antiparticle operators , we find the exact quantum states under the influence of electric pulses and measure the number of pairs of the minkowski particle and antiparticle .
the number of pairs , depending on the configuration of electric pulses , exhibits rich structures in the longitudinal momentum spectrum and undergoes diverse dynamical behaviors at the onset of the interaction but always either converges to a momentum - dependent constant or oscillates around a momentum - dependent time average after the completion of fields . |
the detailed phase - space distribution of cold dark matter haloes can substantially affect prospects for dark matter detection .
direct detection experiments are starting to probe significant fractions of the parameter space of plausible theoretical models , so a first detection of dark matter ( dm ) may be imminent .
only recently has realistic prediction of the phase - space structure near the earth become possible , because very high - resolution numerical simulations are required .
it is now well established that the outer regions of cold dark matter ( cdm ) haloes have a complicated phase - space structure with many subhaloes and tidal streams @xcite .
this raises the question of whether similar structures might affect direct detection probabilities .
could the earth be sitting in a `` hole '' , i.e. a locally very underdense region , in the dm distribution , as might occur in a fractal structure , or if most of the mass near the sun were concentrated in dense , low - mass subhaloes .
some simulators have indeed argued that a significant fraction of the local mass could lie in solar or earth - mass subhaloes ( e.g. * ? ? ?
* ) , although more recent simulations suggest that the local mass fraction in bound subhaloes of any mass is well below 1% @xcite . other possibilities might be for the earth to lie within a subhalo , or within a dense tidal stream created by disruption of an earlier subhalo .
either of these would produce a spike in the velocity distribution of local dark matter particles . quantifying the detailed phase - space structure of a halo requires disentangling its various dm components : the smooth component which is , in fact , a superposition of many fundamental streams @xcite ; compact , self - bound subhaloes ; and tidal streams created by the disruption of such subhaloes .
efficient identification of the tidal streams requires a sensitive and robust structure - finder in 6d phase - space .
recently @xcite presented an algorithm , hierarchical structure finder ( hsf ) , designed specifically for this purpose .
hsf identifies structures in an n - body simulation as coherent , overdense sets of particles in the full 6d position - velocity distribution . in this paper
we use hsf to study the six galaxy - scale halos simulated as part of the aquarius project @xcite , providing a quantitative analysis of the various halo components and focussing , in particular , on the inner halo relevant for detection experiments .
our paper is organised as follows : section [ sec : hsf ] presents a brief description of our hierarchical structure finder .
section [ sec : str ] then explores the numerical convergence of our results by analysing simulations of a single halo at a variety of resolutions .
finally we use the full set of six aquarius halos to study the expected scatter in substructure properties among galaxy - scale halos .
section [ sec : inn ] focusses on substructure in the inner halo in order to assess possible consequences for direct detection experiments .
the final section gives our conclusions .
many different algorithms for identifying ( sub)structure have been applied to n - body simulations of the growth of cosmic structure .
one of the first and most widely used is the friends - of - friends ( fof ) scheme introduced by @xcite ; this defines `` halos '' as disjoint particle sets containing every particle closer than some maximal linking length to at least one other member of its set .
this bounds objects approximately by an isodensity surface , but makes no assumption about their shape or internal structure .
in contrast , the spherical overdensity ( so ) group - finder of @xcite finds high - density peaks in the particle distribution , grows spheres centred on each until the mean enclosed density drops to a specified value ( typically @xmath11 times the cosmological mean ) and then defines halos as the contents of those spheres whose centres do not lie within a more massive halo .
more recent structure - finders [ e.g. skid @xcite , subfind @xcite , voboz @xcite , psb @xcite , adaptahop @xcite , ahf @xcite , hsf @xcite ] typically identify objects as connected self - bound particle sets above some density threshold .
such methods have two steps .
each particle is first linked to a local dm density maximum by following the gradient of a particle - based estimate of the underlying dm density field .
the particle set attached to a given maximum defines a candidate structure . in a second step , particles which are gravitationally unbound to the structure
are discarded until a fully self - bound final object is obtained .
the various methods differ in the way particles are treated when they belong to more than one candidate and in the way unbound particles are redistributed .
most methods produce a hierarchical characterisation of structure where halos contain subhaloes which in turn can contain their own subhaloes .
these methods can be extended to higher dimensions , in particular to 6d phase - space .
the main complication is then that the smoothed density field and its gradient must be estimated from the particle distribution in six dimensions .
the hierarchichal structure finder ( hsf ) presented by @xcite is an algorithm of this type , and can be used , just like the above 3d algorithms , to identify bound subhaloes .
other six dimensional phase - space structure finders have been developed recently by @xcite and ( * ? ? ?
* enlink ) . in the following
we describe the hsf method in somewhat more detail , since this is the method we will use for the rest of this paper . further technical details and tests can be found in @xcite . to find candidate structures we first need to estimate phase - space densities at the positions of all the particles .
furthermore we need to calculate local phase - space density gradients .
hsf does this using a six - dimensional sph smoothing kernel with a local adaptive metric as implemented in the enbid code @xcite .
neighbouring particles can then be used to derive the required gradients . for the sph kernel we use @xmath12 neighbours whereas for the gradient estimate we use @xmath13 neighbours . once phase - space densities have been calculated
, we sort the particles according to their density in descending order . then we start to grow structures from high to low phase - space densities .
while walking down in density we mark for each particle the two closest ( according to the local phase - space metric ) neighbours with higher phase - space density , if such particles exist . in this way
we grow disjoint structures until we encounter a saddle point , which can be identified by observing the two marked particles and seeing if they belong to different structures .
a saddle point occurs at the border of two structures . in the standard setup of hsf , which is used throughout this paper , the masses of the two structures separated by the saddle point
are compared and the smaller one is cut , defining a complete individual structure . all particles below the saddle point whose higher density neighbours are part of the cut object are attached to the other , larger structure .
pursuing this procedure until all particles have been considered divides a halo into a unique disjoint set of substructures , of which the most massive , which also contains the lowest phase - density particles , is the main substructure .
when we wish to isolate self - bound subhaloes , we follow an identical procedure , except that each time we reach a saddle point , we remove all unbound particles iteratively from the smaller structure and attach them provisionally to the larger structure .
once we have followed this algorithm down to the lowest phase - density particle , we are left with a set of self - bound subhaloes and a few particles which are bound to no subhalo , not even the most massive self - bound subhalo which again is the one containing the lowest phase - density bound particles .
both these procedures divide a halo into a disjoint set of phase - space structures , each containing a single phase - space density peak and bounded approximately by a level surface of phase - space density . in the first procedure each structure normally contains both bound and unbound particles , and all halo particles
are assigned to some structure . in the second procedure ,
each structure is self - bound , and some halo particles are not assigned to any structure . to be specific ,
in the following we will refer to all particles inside @xmath0 times the critical value .
halo_. we call the most massive substructure constructed from these particles the _ main halo_. note that by definition it can not extend beyond @xmath0 and that its mean density within @xmath0 will be less than 50 times the critical density . note also that the main halo will change slightly according to whether we do or do not apply the unbinding and reassignment procedures . in the former case
we refer to all other structures as ( self - bound ) _ subhaloes _ , whereas in the latter case we refer to them as _
substructures_. a subhalo is thus always part of a substructure , but a substructure does not necessarily contain a subhalo .
we study the phase - space structure of milky way - sized dm haloes using the high - resolution simulations of the aquarius project @xcite .
the cosmological parameters for these simulations are @xmath14 and @xmath15 @xmath16 .
for this project six galaxy - mass haloes ( aq - a to aq - f ) were selected from a lower resolution version of the millennium - ii simulation @xcite and resimulated with progressively higher particle number and smaller softening length .
the haloes were selected to have no close massive companion at @xmath17 .
when studying differences in phase - space structure between these haloes , we use the second resolution level ( the highest for which results are available for all six objects ) . at this resolution
all haloes have more than @xmath18 particles inside @xmath0 , corresponding to a particle mass @xmath19 . in addition , we use resimulations of the aq - a halo at four different resolution levels to check the numerical convergence of our results . in the final section of this paper
we investigate phase - space structure in the inner halo , defined as @xmath20 kpc . for this purpose
we use three resolution levels of the aq - a halo with the largest one ( aq - a-1 ) having almost @xmath21 particles inside @xmath0 and more than @xmath22 particles inside @xmath23 .
together with the ability of hsf to analyse the full 6d particle distribution , this simulation set allows the first robust and fully general quantification of the various phase - space components predicted by the @xmath24cdm model at @xmath25 kpc where direct detection takes place .
solid lines indicate mass functions for the substructures identified when hsf is executed without unbinding and reassignment procedures ( see the text ) , while dashed lines indicate the corresponding functions for the self - bound subhaloes found when these procedures are implemented .
different colours refer to the different resolution levels as indicated in the plot .
the straight black line is a power - law fit , @xmath26 , to the data for self - bound subhaloes . the data for substructures are instead fit by a power - law of slope @xmath3 ( the horizontal red line ) .
bottom panel : corresponding cumulative mass functions inside @xmath0 , expressed as the total fraction of the enclosed mass in identified substructure .
line styles are the same as in the top panel .
black and red lines are based on the fits in the upper panel with a high mass truncation at @xmath27 of the total mass.,title="fig:",width=321,height=321 ] kpc .
solid lines indicate mass functions for the substructures identified when hsf is executed without unbinding and reassignment procedures ( see the text ) , while dashed lines indicate the corresponding functions for the self - bound subhaloes found when these procedures are implemented .
different colours refer to the different resolution levels as indicated in the plot .
the straight black line is a power - law fit , @xmath26 , to the data for self - bound subhaloes . the data for substructures are instead fit by a power - law of slope @xmath3 ( the horizontal red line ) .
bottom panel : corresponding cumulative mass functions inside @xmath0 , expressed as the total fraction of the enclosed mass in identified substructure .
line styles are the same as in the top panel .
black and red lines are based on the fits in the upper panel with a high mass truncation at @xmath27 of the total mass.,title="fig:",width=321,height=321 ] we begin by analysing the mass functions of substructures and of self - bound subhaloes in the aq - a halo and their dependence on resolution . to be consistent with earlier work
we define the edge of the halo at @xmath28 kpc and we count all objects within this radius , but we note that this is a large radius and , as a result , the counts are dominated by objects beyond 100 kpc , more than an order of magnitude further from the galactic centre than the sun . in the upper panel of fig .
[ mass_res ] we compare the differential mass functions of substructures ( solid curves ) and of self - bound subhaloes ( dashed curves ) at four different resolutions .
the lower panel shows the corresponding cumulative mass functions . in both cases
the mass functions agree quite well between the simulations above their respective resolution limits .
for the self - bound subhaloes , the slope of the differential mass function is close to @xmath29 , as found earlier in the subfind analysis of @xcite . at lower resolution
the distribution is better approximated , particularly in the low mass bins , by a slope close to @xmath30 . for the aq - a-2 halo
, we find that @xmath2 of the halo mass is in self - bound subhaloes , which is 20% higher than the corresponding subfind value ( @xmath31 - * ? ? ?
* ) , reflecting the fact that hsf typically attaches more mass to each identified object than subfind . in the lower resolution simulations , particles from unresolved low - mass substructures are in many cases attached to more massive objects
this explains the shifts in the mass functions at high masses that are well resolved by all simulations ( see the bottom panel of fig.[mass_res ] ) .
the mass fraction in self - bound subhaloes changes from @xmath32 for aq - a-5 to @xmath2 for aq - a-2 .
the power - law behaviour of the mass function of self - bound subhaloes has been known for some time , but there has been controversy over its slope .
if this slope is @xmath3 , then the mass fraction in subhaloes diverges logarithmically at low mass , and is cut off at a mass corresponding to the free - streaming length of the underlying dm particle ( typically earth mass for neutralino candidates ) .
if the slope is -1.9 , however , as found above , then the mass fraction in subhaloes has already effectively converged at the limit of the highest resolution aquarius simulations and hence is @xmath33 within @xmath0 . within smaller
radii this fraction drops dramatically , as we will see below .
hsf makes it possible to find unbound substructures also , and the solid lines in fig .
[ mass_res ] show that within @xmath0 rough numerical convergence is achieved for their mass function . in this case
, however , the slope appears close to @xmath3 and the mass in substructures exceeds that in self - bound subhaloes by more than a factor of 2 at all masses , and by increasingly large amounts at small mass . to the resolution limit of aq - a-2 ,
35% of the mass within @xmath0 is in substructure , showing the total mass detected in unbound tidal streams to be significantly larger than in self - bound subhaloes . with increasing resolution , significantly more
substructures are found , and mass shifts from massive to smaller substructures , as found above for self - bound subhaloes but even more strongly .
the behaviour seen in the lower panel of fig .
[ mass_res ] can not be extrapolated straightforwardly to lower mass .
as we will see below , at the resolution of aq - a-2 , most of the mass in the outer halo is resolved into substructures , but relatively little of the mass in the inner regions . at higher resolution
it will be the transition between these two regimes which controls the total mass fraction in substructure , rather than the increase in resolved substructures at any particular radius . for the 4 lower resolution haloes . for aq - a-1
only the mass within the inner @xmath7 kpc was used .
solid lines represent the substructures and dashed lines the self - bound subhaloes .
the black line is an analytic fit based on data for the full set of 6 aquarius haloes ( see below ) @xmath34 $ ] with parameters @xmath35 and @xmath36 .
the red line is the same function shifted vertically to @xmath37.,width=321,height=321 ] for this same resolution series of aq - a simulations , fig .
[ radius_res ] shows the mass fractions in substructures and in self - bound subhaloes for a set of of 10 disjoint spherical shells extending from 1 kpc to @xmath0 . within 35 kpc
we also show results for the highest resolution simulation aq - a-1 .
although the results for a single simulation are rather noisy , they can be represented reasonably well by @xmath38 , \label{eq : fsub}\ ] ] with parameters @xmath35 and @xmath36 for the subhaloes .
this analytic form was used to fit the radial distribution of subfind subhaloes in @xcite .
we note that the quoted parameters were obtained from a fit to data for the full set of six level 2 aquarius haloes ( see below ) .
the distribution of hsf subhaloes is very similar to that of subfind subhaloes , but hsf attaches slightly more particles to objects near the centre .
we find substructures down to 0.6 kpc in aq - a-1 and down to @xmath39 kpc in aq - a-2 and aq - a-3 simulations .
for aq - a-4 , however , no substructures are found within @xmath40 kpc , demonstrating that at least @xmath41 particles are needed within @xmath0 to begin to study streams around the sun s position .
the red curve shows the prediction of eq .
( [ eq : fsub ] ) for parameters @xmath35 and @xmath37 but , in contrast to the situation with subhaloes , it is clear that the results in the inner regions are not converging with increasing resolution .
the results for aq - a-1 lie well above the red line and should clearly still be considered as a lower limit to the mass fraction contained in tidal streams in these regions . [
cols="^,^,^,^,^,^ " , ] which is a good fit to the differential mass function of the self - bound subhaloes .
the black line shows a @xmath3 power - law which better describes the differential mass function of substructures .
bottom panel : the corresponding cumulative mass functions .
line styles are the same as in the top panel.,title="fig:",width=321,height=321 ] which is a good fit to the differential mass function of the self - bound subhaloes .
the black line shows a @xmath3 power - law which better describes the differential mass function of substructures .
bottom panel : the corresponding cumulative mass functions .
line styles are the same as in the top panel.,title="fig:",width=321,height=321 ] $ ] with parameters @xmath35 and @xmath36 for subhaloes and the same function shifted vertically to @xmath37 for substructures .
the black dashed line shows the best fit for substructures with parameters @xmath42 and @xmath43.,width=321,height=321 ] in this section we quantify the object - to - object scatter in the substructure mass function by analysing all six level 2 haloes of the aquarius project .
two of our haloes aq - a and aq - c did not experience major mergers below redshift @xmath44 .
haloes aq - b and aq - f on the other hand each underwent a major merger below redshift @xmath45 .
more information on the merger history of the aquarius haloes and how representative these haloes are of the population of milky way - like haloes can be found in @xcite and in @xcite .
the top panel of fig .
[ mass_profile ] shows the differential substructure mass function for the six aquarius haloes aq - a-2 to aq - f-2 ; the bottom panel presents corresponding cumulative mass functions .
table [ table2 ] lists values for the total mass fraction in substructure in each halo using different substructure identification methods .
subfind subhaloes @xcite and subhaloes found by hsf follow the same power - law with a slope close to @xmath29 .
halo aq - c-2 evolves in a quiet merger environment and this explains its deficit in substructures . for the general
substructures the slope of the power - law is close to @xmath3 , but it is difficult to measure this value accurately because the low - mass end is contaminated by substructure arising through discreteness noise , particles which are connected by hsf but do not represent physical substructures . in aq - c-2
only @xmath46 of the mass within @xmath0 belongs to self - bound subhaloes , while for aq - e-2 this number is @xmath47 and for the other haloes it is @xmath48
. the subhalo mass of aq - f-2 is dominated by the largest subhaloes .
it is interesting to observe that although aq - c-2 has the fewest self - bound subhaloes , its mass fraction in substructures is about @xmath6 , close to the value for aq - b-2 ( @xmath49 ) which had completely different and much richer merger history .
all the other haloes also have substructure mass fractions around @xmath49 .
the slope of the differential substructure mass function is similar in all haloes except aq - f-2 , where the very recent merger apparently causes a bias towards massive substructures .
[ mass_radius ] shows the fractions of mass in substructures ( solid lines ) and in self - bound subhaloes ( dashed lines ) as a function of distance from halo centre . for most of the haloes
the latter follows eq .
( [ eq : fsub ] ) , indicated as a black solid line .
clearly , hsf identifies substructures close to the centre in all halos .
this is possible because of the high density contrast in 6d phase - space compared to 3d configuration space .
( [ eq : fsub ] ) with a different normalisation ( i.e. a vertical shift , see the solid black line ) also gives a rough fit to the radial dependence of the substructure mass fraction in most haloes , but we note that an independent fit of the same functional form ( the black dashed line ) suggests that the mass fraction in substructures increases relative to that in self - bound subhaloes in the inner regions of the haloes . in the outskirts of the haloes , near @xmath0 ,
up to @xmath6 of the mass is in the form of self - bound subhaloes and up to @xmath5 resides in substructures .
for many years experimenters have been trying to detect dm in laboratory devices .
detector signals are very sensitive to the local dm phase - space distribution , so it is important to study halo structure in detail at the solar position .
this requires high - resolution simulations like those of the aquarius project .
a first phase - space analysis using the aquarius haloes was carried out by @xcite with a focus on the local velocity and spatial distributions and their imprints on direct detection signals . here
we extend this study and focus on phase - space structures at @xmath25 kpc .
we go beyond the self - bound subhalo analysis of @xcite by using hsf , which efficiently identifies gravitationally unbound structures like tidal streams .
such features can have a significant impact on dm experiments , and our goal here is to quantify the total amount of structure in the inner halo , both bound and unbound .
we therefore concentrate on the region within @xmath50 kpc and use the excellent resolution of aq - a-1 to analyse structures near the solar circle .
-@xmath51 plots of aq - a-2 ( top ) and aq - a-1 ( bottom ) .
these were constructed by first calculating the phase - space density at each particle using enbid @xcite .
we then create a pixelised image with @xmath52 bins and for each pixel we calculate the logarithm of the maximum phase - space density over all particles in that pixel .
the phase - space density is measured in units of @xmath53.,title="fig:",width=330,height=396 ] -@xmath51 plots of aq - a-2 ( top ) and aq - a-1 ( bottom ) .
these were constructed by first calculating the phase - space density at each particle using enbid @xcite .
we then create a pixelised image with @xmath52 bins and for each pixel we calculate the logarithm of the maximum phase - space density over all particles in that pixel .
the phase - space density is measured in units of @xmath53.,title="fig:",width=330,height=396 ] to give a first impression of phase - space structure in the inner halo we use the technique of @xcite .
we estimate the phase - space density at the position of each particle with enbid @xcite and plot an @xmath54-@xmath51 phase - space portrait in which each pixel is colour - coded according to the maximum phase - space density of the particles it contains .
the top panel of fig .
[ rvr ] shows the resulting phase - space plot for the inner part of the aq - a-2 halo , while the bottom panel shows the corresponding plot for aq - a-1 .
aq - a-2 has about @xmath55 and aq - a-1 about @xmath22 particles inside @xmath23 .
the increased resolution results in substantially more self - bound structures and tidal streams being visible in the inner regions of aq - a-1 . in the following we quantify these phase - space structures in some detail .
kpc ) at various resolutions .
solid lines show results for substructures and dashed lines for self - bound subhaloes.,width=321,height=321 ] kpc ) at different resolutions .
solid lines represent substructures and dashed lines self - bound subhaloes .
all curves are normalised to @xmath56 , the mass within @xmath23.,width=321,height=321 ] the solid lines in fig .
[ mass_profile_inner ] show the differential mass functions of the substructures found by hsf within @xmath23 for the three highest resolution aq - a haloes .
the number of substructures more massive than @xmath57 is quite small and only for aq - a-1 is the dynamic range sufficient to determine a power - law slope , which is close to @xmath3 . while the differential mass function for substructures within @xmath0 ( the top panel of fig .
[ mass_profile ] ) converges reasonably well with increasing resolution , this is not the case for the inner regions . here
, increasing resolution enables tidal streams to be followed to significantly lower contrast , substantially increasing the mass attached to each one and so the total mass in substructures .
this effect was already visible in fig .
[ radius_res ] and is confirmed by the cumulative mass function shown in fig .
[ mass_cum_inner ] . here
also the curves for different resolutions agree much less well than was the case when we focused on substructures within @xmath0 ( see the bottom panel of fig .
[ mass_profile ] ) .
only @xmath8 of the mass inside @xmath23 is in the form of self - bound subhaloes .
although this number includes only subhaloes resolved in aq - a-1 , it is expected to increase by at most a factor of two if one extrapolates down to the free - streaming length ( see * ? ? ?
the mass fraction in substructures in this same region increases from @xmath58 for aq - a-3 to @xmath9 for aq - a-1 .
thus tidal streams contain almost @xmath59 times as much mass as self - bound subhaloes at the resolution of aq - a-1 , and presumably would contain even more at higher resolution . as fig .
[ mass_radius ] shows , the substructure mass fraction varies as a function of radius . at the solar circle about @xmath60 of the mass
is in self - bound subhaloes and about @xmath61 in tidal streams at the resolution of aq - a-1 .
more than 99% of the mass appears smoothly distributed even at this extremely high resolution and when processed with a state - of - the - art 6d structure finder .
-@xmath62 and @xmath54-@xmath51 projections for aq - a-1 .
we create a @xmath63 image and colour each pixel according to the logarithm of the maximum 6d phase - space density over the enclosed particles as estimated using enbid .
the phase - space density is measured in units of @xmath53 .
the mass of each substructure is given in units of @xmath64 at the bottom right of each row.,title="fig:",width=321,height=56 ] -@xmath62 and @xmath54-@xmath51 projections for aq - a-1 .
we create a @xmath63 image and colour each pixel according to the logarithm of the maximum 6d phase - space density over the enclosed particles as estimated using enbid .
the phase - space density is measured in units of @xmath53 .
the mass of each substructure is given in units of @xmath64 at the bottom right of each row.,title="fig:",width=321,height=658 ] to give a visual impression of the structures found by hsf , fig .
[ sub_rvr ] shows some typical substructures in the inner halo of aq - a-1 .
the top panel presents the main halo with all hsf substructures ( bound and unbound ) removed . in the second row
we show the biggest bound subhalo and its attached tidal streams .
these extend over nearly 60 kpc .
the mass of this biggest substructure is @xmath65 of which @xmath66 is in the self - bound subhalo . in the middle row
we show an example of the same kind of substructure , but here the tidal tails are more pronounced and the stream passes within @xmath67 kpc of halo centre .
the bottom two rows show two typical stream - like structures which have no associated self - bound subhalo .
their masses are similar to those of the biggest subhaloes . in this section
we try to understand how the different components contribute to the dark matter density near the sun .
we ask how likely is it for the solar system to lie within a subhalo or tidal stream of given local density .
this is accomplished by computing probability density distributions for the space density at random points within a thick spherical shell centred at the solar radius , as in @xcite , but separating the dark matter into the different phase - space components identified by hsf .
the result of this procedure is presented in fig .
[ volume ] .
the left panel of fig .
[ volume ] shows the probability distribution function of dm density for various structures in the inner halo . to make this plot we first estimated the density at the position of every particle with radius between 6 and 12 kpc using a standard sph scheme based on 32 neighbours .
as described in @xcite , we then fitted a smooth model to these values assuming the density to be stratified on similar , concentric ellipsoids and to be a power law of radius .
this defines a model density @xmath68 at the position of each particle which can be compared with the directly estimated local density .
this step is crucial to account for the large density gradients in the inner halo so that we can focus on small - scale variations due to substructure .
we then repeat the sph density estimates for the subsets of particles in this radial range corresponding to each individual subcomponent : the main subhalo and each individual self - bound subhalo and substructure . in the following
we plot all density distributions as functions of @xmath69 and we construct volume - weighted probability distributions by histogramming the particles with individual weights @xmath70 , where @xmath71 is the particle mass and @xmath72 the total volume between 6 and 12 kpc . for the total mass distribution and the main subhalo the resulting distributions give the probability that an observer at a random point in this radial range will see local density contrast @xmath73 .
for the self - bound subhalo and substructure components , the distributions show the mean number of self - bound subhaloes or substructures with local density contrast @xmath69 at a random point .
if we consider first the probability distribution of density contrast for the total mass ( i.e. the sum of all the components ) we see that there is a lognormal distribution centred on @xmath74 with an additional low amplitude , power - law tail to high densities .
this result was already given in @xcite .
as they showed , the lognormal part of the distribution reflects discreteness noise in our density estimator .
we demonstrate this again here by plotting as a dashed black line the distribution of analogous density estimates for points sampled from a uniform poisson distribution .
this line can not be distinguished from that corresponding to the main subhalo in fig .
[ volume ] , demonstrating that this component follows our ellipsoidal model very closely .
the power law tail is due to self - bound subhaloes , as is evident from its close agreement with the distribution calculated directly from this component .
this general behaviour was pointed out not only in the analysis of this same simulation by @xcite , but also in the analytic model of @xcite and in the analysis of a different high resolution simulation by @xcite .
if we now consider the density contrast distributions of the individual components , we see that self - bound subhaloes are detectable not only in the high - density tail but also down to contrasts as small as @xmath75 .
this reflects the excellent resolution of the aq - a-1 simulation and , more importantly , the fact that our 6-d structure finder can identify subhalo material even at very low density contrast because of its small internal velocity dispersion .
hsf identifies general phase - space substructure ( e.g. tidal streams ) down to even lower contrasts , of order @xmath76 .
the additional substructure mass which is not part of self - bound subhaloes is almost entirely in this low - contrast regime .
it is interesting that the `` probability '' a random point lies in such a low - density tidal stream reaches values _ much _
larger than one , meaning that hsf has identified multiple structures at each point in 3-space .
note , however , that because low - density tidal streams have an effective spatial dimensionality less than 3 , their @xmath69 values are biased low ( and the mean stream number correspondingly high ) by the spherical kernel of the sph density estimator . in the radial range between 6 and 12 kpc selected for this analysis about @xmath77 of the mass is in the form of self - bound subhaloes and about @xmath78 in substructure .
thus low - density tidal streams account for almost 90% of the substructure detected by hsf .
the right panel of fig .
[ volume ] shows , for the various components , a cumulative plot of the mass at local density contrast exceeding @xmath69 , expressed as a fraction of the total mass between 6 and 12 kpc . here
we see explicitly that the substructure component contains almost ten times as much mass as the bound subhalo component and that this excess lies almost exclusively at contrasts below 0.1 .
this panel also gives similar cumulative data for the five individually most massive self - bound subhaloes and for the five individually most massive substructures .
almost a third of the mass in self - bound subhaloes is contained in these five objects , almost all of it at density contrasts exceeding unity
. however , only the most massive subhalo corresponds to one of the five most massive substructures , accounting for most of its mass .
the other four massive substructures are unbound tidal streams with no associated subhalo .
the maximum density contrast of these unbound streams is @xmath79 ( again this is probably biased low ) .
the five most massive substructures together account for only about 10% of the total substructure mass . not only the mass density , but also the velocity distribution of dm particles in the vicinity of the earth is relevant for direct detection experiments .
@xcite showed that although this velocity distribution is quite well approximated by a smooth trivariate gaussian , potentially measurable features are imprinted on the corresponding energy distribution by the detailed formation history of the milky way s halo . here
we concentrate on the velocities of the different phase - space components in the inner halo . in fig .
[ vrvt ] we show @xmath51-@xmath80 projections of the distribution of all particles in the radial range 6 to 12 kpc ( top ) , of those in substructures ( middle ) , and of those in self - bound subhaloes ( bottom ) .
these plots are two - dimensional histograms , with colour encoding the mass in the corresponding bin as indicated by the colour bar ( in units of solar masses per @xmath39 km / s x @xmath39
km / s pixel ) .
the total mass contributing to each panel is given in its top right - hand corner .
the main halo and so the bulk of the particles lie primarily at velocities below 200 km s@xmath81 , whereas subhaloes and tidal streams are found almost exclusively at higher velocities . as a result ,
the most massive subhaloes are still ( just ) visible in the top panel despite the fact that they contribute less than a tenth of a percent of the mass .
these structures contribute to the high - energy tail of the recoil spectrum in direct dm detection experiments , and so may be visible in high resolution experiments , particularly those with directional sensitivity which can detect the common motion of the substructure particles .
tangential velocity @xmath80 phase - space plots for the @xmath82 kpc region of the aq - a-1 halo . for each plot
a two - dimensional histogram was calculated and colours were set to reflect the mass in each @xmath39 km s@xmath81 x @xmath39 km s@xmath81 pixel as shown by the colour bar , labelled in units of @xmath64 .
the top , middle and bottom panels show the distributions for all particles , for substructures and for subhaloes respectively .
the low velocity region is dominated by the main halo but some substructures / subhaloes are still visible in the the high @xmath51 and @xmath80 velocity regions .
hsf only detects significant substructure in these regions of phase - space .
the total component mass in @xmath64 is indicated in the top right - hand corner of each panel.,title="fig:",width=321,height=56 ] tangential velocity @xmath80 phase - space plots for the @xmath82 kpc region of the aq - a-1 halo . for each plot
a two - dimensional histogram was calculated and colours were set to reflect the mass in each @xmath39 km s@xmath81 x @xmath39 km s@xmath81 pixel as shown by the colour bar , labelled in units of @xmath64 .
the top , middle and bottom panels show the distributions for all particles , for substructures and for subhaloes respectively .
the low velocity region is dominated by the main halo but some substructures / subhaloes are still visible in the the high @xmath51 and @xmath80 velocity regions .
hsf only detects significant substructure in these regions of phase - space .
the total component mass in @xmath64 is indicated in the top right - hand corner of each panel.,title="fig:",width=321,height=680 ]
we study the population of subhaloes and tidal streams in six milky way - like dm haloes taken from the aquarius project .
these structures are identified using the hierarchical structure finder ( hsf * ? ? ?
* ) , a state - of - the - art structure finder which operates in 6-d phase - space .
we find that that the differential mass function of self - bound subhaloes can be well described by a power - law with slope close to @xmath29 .
this agrees with results from an independent analysis using the 3-d structure finder subfind @xcite .
typically hsf attaches slightly more particles to subhaloes than subfind , and also finds slightly more subhaloes above the simulation resolution limit ( see table [ table2 ] ) .
this agrees with previous results described in @xcite .
about 14% of the mass within @xmath0 is in self - bound subhaloes , with significant scatter among the six haloes aq - a to aq - f .
hsf subhalo masses are @xmath83 larger than those found by subfind , although the increase can be larger near halo centre . in most haloes
the total subhalo mass is dominated by the largest objects .
the radial distributions of hsf and subfind subhaloes are almost identical , although hsf can identify subhaloes closer to halo centre due to their enhanced density contrast in phase - space .
the differential mass function for substructures ( i.e. both subhaloes and tidal streams ) is also well described with a power - law , but in this case the slope is close to @xmath3 .
this is independent of simulation resolution and holds approximately for all six haloes , thus appearing robust . for most of the level 2
haloes around @xmath1 of the mass within @xmath0 is assigned to substructures with mass above @xmath84 .
the radial distribution of these objects can be approximated by equation ( [ eq : fsub ] ) introduced in @xcite but with a higher normalisation than applies for self - bound subhaloes . in the inner halo almost @xmath85 times as much mass is detected in unbound tidal streams as in self - bound subhaloes .
this reflects the efficiency with which tidal forces destroy bound subhaloes in the inner halo . in our highest resolution halo ,
aq - a-1 , hsf assigns about @xmath8 of the mass within @xmath7 kpc to self - bound subhaloes , and about @xmath9 to substructures ( subhaloes and tidal streams ) , with masses higher then @xmath86 . in this region
the largest phase - space substructures are either self - bound subhaloes with massive tidal tails stretching across the entire inner @xmath7 kpc region or equally massive tidal streams with no attached subhalo at @xmath87 kpc .
the largest individual substructures in the inner region have masses up to @xmath88 .
@xcite showed that the density field in the radial range from 6 to 12 kpc is very well represented by a simple smooth model where density is stratified on similar concentric ellipsoids and falls as a power law of radius .
fluctuations around this model are small except for a low - amplitude power - law tail to high density contrast which corresponds to self - bound subhaloes .
our hsf analysis is consistent with these results and allows us , in addition , to study the contrast of individual substructures with respect to the smooth background , the main subhalo , in which they are embedded .
hsf is able to identify not only the high - contrast cores of individual self - bound subhaloes , but also their outskirts where the density contrast drops to values as low as @xmath89 .
the maximum contrast of unbound tidal streams is @xmath79 .
since these streams contain @xmath61 of the total mass between 6 and 12 kpc , of order one massive tidal stream is predicted to pass through every point , contributing a few tenths of a percent of the local dm density .
in contrast , only @xmath77 of the mass in this region is contributed by self - bound subhaloes , and the chance that the earth lies in the high - density contrast region of such a subhalo is below @xmath75 .
both subhaloes and tidal streams populate the high - energy tail of the velocity distribution preferentially , and would show up in direct dm detection experiments as a small but significant part of the signal in events with almost identical ( vector ) momenta .
the aquarius project was carried out as part of the activities of the virgo consortium .
we thank the consortium members responsible for setting up and executing the simulations , as well as the consortium as whole for making them available to us .
aubert d. , pichon c. , colombi s. , 2004 , mnras , 352 , 376 boylan - kolchin m. , springel v. , white s. d. m. , jenkins a. , 2010 , mnras , 406 , 896 cole s. , lacey c. , 1996 , mnras , 281 , 716 davis m. , efstathiou g. , frenk c.s .
, white s.d.m . , 1985 , apj , 292 , 371 diemand j. , moore b. , stadel j. , 2004 , mnras , 352 , 535 diemand j. , moore b. , stadel j. , 2005 , nature , 433 , 389 diemand j. , kuhlen m. , madau p. , 2006 ,
apj , 649 , 1 diemand j. , kuhlen m. , madau p. , zemp m. , moore b. et al . , 2008 , nature , 454 , 735 ghigna s. , moore b. , governato f. , lake g. , quinn t. et al . , 2000 , apj , 544 , 616 governato f. , moore b. , cen r. , stadel j. , lake g. , quinn t. , 1997 , newa , 2 , 91 kamionkowski m. , koushiappas s. m. , 2008 , phys .
d , 77 , 103509 kamionkowski m. , koushiappas s. m. , kuhlen m. , 2010 , phys .
d , 81 , 043532 kim j. , park , c. , 2006 , apj , 639 , 600 klypin a. , kravtsov a. v. , valenzuela o. , prada f. , 1999 , apj , 522 , 82 knollmann s. r. , knebe a. , apjs , 2009 , 182 , 608 maciejewski m. , colombi s. , springel v. , alard c. , bouchet f. r. , 2009a , mnras , 396 , 1329 maciejewski m. , colombi s. , alard c. , bouchet f. , pichon c. , 2009b , mnras , 393 , 703 moore b. , ghigna s. , governato f. , lake g. , quinn t. et al .
, 1999 , apj , 524 , l19 neyrinck m. c. , gnedin n. y. , hamilton a. j. s. , 2005 , mnras , 356 , 1222 sharma s. , steinmetz m. , 2006 , mnras 373 , 1293 sharma s. , johnston k.v . , 2009 ,
apj , 703 , 1061 springel v. , white s. d. m. , tormen g. , kauffmann g. , 2001 , mnras , 328 , 726 springel v. , wang j. , vogelsberger m. , ludlow a. , jenkins a. et al .
, 2008 , mnras , 391 , 1685 stoehr f. , white s. d. m. , tormen g. , springel v. , 2002 , mnras , 335 , 84 vogelsberger m. , amina h. , springel v. , white s. d. m. , wang j. et al .
, 2009 , mnras , 395 , 797 vogelsberger m. , white s. d .m . , 2010 , arxiv:1002.3162 wang j. , navarro j. f. , frenk c. s. , white s. d. m. , springel v. et al . , 2010 , | we analyse the coarse - grained phase - space structure of the six galaxy - scale dark matter haloes of the aquarius project using a state - of - the - art 6d substructure finder . within @xmath0
, we find that about @xmath1 of the mass is in identifiable substructures , predominantly tidal streams , but including about @xmath2 in self - bound subhaloes .
the slope of the differential substructure mass function is close to @xmath3 , which should be compared to @xmath4 for the population of self - bound subhaloes .
near @xmath0 about @xmath5 of the mass is in substructures , with about @xmath6 in self - bound subhaloes .
the inner @xmath7 kpc of the highest resolution simulation has only @xmath8 of its mass in self - bound subhaloes , but @xmath9 in detected substructure , again primarily tidal streams .
the densest tidal streams near the solar position have a 3-d mass density about @xmath10 of the local mean , and populate the high velocity tail of the velocity distribution .
[ firstpage ] methods : numerical , cosmology : dark matter |
[ spinhalf ] [ spinone ] [ alpha ] .series expansion coefficients for the spin - stiffness of the spin-1/2 heisenberg model on the anisotropic square lattice for selected values of the anisotropy coupling @xmath16 . [ cols="^,^,^,^,^",options="header " , ] we consider antiferromagnetic heisenberg model on a square - lattice , with spatially anisotropic exchange couplings given by the hamiltonian @xmath17 where the sum over @xmath18 runs over all sites of the square - lattice .
spin - stiffness can be defined by the change in the ground state energy of the system under an applied twist along one of the axessingh1989,hamer1994 . in general
, it can be decomposed into a sum of two parts , a paramagnetic part and a diamagnetic part . for the anisotropic model
, one can define two different twists @xmath19 and @xmath20 depending on whether the twist is applied along the @xmath21 or the @xmath22 axis .
following ref . , the diamagnetic component of the twist for @xmath20 is given by the expression @xmath23 where angular brackets denote expectation value in the ground state of the hamiltonian in eq .
( [ hamiltonian ] ) .
the paramagnetic term is given by the equation @xmath24 where @xmath25 is the coefficient of the @xmath26 term in the ground state energy per site of the hamiltonian in eq .
( [ hamiltonian ] ) with a perturbation @xmath27 in order to calculate these quantities we introduce an ising anisotropyoitmaa2006 by scaling all xy parts of the exchange interactions by a factor @xmath28 . then @xmath19 and @xmath20 can be calculated as a power series in @xmath28 for any value of the coupling anisotropy .
series expansions for selected values of the anisotropy for the spin - half and spin - one model are given in table [ table 1 ] and table [ table 2 ] respectively .
the series are analyzed by integrated differential approximants ( ida)oitmaa2006 . before the analysis , a change of variable of the form @xmath29 has been introduced to remove leading singularities as @xmath30 .
the results for the spin - half model are shown in fig .
[ fig : spinhalf ] and the results for spin - one model are shown in fig . [
fig : spinone ] . in the 1d limit ,
the spin - stiffness constant is know to be @xmath31 from exact calculations by shastry and sutherland@xcite .
this value is clearly larger than the square - lattice case where @xmath32 .
our results are more accurate away from the 1d limit , but they clearly appear to approach the 1d limit in a smooth and monotonic manner . for the spin - one case it is known that the nel order disappears at an anisotropy ratio of approximately @xmath33@xcite
. the transition should be in the universality class of the 3d heisenberg model .
the spin - stiffness should vanish at the transition with an exponent of @xmath34@xcite .
the fit shows that the data agrees well with these expectations .
for the spin - half case we also fit the small anisotropy regime to a linear behavior .
the results are shown in fig .
[ fig : alpha ] where they are compared to the spin - wave results discussed in the next section .
we find that the anisotropy can be expressed as @xmath35 where @xmath36 .
this deviates significantly from the naive expectation @xmath37 .
in this section we consider the self - consistent version of spin - wave theory@xcite . to apply this approach
we subdivide the lattice into sublattices @xmath38 and @xmath39 and use the dyson - maleev representation for spin operators on each sublattice,@xmath40and @xmath41where @xmath42 and @xmath43 are the bose operators .
introducing the operators @xmath44and decoupling the four - boson terms in the hamiltonian into all possible two - boson combinations , we derive ( see refs . )
@xmath45where @xmath46 correspond to the nearest neighbour sites in the @xmath47 and @xmath48 directions , @xmath49are the short - range order parameters and @xmath50is the sublattice magnetization . diagonalizing the hamiltonian ( [ hsswt ] )
one finds the self - consistent equations at @xmath51@xmath52where the antiferromagnetic spin - wave spectrum has the form @xmath53with @xmath54and @xmath55 ( we assume here that the ground state is antiferromagnetically ordered , otherwise a bosonic chemical potential @xmath56 should be introduced in the dispersion ( ek ) to fulfill the condition @xmath57 , see refs . ) .
the parameters @xmath58 are simply related to the spin correlation function at the nearest - neighbor sites by latexmath:[\[\gamma _
{ \delta } = |\langle \mathbf{s}_{i}\mathbf{s}_{i+\delta } \rangle @xmath47 and @xmath48 axes is expressed through these parameters as@xmath60the results obtained according to eqs .
( [ eqsswt ] ) and ( [ rss ] ) are shown in figs .
[ fig : spinhalf ] , [ fig : spinone ] .
while for @xmath61 the spin - wave results are close to those obtained by series expansions over the entire parameter range , for @xmath62 at large anisotropy , the two techniques give results for @xmath63 which are qualitatively different .
this difference is most probably due to topological excitations within the half - integer spin chains , which are not considered by spin - wave theory . for @xmath61 , at the transition to the haldane phase , we find the stiffness exponents to be @xmath64 and @xmath65 , in qualitative agreement with series expansions . for @xmath62 ,
a quantitative discrepancy between spin - wave theory and series expansions is visible already at small anisotropies .
while series expansion and spin - wave stiffness along the weaker exchange couplings axis ( @xmath16 ) remain very close to each other , a discrepancy arises from the stiffness along the stronger coupling direction .
we find the coefficient in the linear fit ( [ linearfit ] ) @xmath66 somewhat lower than what found by series expansion , as shown in fig [ fig : alpha ] .
it is possible that part of the difference is due to numerical inaccuracies or high order effects in @xmath67 .
however , with increasing anisotropy the difference is not just quantitative .
it becomes qualitative and it implies the onset of new physics for the spin - half case associated with the berry phase terms @xcite .
we base our considerations on the theory of underdoped cuprates suggested in ref . .
according to this theory , the ground state of an underdoped uniformly doped cuprate is a spin spiral , spontaneously directed along the ( 1,0 ) or the ( 0,1 ) direction . at
sufficiently small doping , @xmath68 , the spiral has a static component while at @xmath69 it is fully dynamic . here
@xmath21 is the concentration of holes in cuo@xmath9 plane . according to this picture ,
the electronic liquid crystal observed in ref .
is the mostly dynamic spin spiral which may still have some small static component . for sr
doped la@xmath70cuo@xmath71 , the value of the critical concentration is @xmath72 , and for yba@xmath9cu@xmath10o@xmath73 is @xmath74 .
the absolute value of the wave vector of the spin spiral ( static or dynamic ) is given by @xmath75 here @xmath76 is the spin - stiffness of the initial heisenberg model @xcite , @xmath77 is the antiferromagnetic exchange parameter of the model , and @xmath78 is the coupling constant for the interaction between mobile holes and spin waves .
we set the spacing of the tetragonal lattice equal to unity , so the the wave vector @xmath79 is dimensionless .
to fit the neutron scattering experimental data to the position of the incommensurate structure in sr doped single layer la@xmath70cuo@xmath71 , we need to set @xmath80 , and to fit similar data for double layer yba@xmath9cu@xmath10o@xmath81 , we need to set @xmath82 .
it is not clear yet why the values of @xmath78 for these compounds are slightly different , but for purposes of the present work this difference is not important .
the coupling constant @xmath78 was calculated within the extended t - j model @xcite .
the result is @xmath83 where @xmath84 is the nearest site hopping matrix element and @xmath85 is the quasihole residue .
it is known that in cuprates @xmath86 and @xmath87 .
thus the calculated value of the coupling constant , @xmath80 , agrees well with that found by fitting of experimental data .
the ground state energy of the spin spiral state consists of two parts .
the spin spiral with the wave vector @xmath79 gives rise to the gain @xmath88 in the kinetic energy of a single hole .
on the other hand the spiral costs the spin elastic energy @xmath89 .
so the total balance is @xmath90 , and minimization with respect to @xmath79 gives the wave vector ( [ q ] ) and the energy per elementary cell @xmath91 there are also quantum corrections to this energy , but they are small and hence not important for our purposes@xcite .
note , that eq .
( [ e1 ] ) is valid for both @xmath92 and @xmath93 , assuming that @xmath21 is not large . up to now
we have disregarded the anisotropy assuming a perfect square lattice . to analyze anisotropy in the spiral direction we have to replace @xmath94 where @xmath95 is due to the lattice deformation , so @xmath96 and @xmath97 .
the antiferromagnetic exchange , @xmath98 , also becomes anisotropic , @xmath99 , @xmath100 .
hence the spin - stiffness is replaced by @xmath101 , where @xmath102 has been calculated above .
now we can see how the lattice deformation influences the spiral energy ( [ e1 ] ) . in the case when @xmath103 is directed along the @xmath104
we have to replace in ( [ e1 ] ) @xmath105 and @xmath106 ; and in the case when @xmath103 is directed along the @xmath107 we have to replace in ( [ e1 ] ) @xmath108 and @xmath109
. note that the quasiparticle residue @xmath85 is a scalar property and therefore it is independent of direction of @xmath103 .
altogether , with account of the anisotropy , the energy ( [ e1 ] ) is replaced by @xmath110 \ .\ ] ] the minus sign corresponds to @xmath111 directed along the @xmath104 axis , and the plus sign corresponds to @xmath111 directed alone the @xmath112 axis .
interestingly , without the spin - quantum - fluctuations effect ( i. e. if @xmath113 ) the anisotropy in energy disappears . since @xmath114 it is most natural to assume that @xmath115 , this means that @xmath116 .
this point is supported by the lda calculation performed in ref . .
in this case , according to eq .
( e2 ) , the energy of the state with @xmath111 along the @xmath13-axis is higher than that with @xmath111 along the @xmath117-axis .
this disagrees with the experimental data in ref . .
however , the anisotropy of the hopping matrix element @xmath84 is not straightforward .
there are two competing contributions to @xmath95 .
the first one is related to the lattice deformation and is positive .
the second one is related to oxygen chains that are present in yba@xmath9cu@xmath10o@xmath11 and is negative . in principle
it is possible for the negative contribution to win , making @xmath95 negative @xcite .
the neutron scattering anisotropy has been previously discussed within the pomeranchuk instability scenario @xcite .
this is probably not sufficient to explain the newest data , see discussion in ref .
yamase2 . however , it is interesting to note that , to explain the sign of the pinning , the pomeranchuk scenario also requires a negative @xmath95 .
anyway , for further numerical estimates , we will assume @xmath118 the absolute value is consistent with the 1% lattice deformation and the sign has been discussed above . in this case , according to eq .
( [ e2 ] ) , the energy of the state with @xmath111 along the @xmath13-axis is lower and this is consistent with experimental data .
the direction of pinning energy at @xmath119 reads @xmath120 this is the pinning energy per cu site and it is a pretty strong pinning . for comparison ,
the pinning energy of spin to the orthorhombic b - direction in undoped la@xmath70cuo@xmath71 is just @xmath121 .
assuming that the correlation length is at least comparable with the period of the spin spiral , @xmath122 , we find that the total pinning energy per correlation unit is @xmath123 , which is a significant energy scale .
in this paper , we have studied the spin - stiffness constants for spatially anisotropic spin - half and spin - one heisenberg models using series expansions and self - consistent spin - wave theory .
the theoretical results have been of interest in themselves and show the importance of berry phase interference terms in anisotropic square - lattice models .
our primary motivation for the study has been to understand the phenomena of electronic liquid crystal and its pinning in high temperature superconductors .
we find that quantum interference effects significantly enhance the spin - stiffness anisotropy and this can provide the primary mechanism for the pinning of the liquid crystal direction .
we have provided a detailed quantitative account of the pinning energy in ybco . | using series expansions and spin - wave theory we calculate the spin - stiffness anisotropy @xmath0 in heisenberg models on the square lattice with anisotropic couplings @xmath1 .
we find that for the weakly anisotropic spin - half model ( @xmath2 ) , @xmath0 deviates substantially from the naive estimate @xmath3 .
we argue that this deviation can be responsible for pinning the electronic liquid crystal direction , a novel effect recently discovered in ybco . for completeness
, we also study the spin - stiffness for arbitrary anisotropy @xmath4 for spin - half and spin - one models . in the limit of @xmath5 , when the model reduces to weakly coupled chains , the two show dramatically different behavior . in the spin - one model , the stiffness along the chains goes to zero , implying the onset of haldane - gap phase , whereas for spin - half the stiffness along the chains increases monotonically from a value of @xmath6 for @xmath7 towards @xmath8 for @xmath5 .
spin - wave theory is extremely accurate for spin - one but breaks down for spin - half presumably due to the onset of topological terms .
this work is motivated by the recent discovery @xcite of the electronic liquid crystal in underdoped cuprate superconductor yba@xmath9cu@xmath10o@xmath11 .
the electronic liquid crystal manifests itself in a strong anisotropy of the low energy inelastic neutron scattering .
the liquid crystal picture implies a spontaneous violation of the directional symmetry : the `` crystal '' can be oriented either along the ( 1,0 ) or along the ( 0,1 ) axes of the square lattice .
the yba@xmath9cu@xmath10o@xmath11 compound has a tetragonal lattice with tiny in - plane lattice anisotropy , @xmath12 .
this tiny anisotropy is sufficient to pin the orientation of the electronic liquid crystal along the @xmath13-axis . as a result
the low energy neutron scattering @xcite demonstrates a quasi-1d structure along @xmath13 .
to understand the pinning mechanism of the electronic crystal , in the present work , we study the anisotropic heisenberg model .
we calculate the in - plane anisotropy of the spin - stiffness and demonstrate that this is strongly enhanced by quantum fluctuations .
we argue that the enhancement is sufficient to provide a pinning mechanism for the initially spontaneous orientation of the electronic liquid crystal and suggest a specific mechanism for the pinning .
the anisotropic heisenberg model has previously attracted a lot of theoretical interest@xcite .
however , most theoretical studies have focussed on the regime of strong anisotropy , where the system reduces to one of weakly coupled spin - chains , and the most significant issue there is that of dimensional crossover and the onset of long range antiferromagnetic order . to the best of our knowledge
the anisotropy of spin - stiffness has not been studied before .
this is an important theoretical problem in itself and therefore we extend our study to the case of arbitrary strong anisotropy .
we consider both spin - half model , where in the limit of strong anisotropy we come to the situation of weakly coupled heisenberg s=1/2 chains , and also spin - one model , where in the limit of strong anisotropy we come to the situation of weakly coupled haldane chains .
our series expansion results show that the spin - stiffness indeed behaves very differently in the two cases . for spin - one
, the stiffness along the chains vanishes at an anisotropy ratio of @xmath14 .
self - consistent spin - wave theory remains highly accurate in this case all the way down to the transition . on the other hand , for spin - half ,
series expansions show that the stiffness along the chains increases from @xmath6 in the isotropic limit towards the known@xcite 1d result of @xmath8 as @xmath15 . in this case spin - wave theory
clearly breaks down with increasing anisotropy , presumably due to the onset of berry phase interference@xcite .
the structure of the paper is as follows : in section [ se ] we calculate the spin - stiffness using series expansions .
this is probably the most accurate method that is valid from small to very large anisotropy . in section [ sw ]
we calculate the same spin - stiffness using spin - wave theory .
this method is valid as long as one is not close to 1d limit . in section
[ pinning ] we discuss the application of our results to the explanation of the electronic liquid crystal pinning in yba@xmath9cu@xmath10o@xmath11 .
finally , in section [ conclusions ] , we draw our conclusions . |
the use of chiral perturbation theory ( @xmath7pt ) to describe few - nucleon systems is a problem that has received much attention over the past twenty years .
this effort began with the seminal papers of weinberg @xcite , and has recently been reviewed in refs .
@xcite . in this approach
the nn potential is computed up to some fixed order , @xmath8 , in the chiral expansion in powers of @xmath9 . here
@xmath10 is the nn c.m .
momentum , @xmath11 the pion mass , and the breakdown scale @xmath12 is nominally @xmath13 , but in reality is somewhat lower for reactions involving baryons .
this nn potential is then iterated , using the schrdinger or lippmann - schwinger equation , to obtain the scattering amplitude .
this approach has come to be known as chiral effective field theory ( @xmath7eft ) as it encodes the consequences of qcd s pattern of chiral - symmetry breaking for few - nucleon systems and is built on a systematic expansion in powers of @xmath14 , while resumming the non - perturbative effects that lead to the existence of nuclear bound states .
electromagnetic reactions provide particularly fertile ground for the application of @xmath7eft .
a few - nucleon electromagnetic current operator , @xmath15 , can also be derived from @xmath7pt ko09,ko11,pa08,pa09,pa11,pi12 , and matrix elements are then constructed via : @xmath16where @xmath17 and @xmath18 are solutions of the schrdinger equation for the @xmath7eft potential @xmath19in principle at order @xmath8 .
@xmath7pt predicts that short - distance effects will enter @xmath20 eventually , but minimal substitution generates the first several orders in the expansion for @xmath15 , especially in the case of the charge operator .
this enables predictions for charge form factors of @xmath21 pc99,wm01,ph03,ph07,pi12 and @xmath22 @xcite systems to be obtained as pure predictions up to order @xmath23 relative to leading order .
these predictions agree well with data up to at least @xmath24 gev@xmath25 . in the magnetic response
the first short - distance operator not determined by nn scattering or single - nucleon properties enters considerably earlier , but accurate predictions for form factors can still be made @xcite .
indeed , the existence of this additional short - distance operator allows enhanced understanding of magnetic moments and m1 transitions in systems up to @xmath26 , with reactions that had not previously been understood now being explained , and correlated , through the presence of the same short - distance @xmath27nn@xmath28nn operator @xcite . in this work we focus on the application of this formalism to electrodisintegration of deuterium , and specifically the computation of the longitudinal response function , @xmath29 .
@xmath29 is proportional to the square of the matrix element of the nn charge operator between the deuteron wave function and the wave function of a continuum nn state .
@xmath30 is given solely by a single - nucleon operator up to @xmath31 relative to leading order in the @xmath7eft expansion , so calculations of @xmath29 can take the information gleaned on nn interactions from scattering experiments and use it to make quite accurate predictions for an electromagnetic observable .
deuteron electro - disintegration has been employed as a testing ground for nn models for a long time aren , aren1,aren2,alt00,alt02,alt05 .
since the 4-momentum in the breakup process is given by a virtual photon , a richer set of kinematics can be probed as compared to the photo - disintegration process . and , in addition to the deuteron wave function , the disintegration process probes both the on - shell and off - shell nn t - matrix through the final - state interaction .
this means that the solutions for the deuteron wave function @xmath32 , and the nn t - matrix , are input to our calculation . to obtain them from @xmath0eft potentials we need to impose a cutoff on the intermediate states , @xmath33 .
the low - energy constants ( lecs ) multiplying contact interactions in the nucleon - nucleon part of the chiral lagrangian should then be adjusted to eliminate any cutoff dependence @xmath34 or greater in the effective theory s predictions for low - energy observables .
if this is not possible we conclude that @xmath0eft is unable to give reliable predictions . at leading order
the @xmath0eft potential consists of a zero - derivative contact interaction that is operative only in nn partial waves with @xmath35 together with one - pion exchange , which is active in all partial waves . in an earlier work
we showed how to subtractively renormalize " the lo equations for nn scattering in these two channels @xcite .
this technique eliminates the contact interaction in favor of a low - energy observable ( e.g. the relevant nn scattering length ) .
this makes it straightforward to take the limit @xmath36 : no fine - tuning of the pertinent lec is necessary . the resulting phase shifts ( and one mixing parameter ) do not provide anything like a precision description of nn data , but they are , at least , a well - defined , renormalized lo calculation .
however , several papers have demonstrated that a lo @xmath0eft calculation does not produce reliable predictions in partial waves with @xmath37 once @xmath33 is sufficiently large @xcite .
this is because only one - pion exchange is present in these waves at lo , and the resulting singular potential has no nn lec that permits renormalization .
we examined this problem within the context of subtractive renormalization and confirmed the conclusion of ref .
@xcite , i.e. any partial wave with @xmath37 where one - pion exchange is attractive does not have stable lo results in @xmath0et @xcite .
we also showed that this problem is not removed at @xmath38 or @xmath39 . in particular , at @xmath40 two - pion exchange produces a highly singular , attractive potential .
the nn contact interactions needed to renormalize this potential are not present , e.g. in the @xmath41p@xmath42-@xmath41f@xmath42 channel .
the resulting lack of stability with @xmath33 of the nn phase shifts occurs once @xmath33 is larger than @xmath43 gev . in ref
. @xcite we used subtractive renormalization to calculate s - wave nn phase shifts from @xmath38 and @xmath39 @xmath0et potentials . here
too we found that the phase shifts are not stable once @xmath44 gev . in this case
part of the problem is that the momentum - dependent contact interaction that appears at @xmath38 has limited effect as @xmath36 wigner , pc96,sc96 .
thus the nn potential obtained by straightforward application of @xmath0pt can not be used over a wide range of cutoffs : @xmath0eft as formulated above is not properly renormalized , i.e. the impact of short - distance physics on the results is not under control . nevertheless , in refs .
@xcite ( refs .
@xcite ) @xmath19 was computed to @xmath38 and @xmath39 ( @xmath3 ) , and the several nn lecs which appear in @xmath19 were fitted to nn data for a range of cutoffs between 500 and 800 mev .
the @xmath3 predictions contain very little residual cutoff dependence in this range of @xmath33 s , and describe nn data with considerable accuracy .
this suggests that @xmath0eft may be a systematic theory of few - nucleon systems if we employ @xmath33 s in the vicinity of @xmath45 .
since the short - distance physics of the effective theory for @xmath46 is different to the short - distance physics of qcd itself , some authors argue that considering @xmath47 does not yield any extra information about the real impact of short - distance physics on observables @xcite .
using low cutoffs has the advantage that relevant momenta are demonstrably within the domain of validity of @xmath0pt .
discussion about whether such a procedure results in the omission of some operators is ongoing @xcite . in what follows
we employ wave functions obtained with the standard @xmath0eft counting and our subtractive renormalization method @xcite , but we do so only for cutoffs @xmath33 up to the maximum value where we found reasonable results in refs .
@xcite , i.e. @xmath48 mev .
our deuteron electrodisintegration calculation provides an opportunity to test this strategy , by comparing its predictions for deuteron structure and final - state interactions to experimental data
. however , experimental data on @xmath29 are somewhat limited especially in the low - energy and low-@xmath49 region of most interest to @xmath7eft .
much of the data on @xmath29 that do exist have been averaged over spectrometer acceptances , which makes comparison with theory not only complicated , but also , in some cases , ambiguous . therefore , although we do compare with data from refs .
vds91,du94,jo96 , we use the bonn - potential calculations of arenhvel _ et al .
_ as a proxy for data .
these calculations have been quite successful in describing data : non - l / t separated @xcite , l / t - separated @xcite , and even on the interference response functions , @xmath50 and @xmath51 @xcite , taken at many different laboratories over a wide range of relevant kinematic conditions .
( note that we do not list data at squared momentum transfers @xmath52 gev@xmath25 here because those are well outside the reach of @xmath7eft , and even beyond the scope of the semi - relativistic treatment of arenhvel _
et al._. for recent progress on high-@xmath53 electrodisintegration see refs .
@xcite and references therein . )
a necessary condition for the @xmath0eft predictions to be considered reliable is that they show minimal dependence on the cutoff @xmath33 .
we will use this criterion to diagnose situations in which the final - state interaction matrix - element computation has significant sensitivity to short - distance physics .
deuteron photodisintegration has been studied ( albeit with an incomplete current operator ) in @xmath0eft by rozdzepik _
_ using a similar strategy and the wave functions of refs
. @xcite @xcite .
since this process involves real photons , it is sensitive to the deuteron transverse response function , @xmath54 , and has no dependence on the response function computed here .
christlmeier and griehammer computed deuteron electrodisintegration at very low energies and momentum transfers in the pionless effective field theory @xcite .
they demonstrated the incompatibility of the data on the mixed response function @xmath50 , published by von neumann - cosel _ et al .
_ with the low - energy nn phase - shift data and our knowledge of other electromagnetic transitions in the nn system @xcite .
this helped them diagnose a flaw in the analysis of ref .
pionless eft was also successfully applied to the @xmath55 data taken in a later experiment at s - dalinac re08 .
the paper is structured as follows . in sec .
[ sec - nnj0op3 ] we review the facts about the nn charge operator which are relevant for our study . in sec .
[ sec - d1 ] , we introduce the general set up of the electrodisintegration problem and lay out the basic formulae . in sec .
[ sec - d2 ] we evaluate the matrix element of @xmath56 which enters @xmath1 and derive an explicit expression for it in terms of partial waves .
we also present the nn input used , i.e. results for deuteron wave functions and nn phase shifts from refs .
ya09a , ya09b . in section [ sec - d4 ]
we present our results for the longitudinal response function .
we pay particular attention to the kinematic regions where significant dependence of @xmath1 on the momentum cutoff @xmath57 is and is not present .
we present our conclusions in sec .
sec - conc .
in this section we analyze the nn charge operator @xmath30 , expanding it in powers of the @xmath7eft small parameter @xmath14 .
since our wave functions are computed up to @xmath59 relative to leading order , we also need the two - nucleon charge operator @xmath30 up to the same relative accuracy .
we denote the leading order for @xmath30 as @xmath60 .
a @xmath30 that is consistent with chiral potential @xmath19 up to @xmath61 ( and obeys appropriate ward identities ) has recently been obtained by several authors @xcite .
we now summarize the results needed for this study .
the analysis proceeds by dividing the charge operator into an isoscalar part @xmath62 , and an isovector one @xmath63 . up to corrections of @xmath61 in the chiral expansion we have , on the basis of states of nn relative momentum in units of @xmath64 , since these factors of the proton charge are incorporated in the expression for the electrodisintegration cross section.]:@xmath65 g_{e}^{(s)}(q^{2 } ) \label{eq : j0s}\ ] ] where @xmath66 is the isoscalar nucleon form factor , and we have summed over the contributions of nucleons one and two . in the case of the deuteron we may use the symmetry of the state ( only even partial waves are present ) to prove that the two terms in square brackets are equal .
meanwhile for @xmath63 we have : @xmath67 g_{e}^{(v)}(q^{2 } ) , \label{eq : j0v}\]]where @xmath68 is the isovector nucleon form factor .
this operator does not contribute to the matrix elements for elastic scattering , but it is relevant for electrodisintegration . @xmath66 and @xmath68 are related to the proton ( neutron ) charge form factor @xmath69 ( @xmath70 ) by@xmath71where the @xmath72 sign applies to the @xmath73 case .
@xmath7pt does a reasonable job describing @xmath68 for squared 4-momentum @xmath74 gev@xmath25 , but its description of isoscalar nucleon structure @xmath66 is of limited utility , even in this low-@xmath53 domain @xcite .
our goal here is to look at higher @xmath75 , and we do not wish to be limited in our pursuit of that goal by @xmath7pt s description of single - nucleon electromagnetic structure .
there are no two - body corrections to @xmath56 at @xmath76 .
it might seem that there will be an effect in @xmath77 , because the @xmath78 photon can couple directly to the exchanged pion .
however , in the static limit used to obtain nn potentials and charge operators that pion line carries no energy .
thus two - body corrections to @xmath77 that contain such an effect are deferred until @xmath79 , because we follow the counting of ref .
@xcite and count @xmath80 while @xmath81 . at @xmath82 an nn contribution to @xmath56 must be built out of a vertex from @xmath83 , one from @xmath84 and a pion propagator .
however , the only term in @xmath83 that couples an @xmath78 photon to a nucleon and a pion has a fixed @xmath85 coefficient , so this effect is also deferred until @xmath79 .
therefore , in the counting where @xmath86 , the nn charge operator is given by eqs .
( [ eq : j0s ] ) and ( [ eq : j0v ] ) , up to corrections of @xmath79 .
the usual expression for the differential cross section for deuteron electro - disintegration is ( see , for example , ref . @xcite)@xmath87+\rho
_ { tt}f_{tt}\cos [ 2\phi ]
\}. \label{eq:6.1}\]]@xmath88 describe the lepton ( hadron ) tensor .
the kinematics is to be visualized as in fig .
[ fig - dkin ] . here
the virtual photon gives its 4-momentum ( @xmath89 ) to the deuteron , with these quantities determined by the initial electron energy and @xmath90 , the electron scattering angle .
@xmath91 is the angle between @xmath92 and the momentum @xmath93 of the outgoing proton .
( here and in what follows , unless otherwise stated , we work in the c.m
. frame of the final proton - neutron pair .
if necessary , the superscript lab is used to denote quantities in the lab . frame . )
meanwhile , @xmath94 is the angle between the scattering plane containing the two electron momentum vectors and the plane formed by the outgoing proton and neutron . finally , in fig .
[ fig - dkin ] we have defined @xmath92 to be along the z - axis . hence @xmath95 @xmath96 .
meanwhile , @xmath97where @xmath98 is the fine structure constant , @xmath99 is the absolute value of the incoming ( outgoing ) electron 3-momentum in the laboratory frame , and @xmath100 is the 4-momentum - squared of the virtual photon . in the ( final - state ) c.m .
frame we have @xmath101here @xmath102 represents the rest mass of the nucleon ( deuteron ) and @xmath103 mev is the deuteron binding energy . from energy conservation : @xmath104 .
the quantities in the lab .
frame can be easily related to those in c.m .
frame by a lorentz boost @xcite by an amount @xmath105 i.e. , @xmath106 ) , which is thus rewritten as@xmath107where the lepton tensor@xmath108and @xmath109 is defined as @xcite:@xmath110 in eq .
( [ eq:6.4 ] ) @xmath111 is the nn final state with the total spin quantum number @xmath112 and its projection on the z - axis , @xmath113 , both specified , @xmath93 represents that the final proton has 3-momentum @xmath93 , and @xmath114 labels the polarization index of the virtual photon .
we consider only @xmath115 here .
the deuteron state @xmath116 has total angular momentum 1 , and @xmath117 labels the z - projection of its total angular momentum . the angular dependence of eq .
( [ eq:6.4 ] ) can be separated into two parts , i.e. , @xmath118the longitudinal structure function @xmath29 is obtained from the @xmath91-dependent part of @xmath119 , i.e.,@xmath120 and so @xmath121 , with the proportionality determined solely by kinematic factors .
from eqs . ( [ eq:6.4 ] ) and ( [ eq:6.5a ] )
, one sees that to obtain @xmath122 the matrix element @xmath123 needs to be evaluated , which , up to @xmath61 , can be represented by @xmath124with @xmath56 given by eqs .
( [ eq : j0s ] ) and ( [ eq : j0v ] ) , and @xmath125 and @xmath126 the nn t - matrix and free green s function . here
we have used the @xmath127 to represent the fact that the final - state wavefunction is isospin dependent and introduced a @xmath128 in the kets and the @xmath127 in the two bras on the right - hand side to indicate the isospin of those states . the first term on the right - hand side of eq .
( [ eq:6.6a ] ) is the plane - wave impulse approximation ( pwia ) , and the second term is the final - state interaction ( fsi ) .
the dynamics of the pwia part can be described by fig .
[ dd ] : there the final - state proton ( neutron ) has 3-momentum @xmath129 in the final c.m .
frame , while before the proton ( neutron ) is struck , it has 3-momentum @xmath130 @xmath131 in the deuteron s c.m . frame . by inserting a complete set of isospin states and using the identity : @xmath132 .\end{aligned}\]]we
obtain @xmath133 \langle \mathbf{p}sm_{s}t|j_{0}(\mathbf{q})|m_{j}0\rangle , \label{eq:6.10aa}\end{aligned}\]]with @xmath134 and @xmath135 denotes the principal value .
note that in eq .
( [ eq:6.6a ] ) the magnitude of @xmath93 is restricted to be that of the proton in the final - state c.m .
frame , while in eq .
( [ eq:6.10aa ] ) @xmath136 is the integration variable .
note also that in eq . ( [ eq:6.10aa ] ) , and in what follows , we have dropped the @xmath137 label , since all states have @xmath138 : there are no interactions present that change this quantum number . the matrix element @xmath139 needs to be evaluated so that we can calculate eq .
( eq:6.6a ) .
the matrix element can be evaluated as @xmath140where the @xmath72 sign applies to the @xmath141 case , and the second factor on each line is now purely a matrix element in isospin space . here
@xmath142 is the momentum of particle 1(2 ) in the final c.m .
note that particle 1(2 ) can be either a proton or neutron . substituting eqs .
( [ form ] ) and ( [ eq:6.13 ] ) into eq .
( eq:6.6a ) and eq .
( [ eq:6.10aa ] ) , we get the final expression @xmath143 \notag \\ & & \quad \times \left [ \langle \mathbf{p}-\frac{\mathbf{q}}{2}sm_{s}|m_{j}\rangle + ( -1)^t \langle -\mathbf{p}-\frac{\mathbf{q}}{2}sm_{s}|m_{j}\rangle\right](g_{e}^{(p)}(\mathbf{q}^{2 } ) + ( -1)^t g_{e}^{(n)}(\mathbf{q}^{2 } ) ) . \label{eq:6.14}\end{aligned}\ ] ] note that the final - state interaction piece in eq .
( [ eq:6.14 ] ) itself has two parts : one for @xmath144 and one for @xmath145 .
each part consists of a t - matrix , the free green s function and the deuteron wave function with the nucleon form factors .
here we need to integrate over @xmath136 , thus the deuteron wave function is multiplied by the half - shell t - matrix @xmath146(@xmath147 ) . in principle , arbitrarily high values of @xmath148 contribute to the integration which yields the electrodisintegration amplitude .
physically , this means that the virtual photon can strike one of the nucleons in a state with arbitrarily large 3-momentum @xmath149 ( the other nucleon will have momentum @xmath150 in the final proton - neutron c.m .
the two nucleons then exchange momentum to reach their final state through the fsi .
this is in contrast with the pwia , where , to reach a given final state , the virtual photon must strike the nucleon at a specific momentum .
however , in practice , the high-@xmath148 component of the deuteron wave function is small , so the high - momentum part of the fsi integral will be suppressed . at this point , we have an expression for the sum of a plane - wave - impulse - approximation piece and the final - state interaction in terms of the 3-momentum of the measured proton @xmath93 .
the next step is to express eq .
( [ eq:6.14 ] ) in terms of partial waves .
first , we perform the partial - wave decomposition of the pwia part of eq .
( [ eq:6.14 ] ) . to do this we insert @xmath151into @xmath152 , where @xmath153 .
note that the normalization adopted here is@xmath154and hence @xmath155meanwhile , the deuteron wave function appears as matrix elements @xmath156 where the s- ( @xmath35 ) and d- ( @xmath157 ) wave components of the deuteron wave function satisfy@xmath158=1.\ ] ] the momentum - space wave functions we employ are obtained using the methods discussed in ref.@xcite , and are shown , together with those of the cd - bonn potential , in fig .
[ drnlow ] .
( p ) is the @xmath159s@xmath160 wave function and @xmath161(p ) denotes the @xmath159d@xmath160 wave function .
these wave functions are obtained from the dimensional - regularization ( dr ) two - pion - exchange potential ( tpe ) up to next - to leading order ( nlo ) , i.e. , @xmath162 ( black solid line ) , dr tpe up to next - to - next - to leading order ( nnlo ) , i.e. , @xmath163 ( green dashed line ) , and the spectral - function - regularization ( sfr ) tpe up to nnlo ( blue dash - dotted line ) , where the cutoff in the lippmann - schwinger equation ( lse ) is @xmath164 mev , and the sfr intrinsic cutoff is @xmath165 mev .
the red dots indicate the corresponding wave functions obtained from the cd - bonn potential .
, width=529 ] putting this all together and using the plane - wave expansion formula ( with s=1 for deuteron ) @xmath166 the two contributions to the pwia matrix element become : @xmath167 here @xmath168 is the clebsch - gordan coefficient and @xmath169 is the angle between @xmath170 and @xmath171 .
one can follow the same logic as for the pwia piece to perform the partial - wave decomposition of the fsi term .
note that the total spin ( @xmath112 ) and isospin ( @xmath127 ) are conserved in the nn interaction .
thus @xmath172\langle \mathbf{p}sm_{s}^{\prime } t|j_0(\mathbf{q})|m_{j } 0\rangle . \notag \\\end{aligned}\ ] ] the matrix element @xmath173 is evaluated via the first line of eq .
( eq:6.14 ) and eq .
( [ eq:6.20 ] ) .
the 3-dimensional t - matrix @xmath174 can be constructed from @xmath175 .
since the deuteron has @xmath176 , we need only @xmath177 , for which : @xmath178 upon insertion of these results into eq .
( [ eq : fsime ] ) we find that we need to perform the angular part of the integral in the following form@xmath179here @xmath180 is the solid angle between @xmath170 and @xmath181 .
we now denote the angle between @xmath170 and @xmath182 as ( @xmath183 ) , and@xmath184 , \label{eq:4}\]]with @xmath185 the solid angle between @xmath186 and @xmath187 . similarly , the angle between @xmath170 and @xmath188 , which we denote as ( @xmath189 ) , is@xmath190 .
\label{eq:5}\ ] ] thus , to evaluate @xmath191 we need to do the integral over the angles @xmath91 and @xmath94 .
the integral over @xmath94 can be reduced by taking advantage of the property of spherical harmonics:@xmath192 with the aid of eq .
( [ eq:6 ] ) , we can then perform the numerical integration over @xmath91 and @xmath10 to obtain the final - state interaction contribution to the longitudinal response function . in doing this
it is clearly important to have a description of the nn interaction that agrees with data for nn final - state energies of interest .
in fact , the lecs of the nn t - matrix we adopted in our calculation of @xmath1 are those which generate phase shifts that agree with the nijmegen phase - shift analysis @xcite for @xmath193 mev ( as can be seen in figs .
[ figfullbest ] and [ fig - fig200 ] ) .
this kinematics corresponds to @xmath194 mev .
, with the latter chosen to be the sfr tpe up to nnlo ( with intrinsic cutoff @xmath165 mev ) . here the generalized scattering lengths @xmath195 are adjusted to give the best fit in the region @xmath196 mev .
the nijmegen phase - shift analysis nnonline , st93 is indicated by the open diamonds . this graph is adapted from our previous publication @xcite.,width=529 ] s@xmath197d@xmath160 phase shifts as a function of the laboratory kinetic energy for different cutoffs @xmath57 ranging from 0.6 to 1 gev .
the potentials employed are the sfr nnlo ( with intrinsic cutoff @xmath165 mev ) .
the values of the nijmegen phase - shifts @xcite are indicated by the open triangles .
, width=377 ]
in this section we present our results for the longitudinal response function of deuteron electro - disintegration . for the nucleon form factors we adopt the results listed in ref .
@xcite . for electro - disintegration
, one needs to specify two kinematic variables , e.g. , ( @xmath89 ) to describe the whole process .
we adopted the following kinematic variables in order to compare our results with those obtained with the bonn potential in ref .
@xcite : first , the final energy of the proton - neutron system , which hereafter is labeled @xmath198 ( previously it was denoted @xmath199 ) i.e.,@xmath200second , the 3-momentum of the virtual photon ( also in the system s final c.m .
frame ) @xmath201 . with
@xmath202 and @xmath201 specified , the energy of the virtual photon can be calculated to be@xmath203 the experimental data of refs .
@xcite are presented in terms of the lab .
frame value of @xmath75 and the value of the missing momentum " , @xmath204 with @xmath205 the momentum of the detected proton .
all these data were taken in kinematics such that @xmath205 and @xmath206 are aligned , and so @xmath207 . from this information , and knowledge of the virtual - photon energy , @xmath208
, we can compute the kinetic energy of the np pair in the lab frame in two different ways : @xmath209 lorentz transformation of this quantity to the cm frame according to @xmath210 with @xmath211 given by eq .
( [ eq : beta ] ) , yields the @xmath202 which we quote in our results .
alternatively , @xmath212 can be obtained by energy conservation , applied in the cm frame : @xmath213 where @xmath214 and @xmath215 are obtained from @xmath208 and @xmath206 using eq .
( [ eq : lt ] ) . before presenting the results
, we introduce one kinematics which is of particular interest : the so - called quasi - free ridge .
the quasi - free ridge occurs when @xmath216 .
physically , this means that the virtual photon hits one of the nucleons and gives just enough 3-momentum to put it on - mass - shell .
the other nucleon remains at rest in the laboratory frame .
this occurs when : @xmath217consequently , on the quasi - free ridge @xmath202 ( in mev)@xmath218 ( in @xmath5 ) .
we now present our results .
first , we adopt the nn t - matrix and deuteron wave function generated with spectral - function regularization ( sfr ) applied to the two - pion - exchange potential up to nnlo , with the sfr cutoff @xmath219 set to @xmath220 mev .
[ figex ] shows the longitudinal structure function @xmath29 versus angle @xmath91 , i.e. , the angle between @xmath93 and @xmath221 for @xmath222 mev and @xmath223 @xmath5 .
the @xmath0eft pwia result is denoted by the red dash - double - dotted line , with error bars indicating the effect of varying the cutoff in the lippmann - schwinger equation ( lse ) from @xmath224 mev .
it is obtained using the full deuteron wave function obtained from the sfr tpe potential , evaluated at the pertinent three - momentum , see eq .
( [ eq:6.20 ] ) . for the final - state interaction ( fsi ) , one needs to sum over partial - waves in order to obtain the 3-dimensional t - matrix .
we have summed over partial - waves up to @xmath225 and the results are denoted as blue dashed ( @xmath226 ) , green dash - dotted ( @xmath227 up to @xmath228 ) and black double - dash - dotted ( @xmath227 up to @xmath229 ) line in fig .
[ figex ] , there is no contact term associated with the long range part of the tpe in the standard weinberg s power counting .
here we adopt the born approximation instead of iterating the sfr tpe in the lse . ] .
in general , our results converge once we include partial waves with @xmath230 in our calculation of the fsi .
here and below we compare our calculations to the calculations of arenhvel and collaborators @xcite .
these calculations are done using the bonn - b potential @xcite and include pwia and fsi pieces .
this allows a direct comparison with the pwia and fsi - included @xmath0eft calculations that are the subject of this work .
the pwia result obtained with the bonn - potential wave function is indicated by the red dotted line .
when both @xmath202 and @xmath201 are low , i.e. , @xmath222 mev and @xmath231 @xmath5 , the results agree very well .
as @xmath201 becomes larger , the pwia results obtained from the two potentials start to deviate from each other . as a function of @xmath232 for @xmath222 mev and four different values of @xmath201 .
the error bars are obtained from varying the cutoff @xmath224 mev in the lse .
the convergence of the calculation as additional partial waves are included in the computation of the fsi is also shown .
legend as indicated . here
the intrinsic cutoff of sfr tpe is @xmath165 mev.,width=529 ] we first discuss the quasi - free ridge case shown in fig . [ figex ] , i.e. , @xmath222 mev and @xmath233 @xmath5 . out of all four panels in fig .
[ figex ] , @xmath29 receives the least correction from the fsi here , and , as shown in the other three panels , the further away we move from the quasi - free ridge the larger the fsi correction becomes .
this can be explained easily by the fact that , at the quasi - free ridge , both nucleons in the deuteron are on the mass shell after being struck by the virtual photon , and no fsi is needed in order to make the final - state particles real . on the other hand ,
as we move further kinematically from the quasi - free ridge , the fsi must provide a larger energy - momentum transfer to make the proton and neutron become real particles in the final state , and so it becomes more important .
moreover , for this particular np final - state energy , the quasi - free ridge is the last @xmath234 where all the pwia and fsi results from the two potentials agree .
as we increase @xmath201 to 4 @xmath5 and above , both our pwia and fsi results start to diverge away from the corresponding bonn potential results .
the error bars also grow quite significantly for @xmath235 @xmath5 , i.e. @xmath236 mev , particularly in the fsi . there , where both the deuteron wave functions and the nn t - matrix enter the calculation ,
the results become highly cutoff - dependent .
we now assess how this uncertainty in the @xmath0eft @xmath29 prediction comes from the uncertainty of the @xmath0eft deuteron wave function and nn t - matrix .
let s first look at the quasi - free ridge . from eq .
( [ eq:6.00 ] ) and eq .
( [ eq:6.29 ] ) , we infer that @xmath237at the quasi - free ridge , where the dominant element in the calculation is the deuteron wave function @xmath238 . at the quasi - free ridge ,
the value of the wave - function argument achieves its lowest possible value ( for a given @xmath202 ) : it is @xmath128 ( for @xmath239 ) increasing to @xmath240 ( for @xmath241 ) .
[ drnlow ] shows that the deuteron wave function @xmath242 given by both the sfr and dr tpe up to nnlo agrees with the one given by the bonn potential at least up to wave function arguments @xmath243 mev , and dies off quickly at higher momentum .
since the high - momentum component of the wavefunction is almost zero wavefunction in momentum - space @xmath244 dies off at a higher momentum , i.e. , @xmath245 mev .
however , it is at least 10 times smaller in amplitude than the @xmath246 wavefunction @xmath247 . ]
, this suggests that @xmath29 calculated from these two potentials should agree with each other at the quasi - free ridge .
in fact , as shown in fig . [ quasi ] , in quasi - free kinematics the @xmath29 given by the sfr tpe up to nnlo does agree with those given by the bonn potential all the way up to @xmath248 mev . as a function of @xmath232 in various kinematics on or around the quasi - free ridge " . here
the blue line represents the result given by the bonn potential with fsi included , and the black double - dash - dotted line denotes the same results given by the sfr tpe up to nnlo .
the error bars are obtained from varying the cutoff @xmath224 mev in the lse .
the sfr intrinsic cutoff is @xmath165 mev.,width=529 ] to see where the two wave functions start to disagree , we use fig .
plot_en10 as an example ( @xmath222 mev again , now at more @xmath234 values ) . for @xmath249 @xmath5
, there is no significant difference between results obtained by @xmath0eft and the bonn potential . at @xmath250 @xmath5 ,
the shift due to fsi is roughly the same ( @xmath251 fm at @xmath239 ) for both the nnlo sfr tpe and the bonn potential . in other words ,
the fsi has almost the same effect for the two potentials , and the disagreement in the total @xmath1 comes ( mostly ) from the pwia part .
the @xmath252 difference in the pwia amplitude originates from the difference between the deuteron wave functions generated by nnlo sfr tpe and the bonn potentials at around @xmath253 mev and is not significantly enlarged by the fsi piece where the deuteron wave function is integrated against the nn t - matrix . as a function of @xmath232 for @xmath222 mev , and @xmath254 ranging from 0.5 @xmath5 to 16 @xmath5 . here
the red dash - double - dotted ( dotted ) line represents the pwia results given by the sfr tpe up to nnlo ( bonn ) , and the black double - dash - dotted ( solid ) line denotes the fsi results given by the sfr tpe up to nnlo ( bonn ) .
the error bars are obtained from varying the cutoff @xmath224 mev in the lse .
the intrinsic cutoff is @xmath165 mev for the sfr tpe up to nnlo.,width=529 ] on the other hand , as we increase @xmath201 to @xmath255 @xmath5 ( with @xmath222 mev ) @xmath29 given by the two different potentials starts to have a larger difference in the fsi than in the pwia see the lower panels of fig .
[ plot_en10 ] .
although this final - state energy is well within the range that is fit by our nn potential , one must remember that the deuteron wave function that enters the fsi integral is largest when @xmath256 , and the phase - shift data where we perform best fit up to @xmath257 mev only validates our computation of @xmath258 for @xmath259 mev and @xmath10 up to about @xmath260 mev .
we infer that it is important for @xmath258 to at least accurately describe data for the on - shell kinematics corresponding to both @xmath261 and @xmath262 .
if either of these is greater than @xmath260 mev , then the difference in the nn t - matrix generated by the sfr tpe up to nnlo and the bonn potential enters the fsi calculation in addition to any differences in @xmath263 .
as a function of @xmath232 for @xmath264 mev , and @xmath254 ranging from 0.5 @xmath5 to 16 @xmath5 .
legend as in fig .
[ plot_en10].,width=529 ] in fig .
[ plot_en30 ] we show a similar set of panels to those in fig .
plot_en10 , but at @xmath264 mev .
the agreement between @xmath0eft and bonn results is again quite good at low @xmath4 , although there is some disagreement at backward angles once fsi is included .
this trend in the final result for @xmath1 diminishes as we move towards the quasi - free ridge . at the quasi - free ridge
the cutoff variation of the @xmath0eft calculation is small smaller than for @xmath222 mev , because the fsi plays less of a role at this higher energy .
the agreement with the bonn potential is also quite good there . immediately above the quasi - free ridge
these features persist , until @xmath265 @xmath5 , at which point @xmath266 becomes large enough , and the fsi important enough , that agreement can not be maintained . as a function of @xmath232 for @xmath6 mev , and @xmath254 ranging from 0.5 @xmath5 to 16 @xmath5 .
legend as in fig .
[ plot_en10].,width=529 ] if we now examine @xmath6 mev we are already in a regime where the fsi is not trustworthy even if @xmath254 is low .
this is reflected in the failure of the fsi - included result to encompass the bonn - potential answer , even within error bars , at @xmath267 @xmath5 ( upper - left panel of fig .
[ plot_en60 ] ) .
but , already by @xmath268 @xmath5 , the bonn - potential and @xmath0eft predictions ( both with fsi included ) are within 10% of each other , with the difference entirely accounted for by the @xmath0eft result s variation with the lse cutoff @xmath33 .
as we move to the quasi - free ridge , and fsi becomes less important , this variation becomes less of a component of the full answer for @xmath1 . in consequence
there is a window , up to @xmath269 @xmath5 , where the cutoff dependence of the final prediction for @xmath1 is not sizable .
the agreement between @xmath0eft and the bonn potential for @xmath1 is also good through much of this range .
however as @xmath254 approaches 10 @xmath5 the bonn - potential and @xmath0eft pwia answers start to differ , especially at small angles .
since the fsi is small in both calculations in this range , that difference is not ameliorated in the full calculation .
we will now summarize the results of the calculations we carried out for @xmath270 mev and @xmath271 @xmath5 .
the results are generically similar to those displayed above , in that , if we define@xmath272then our results show that for the kinematic region @xmath273 mev and @xmath274 @xmath275 @xmath5 , the calculations using @xmath0eft up to nnlo have @xmath276% variation with respect to the cutoff in the lse . the @xmath0eft results and those found with the bonn potential also agree within 10% if @xmath274 @xmath277 @xmath5 , for @xmath202 at least as high as @xmath278 mev .
somewhat remarkably this agreement is possible even in cases where both @xmath202 and @xmath254 appear to be outside the range of validity of our calculation as long as we are close to the quasi - free ridge and so details of the fsi remain unimportant . on the other hand
, we emphasize that the @xmath0eft fsi is reliable for low @xmath266 and low @xmath202 , see , e.g. first two panels of fig
. plot_en10 .
indeed , in the first panel there one might be concerned about the accuracy of our description of the deuteron wave function .
the lecs in our chiral potential are obtained through the best fit to the nn phase shifts for @xmath196 mev . in general this
does not give the best possible deuteron wavefunction which is crucial for the pwia .
an alternative would be to perform the renormalization of the lecs in the @xmath279 channel so that some of the deuteron properties , e.g. , the binding energy , are very accurately reproduced .
we have verified that , by doing this , for @xmath274 @xmath275 @xmath5 the 10% uncertainty due to the variation of the cutoff in the lse can be reduced to 5% , but only for lower values of @xmath280 mev@xmath281 . aside from that
, it does not improve the discrepancy between the results generated by @xmath0eft and the bonn potential .
moreover , for @xmath282 mev , the variation of the results with respect to the cutoff becomes larger than before , due to the fact that the nn @xmath279 t - matrix now has worse convergence with respect to the cutoff in the higher energy / momentum region .
we conclude that the uncertainty of @xmath29 will remain roughly the same unless higher orders in the chiral potential are included .
.the first two columns give kinematics quoted in the relevant papers : the missing momentum , and the three - momentum transfer to the nucleus as measured in the lab . frame .
the next two columns give the ( c.m . frame ) quantities we employ for our calculation .
the fifth column shows results for @xmath283 were extracted from plots given in refs .
the first error is the statistical error , while the second is the systematic quoted in the publication .
the last column shows @xmath284 evaluated with the paris potential .
the units for the jordan et al .
@xmath1 measurement are fm . [ cols="<,^,^,^,^,^,^",options="header " , ] finally , we compare our results with data for the longitudinal structure function published in refs .
the kinematics and data for @xmath1 are listed in table [ tab - data ] .
the experiments of van der schaar et al .
@xcite and ducret et al .
@xcite give their data as ratios between the measured longitudinal response and that predicted by a paris - potential @xcite impulse - approximation calculation .
for those two experiments we have calculated @xmath285 at the pertinent momentum using the paris - potential deuteron wave - function parameterization of ref .
the result for @xmath1 given in table [ tab - comparison ] is then obtained by multiplying the final two columns in table [ tab - data ] together ( with the obvious exception of the single data point of jordan et al .
@xcite , where the publication gives @xmath1 directly ) .
table tab - comparison compares these experimental results for @xmath1 with those from our @xmath0eft calculations using the sfr nnlo potential and cutoffs of 0.6 to 1 gev .
in fact , we find significant sensitivity of both the paris ia and the @xmath0eft result to the value of @xmath202 chosen .
the value of @xmath202 listed here is obtained from the initial - state kinematics given in the publications , using the relativistic energy - momentum relation . using the value of @xmath202 found from the final - state kinematics to compute
@xmath285 produces results that differ by more than the quoted uncertainties on @xmath286 .
this makes us think that further details of the experiments , e.g. , acceptances , are needed in order to provide a completely meaningful comparison between theory and data
. nevertheless , some trends are already clear from the straightforward comparison between theory and data that can be made without detailed knowledge of the experiments . for the @xmath287 mev data ( ducret et al .
) the @xmath0eft calculations are in good agreement with the data , once the experiment s @xmath288% systematic is accounted for .
for example , multiplying the ducret et al .
data by 1.015 , we find that the @xmath0eft prediction is within 1.3 ( combined theory and statistical uncertainty of the measurement ) @xmath289 for all but the @xmath290 mev data point .
the agreement for @xmath291 mev is also very good even at the highest @xmath234 of almost 6 @xmath5 . for the rest of the ducret et al .
data set , taken at @xmath292 mev , the @xmath0eft calculation is systematically below the data , and this can not be attributed to normalization uncertainty at least not within the quoted systematic . similarly , the van der schaar et al .
data are ( with one exception ) all underpredicted by @xmath0eft at this order .
this problem persists even if the entire 7% systematic uncertainty quoted in ref .
@xcite is assigned to the experimental normalization .
the disagreement worsens as @xmath234 increases , with the only experimental point for @xmath293 @xmath5 agreeing with the @xmath0eft prediction while points in the range @xmath294 have @xmath0eft 1 - 3@xmath289 below the data ( depending on how the experimental normalization is treated ) .
the disagreement between @xmath0eft and data becomes dramatic at higher @xmath234 .
similar under - prediction also occurs when arenhvel s bonn - potential calculation is compared to these data , so the difficulty appears generic to calculations employing only the one - body operator for @xmath56
. it will be interesting to explore whether higher - order corrections to the nn charge operator in @xmath0eft can redress the difference seen here between theory and experiment .
it is also worth noting that the uncertainty due to cutoff variation in the @xmath0eft result is very small for most cases .
this leads us to question whether varying the cutoff between 600 mev and 1 gev adequately estimates the uncertainty in the @xmath0eft result due to higher - order effects .
we have computed the longitudinal structure function of the deuteron , @xmath29 , in @xmath7eft , including effects up to @xmath59 relative to leading order in the standard counting .
comparison with calculations of arenhvel _ et al .
_ , which use the bonn potential , indicates good agreement at low energies ( @xmath295 mev ) and for @xmath49 which are not very distant ( @xmath275 @xmath5 ) from the quasi - free ridge .
we also use the @xmath57 dependence of the @xmath7eft calculation as a diagnostic , to find kinematic regions where the final result for @xmath29 is sensitive to short - distance components in the evaluation of the final - state interaction .
if significant @xmath57 dependence is present it suggests sensitivity to such effects , which may mean that both the @xmath7eft calculation and the bonn - potential calculation do not capture the full dynamics present in @xmath296 . with this caveat in mind
, we find that our calculations are able to describe data from saclay , nikhef , and bates , on the longitudinal structure function , either within a more restricted kinematic region ( compared to the region where our results agree with bonn potential ) , or by allowing somewhat expanded combined ( statistical + systematic + theory ) error bars .
we also notice that both @xmath7eft and bonn potential give a similar trend of under - prediction of the experimental @xmath29 with increasing @xmath201 .
further studies are needed to understand the origin of this discrepancy .
in fact , the @xmath0eft calculation is in good agreement with the bonn potential and with the data to surprisingly high energies and @xmath4 as long as one stays on ( or near ) the quasi - free ridge .
this suggests that a different counting may be needed for this reaction , one where the expansion is not in the naive kinematic variables , @xmath297 and @xmath75 , but instead , perhaps where some account is taken of the dominance of the pwia part of the matrix element for sufficiently high energies ( @xmath298 mev ) and sufficiently low values of the missing momentum " @xmath299 . in the future
, the recent @xmath7eft calculations of elastic scattering on tri - nucleons @xcite could be extended to electrodisintegration , and compared with data on that reaction @xcite . however , a more obvious and immediately necessary next step is to extend this @xmath7eft calculation to other structure functions .
the three - current operator @xmath300 has also been computed to three orders relative to leading order @xcite , and comparison of @xmath301 , @xmath50 , and @xmath51 with data ( and bonn - potential calculations ) would be an important test of the domain of validity of the @xmath7eft expansion for that object . in closing , we reiterate that such a test can be rendered unambiguous because the results presented here show the regions in which the @xmath0eft expansion for the pwia and fsi pieces of the deuteron electrodisintegration process are under control .
the absence of any nn mechanisms in the charge operator up to the order considered makes the @xmath1 presented here a prediction once the nn potential is fixed by the fit to nn data .
the fact that we find good agreement with both @xmath1 data and other theories in a broad kinematic range provides further reassurance ( see also refs .
ph07,ko12 ) that @xmath0eft does a good job of describing deuteron structure for internal relative momenta @xmath302 gev .
we thank h. arenhvel for providing his bonn potential results for comparison .
yang thanks b. barrett , c. elster , s. fleming and u. van kolck for vaulable support .
this work is supported by the us nsf under grant phys-0854912 and us doe under contract no .
de - fg02 - 04er41338 and de - fg02 - 93er40756 .
99 s. weinberg , phys .
b * 251 * , 288 ( 1990 ) ; nucl .
b * 363 * , 3 ( 1991 ) .
e. epelbaum , h. -w .
hammer and u. -g .
meiner , rev .
* 81 * , 1773 ( 2009 ) .
e. epelbaum and u. -g .
meiner , ann .
nucl . part .
sci . * 62 * 159 - 185 ( 2012 ) .
s. klling , e. epelbaum , h. krebs and u .- g .
meiner , phys .
c * 80 * , 045502 ( 2009 ) .
s. klling , e. epelbaum , h. krebs and u .- g .
meiner , phys .
c * 84 * , 054008 ( 2011 ) .
s. pastore , r. schiavilla and j. l. goity , phys .
c * 78 * , 064002 ( 2008 ) .
s. pastore , l. girlanda , r. schiavilla , m. viviani and r. b. wiringa , phys .
c * 80 * , 034004 ( 2009 ) . s. pastore , l. girlanda , r. schiavilla and m. viviani , phys .
c * 84 * , 024001 ( 2011 ) .
m. piarulli , l. girlanda , l. e. marcucci , s. pastore , r. schiavilla and m. viviani , phys .
c * 87 * , 014006 ( 2013 ) .
d. r. phillips , j. phys .
g * 34 * , 365 - 388 , ( 2007 ) .
d. r. phillips and t. d. cohen , nucl .
a * 668 * , 45 ( 2000 ) .
m. walzl and u. g. meiner , phys .
b * 513 * , 37 ( 2001 ) .
s. klling , e. epelbaum and d. r. phillips , phys .
c * 86 * , 047001 ( 2012 ) .
l. girlanda , a. kievsky , l. e. marcucci , s. pastore , r. schiavilla and m. viviani , phys .
* 105 * , 232502 ( 2010 ) .
s. pastore , s. c. pieper , r. schiavilla and r. b. wiringa , phys .
c * 87 * , 035503 ( 2013 ) .
h. arenhvel , w. leidemann and e. tomusiak , phys .
c * 46 * , 455 ( 1992 ) .
h. arenhvel , w. leidemann and e. l. tomusiak , few body syst .
* 28 * , 147 ( 2000 ) .
h. arenhvel , w. leidemann and e. l. tomusiak , eur .
j. a * 14 * , 491 ( 2002 ) .
h. arenhvel , w. leidemann and e. l. tomusiak , eur .
phys . j. a * 23 * , 147 ( 2005 ) .
yang , ch .
elster and d. r. phillips phys .
c * 77 * , 014002 ( 2008 ) .
d. eiras and j. soto , eur . phys .
j. a * 17 * , 89 ( 2003 ) .
a. nogga , r.g.e .
timmermans and u. van kolck , phys .
c * 72 * , 054006 ( 2005 ) .
m. c. birse , phys .
c * 74 * , 014003 ( 2006 ) .
m. pavon valderrama and e. ruiz arriola , phys .
c * 74 * , 064004 ( 2006 ) [ erratum - ibid .
c * 75 * , 059905 ( 2007 ) ] .
yang , ch .
elster and d. r. phillips , phys .
c * 80 * , 034002 ( 2009 ) .
d. r. phillips and t. d. cohen , phys .
b * 390 * , 7 ( 1997 ) .
k. a. scaldeferri , d. r. phillips , c. w. kao and t. d. cohen , phys .
c * 56 * , 679 ( 1997 ) . c. ordonez , l. ray and u. van kolck , phys .
c * 53 * , 2086 ( 1996 ) .
d. r. entem and r. machleidt , phys .
c * 68 * , 041001(r ) ( 2003 ) .
e. epelbaum , w. glckle and u - g .
meiner , nucl .
a * 747 * , 362 ( 2005 ) .
g. p. lepage , arxiv : nucl - th/9706029 .
e. epelbaum and ulf - g .
meiner , arxiv : nucl - th/0609037 .
m. c. birse , pos cd * 09 * , 078 ( 2009 ) [ arxiv:0909.4641 [ nucl - th ] ] .
d. r. phillips , pos cd * 12 * , 013 ( 2012 ) [ arxiv:1302.5959 [ nucl - th ] ] .
m. van der schaar , h. arenhovel , t. s. bauer , h. p. blok , h. j. bulten , m. daman , r. ent and e. hummel _ et al .
_ , phys .
* 66 * , 2855 ( 1991 ) .
j. e. ducret , m. bernheim , j. f. danel , l. lakehal - ayat , j. m. le goff , a. magnon , c. marchand and j. morgenstern _ et al .
_ , phys .
c * 49 * , 1783 ( 1994 ) .
d. jordan , t. mcilvain , r. alarcon , r. beck , w. bertozzi , v. bhushan , w. boeglin and j. p. chen _ et al .
_ , phys .
* 76 * , 1579 ( 1996 ) .
m. bernheim _ et al .
_ , nucl .
a * 365 * , 349 ( 1981 ) .
s. turck - chieze , p. barreau , m. bernheim , p. bradu , z. e. meziani , j. morgenstern , a. bussiere and g. p. capitani _ et al .
_ , phys .
b * 142 * , 145 ( 1984 ) .
h. breuker , v. burkert , e. ehses , u. hartfiel , g. knop , g. krosen , j. langen and m. leenen _ et al .
_ , nucl .
a * 455 * , 641 ( 1986 ) .
k. i. blomqvist , w. u. boeglin , r. bohm , m. distler , r. edelhoff , i. ewald , r. florizone and j. friedrich _ et al .
_ , phys .
b * 424 * , 33 ( 1998 ) .
n. ryezayeva _ et al .
_ , phys .
* 100 * , 172501 ( 2008 ) .
m. van der schaar , h. arenhovel , h. p. blok , h. j. bulten , e. hummel , e. jans , l. lapikas and g. van der steenhoven _
et al . _ ,
lett . * 68 * , 776 ( 1992 ) .
t. tamae , h. kawahara , a. tanaka , m. nomura , k. namai , m. sugawara , y. kawazoe and h. tsubota _ et al .
_ , phys .
lett . * 59 * , 2919 ( 1987 ) . f. frommberger , d. durek , r. gothe , g. happe , c. hueffer , d. jakob , m. jaschke and g. knop _ et al .
_ , phys .
b * 339 * , 17 ( 1994 ) .
kasdorp _ et al .
_ , phys .
b * 393 * , 42 ( 1997 ) .
z. l. zhou _ et al .
_ [ mit - bates oops collaboration ] , phys .
lett . * 87 * , 172301 ( 2001 ) .
w. u. boeglin , h. arenhovel , k. i. blomqvist , r. bohm , m. distler , r. edelhoff , i. ewald and r. florizone _ et al .
_ , phys .
c * 78 * , 054001 ( 2008 ) .
e. ulmer , k. a. aniol , h. arenhovel , j. p. chen , e. chudakov , d. crovelli , j. m. finn and k. g. fissum _ et al .
_ , phys .
* 89 * , 062301 ( 2002 ) .
w. u. boeglin _ et al . _ [ hall a collaboration ] , phys .
lett . * 107 * , 262501 ( 2011 ) .
s. jeschonnek and j. w. van orden , phys .
c * 78 * , 014007 ( 2008 ) .
s. jeschonnek and j. w. van orden , phys .
c * 81 * , 014008 ( 2010 ) .
d. rozpedzik , j. golak , s. klling , e. epelbaum , r. skibinski , h. witala and h. krebs , phys .
c * 83 * , 064004 ( 2011 ) .
s. christlmeier and h. w. griehammer , phys .
c * 77 * , 064001 ( 2008 ) .
r. machleidt , k. holinde and c. elster , phys .
rept . * 149 * , 1 ( 1987 ) .
m. lacombe , b. loiseau , j. m. richard , r. vinh mau , j. cote , p. pires and r. de tourreil , phys .
c * 21 * , 861 ( 1980 ) .
m. lacombe , b. loiseau , r. vinh mau , j. cote , p. pires and r. de tourreil , phys .
b * 101 * , 139 ( 1981 ) .
keizer , et al .
b * 157 * 255 ( 1985 ) ; p.h.m . keizer , et al .
b * 197 * 29 ( 1987 ) ; p.h.m . keizer , ph.d .
thesis , amsterdam 1986 . | we use chiral effective field theory ( @xmath0eft ) to make predictions for the longitudinal electromagnetic response function of the deuteron , @xmath1 , which is measured in @xmath2 reactions . in this case
the impulse approximation gives the full @xmath0eft result up to corrections that are of @xmath3 relative to leading order . by varying the cutoff in the @xmath0eft calculation between 0.6 and 1 gev
we conclude that the calculation is accurate to better than 10% for values of @xmath4 within 4 @xmath5 of the quasi - free peak , up to final - state energies @xmath6 mev . in these regions
@xmath0eft is in reasonable agreement with predictions for @xmath1 obtained using the bonn potential .
we also find good agreement with existing experimental data on @xmath1 , albeit in a more restricted kinematic domain . |
the traditional _ trapping _ problem involves diffusive ( brownian ) particles ( @xmath0 ) that wander in a medium doped with static traps ( @xmath1 ) and disappear when they meet @xcite . in the traditional _ target _ problem @xcite , on the other hand , one has static @xmath0 particles and mobile traps . both of these problems are described by the reaction " @xmath2 , but in one case the @xmath0 s move and the @xmath1 s stand still , while in the other the @xmath1 s move while the @xmath0 s are stationary .
both of these problems have a long and active history in the literature .
they not only represent experimentally observable phenomena , but they have served as a testbed for theoretical and numerical studies and as a starting point for the formulation of models for more complex systems that have only recently been successfully solved analytically . for example , the survival probability of an @xmath0 particle in a medium of @xmath1 particles when _ both _ species are diffusive , first investigated numerically in the seminal work of toussaint and wilczek @xcite , was only partially solved analytically @xcite until the recent full ( asymptotic ) solution in one dimension @xcite .
these results have also recently been generalized to subdiffusive species @xcite .
the survival probability of @xmath0 particles in the reactions @xmath3 and @xmath4 in one dimension when @xmath0 is mobile is also of relatively recent vintage in the history of such analytic solutions @xcite .
the purpose of this paper is to extend the one - dimensional target problem calculations for both diffusive and subdiffusive traps to the case of traps that themselves disappear in time according to some survival probability function of their own ( e.g. , exponential or power law ) .
the decay of the moving traps with time of course increases the survival probability of the stationary target , and the interesting questions concern the interplay of the time dependences of the movement and decay of the traps .
a related problem was considered in @xcite , where diffusive particles @xmath0 and traps @xmath1 and @xmath5 undergoing the explicit reactions ( a ) @xmath6 , @xmath7 , and ( b ) @xmath6 , @xmath8 were considered using entirely different methods .
our methods are equally applicable to trap densities that increase with time , but this problem is less interesting because it necessarily leads to the eventual demise of the target .
a common characterization of the diffusive motion of a particle is through its mean square displacement for large @xmath9 , @xmath10 here @xmath11 is the ( generalized ) diffusion constant , and @xmath12 is the exponent that characterizes normal ( @xmath13 ) or anomalous ( @xmath14 ) diffusion . in particular , the process is sudiffusive when @xmath15 .
subiffusive processes are ubiquitous in nature @xcite , and are particularly useful for understanding transport in complex systems @xcite .
the problem considered in this paper is a special case of a broad class of reaction-_subdiffusion _ processes that have been studied over the past decades .
one approach that has been used to study these processes is based on the continuous time random walk ( ctrw ) theory with waiting - time distributions between steps that have broad long - time tails and consequently infinite moments , @xmath16 for @xmath17 with @xmath15 .
another approach is based on the fractional diffusion equation , which describes the evolution of the probability density @xmath18 of finding the particle at position @xmath19 at time @xmath9 by means of the fractional partial differential equation ( in one dimension ) @xcite @xmath20 where @xmath21 is the riemann - liouville operator , @xmath22 in this paper we study the one - dimensional target problem for a static particle @xmath0 subject to attack by diffusive or subdiffusive traps @xmath1 that may die before reaching the target @xmath0 @xcite . for this purpose
, we generalize the ideas of bray and blythe @xcite , and of our own work @xcite based on a fractional diffusion equation approach .
while recent work shows that a simple generalization of reaction - diffusion to reaction - subdiffusion equations in which the reaction and subdiffusion terms are assumed to enter additively is not valid in some cases @xcite , this is not a difficulty in our particular application .
the difficulties do not arise when the reaction process can be translated into a static boundary value problem , which is the case for the target ( as well as the trapping ) problem @xcite . in some cases ,
asymptotic anomalous diffusion behavior can be found from corresponding results for normal diffusion via the simple replacement of @xmath9 by @xmath23 .
this can be understood from a ctrw perpective because the average number of jumps @xmath24 made by a subdiffusive walker up to time @xmath9 scales as @xmath25 and , in many instances the number of jumps is the relevant factor that explains the behavior of the system .
the simple replacement result is evidence of `` subordination '' ( see secs . 5 and 7.2 of @xcite ) .
however , there are other instances where the behavior of subdiffusive systems can not be found in this way .
a simple example is the survival probability of subdiffusive particles in the trapping problem ( see sec .
5 of @xcite ) . in particular , for systems where competing processes ( motion toward target and death ) occur according to different temporal rules , such a replacement becomes ambiguous .
this is the case for the problem considered here .
while our analytic results are based on the fractional diffusion equation formalism , our numerical simulations are based on a ctrw algorithm .
these two renditions of the problem are expected to differ if trapping events are likely in a small number of steps , that is , if the initial density of traps is too high . on the other hand ,
if the initial trap density is too low , then the simulations to produce valid statistics would take inordinately long because trapping events are rare and because the system has to be sufficiently large to include many particles .
we note this as a caveat for our subsequent comparisons . in sec .
[ survival1 ] we present an integral equation for the survival probability , which we reduce to quadrature in sec . [ survival2 ] .
the resulting integral is explicitly evaluated for exponentially decaying trap densities ( including a stretched exponential decay ) , as well as for trap densities that decay as a power law .
not surprisingly , we find that a sufficiently rapid decay of the trap density leads to a finite asymptotic survival probability of the target . comparisons of our results with numerical simulations are also shown in this section . a summary and some conclusions
are presented in sec .
[ conclusions ] .
we consider a finite interval @xmath26 containing @xmath27 mobile traps @xmath1 of constant density @xmath28 initially distributed at random , and a single immobile @xmath0 particle at the origin . following the approach of bray et al .
@xcite for diffusive traps and our generalization of this approach to the subdiffusive case @xcite , we write the survival probability of @xmath0 as @xmath29 , where @xmath30 is to be determined . to find this function ,
one calculates in two ways the probability density to find a @xmath1 particle at the origin at time @xmath9 , @xmath31 that the left side is this probability density is obvious . on the right side
one has the renewal theory expression where @xmath32 is the probability that a @xmath1 particle intersected @xmath0 in the time interval @xmath33 for the first time , and the propagator @xmath34 is the probability density for this particular @xmath1 to be at the origin at time @xmath9 . in one dimension
it is given by @xcite @xmath35 \nonumber\\ & = & \frac{1}{\sqrt{4k_\gamma t^\gamma } \gamma\left(1-\frac{\gamma}{2}\right)},\end{aligned}\ ] ] where @xmath36 is fox s @xmath37-function , whose value at the given arguments we have used to write the last equality . in a different context than the target problem , bray et al .
@xcite generalized their approach to a time - dependent density @xmath38 of @xmath1 .
they argue that in place of eq .
( [ fundamental ] ) one now has @xmath39 that is , @xmath40 this is the basic equation to be considered in this paper .
to calculate the survival probability of particle @xmath0 , we rewrite eq .
( [ ecuintegral1 ] ) explicitly as @xmath41 where we have multiplied both sides by @xmath42 , and where we have introduced @xmath43 equation ( [ ecuabel ] ) is an equation of abel of the first kind @xcite , @xmath44 with @xmath45 .
the solution of this classic equation is well known ( see sec . 12 in @xcite or eqs .
( 2.5a ) and ( 2.5b ) in @xcite ) , @xmath46 here , as earlier , @xmath47 is the riemann - liouville fractional derivative . in our case
@xmath45 is constant , so that @xmath48 it then follows that @xmath49 which provides a general solution to our problem for any @xmath38 .
while eq .
( [ musolinte ] ) applies to trap densities that grow or that decrease or even oscillate in time , the case of evanescent traps is the more interesting and the one we choose to focus on . as a reminder
, we note that for traps of a constant density @xmath50 the survival probability is given by @xmath51 as a benchmark , we show in fig . [
fig : mu0_ga0.5_rho0.01_a0 ] a typical comparison of this result with simulation results .
the agreement is clearly good , although a lower initial density run for a longer time would lead to even better agreement .
vs @xmath9 for non - evanescent traps as given in eq .
( [ nonevan ] ) ( solid line ) and simulations ( symbols along with error bars ) .
parameter values are @xmath52 , @xmath53 , @xmath54 .
[ fig : mu0_ga0.5_rho0.01_a0 ] ] suppose that the traps have a finite lifetime @xmath55 and decay exponentially , as in a unimolecular reaction , @xmath56 .
the integral in eq .
( [ musolinte ] ) immediately leads to the solution @xmath57 where @xmath58 is an incomplete gamma function , and @xmath59 when @xmath13 , i.e. , when the traps are diffusive , this reduces to @xmath60 for arbitrary @xmath61 , the survival probability of the target in the presence of the subdiffusive traps with finite lifetime thus is @xmath62 . \label{forfig3}\ ] ] the interesting result here is that the funtion @xmath30 _ goes to the constant _ @xmath63 and not to infinity as @xmath64 .
therefore the survival probability does not vanish with increasing time , @xmath65.\ ] ] we note that @xmath66 is a characteristic distance that measures the root mean square displacement of the traps during their decay time @xmath55 .
therefore @xmath67 is the ratio of this average displacement to the average initial distance @xmath68 between traps .
this finite asymptotic survival probability , @xmath69 , displays reasonable qualitative features : it increases with decreasing trap lifetime @xmath55 , and it decreases with increasing initial trap density @xmath70 . that there is a finite asymptotic survival probability reflects the fact that if the traps disappear sufficiently rapidly ( which they do if they disappear exponentially while the traps move diffusively or subdiffusively ) ,
then many traps disappear before they can reach the particle , and there is a finite probability that the particle remains forever safe . "
the next two figures show the comparison of simulation results with our analytic outcome .
first , in fig .
[ fig : mu0_rho0.1_a0.01 ] we illustrate our earlier caveat , that agreement can not be expected if the initial density of traps is too high and the extinction rate of the traps is large , and that the agreement improves with lower initial density .
the disagreement is clear and can be traced exactly to the early time trapping events that cumulatively affect the survival probability .
[ fig : mu0_ga0.5_rho0.01_a1e-8 ] shows typical results for the lower initial density of traps and a more slowly decaying trap density , where the agreement between analytic results and simulations is clearly very good . vs @xmath9 for exponentially evanescent traps .
solid line : eq .
( [ forfig2 ] ) .
squares : simulation results for a high initial density @xmath71 .
@xmath72 s : simulation results for a lower initial density @xmath53 .
other parameter values are @xmath13 , @xmath73 , and @xmath74 .
asymptotic value : @xmath75 .
[ fig : mu0_rho0.1_a0.01 ] ] vs @xmath9 for exponentially evanescent traps of a lower initial density @xmath53 .
solid line : eq .
( [ forfig3 ] ) . symbols : simulation results ( there are error bars on the symbols but they are too small to see clearly ) .
other parameter values are @xmath52 , @xmath76 , and @xmath77 .
asymptotic value : @xmath78 .
[ fig : mu0_ga0.5_rho0.01_a1e-8 ] ] finally , it is straightforward to extend the results of this section to trap densities that decay as a stretched exponential , @xmath79 $ ] .
the integral is still straightforward and gives @xmath80 which reduces to eq .
( [ mu0 ] ) when @xmath81 .
the asymptotic finite survival probability then is @xmath82.\ ] ] an interesting interplay of @xmath12 and @xmath83 should be noted : there are values of @xmath83 and @xmath12 for which the survival probability of the target when the trap density decays as a stretched exponential ( @xmath84 ) is actually greater than with an exponential decay ( @xmath81 ) .
this seemingly counterintuitive behavior is connected with the reversal of time inequalities , i.e. , with the fact that @xmath85 is greater ( smaller ) than @xmath86 when @xmath9 is smaller ( greater ) than @xmath55 .
suppose now that the trap density decays as a power law as might happen , for instance , if there is a process of trap - trap annihilation .
the trap density at long times then decreases as @xmath87 and it is to be expected that the target survival probability ( and , in particular , whether it is asymptotically vanishing or finite ) depends sensitively on the relation between the exponents @xmath88 and @xmath12 . we expect that for sufficiently large @xmath88 the target will again have a finite probability of surviving forever
. to find a closed expression for the survival probability we need to specify @xmath38 for all times , not just asymptotically , and we choose @xmath89 with this form , the integral ( [ mu0 ] ) can be carried out exactly , to give @xmath90 for all @xmath88 , where @xmath1 is the incomplete beta function @xcite @xmath91 equation tells us that the typical length explored by a ( living ) trap grows with time as @xmath92 .
on the other hand , the mean distance between traps grows as @xmath93 .
it thus stands to reason that the asymptotic survival probability depends sensitively on the relative magnitudes of @xmath88 and @xmath94 . to present more explicit results in this long - time regime
we distinguish three cases .
* case 1 : @xmath95*. in this case eq .
( [ general ] ) can be written as @xmath96 here @xmath97 is the beta function ( where the requirement @xmath98 and @xmath99 places us in the case 1 " regime ) , and @xmath100 is the regularized incomplete beta function as defined in sec .
6.6.2 ( pg .
263 ) of @xcite . using the property 6.6.3 in
@xcite we can set @xmath101 , and applying the relation 26.5.5 in @xcite we can then write the asymptotic result @xmath102 consequently , recognizing the relation between the beta function and the gamma function , as @xmath103 we arrive at the asymptotic result @xmath104 the survival probability thus approaches ( via a power law decay of the exponent ) the finite asymptotic value @xmath105 figure [ fig : mu_ga0.75_b1e-6beta0.8_l1e4_rho1e-2 ] illustrates this result along with numerical simulations for comparison . vs @xmath9 for power law evanescent traps with @xmath106 ( case 1 " ) .
parameter values are @xmath107 , @xmath108 , @xmath53 , @xmath109 , and @xmath77 .
asymptotic value : @xmath110 .
solid line : eq .
( [ general ] ) .
symbols : simulation results .
[ fig : mu_ga0.75_b1e-6beta0.8_l1e4_rho1e-2 ] ] * case 2 : @xmath111*. in this case the integrand in eq .
goes to zero more slowly than @xmath112 for @xmath113 , so that a simple asymptotic analysis of the integral readily establishes that @xmath30 goes to infinity with increasing time as @xmath114 so that the survival probability vanishes at long times as a stretched exponential , @xmath115 analytic and simulation results for this case are shown in fig .
[ fig : mu_ga0.8_b1e-6_beta0.2_l1e4_rho0.01 ] .
vs @xmath9 for power law evanescent traps with @xmath116 ( case 2 " ) .
parameter values are @xmath117 , @xmath118 , @xmath53 , @xmath109 , and @xmath77 .
solid line : eq .
( [ general ] ) .
symbols : simulation results along with error bars .
[ fig : mu_ga0.8_b1e-6_beta0.2_l1e4_rho0.01 ] ] * case 3 : @xmath119*. this is the marginal case , and the incomplete beta function ( [ ibf ] ) can be rewritten as a hypergeometric function , @xmath120 the survival probability thus decays as an inverse power , @xmath121 results for the marginal case are shown in fig .
[ fig : mu_ga0.8_b1e-6_beta0.4_l1e4_rho0.01 ] .
vs @xmath9 for power law evanescent traps with @xmath122 ( case 3 " ) .
parameter values are @xmath117 , @xmath123 , @xmath53 , @xmath109 , and @xmath77 .
solid line : eq .
( [ general ] ) .
symbols : simulation results .
[ fig : mu_ga0.8_b1e-6_beta0.4_l1e4_rho0.01 ] ]
we have calculated the survival probability of a stationary target in a one - dimensional system in which diffusive or subdiffusive traps that eliminate the target upon encounter themselves disappear according to a survival probability .
the root mean square displacement of the traps grows with time as @xmath124 , that is , diffusively when @xmath13 and subdiffusively when @xmath61 .
the survival probability of the target depends sensitively on the interplay of two temporal events , namely , the motion of the traps as characterized by the exponent @xmath12 and their disappearance .
when the motion of the traps is diffusive or subdiffusive and the traps do not decay in time , the survival probability goes to zero as a stretched exponential , eq .
( [ nonevan ] ) .
when the traps undergo exponential decay or stretched exponential decay , the target has an asympototic safety margin , that is , a finite probability of surviving forever , cf .
( [ pas ] ) and ( [ pass ] ) .
when the traps are diffusive or subdiffusive and disappear according to a power law survival probability @xmath125 , the survival of the target depends sensitively on the relation between @xmath12 and @xmath88 .
if the traps move sufficiently rapidly relative to their disappearance , that is , if @xmath126 , the target is trapped with certainty at long times , its survival probability going to zero again as a stretched exponential , cf .
( [ case2 ] ) .
if the traps move slowly , @xmath127 , then the target has a chance of eternal survival , cf .
( [ case1 ] ) . at the critical relation @xmath128
the survival probability goes to zero as an inverse power of time , cf .
( [ marginal ] ) .
if in fact the trap density increases with time , the survival probability of the target necessarily vanishes asymptotically .
in this paper we have calculated the survival probability of a target particle in the presence of evanescent subdiffusive traps of given time - dependent density .
we could equally consider the inverse problem , namely , that of finding the time dependence of the density of traps to obtain a particular survival probability function .
for this purpose we need only invert " eq .
( [ solomega ] ) , @xmath129 an exponentially decaying survival probability of the form @xmath130 requires a density that decays as @xmath131 .
this is included in and consistent with case 2 in sec .
[ powerlawdec ] with @xmath132 .
similarly , for an inverse power decay of the form @xmath133 we require that @xmath134 consistent with case 3 in the same section .
this work has focused on the survival probability of a stationary target .
the survival probability of a moving target , diffusive or subdiffusive , surrounded by non - evanescent diffusive or subdiffusive traps has been considered recently in a number of papers @xcite .
extension of our work with evanescent traps to the case of a diffusive or subdiffusive target is in progress @xcite . the research of s.b.y . and j.j.r.l has been supported by the ministerio de educacin y ciencia ( spain ) through grant no.fis2004-01399 ( partially financed by feder funds ) and by the european community s human potential programme under contract hprn - ct-2002 - 00307 , dyglagemem .
is supported in part by the national science foundation under grant phy-0354937 .
g. h. weiss , _ aspects and applications of the random walk _ ( north - holland , amsterdam , 1994 ) ; f. den hollander and g. h. weiss , in _ contemporary problems in statistical _ , g. h. weiss , ed .
( siam , philadelphia , 1994 ) p. 147 .
j. m. sancho , a. lacasta , k. lindenberg , i. m. sokolov , and a. romero , phys .
lett . * 92 * , 250601 ( 2004 ) ; a. m. lacasta , j. m. sancho , a. h. romero , i. m. sokolov , and k. lindenberg , phys .
e * 70 * , 051104 ( 2004 ) ; j. m. sancho , a. m. lacasta , k. lindenberg , i. m. sokolov , and a. h. romero , phys .
lett . * 94 * , 188902 ( 2005 ) .
r. gorenflo and f. mainardi , integral and differential equations of fractional order , " in _ fractals and fractional calculus in continuum mechanics _ , edited by a. carpinteri and f. mainardi ( springer verlag , wien and new york , 1997 ) pp .
223 - 276 . | we calculate the survival probability of a stationary target in one dimension surrounded by diffusive or subdiffusive traps of time - dependent density .
the survival probability of a target in the presence of traps of constant density is known to go to zero as a stretched exponential whose specific power is determined by the exponent that characterizes the motion of the traps .
a density of traps that grows in time always leads to an asymptotically vanishing survival probability .
trap evanescence leads to a survival probability of the target that may be go to zero or to a finite value indicating a probability of eternal survival , depending on the way in which the traps disappear with time . |
recent theoretical studies reveal that in two dimensions strongly interacting two - component bosons in a magnetic field can realize a bosonic integer quantum hall ( biqh ) state @xcite .
the biqh phase characterized by hall conductivity quantized to an even integer @xcite is protected by a global @xmath1-symmetry and the real - space entanglement spectrum of this state hosts two counter propagating chiral modes .
recently , two different lattice versions of biqh states have been proposed at integer filling @xmath0 of the lowest topological flat - band with chern number @xmath2 .
the optical flux lattice has been studied by exact diagonalization of the projected hamiltonian in momentum space @xcite and the correlated haldane - honeycomb lattice has been studied by infinite density matrix renormalization group of hardcore boson in real space @xcite .
this is different from the two dimensional topological @xmath3 band filled by hardcore bosons at @xmath0 , which is believed to exhibit the fermi - liquid - like state @xcite .
indeed , for @xmath4 , a series of color - entangled abelian topological states have been suggested at various filling numbers under repulsive two - body interaction @xcite . in harper - hofstadter model with topological @xmath2 band ,
different emergent topological states including the symmetry protected biqh state , can be understood by an insightful approach from streda formula of composite fermion @xcite .
so far , the study on biqh state for single component bosons on topological hofstadter lattice is still lacking .
it is interesting to compare the biqh in such a system with other lattice realizations of topological flatbands with @xmath2 .
more specifically , it is interesting to address the issue what conditions can make the biqh stable against possible competing phases , like charge density wave and bosonic superfluid .
if the system can host other competing phases under certain conditions , it opens a door to explore quantum phase transition between biqh and other phases . in refs .
@xcite , the low energy theory describing continuous phase transitions between superfluid and biqh is constructed from fermionic parton . in this paper , we study the generalized hofstadter model and address the stability of the biqh against other phases , taking into account the effects of interaction strength , band topology and their interplay within the full real space hamiltonian .
we find that the many - body ground state is indeed biqh for hardcore bosons at integer filling of @xmath2 band with a robust spectrum gap , due to onsite hubbard repulsion . without onsite hubbard repulsion ,
the softcore bosons would undergo a bose - condensation into the lowest single particle orbit .
increasing the nearest neighbor interaction to strong repulsion , a charge density wave state would dominate over the biqh state .
when tuning the next nearest neighbor hopping down to zero , the chern number of the lowest band becomes @xmath5 and we show that the ground state would have a transition into a metallic liquid - like phase in this case .
thus , the emergence of biqh phase in single component bosons on lattice model is ultimately related to the interplay of interaction and band topology .
experimentally , the bosonic hofstadter model has been realized by laser - assisted tunneling in cold atoms @xcite , and a bose - einstein condensation ( bec ) is observed in this cold atom setup @xcite . in relation to current experiments ,
we discuss an experimental prospect toward realization of biqh state in optical lattices and further predict that once a stronger next nearest neighbor hopping can be implemented in this setup , one may realize the many - body biqh state in such a generalized hofstadter model and study the quantum phase transition between bose condensate and biqh phase by tuning the onsite interaction between bosons through lattice potential or feshbach resonances @xcite . this paper is organized as follows . in sec .
[ model ] , we give a description of the bose - hubbard model on the generalized hofstadter band with topological invariant @xmath2 . in sec .
[ mbgs ] , we explore the many - body ground state at infinite onsite repulsion ( namely , hardcore boson ) and present a detailed proof of biqh state by exact diagonalization at filling @xmath0 . in sec .
[ phasetransition ] , we explore other possible competing phases like bosonic superfluid , and discuss their transitions into biqh state by varying interactions and lattice parameters . finally , in sec . [ summary ] , we summarize our results and discuss the prospect of investigating nontrivial topological states in cold atoms systems .
here , we consider the interacting bosons on the generalized hofstadter lattice which describes the motion of charged particle under a uniform magnetic field @xmath6 on a square lattice with lattice constant @xmath7 @xcite .
take the gauge vector @xmath8 , such that the particle hopping between sites @xmath9 and @xmath10 has a magnetic phase @xmath11 @xcite .
in cold atoms , this spatially dependent gauge field can be artificially engineered in laser - assisted tunneling . when both nearest neighbor and next nearest neighbor hoppings are included , the noninteracting generalized hofstadter model , is described by the hamiltonian @xcite @xmath12 where @xmath13 is the bosonic annihilating operator at site @xmath14 , @xmath15 the unit vector along the @xmath16-direction and the magnetic flux through each plaquette @xmath17 .
now we consider that the bosons interact with each other via : @xmath18 where @xmath19 is the boson number operator and @xmath20 denote nearest neighbor pairs of sites . here
we take lattice parameters @xmath21 , such that the lowest band has a topological invariant @xmath2 . when @xmath22 , the lowest band has a topological invariant @xmath5 .
we can choose three sites in the x direction as a magnetic unit cell and the total number of lattice sites is @xmath23 , @xmath24 is the number of unit cells .
we explore the many - body ground state of @xmath25 by exactly diagonalizing a finite @xmath26-particle system at fixed integer filling @xmath27 . with periodic conditions we identify each many - body state using its total momentum sectors @xmath28 due to translation symmetry .
we first look at the limiting case that @xmath29 and @xmath30 , where each site can be occupied by one boson at most . the low energy spectrum is plotted in figs .
[ energy](a - b ) , for systems with different aspect ratios @xmath31 .
we find that there always exists a single gapped ground state with total momentum @xmath32 , separated from the excited states by a large gap , for both even and odd @xmath33 .
( color online ) numerical results for generalized hofstadter model with infinite two - body interaction @xmath29 : ( a - b ) low energy spectrum of different system sizes at filling @xmath0 ; ( c ) the berry curvatures for the @xmath34 ground state of @xmath35 system at @xmath36 mesh points ; ( d ) low energy spectrum of one or two quasiholes by removing particles .
the number of quasihole energy manifold under the red dashed line matches that of biqh state.,width=326,height=288 ] the biqh state is characterized by a finite spectrum gap . to explore the stability of this phase
, we use the twisted boundary conditions @xmath37 , and inspect the spectrum gap .
the energy gap @xmath38 is defined as the minimum value of the difference of the ground energy and the first excited energy on the full @xmath39 parameter plane in order to explore the stability of the phase . if the ground state mixes with other levels during the change of the boundary phases , we take @xmath40 .
our calculations confirm that the obtained ground state does not mix with the higher energy levels , demonstrating itself as the unique ground state under the insertion of the flux @xmath41 .
the many - body chern number of the ground state wavefunction @xmath42 at @xmath34 is given by @xmath43 where @xmath44 is the berry curvature . by numerically calculating the berry curvatures using @xmath45 mesh points in boundary phase space with @xmath46 as shown in fig .
[ energy](c ) , we find that the many - body chern number @xmath47 indeed converges to a quantized value 2 .
in addition , we calculate the density structure factor @xmath48 for the ground state , and we find no evidence of finite momentum bragg peaks in @xmath49 .
thus we can rule out the possibility of charge density wave ( cdw ) as the competing ground state . to distinguish the biqh phase from the usual topological ordered phase
, we investigate the quasiparticle excitations in the bulk . in a topological ordered phase such as fractional quantum hall states ,
the existence of quasihole excitations carrying fractional charge is the key evidence for the nontrivial nature of the topological state . here ,
we utilize two different methods .
first , we generate a quasihole by inserting a single flux quantum , namely we change the flux quanta of the lattice to @xmath50 . here as shown in fig .
[ energy](d ) , the counting number of low energy states of @xmath26 particles in @xmath33 orbits is simply given by @xmath51 .
second , by introducing an onste impurity potential @xmath52 at site @xmath53 of a given unit cell @xcite , one can pin the quasihole near the impurity location , and define the excess charge @xmath54 as the quasihole charge , where @xmath55 is the uniform density of biqh state .
for example , we add a @xmath56-impurity potential to the original periodic system @xmath57 and obtain @xmath58 to high precision , thus we obtain the exciations carrying integer charge unit , in against to the fracional value as expected for topological ordered phases .
these numerical results match the theoretical predictions of biqh state without any bulk topological order .
( color online ) numerical results for generalized hofstadter model with softcore two - body interaction @xmath59 : ( a ) the evolutions of the energy difference between the @xmath34 ground energy and the excited levels at @xmath60 ; ( b ) the low energy spectra under the insertion of @xmath61 at fixed @xmath62 and @xmath63 ; ( c ) the diffraction peak of the visibility @xmath64 and the related superfluid phase stiffness @xmath65 ; ( d ) the energy gap @xmath38 and many - body chern number of the groundstate at @xmath34 calculated using @xmath66 and @xmath67 mesh points in boundary phase space for berry curvature .
the red dashed line in ( d ) panel is the quantized value @xmath2 .
the inset panel shows the one - particle occupancy entropy.,width=326,height=268 ] having established biqh phase in the condition of @xmath29 and @xmath30 , we continue to discuss the competing phases in the nearby parameter space by varying interaction strength @xmath59 , @xmath68 and the band parameter @xmath69 , respectively .
the corresponding phase transitions will also be addressed . following the last section
, we further consider the effect of a finite repulsion @xmath59 and small @xmath69 on possible competing phases with zero nearest interaction @xmath30 .
as it is well - known , without onsite interaction , the free bosons would undergo a condensation into the lowest single - particle energy orbit for unfrustrated energy bands .
the emergence of biqh state in the strongly interacting regime signifies the important role of the band topology and interaction . in fig .
[ phaseu](a ) , we plot the energy variation of @xmath70 against @xmath59 at @xmath60 .
the @xmath34 ground state evolves adiabatically with @xmath59 . at small @xmath71 , the low energy spectrum
is shown in fig .
[ phaseu](b ) , and the @xmath34 ground state mixes with excited levels under the insertion of @xmath61 , demonstrating its metallic ( gapless ) nature . on one hand , bose condensate can be characterized by the off - diagonal long range order @xmath72 , such that the condensation momentum can be identified by the peak position of the diffraction pattern @xmath73 @xcite , which is defined as : @xmath74 for several degenerate lowest single - particle orbitals @xmath75 $ ] , take @xmath76 .
the reference background signal is the average of @xmath73 over entire brillouin zone @xmath77 , and the condensed visibility is defined by the difference @xmath64 . as shown in fig .
[ phaseu](c ) , at @xmath71 , @xmath73 has sharp peaks at momenta @xmath75 $ ] , while it vanishes at other momenta , and @xmath78 is almost a constant .
otherwise , we also diagonalize the @xmath79-matrix @xmath80 and obtain one particle eigenstates @xmath81 where @xmath82 ( @xmath83 ) are the natural orbitals and @xmath84 ( @xmath85 ) are interpreted as occupations . for @xmath71 , we find that the occupations @xmath86 for @xmath87 , while @xmath88 for @xmath89 , namely , a bose condensate occurs @xcite . by increasing the interaction , @xmath80 at @xmath90 gradually decreases to a small value . for strong interaction @xmath91 , @xmath92 for @xmath93 , while @xmath88 for @xmath94 .
this is consistent with our observation that @xmath95 should be vanishingly small . on the other hand , to evaluate its phase coherence , we impose a phase gradient @xmath96 through twisted boundary conditions @xcite , and define the bosonic superfluid phase stiffness as @xmath97 as illustrated in fig .
[ phaseu](c ) , for weak interaction @xmath98 , @xmath65 has a finite large value indicating the superfluidity of the ground state , and begins to drop with the increase of @xmath59 .
finally consistent with the diffraction peak , @xmath65 decreases to a small value for @xmath91 . by finite - size scaling with increasing particle numbers up to ten for @xmath29 , exact
diagonalization confirms that @xmath65 becomes vanishingly small . in fig .
[ phaseu](d ) , the energy gap and many - body chern number is plotted . for small @xmath99
, @xmath38 is zero .
@xmath38 is quite small for @xmath100 , while it saturates to a value of the order of the band gap for strong interaction @xmath91 .
meanwhile , the many - body chern number of the ground state is quantized to @xmath101 for @xmath102 and @xmath46 .
however , for small @xmath59 , its many - body chern number is not well - defined due to the level crossing , and we do not plot it . for @xmath103 ,
the many - body wave function for bose condensate is a product of single particle orbits in the lowest @xmath2 band , @xmath104 , where @xmath105 is the single - particle bloch state in the lowest band and @xmath106 $ ] .
thus its many - body chern number @xmath107 , such that @xmath47 usually does not exhibit a quantized behavior ( for instance , @xmath108 for @xmath109 ) and may change with the interaction . in recent experiments , the non - quantized hall response of a bose condensate
is observed in transport properties @xcite and charge pumping @xcite . in order to identify the fluctuations in the crossover ,
we consider the one - particle occupation entropy @xmath110 . in the inset of fig .
[ phaseu](d ) , we show the variance of the entropy as a function of interaction . in the limit @xmath103
, the entropy approaches the negative value @xmath111 .
in contrast , the entropy in the biqh phase is a much smaller positive value .
( color online ) numerical results for generalized hofstadter model versus next nearest hopping @xmath69 at infinite @xmath59 : ( a ) the evolutions of the energy difference between the @xmath34 ground energy and the excited levels at @xmath60 ; ( b ) the low energy spectra flux under the insertion of @xmath61 at fixed @xmath22 and @xmath63 ; ( c ) the diffraction peak of the visibility @xmath64 and the related superfluid phase stiffness @xmath65 ; ( d ) the energy gap @xmath38 and many - body chern number of the ground wavefunction at @xmath34 for @xmath67 mesh points of berry curvature .
the red dashed line in ( d ) panel is the quantized value @xmath2.,width=326,height=268 ] ( color online ) numerical results for generalized hofstadter model on the @xmath112-plane : ( a ) the intensity plot of the energy gap of @xmath113 system ; ( b ) the diffraction peak of the visibility @xmath64 at @xmath22.,width=326,height=163 ] similarly , in fig .
[ phaset](a ) , we plot the evolution of the @xmath34 ground state for hardcore boson against next nearest hopping @xmath69 . for small @xmath69 near @xmath22 , the @xmath34 ground state no longer maintain its robustness under the insertion of flux quantum , as indicated in fig .
[ phaset](b ) .
meanwhile , its superfluid stiffness and diffraction pattern show featureless behavior during the transition , implying no possible superfluid phase . in fig .
[ phaset](d ) , one can see that the energy gap gradually drops to zero as @xmath69 goes to zero , and the many - body chern number changes rapidly for small @xmath69 where the band topology structure is altered , and then drops to a negative value around @xmath114 . a typical picture of energy spectrum gap @xmath38 in the @xmath112-plane is shown in fig .
[ phasediagram](a ) . the biqh phase characterized by a finite gap
is located at the right lower region where @xmath115 .
the diffraction peak @xmath116 of the ground state at @xmath60 is shown in fig . [ phasediagram](b ) .
the left region where @xmath117 is characterized by a finite near constant condensation in single - particle orbits , regardless of the band topology , which is a superfluid phase .
the right region where @xmath91 implies no off - diagonal long range order .
all these phases do not host any bragg peak .
moreover , upon changing twisted boundary angles , except for biqh , the ground state evolves into higher energy levels , indicating its metallic nature . for strongly interacting bosons in the right upper region where the lowest band topology changes to a band with @xmath3 , the many - body chern number of its ground state usually does not host
integer quantized value once the disorder is introduced @xcite . in comparison
, we also present numerical results of hardcore bosons at @xmath0 on the topological checkerboard and honeycomb lattice models whose lowest band possesses topological invariant @xmath3 @xcite . as shown in fig .
[ metal ] , the low energy states in the same @xmath118 sector evolve into each other under the insertion of flux quantum .
we find that the berry curvatures are vanishing small and the many - body chern number of the ground state @xmath119 .
both the intra - sublattice and inter - sublattice structure functions of density correlations @xmath120 have only a zero - momentum peak .
one conjecture is that this phase would be a metallic fermi - liquid like phase at @xmath0 .
we leave more details of this phase to be addressed in future studies .
( color online ) numerical results for hardcore bosons on two different topological lattices at filling @xmath121 of the lowest chern band @xmath3 : ( a ) the low energy spectrum and ( b ) the low energy spectrum vs. flux for checkerboard lattice ; ( c ) the low energy spectrum and ( d ) the low energy spectrum vs.flux for honeycomb lattice .
the checkerboard lattice hopping parameters are the same as ref .
@xcite , while the honeycomb lattice hopping parameters are the same as refs .
@xcite , width=326,height=288 ]
( color online ) numerical results for generalized hofstadter model at @xmath29,@xmath60 and @xmath122 : ( a ) the evolutions of low energy states , ( b ) the energy gap @xmath38 and fidelity susceptibility @xmath123 and ( c ) the peaks of density structure factor @xmath49 and diffraction pattern versus nearest neighbor interaction @xmath68 for @xmath124 ; ( d ) the density structure factor at @xmath125 for @xmath126.,width=326,height=288 ] finally , when the nearest neighbor repulsion is taken into account , we plot the variations of low energy states with @xmath68 in fig .
[ phasev](a ) from @xmath30 to @xmath127 .
the dependence of @xmath38 on small @xmath68 is quite complicated , and it does not show a monotonic behavior .
however for large @xmath68 , the energy gap would collapse to zero indeed , as shown in fig .
[ phasev](b ) .
we calculate the fidelity susceptibility @xmath123 of the ground @xmath34 state , defined by @xmath128 @xmath123 exhibits a peak near the point where @xmath38 collapses , which serves as a signal of quantum phase transition @xcite . in order to identify this transition
, we plot the evolution of the peaks of its density structure factor and diffraction pattern in fig .
[ phasev](c ) . for strong repulsion @xmath129 , the peak of density structure factor
@xmath49 emerges at finite vector @xmath130 , while the diffraction peak does not exhibit any upward jump behavior . by comparing @xmath49 for larger sizes , e.g. , exact diagonalization of up to ten particles shown in fig .
[ phasev](d ) and density matrix renormalization group of up to 18 particles , we confirm the existence of the bragg peak .
thus the phase for @xmath129 is found to be a charge density wave phase , instead of a superfluid or supersolid phase .
in summary , we have studied the bose - hubbard model in two dimensional generalized hofstadter band with @xmath2 , and demonstrated that strongly interacting bosons at filling @xmath0 can host biqh state with hall conductivity exactly quantized to 2 for strong onsite repulsion . the phase transition from bosonic superfluid to biqh state driven by tuning the onsite hubbard repulsive interaction
is revealed by calculating superfluid phase stiffness and the diffraction pattern of the off - diagonal long range order .
tuning next nearest neighbor hopping down to zero changes the band topology , and the possible fermi - liquid - like phase at at filling @xmath0 of @xmath3 chern band is characterized by no phase coherence or well - defined many - body chern number , gapless spectrum flux and featureless structure factors .
strong nearest neighbor repulsion would lead to a charge density wave . in current @xmath131 experiments
@xcite , magnetic flux @xmath132 can be tunable over the range @xmath133 , onsite hubbard interaction @xmath59 can be adjusted from zero up to @xmath134hz by varying lattice potential depth and the nearest neighbor hopping @xmath135hz , but the next nearest neighbor hopping is much smaller than nearest neighbor hopping .
experimental observations of these topological phases may be possible by further enhancement of next nearest neighbor hopping in the future , or using @xmath136 with only small next nearest neighbor hopping where the lowest band also hosts chern number two @xcite . in our calculation , the energy gap @xmath38 can be of the order of @xmath137 , which is close to the cooling temperature limit .
we believe that this hofstadter model should provide a good platform for future study of topological phases .
99 t. senthil and m. levin , phys .
* 110 * , 046801 ( 2013 ) .
s. furukawa and m. ueda , phys .
lett . * 111 * , 090401 ( 2013 ) .
n. regnault and t. senthil , phys .
b * 88 * , 161106(r ) ( 2013 ) . y .-
h . wu and j. k. jain , phys .
b * 87 * , 245123 ( 2013 ) .
t. gra , d. ravents , m. lewenstein , and b. juli - daz , phys .
b * 89 * , 045114 ( 2014 ) . y .-
lu and a. vishwanath , phys .
b * 86 * , 125119 ( 2012 ) .
z .- x . liu and x .- g .
wen , phys .
lett . * 110 * , 067205 ( 2013 ) .
s. d. geraedts and o. i. motrunich , ann .
phys . * 334 * , 288 ( 2013 ) .
m. mulligan and m. p. a. fisher , phys .
b * 89 * , 205315 ( 2014 ) .
a. sterdyniak , nigel r. cooper , and n. regnault , phys .
115 * , 116802 ( 2015 ) .
c . he , s. bhattacharjee , r. moessner , and f. pollmann , phys .
lett . * 115 * , 116803 ( 2015 ) .
n. read , phys .
b * 58 * , 16262 ( 1998 ) .
z. liu , e. j. bergholtz , h. fan , a. m. luchli , phys .
lett . * 109 * , 186805 ( 2012 ) .
wang , h. yao , c .- d .
gong , and d. n. sheng , phys .
b * 86 * , 201101(r ) ( 2012 ) .
s. yang , z .- c .
gu , k. sun , and s. das sarma , phys .
b * 86 * , 241112(r ) ( 2012 ) .
a. sterdyniak , c. repellin , b. a. bernevig , and n. regnault , phys .
b * 87 * , 205137 ( 2013 ) .
d. wang , z. liu , j. cao , and h. fan , phys .
* 111 * , 186804 ( 2013 ) .
g. mller and n. r. cooper , phys .
115 * , 126401 ( 2015 ) .
t. grover and a. vishwanath , phys .
b * 87 * , 045129 ( 2013 ) .
lu , d .- h .
lee , phys .
b * 89 * , 195143 ( 2014 ) .
m. aidelsburger , m. lohse , c. schweizer , m. atala , j. t. barreiro , s. nascimbne , n. r. cooper , i. bloch , n. goldman , nature phys .
* 11 * , 162 ( 2014 ) ; m. aidelsburger , m. atala , m. lohse , j. t. barreiro , b. paredes , and i. bloch , phys . rev . lett . * 111 * , 185301 ( 2013 ) . h. miyake , g. a. siviloglou , c. j. kennedy , w. c. burton , and w. ketterle , phys . rev . lett . * 111 * , 185302 ( 2013 ) . c. j. kennedy , w. c. burton , w. c. chung , and w. ketterle , nature phys . *
11 * , 859 ( 2015 ) .
p. courteille , r. freeland , d. heinzen , f. van abeelen , and b. verhaar , phys .
lett . * 81 * , 69 ( 1998 ) .
d. r. hofstadter , phys .
b * 14 * , 2239 ( 1976 ) .
e. kapit and e. mueller , phys .
lett . * 105 * , 215303 ( 2010 ) .
z. liu , r. n. bhatt , and n. regnault , phys .
b * 91 * , 045126 ( 2015 ) .
f. gerbier , a. widera , s. flling , o. mandel , t. gericke , and i. bloch , phys .
lett . * 95 * , 050404 ( 2005 ) ; k. gnter , t. stferle , h. moritz , m. khl , and t. esslinger , _
ibid_. * 96 * , 180402 ( 2006 ) ; s. ospelkaus , c. ospelkaus , o. wille , m. succo , p. ernst , k. sengstock , and k. bongs , _ ibid_. * 96 * , 180403 ( 2006 ) .
e. j. mueller , t .- l .
ho , m. ueda , and g. baym , phys .
a * 74 * , 033612 ( 2006 ) .
r. g. melko , a. paramekanti , a. a. burkov , a. vishwanath , d. n. sheng , and l. balents , phys . rev .
* 95 * , 127207 ( 2005 ) .
l. j. leblanc , k. jimenez - garcia , r. a. williams , m. c. beeler , a. r. perry , w. d. phillips , and i. b spielman , proc .
usa * 109 * , 10811 ( 2012 ) .
hsin - i lu , m. schemmer , l. m. aycock , d. genkina , s. sugawa , and i. b. spielman , arxiv:1508.04480 ( 2015 ) .
d. n. sheng , x. wan , e. h. rezayi , k. yang , r. n. bhatt , and f. d. m. haldane , phys .
lett . * 90 * , 256802 ( 2003 ) ; s. yang , k. sun , and s. das sarma , phys .
b * 85 * , 205124 ( 2012 )
. d. n. sheng , z. gu , k. sun , and l. sheng , nat .
* 2 * , 389 ( 2011 ) .
t. neupert , l. santos , c. chamon , and c. mudry , phys .
lett . * 106 * , 236804 ( 2011 ) .
e. tang , j .- w .
mei , and x .-
wen , phys .
* 106 * , 236802 ( 2011 ) .
k. sun , z. gu , h. katsura , and s. das sarma , phys .
lett . * 106 * , 236803 ( 2011 ) .
n. regnault and b. a. bernevig , phys .
x * 1 * , 021014 ( 2011 ) .
n. y. yao , a. v. gorshkov , c. r. laumann , a. m. luchli , j. ye , and m. d. lukin , phys .
lett . * 110 * , 185302 ( 2013 ) ; n. y. yao , c. r. laumann , a. v. gorshkov , s. d. bennett , e. demler , p. zoller , and m. d. lukin , _
ibid_. * 109 * , 266804 ( 2012 ) .
zeng and l. yin , phys .
b * 91 * , 075102 ( 2015 ) . y .-
wang , h. yao , z .- c .
gu , c .- d . gong , and d. n. sheng , phys .
* 108 * , 126805 ( 2012 ) .
wang , z .- c .
gu , c .- d .
gong , and d. n. sheng , phys .
lett . * 107 * , 146803 ( 2011 ) . s .-
b * 24 * , 4371 ( 2010 ) . | we study the interacting bosons in topological hofstadter bands with chern number two . using exact diagonalization
, we demonstrate that bosonic integer quantum hall ( biqh ) state emerges at integer boson filling factor @xmath0 of the lowest chern band with evidences including a robust spectrum gap and quantized topological hall conductance two .
moreover , the robustness of biqh state against different interactions and next - nearest neighbor hopping is investigated .
the strong nearest neighbor interaction would favor a charge density wave . when the onsite interaction decreases , biqh state undergoes a continuous transition into a superfluid state . without next - nearest neighbor hopping ,
the ground state is possibly in a metallic fermi - liquid - like phase . |
the galactic center region ( gcr ) has the highest star formation rate density in the milky way , more than @xmath42 orders of magnitude above the average value of the galactic disk ( @xcite @xcite ) .
this is considered to be a consequence of the large reservoir of molecular gas that fills the central half - kiloparsec of the galaxy , where @xmath410% of the galaxy s molecular gas is concentrated . with respect to the physical properties that are typical of molecular gas further out in the galactic disk ,
the gcr medium is significantly hotter , more turbulent , and more highly magnetized .
it is also subject to mechanical shearing by the strong tidal field , and repeated compression from cloud - cloud collisions , stellar winds , and supernova shock waves ( @xcite @xcite ) .
it has been suggested that the environmental conditions in the gcr result in an elevated jeans mass and an enhanced rate of massive - star production .
at least three extraordinary clusters of massive stars have formed within the unique gcr medium during the past several myr , including the arches cluster ( ) and the quintuplet cluster ( ; ; ) , both of which lie @xmath430 pc , in projection , from the center . within the inner few parsecs of the galaxy
lies the central cluster ( ) , which is gravitationally bound to the galactic black hole , sgr a@xmath3 ( ; ) .
all three of the known clusters are massive ( @xmath5 ) , dense , containing hundreds of o stars , and dozens of evolved wolf - rayet stars ( wrs ) and ob supergiants ( , ; ) .
the collection of three such extraordinary clusters in the gcr is evidence favoring the hypothesis that the predominant mode of star formation in the region gives rise to dense , massive clusters .
however , surveys aimed at identifying massive stars throughout the greater gcr have revealed a population of several dozen massive stars that have no apparent spatial association with the known stellar clusters ( ; ; ; ; ; , ) .
the discovery of these isolated " massive stars has motivated the continued search for additional such cases , with the goal of determining the total extent of the inter - cluster population of massive stars and their physical relation to , or independence from , starburst clusters like the arches and quintuplet .
the demography and kinematics of these isolated stars will yield valuable insight into the mode of massive star formation in the gcr and the dynamical evolution of massive stellar clusters within galactic nuclei .
the identification of isolated massive stars in the gcr is a difficult task .
heavy extinction limits observations of the galactic center starlight to the infrared part of the electromagnetic spectrum .
infrared measurements sample the rayleigh - jeans tail of stellar spectral energy distributions where there is a degeneracy of stellar color with effective temperature .
this renders early - type and late - type stars photometrically indistinguishable .
narrowband photometry aimed at detecting emission - line excesses from hot supergiants can be very effective but requires a highly stable point - spread function ( psf ) in order to accurately subtract the stellar continuum emission from the spectral line emission .
this is difficult to achieve in seeing limited conditions , especially when extreme stellar confusion is an issue , as it typically is when observing the gcr .
therefore , infrared narrowband surveys are best conducted from the stable environment of space , where low - background , diffraction - limited observations can be executed without the nuisance of psf variability .
furthermore , space - based narrowband observations enable measurement of the paschen-@xmath0 ( p@xmath0 ) transition ( @xmath6=43 ) of h i , which is the strongest emission line of this species in the infrared .
the @xmath71.87 wavelength of this transition is also shared with several transitions of he i and he ii , which is important in the interest of detecting hydrogen - poor wrs that exhibit strong helium emission . in this work ,
we report discoveries of isolated " emission - line stars in the gcr via p@xmath0 narrowband imaging from space and subsequent infrared spectroscopy from the ground .
the new additions add to our recent identification of a new luminous blue variable in the gcr ( ) , which was the first reported discovery of the spectroscopic survey .
the successful detection of these stars highlights the significant advantage that space - based narrowband searches using the p@xmath0 line have over ground - based surveys centered on other atomic transitions ( e. g. , ; ) which have identified only a fraction of the isolated emission - line stars .
we present here the spectra of new identifications , their photometric properties , and discuss the implications that these objects have for massive star formation in the gcr .
our target sample was assembled following the completion of a p@xmath0 narrowband imaging survey of the gcr , performed with the _ hubble space telescope _ and the near - infrared camera and multi - object spectrometer ( nicmos ; @xcite@xcite ) .
the survey , requiring 144 orbits of observations , covered the central @xmath8 arcmin@xmath9 around sgr a@xmath3 , which corresponds to a total projected physical area of @xmath10 pc@xmath11 , assuming a distance of 8 kpc to sgr a@xmath3 ( and references therein ) .
the nic3 camera was utilized and narrowband images were obtained through the f187n ( line ; @xmath12 ; fwhm=0.019 ) and f190n ( continuum ; @xmath13 ; fwhm=0.017 ) filters .
the survey resulted in the detection of @xmath4150 point - like sources of p@xmath0 line excess ( i.e. , sources with f187n - f190n values @xmath145@xmath15 above the mean of all point sources in the survey area ; dong et al .
, in prep ) .
approximately half of the sources are located outside of the three known stellar clusters , and nearly half of those were unidentified prior to this work .
we have undertaken a near - infrared spectroscopic campaign to characterize the unidentified sources of p@xmath0 line excess .
focusing on the sources having the strongest line excess and the brightest near - infrared counterparts ( @xmath16 mag ) , we have spectroscopically identified 19 new , hot , massive stars in the gcr .
so far , our spectroscopic search has had a 100% success rate for detecting emission - line stars .
table 1 lists basic data for these sources , including their positions , corresponding @xmath17 photometry , and narrowband f187n and f190n measurements .
the sources are named based on their galactic coordinates , but will hereafter be referred to by their numerical entry in table 1 .
the positions and @xmath17 photometry were taken from the catalog of the sirius survey of the gcr ( see and references therein ) , which generated a catalog of point sources having a 10@xmath15 limiting @xmath18 magnitude of 15.6 .
the f187n and f190n measurements are from dong et al .
( 2011 , in preparation ) .
we note that the method of extraction for the narrowband photometry has been refined since our report on the lbv [email protected] ( ) , where we presented our narrowband measurements for this source and the known lbvs qf362 ( aka fmm 362 ; ) and the pistol star ( ) ; so , we take this opportunity to include the refined narrowband measurements for all three of these lbvs in table 1 .
spectra of four stars were obtained using the spex medium - resolution spectrograph on the 3 m infrared telescope facility ( irtf ) telescope ( ) , located on the summit of mauna kea in hawaii .
spex was used in short cross - dispersed mode ( sxd ) , utilizing a slit width of 05 , providing a spectral resolution of @xmath20 .
flat - field images and a wavelength calibration source were provided by obtaining spectra of continuum and argon lamp sources .
the data were reduced and extracted using the -based software package , specially designed for the reduction of data obtained with spex on the irtf ( ) .
the 3 m united kingdom infrared telescope ( ukirt ) on mauna kea was used to obtain spectra of five stars . the ukirt 15 imager spectrometer ( uist ; )
was used in service mode as part of the ukirt service programme . the short-@xmath21 grism and 4-pixel slit ( 048 )
were used , providing a spectral resolution of @xmath22 and a wavelength range of @xmath72.012.26 @xmath23 .
flat - fields and a wavelength calibration source were provided by obtaining spectra of continuum and argon lamps .
the data images were reduced using the pipeline .
the spectra were extracted using the routine .
spectra of four stars were obtained using the 4.1 m anglo - australian telescope ( aat ) on siding spring mountain ( mount woorat ) in new south wales , australia .
the iris2 instrument ( ) provided a spectral resolution of @xmath24 in the @xmath21 band , using the 1 slit .
flat - field images and a wavelength calibration source were provided by obtaining spectra of continuum and neon lamp sources .
the data images were reduced and spectra extracted with the same programs used for the ukirt data .
spectra of six stars were obtained at the southern observatory for astrophysical research ( soar ) , located on cerro pachon in chile . the ohio state infrared imager spectrometer ( osiris ; )
was used in cross - dispersed mode for five of these stars .
this mode provides a spectral resolution of @xmath25 .
wavelength calibration was achieved using the background oh emission lines from the sky . for one star ,
osiris was used in the higher resolution longslit mode to obtain a @xmath21-band spectrum with a resolution @xmath26 . for this observation
, wavelength calibration was performed using spectra of the internal neon and argon lamps .
all of these data were reduced and extracted using the standard routines and . for all observations
the spectra were acquired in an abba " nodding sequence in order to subtract the sky background , and to suppress the contribution of bad pixels .
spectra of a0v standard stars were obtained at airmasses similar to those of the science targets in order to derive a telluric absorption spectrum .
the telluric corrections were applied using the idl package ( ) , which removes model h i absorption lines from the a0v standard star before application to the science data .
table 2 summarizes the facilities used for the observations of each star .
in the following sections , the spectroscopic criteria we use to classify these stars were adopted from previous spectroscopic studies of massive stars in the near - infrared , conducted by ( ) , ( ) , and ( ) . the central wavelengths of all spectral lines referred to in the text are adopted from these references .
the @xmath21-band spectra of stars 6 and 7 are presented in figure 1 .
the spectra are dominated by a complex of blended emission lines near @xmath72.1122.115 , which includes contributions from he i , n iii , c iii and o iii .
blueward of this line complex are emission lines of c iv at @xmath72.069 and @xmath72.078 .
br@xmath27 emission is marginally detected in the spectrum of star 6 , although it may be a background nebular feature .
the spectra of both stars exhibit he ii absorption at @xmath72.189 , although it is very near the noise level .
these features are consistent with the @xmath21-band spectra of early o supergiants , specifically in the range of o46i ( e. g. , see ) .
the spectra of stars 16 , 15 , 12 , 9 , and 10 are presented in figure 2 .
the spectra of these stars are dominated by emission lines of br@xmath27 and the @xmath72.1122.115 complex of he i , n iii , c iii and o iii .
stars 16 , 15 , and 12 exhibit the @xmath72.1122.115 complex in fairly strong emission , while 9 and 10 , instead , exhibit a he i absorption line accompanied by a relatively weak emission component .
weak c iv emission is also apparent in the spectra of stars 16 , 15 , and 12 .
the br@xmath27 lines of all of these stars exhibit an asymmetry on their blue side owing to a contribution from he i at @xmath72.1647 .
an absorption line of he i at @xmath72.058 is present for four of the stars , while this feature appears in emission for two of them .
weak n iii emission appears at @xmath72.247 in the spectrum of star 16 .
three of the stars also exhibit weak he ii absorption at @xmath72.189 , while this transition is not detected in the spectra stars 9 and 10 .
these features are characteristics of oif@xmath28 designation is given to o stars exhibiting s iv emission ( @xmath294089 and 4116 ) , as well as n iii and he ii @xmath74686 emission in their optical spectra ( ) . ] supergiants and late - type wn stars ( wnl ) , which are commonly referred to as hydrogen - rich wrs ( wnh stars ) .
distinguishing between these two types can be very difficult indeed . according to ( ) , oif@xmath28 stars exhibit br@xmath27 emission that has approximately equivalent strength to the @xmath72.1122.115 complex , while wn89h stars exhibit a relatively dominant br@xmath27 emission line . based on the dominance of br@xmath27 in the spectra of all stars in figure 2
, we may be inclined to classify them all as wn89h stars . however , the weakness of the @xmath72.1122.115 complex in the spectra of stars 9 and 10 is more typical of the o46if@xmath30 stars in ( ) than the wn89h stars , which always have a strong @xmath72.1122.115 emission complex .
furthermore , wn89h always exhibit he ii emission or absorption at @xmath72.189 , but this transition is not detected in the spectra of stars 9 and 10 .
thus , we classify stars 16 , 15 , and 12 as wn89h stars , and assign the o46if@xmath28 designation to stars 9 and 10 .
the spectra of stars 13 , 18 , 5 , and 1 are presented in figure 3 . all of these stars exhibit prominent br@xmath27 emission .
stars 13 and 18 also exhibit strong he i emission at @xmath72.058 with a p cygni profile , and weak he i absorption at @xmath72.112 that may also include a weak n iii emission component on the red side of the he i absorption line .
the br@xmath27 line of star 13 exhibits an asymmetry on its blue side owing to a contribution from he i. star 1 exhibits he i at @xmath72.058 in emission , although it is relatively weak .
star 5 exhibits the he i @xmath72.058 feature as well , but it is in weak absorption .
the spectra of stars 5 and 1 exhibit weak emission from the na i doublet near @xmath72.209 , and in the case of star 5 , weak fe ii emission near @xmath72.09 .
the presence of these low - ionization emission features indicates that stars 1 and 5 are probably of the b spectral class , with subtypes in the range of b0b2 ( e.g. , see hanson et al .
their classification as supergiants is justified in section 4.2 . by comparison , stars 13 and 18 must be hotter , as indicated by the presence of he i absorption , while the strong p cygni emission lines of he i at @xmath72.058 indicate relatively strong winds . based on these features , we conclude that stars 13 and 18 can be classified as p cygni - type o supergiants . "
figure 4 presents spectra of stars 11 and 17 .
these stars exhibit broad emission lines , indicative of fast , extended stellar winds . for star
11 spectral emission features include br@xmath27 , he i at @xmath72.058 and @xmath72.112 , and he ii at @xmath72.189 .
the latter of these features is probably the superposition of a he ii @xmath72.189 p cygni profile and a blend of he i emission at @xmath72.185 .
weak emission from n iii is also present at @xmath72.247 , as well as a weak c iii / n iii blend at @xmath72.104 .
the weakness of he ii relative to br@xmath27 and he i at @xmath72.112 , and the strong p cygni emission component of he i at @xmath72.058 , are characteristics of late - type , hydrogen - rich , nitrogen - type stars , specifically wn89h stars (; ) .
the dominant emission feature in the spectrum of star 17 is he ii at @xmath72.189 , which indicates that it is of an early wn ( wne ) subtype .
weak but broad he i emission at @xmath72.058 is present with a deep and highly blueshifted p cygni absorption component .
the large blueshift of this absorption component relative to the peak of the emission component highlights the high wind velocity of star 17 ( @xmath31=0.015 @xmath32 @xmath33 km s@xmath34 ) relative to star 11 ( @xmath31=0.006 @xmath32 @xmath35 km s@xmath34 ) .
furthermore , the emission lines of star 17 are exceptionally broad , having fwhm=155 . according to ( ) , wn stars with fwhm(he ii @xmath72.189 ) @xmath36130 are classified as broad - lined wn stars ( wnb ) .
star 17 satisfies the broad criterion , while the relative strength of he ii relative to br@xmath27 and he i is consistent with a wn5 spectral type .
thus , we classify star 17 as a wn5b star . figure 5 presents the @xmath21-band spectra of star 8 , 19 , 14 , 3 , 2 , and 4 .
each of these stars exhibit emission from c iii at @xmath72.115 , he i , and he ii near @xmath72.058 , c iii , and c iv near @xmath72.072.08 , and weak he ii emission near @xmath72.189 .
these spectral characteristics are consistent with those of late , carbon - type ( wc ) wrs . a subtype diagnostic
is provided by the strength ratio of c iv at @xmath72.08 relative to c iii at @xmath72.115 ( ) , which is less than unity for all stars , consistent with wc9 subtype .
luminous wr and o stars generally exhibit infrared excess from the free - free emission generated in their ionized winds , and also from strong line emission ( ; ) .
thermal continuum excess from hot dust emission is also not uncommon to late - type wc ( wcl ) stars .
color - color diagrams are useful in assessing the infrared excess of wrs and the likely causes for it .
figure 6 shows a color - color diagram of @xmath37 versus @xmath38 for the stars in our sample that were detected in all three near - infrared bands .
the interstellar reddening vector is illustrated by the linear trend traced by random field stars that lie toward the gcr , most of which are presumably late - type giants .
stars 1 and 7 lie near the lower left of figure 6 , well separated from most of the other massive stars .
the fact that they suffer relatively low extinction implies that they are in the foreground toward the gcr .
they also do not appear to exhibit evidence for significant , intrinsic excess .
while the ob and wn stars in the gcr have color excesses typical for their class , several wc stars in figure 6 appear to have color excesses that are significantly higher than those of the o and wn stars and 1 wc star ( note that only four of the wc stars from our sample are included in figure 6 , since two others have not been detected in the @xmath39 band ) .
the colors of star 19 are consistent with the excess being from free - free emission alone , and this might also be true for star 14 .
however , the relatively strong excess emission from stars 2 and 4 ( the two wc stars that are furthest from the reddening trend in figure 6 ) indicates that there is an additional excess component , which is likely the result of thermal emission from hot dust .
late - type wc stars ( wc89 ) commonly exhibit thermal emission from hot dust , and it is generally accepted that the production of dust from these stars is a consequence of binary evolution , whereby the dust is created via the collision of hydrogen - rich and carbon - rich winds from the respective ob and wc stars ( ) .
several likely examples of this phenomenon are already known in the gcr , and they are typically detectable as hard x - ray sources ( e.g. , see and references therein ) . since stars 2 and 4 were not detected in x - ray surveys of their respective fields ( ) ,
there is currently no evidence that they are members of colliding - wind binary systems .
nonetheless , the relatively strong excess emission from stars 2 and 4 is consistent with thermal dust emission .
thus , we tentatively classify stars 2 and 4 and wc9?d stars .
we derived the extinction and absolute photometry for each source by comparing the observed @xmath40 and @xmath37 colors with the intrinsic colors of stars of the same spectral types , which were adopted from the literature ( ; ) .
we used the extinction relation of ( ) for stars near the galactic center , which is given by @xmath41 and @xmath42 where @xmath43 , and similarly for the @xmath40 colors .
the two values were then averaged to obtain a final extinction estimate , @xmath44 .
the results are listed in table 3 .
the values are similar to other known massive stars in the gcr ( e.g. , see ) . as expected from their positions in figure 6 , stars 7 ( o46i ) and 1 ( b0i b2i ) have relatively low @xmath45 values of 1.16 and 0.95 mag , respectively , which indicates that they lie in the foreground . assuming a distance of 8 kpc for the galactic center ( )
, we derived @xmath46 from the extinction - corrected values .
to calculate the bolometric luminosities of the stars , bolometric corrections for the appropriate spectral types were adopted from the same literature sources used to obtain the intrinsic colors , except for the wn89h stars , whose bolometric corrections were derived from the @xmath47 and luminosity values of the wn89h stars in ( ) .
these results are also listed in table 3 , and they are more or less consistent with luminosities that are typical for these spectral types .
the luminosity values derived for the other o and b stars are also consistent with supergiant luminosity class .
we note that the luminosity values obtained for wc9?d stars should be met with caution , since thermal dust emission can have a significant effect on the observed infrared colors , leading to an overestimate of the extinction toward the source .
nonetheless , the luminosity values that we obtain are consistent with those of other wc9 stars .
for the foreground o4i star 7 , we assumed that it shares the same value of @xmath47 as star 6 , since they appear to be identical in spectral type .
thus , we derive a distance modulus of 12.79 mag ( 3.6 kpc ) to star 7 , which places it in the norma arm of the galaxy ( ) . for the b0i b2i star 1 we have no information on its luminosity class , so we are unable to derive its foreground distance .
however , its extinction value is very comparable to star 7 ; so , we could infer that it also lies at roughly the same distance , also in the norma arm .
if so , then the associated distance modulus of 12.79 mag indicates that star 1 has @xmath48 , which implies supergiant luminosity class .
the o46i stars 6 and 7 have x - ray counterparts in the master catalog of ( ) , which contains 9017 x - ray point sources detected in all combined observations of the gcr with the _ chandra x - ray observatory_. the x - ray photometric data for these two sources is presented in table 4 .
the details of the extraction process are described in ( ) .
candidate near - infrared counterparts to the population of gcr x - ray sources were identified in ( ) ; the o46i star 6 is source 432 from table 3 of those authors , and is associated with the _ chandra _ source cxogc j174531.4@xmath19285716 . on the other hand , the foreground o46i star 7 that is associated with _
chandra _ source cxogc j174537.9@xmath19290134 is not present in ( ) , since it was flagged as having erroneous photometry in the sirius near - infrared catalog that was used for that work ; so , for this star we relied on photometry from the two micron all sky survey ( 2mass ; cutri et al .
table 4 includes x - ray source astrometry , total on - source integration time , photon number counts in the hard ( 2.08.0 kev ) and soft bands ( 0.52.0 kev ) , broadband photon flux , hardness ratios for soft and hard energy bands , and the average energy of all detected photons . in order to constrain the distance and nature of the x - ray sources
, it is useful to examine the hardness ratio , defined as hr=@xmath49 , where @xmath50 and @xmath51 are the fluxes in the hard and soft energy bands , respectively .
the soft color , hr0 , is defined by letting @xmath50 and @xmath51 be the fluxes in the respective 0.52.0 kev and 2.03.3 kev energy bands .
the hard color , hr2 , is defined by letting @xmath50 and @xmath51 be the fluxes in the 4.78.0 kev and 3.34.7 kev energy bands , respectively .
galactic center sources , which typically lie behind a hydrogen absorption column of @xmath52 @xmath53 , will have most of their soft x - ray photons absorbed by the interstellar medium ( ism ) , and will , thus , exhibit @xmath54 ( ) . star 6 , with hr0=0.56 , definitely satisfies the criterion for galactic center location , while star 7 , with [email protected] barely makes the cut . compared with hr0 ,
the more energetic photons used to calculate hr2 suffer significantly less absorption from intervening gas and dust , so they are more useful for constraining the intrinsic x - ray hardness .
we compared the hr2 colors of these sources with those of the several dozen x - ray emitting massive stars in the gcr that were presented in (; their figure 11 ) .
star 6 is among the faintest x - ray sources that has a confirmed infrared counterpart .
it was not significantly detected at energies above 4.7 kev , so its hr2 value is @xmath55 .
this contrasts with most of several dozen known , massive stellar x - ray sources in the gcr , many of which have significant detections at energies up to 8.0 kev .
this feature of the general population is presumed to be the result of emission from a relatively hard thermal plasma with @xmath56 kev . by contrast , the practically null detection of x - rays above 4.7 kev energy from star 6 implies that it is a relatively faint and soft x - ray source ( @xmath57 kev , if it is thermal ) . unfortunately , there are not enough counts to effectively model the x - ray spectrum .
but if we assume a thermal plasma and guess a reasonable temperature , we can at least get an order - of - magnitude estimate of the x - ray luminosity . to do this
, we used the derived extinction values in table 3 to calculate the corresponding hydrogen column density ( @xmath58 ) , using the @xmath59 ratio from ( ) , assuming a negligible difference between @xmath60 and @xmath45 , and the @xmath61 ratio of ( ) , which gives @xmath62 @xmath53 mag@xmath34 .
we obtained @xmath63 @xmath53 for star 6 , which is consistent with the canonical average value of @xmath64 @xmath53 that is typical for galactic center x - ray sources . using the observed 0.58.0 kev energy flux of @xmath65 erg s@xmath34 @xmath53 for this source , we employed the simulator to calculate the absorption - corrected x - ray flux .
we assumed an intrinsic thermal plasma energy value range of @xmath661 kev , which is appropriate ; after all , the faintness of this source and its lack of high energy emission ( low hr2 value ) , compared with the other gcr x - ray sources of ( ) , imply that the plasma temperature is unlikely to be any hotter than this .
predicts an unabsorbed x - ray flux in the range of @xmath67@xmath68 erg s@xmath34 @xmath53 for star 6 .
accounting for the distance of 8 kpc , this implies an x - ray luminosity in the range of @xmath69@xmath70 erg s@xmath34 in the 0.58.0 kev energy band .
thus , assuming @xmath71 is between 0.6 and 1 kev , the x - ray to bolometric luminosity ratio of star 6 is in the range of @xmath72 to @xmath73 .
the other o46i star , 7 , has a relatively large x - ray flux in the soft energy range ( 0.52.0 kev ) compared with star 6 and the other massive stellar x - ray sources in the gcr .
it is also relatively bright overall .
this is not surprising , since both the photometry and the extinction value that we calculated for the stellar counterpart in table 3 imply a foreground distance of 3.6 kpc .
so , if the x - ray - detected o46i stars 6 and 7 have the same intrinsic nature , then we should expect the foreground star 7 to appear brighter .
fortunately , star 7 has enough counts to warrant a fitting of its x - ray spectrum , which can better constrain the physical parameters of the x - ray emission . to do this
, we used the program ( ) .
first , the x - ray spectral data were binned to have at least 30 counts in each energy bin . then , we performed least squares fits to the data by trying out two different emission models , including a power - law model and a thermal plasma model .
the power - law fit required an unrealistically steep photon index of @xmath74 , so we do not regard this as a viable model for star 7 . however , the thermal plasma model fit the data reasonably well .
the parameters fit by the model were @xmath58 , @xmath71 , and a normalization factor that is proportional to the emission measure .
for the iterative fitting procedure , we used initial guess values of @xmath75 @xmath53 , derived from the infrared photometry , and @xmath76 kev .
after 30 fitting iterations , returned values of @xmath77 @xmath53 , @xmath78 kev , and an energy flux of @xmath79 erg s@xmath34 @xmath53 .
the reduced @xmath80 of the fit was 0.7 for 13 degrees of freedom . using these parameters we calculated an absorbtion - corrected x - ray flux of @xmath81 erg s@xmath34 @xmath53 . using the derived distance of 3.6 kpc , we calculated an x - ray luminosity of @xmath82 erg s@xmath34 for star 7 , which implies an x - ray to bolometric luminosity ratio of @xmath72 , the estimated plasma temperatures and x - ray luminosities of stars 6 and 7 are consistent with the values that have been measured for the gcr o46i stars in the sample of ( ) , and the o46i stars in the cygnus ob2 sample of ( ) .
o stars typically exhibit @xmath83 , and this is true for both single stars and those in massive binaries ( ) .
this relation is a result of the connection between the strength of radiatively driven winds and the temperature of the resulting shocks , which may occur inside the stellar wind via turbulence - induced micro shocks , or from wind collision if the o star is in a massive binary .
however , the plasma temperatures reached via the inter - wind micro - shock process typically have @xmath57 kev , while shocks in colliding - wind binaries generate higher temperatures ( @xmath842 kev ) and produce harder x - ray spectra .
our derived value of @xmath78 kev is insufficient to effectively discriminate between these two mechanisms .
although the average photon energy of both sources is @xmath853.0 kev , these photons could simply be the tail of a much softer intrinsic spectrum whose lower energy photons were absorbed by the ism .
alternatively , our assumption of a single temperature plasma for star 7 might be too simplistic , and the highest energy photons detected from this source could actually be from an intrinsic hard component with @xmath842 kev . but without more counts in the softer energy bands , we can not reliably fit a multi - temperature plasma model to the data , which is generally the case for high x - ray flux o stars in cygnus . in conclusion ,
the x - ray properties of stars 6 and 7 are consistent with thermal emission from either single o stars or those in colliding - wind binaries , although we currently have no good reason to invoke binarity for either source .
there are currently 92 identified wrs in the gcr , including all known wne ( wn46 ) , wnl ( wn79 ) , wce ( wc47 ) and wcl ( wc89 ) stars identified in the field ( including those identified outside of the p@xmath0 survey area ) , and those identified in the arches , quintuplet , and central clusters .
their subtype statistics are presented in table 5 .
the most noteworthy characteristic of the wr subtype distribution is the high late - type to early - type ratio for both wc and wn stars ; out of the 55 wn stars , only 5 are wne types ; and out of 36 wc stars , only 1 is a wce type .
wc stars comprise 40% of the total wr population ( 47% if we exclude the arches cluster , which is too young to have produced wc stars ) .
figure 7 shows a plot of p@xmath0 line strength ( defined here as the f187n / f190n flux ratio ) versus @xmath18-band magnitude for known wrs in the gcr and the remaining unidentified p@xmath0-excess candidates in the survey area (; dong et al .
2011 , in preparation ) .
the known sources in figure 7 include emission - line stars from the gcr field , and from the arches , quintuplet , and central parsec clusters .
the @xmath18-band values of unidentified sources were obtained by cross - correlating the positions of the p@xmath0-excess candidates with sources from the sirius and 2mass catalogs ( we make a note of caution that a few of the unidentified p@xmath0 sources having the faintest @xmath18 matches might be spurious associations in this crowded region , and that sirius magnitudes may have significant systematic errors because of source confusion , especially near the central parsec cluster ) .
the data in figure 7 suggest that all but a few of the remaining unidentified p@xmath0 candidates occupy a brightness range of @xmath18 and f187n / f190n that is systematically offset from the range occupied by confirmed wrs .
thus , an important question naturally arises as to the completeness of the wr sample at the inner @xmath86 pc@xmath11 around sgr a@xmath3 ( the survey area ) .
wnl stars , which almost completely dominate the current wn sample , have f187n / f190n ratios between 1.2 and 2.9 , and have bright counterparts with @xmath8712 mag .
wnl types are expected to generally have @xmath88 mag when observed in the gcr ( e.g. , see , their figure 25 ) .
thus , we suspect that only a few wnl stars , at most , could remain unidentified in the survey area .
this also appears to be the case for wne types ; those confirmed thus far , all of which have subtypes in the wn56 range , have several of the highest f187n / f190n ratios in the entire sample of wrs . none of the unidentified p@xmath0-excess candidates have f187n / f190n ratios this large .
earlier wne stars of subtype wn34 have not yet been found in the gcr .
although they are intrinsically the faintest of the wn stars ( @xmath89 mag ; ) , having expected @xmath21-band magnitudes as faint as @xmath414 in the gcr , wn34 stars will exhibit p@xmath0 excess similar to the wn56 stars .
therefore , they should still be easily detectable as relatively strong sources of p@xmath0 excess spectra of wn36 and wc46 stars obtained from the ground ( not shown here ) .
the spectra of the wne stars exhibit large flux at @xmath71.87 that can penetrate the severe atmospheric absorption at this wavelength , which implies that such stars will generally have high f187n / f190n ratios . ] , modulo the presence of a bright , diffuse background .
thus , although figure 7 indicates that several wn stars could remain unidentified , we suspect the wn sample in the survey area is near completion nonetheless .
this implies that wnl types heavily dominate the gcr wn population , and that wne subtypes are intrinsically rare in the region .
the situation is less clear for wc stars .
their f187n / f190n ratios span a range of values similar to the distribution of wnl types , but the faint end of their expected brightness distribution in the gcr extends down to @xmath90 mag for the earliest wce subtypes ( ) .
this has been confirmed by photometry of the wc56 star irs 3e of the central parsec cluster ( e58 from ) , which has @xmath91 mag and is the only wce star identified in the gcr to date .
unfortunately , the f187n / f190n ratio for this source could not be measured because of source confusion and the strong , diffuse background emission from sgr a west .
even with relatively large intrinsic f187n / f190n ratios , however , sources as faint as irs 3e will have large photometric uncertainties that could render them weak or undetectable sources of p@xmath0 excess in the _ hst_/nicmos survey ( dong et al .
2011 , in preparation ) .
thus , it is difficult to ascertain the detectability of a hypothetical population of wce types in the gcr , and we can not reliably estimate the potential contribution of such stars to the sample of unidentified sources in figure 7 . the same is true for wcl types , which can also be very faint ( @xmath92 mag ) and whose line strengths can be heavily diluted in cases where there is significant flux contribution from a host dust continuum .
extremely dusty wcl stars ( dwcl ; ) , like the quintuplet proper members , whose emission lines have been completely diluted by a thermal continuum or obscuring dust screen , are not detectable p@xmath0 excess sources at all ; so , some of these stellar types could remain unidentified .
however , dwcl stars are very bright infrared sources ( @xmath937 mag ) and typically have x - ray counterparts . based on these properties ,
the statistics of x - ray / infrared matches from ( ) , combined with the spectroscopic survey results of ( ) , imply that most , if not all , dwcl stars in the survey area should have been found already .
still , less extreme wc9d stars with diluted emission lines that are not detectable x - ray sources might remain elusive .
the near completeness of the wn stars in the survey area implies that apparent dominance of wnl stars over wne stars is an intrinsic property of the gcr .
although the sample of wc stars is probably not complete , we also suspect that wcl stars are intrinsically dominant over wce types .
the reasons for this are as follows .
it has been well established that wr subtype distributions are correlated with environmental metallicity ; this has been established for the milky way and other galaxies of the local group ( ) .
wne and wce stars appear to be rare in relatively high - metallicity environments ( @xmath94 ) , such as the inner galaxy , but are common in low - metallicity environments , such as the outer galaxy ( @xmath95 ) , and the large magellanic cloud ( @xmath96 ) .
this observed trend has been interpreted as a result of the metallicity dependence of wr winds ( see and references therein for the following discussion ) . for wn stars ,
the higher wind density that results from increased metallicity leads to more efficient atomic recombination from high ionization stages ( e.g. , n@xmath97 ) to lower ions ( e.g. , n@xmath98 ) , which effectively increases the trend toward later spectral types .
alternatively , decreased recombination efficiency and reduced nitrogen content at lower metallicity results in a trend toward earlier wn subtypes .
for wc stars in high - metallicity environments , higher wind densities lead to an increased emission strength of c iii with respect to c iv .
since wcl and wce stars are classified as such based upon the relative strengths of c iii and c iv emission lines , wc stars in environments with @xmath94 will tend to manifest as wcl subtypes ( c iii stronger than c iv ) .
although this effect was originally suggested by ( ) to be an important factor for the optical classification of wc stars ( specifically for c iii at @xmath75696 ) , the effect of elevated c iii relative to c iv for increasing wind density is likely to be mirrored in the infrared spectrum of wc stars as well .
thus , the dominance of wnl and wcl stars over wne and wce stars in the gcr is a natural consequence of the fact that the metallicity of the region is at least solar ( e.g. , see , and references therein ) , if not supersolar ( ; ) . figures 8 and 9 show the distribution of isolated " massive stars very near the arches and quintuplet clusters . for the arches ,
there are three wnh stars that lie within @xmath42 pc , in projection , of the cluster .
the magnitudes and spectral types of these outer wnh stars are very similar to the brightest arches members , some of which have estimated initial masses above 100 ( ) . based on the spectral similarities and the close spatial proximity , we suggest that the three massive stars surrounding the arches cluster are coeval with the system .
the main uncertainty is with regard to their current kinematical state .
one might expect that mass segregation via dynamical friction should concentrate massive stars like these toward the cluster center .
however , higher order processes occurring at the cluster center might result in an alternative effect .
indeed , the arches cluster is the most dense system known in the galaxy ( @xmath99 pc@xmath100 ; ) . as such
, the central region of this cluster must be the site of frequent gravitational encounters between massive stars and binaries , including binary captures , disruptions , and exchanges . through such interactions
, one might expect a few massive stars to be gravitationally kicked " from time to time beyond the tidal radius , which is @xmath41 pc for the arches and quintuplet ( )
. this could be the case for stars 11 and 12 of this work , and [email protected] from ( ) , all of which lie well beyond the arches cluster s tidal radius .
given that the internal velocity dispersion of the arches cluster is @xmath10210 km s@xmath34 ( ) , a massive star would not require an unusually large gravitational kick " to escape the cluster .
alternatively , one or more of these stars could have formed at the same time and from the same cloud as the main cluster but never became bound to the system .
if this is the case then the stars could simply be following a path that is in tandem with that of the cluster , without orbiting the cluster center .
these possibilities may also be applicable to the massive stars surrounding the quintuplet cluster .
( ) suggested that the o6if@xmath28 x - ray source cxogc j174617.0@xmath19285131 , a suspected colliding - wind binary , may have originated in the quintuplet , while the same might be true for the nearby lbv [email protected] ( ) .
observationally distinguishing between the various possible origins of the stars surrounding the arches and quintuplet clusters would be very difficult using radial velocities alone , since both ejected stars _ and _ comoving stars that were never bound to the clusters would exhibit similar radial velocities , perhaps differing on the order of only 10 km s@xmath34 .
proper motion measurements , however , would allow for the derivation of a three - dimensional space velocity vector , which would better elucidate the association between these stars and the clusters , and could provide evidence for ejection , specifically .
such measurements are within the capabilities of 10 m class telescopes utilizing adaptive optics . for example , ( ) used proper motion measurements over a 4.3 year baseline to derive a three - dimensional space velocity of @xmath103 km s@xmath34 for the arches cluster . as coeval siblings of the arches cluster stars
, we might expect stars 11 , 12 , and [email protected] to have a similar velocity , whether they were ejected from the cluster or are just comoving with it .
multiple epoch measurements with longer time baselines , however , could discriminate between ejection and comoving scenarios by achieving the required accuracy to obtain a velocity _ difference _ vector between the individual star vectors and the average cluster vector .
if the stars were ejected from the cluster , we would expect these difference vectors to point away from the cluster .
the continuum - subtracted p@xmath0 mosaic of the inner @xmath86 pc@xmath11 of the gcr from ( ) is shown in figure 10 , where the positions of the known clusters and the isolated " massive stars are illustrated . aside from the several stars potentially associated with the arches and quintuplet clusters , some of the other new discoveries appear to reside in regions that were heretofore unidentified as potential associations .
there is a noteworthy association of 3 wc9 stars , a wn7 , and an ob supergiant at @xmath104 .
several promising unidentified candidates having @xmath105 and @xmath106 mag and f187n / f190n ratios consistent with them being wc or ob stars also lie in this region .
this area is far from any of the known clusters , so , these stars might be the brightest representative members of a bona fide association .
there is a more highly concentrated group containing a wc9d , an ofpe / wn9 , and a b2 ia star at @xmath107 , originally identified as a potentially coeval system by ( ) . in addition
, there appears to be a noteworthy collection of 2 wc9 stars and an o supergiant near @xmath108 , and a significant concentration of wn stars near @xmath109 and @xmath110 , including the central stars of the radio h ii regions h1 , h2 , h5 , and h8 ( ) .
could these associations of massive stars represent the tip of the iceberg " of stellar clusters that are less compact than the arches and quintuplet ? there is no measured stellar density enhancement in the vicinity of isolated massive stars in the gcr ( ; dong et al .
2011 , in preparation ) .
but because of the strong tidal field in the gcr , stellar clusters like the arches and quintuplet will dissolve on a timescale of @xmath410 myr ( ) , while the time it takes for less extreme stellar clusters to fade into the background may be shorter than this . according to ( )
the cluster dissolution timescale is a sensitive function of galactocentric radius , initial cluster mass , and initial density , such that less compact stellar clusters could become invisible as stellar density enhancements on a shorter timescale than the arches or quintuplet ages , or could even persist as undetectable entities for their entire lives . in such cases , an earth - based observer undertaking
a narrowband imaging survey of the gcr might see a small group of evolved emission - line stars with no obvious surrounding stellar cluster
. this could be the case for the associations that we have noted above . as an alternative explanation for the isolated " massive stars ,
there may be two modes of massive star formation operating in the gcr , one that produces dense starburst systems like the arches and quintuplet , and one that produces massive stars in small , sparse associations or in total isolation .
currently , we can not discriminate between cluster and isolated star formation origins ; but continued surveys for coeval massive stars in the vicinity of the new emission - line stars could provide inroads to solving this problem . the project is partly supported by nasa through a grant hst - go-11120 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 .
llrrrrrrc 1 & [email protected] & 266.247738 & @xmath1929.090525 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 + 2 & [email protected] & 266.279460&@xmath116&@xmath117&@xmath118&@xmath119 & @xmath120&@xmath121 + 3 & [email protected] & 266.341231&@xmath122 & & @xmath123&@xmath124 & @xmath125&@xmath126 + 4 & [email protected] & 266.290860&@xmath127&@xmath128&@xmath129&@xmath130 & @xmath131 & @xmath132 + 5 & [email protected] & 266.350673&@xmath133 & & @xmath134&@xmath135 & @xmath136&@xmath137 + 6 & [email protected] & 266.381294&@xmath138&@xmath139&@xmath140&@xmath141 & @xmath142&@xmath143 + 7 & [email protected] & 266.408263&@xmath1929.026258 & @xmath144 & @xmath145&@xmath146 & @xmath147&@xmath148 + 8 & [email protected] & 266.385493&@xmath149 & & @xmath150&@xmath151 & @xmath152&@xmath153 + 9 & [email protected] & 266.422032&@xmath154&@xmath155&@xmath156&@xmath157 & @xmath158&@xmath159 + 10 & [email protected] & 266.426394&@xmath160&@xmath161&@xmath162&@xmath163 & @xmath164&@xmath165 + 11 & [email protected] & 266.452572&@xmath166 & & @xmath167&@xmath168 & @xmath169&@xmath170 + 12 & [email protected] & 266.472510&@xmath171&@xmath172&@xmath173&@xmath174 & @xmath175&@xmath176 + 13 & [email protected] & 266.506989&@xmath177&@xmath178&@xmath179&@xmath180 & @xmath181&@xmath182 + 14 & [email protected] & 266.510914&@xmath183&@xmath184&@xmath185&@xmath186 & @xmath187&@xmath188 + 15 & [email protected] & 266.541810&@xmath189&@xmath190&@xmath191&@xmath192 & @xmath193&@xmath194 + 16 & [email protected] & 266.573243&@xmath195&@xmath196&@xmath197&@xmath198 & @xmath199&@xmath200 + 17 & [email protected] & 266.599325&@xmath201&@xmath202&@xmath203&@xmath204 & @xmath205&@xmath206 + 18 & [email protected] & 266.615120&@xmath207&@xmath208&@xmath209&@xmath210 & @xmath211&@xmath212 + 19 & [email protected] & 266.460700&@xmath213&@xmath214&@xmath215&@xmath216 & @xmath217&@xmath218 + & [email protected] & 266.523436 & @xmath1928.858866 & @xmath219 & @xmath220&@xmath221 & 918@xmath22214 & @xmath223 lccc 1 & 2009 aug 4 07:31 & irtf / spex & 2000 + 2 & 2009 june 13 07:20 & soar / osiris & 1200 + 3 & 2009 aug 4 08:06 & irtf / spex & 1200 + 4 & 2009 june 13 05:29 & soar / osiris & 1200 + 5 & 2007 aug 1 08:30 & ukirt / uist & 2133 + 6 & 2010 may 24 06:52 & soar / osiris & 1200 + 7 & 2010 may 27 10:23 & soar / osiris & 1200 + 8 & 2009 aug 4 07:42 & irtf / spex & 1200 + 9 & 2007 aug 1 09:05 & ukirt / uist & 2133 + 10 & 2007 aug 1 06:36 & ukirt / uist & 2133 + 11 & 2008 jul 31 09:10 & ukirt / uist & 2133 + 12 & 2007 aug 1 05:57 & ukirt / uist & 2133 + 13 & 2008 jun 17 09:19 & soar / osiris & 3000 + 14 & 2008 may 16 14:00 & aat / iris2 & 2400 + 15 & 2009 aug 4 08:40 & irtf / spex & 1200 + 16 & 2008 may 14 19:37 & aat / iris2 & 2400 + 17 & 2008 may 16 15:45 & aat / iris2 & 2400 + 18 & 2008 jun 13 01:10 & soar / osiris & 1200 + 19 & 2009 may 16 13:41 & aat / iris2 & 2400 llrrrcccccc 1 & b0i b2i & 7.03 & @xmath190.21 & @xmath190.10 & 0.94 & 0.96 & 0.95 & @xmath196.71 & @xmath193.7 & 6.06 + 2 & wc9?d & 11.11 & 0.23 & 0.26 & 2.91 & 2.86 & 2.89 & @xmath196.28 & @xmath193.6 & 5.85 + 3 & wc9 & 12.70 & 0.23 & 0.26 & & 4.06 & 4.06 & @xmath195.86 & @xmath193.6 & 5.68 + 4 & wc9?d & 11.14 & 0.23 & 0.26 & 2.21 & 2.15 & 2.18 & @xmath195.54 & @xmath193.6 & 5.56 + 5 & b0i b2i & 11.36 & @xmath190.21 & @xmath190.10 & & 3.16 & 3.16 & @xmath196.30 & @xmath193.7 & 5.90 + 6 & o46i & 11.37 & @xmath190.21 & @xmath190.10 & 1.96 & 2.05 & 2.01 & @xmath224 & @xmath194.3 & 5.68 + 7 & o46i & 8.90 & @xmath190.21 & @xmath190.10 & 1.17 & 1.14 & 1.16 & @xmath224 & @xmath194.3 & 5.68 + 8 & wc9 & 12.02 & 0.23 & 0.26 & & 3.12 & 3.12 & @xmath225 & @xmath193.6 & 5.58 + 9 & o46if@xmath28 & 9.86 & @xmath190.21 & @xmath190.10 & 2.54 & 2.67 & 2.61 & @xmath197.25 & @xmath194.3 & 6.52 + 10 & o46if@xmath28 & 10.11 & @xmath190.21 & @xmath190.10 & 2.37 & 2.40 & 2.39 & @xmath196.78 & @xmath194.3 & 6.33 + 11 & wn89h & 11.14 & 0.13 & 0.11 & & 3.41 & 3.41 & @xmath196.77&@xmath194.2 & 6.29 + 12 & wn89h & 11.00 & 0.13 & 0.11 & 2.09 & 2.05 & 2.07 & @xmath195.57 & @xmath194.2 & 5.81 + 13 & p cyg - type oi & 9.12 & @xmath190.21 & @xmath190.10 & 2.24 & 2.41 & 2.33 & @xmath197.71&@xmath193.7 & 6.46 + 14 & wc9 & 11.61 & 0.23 & 0.26 & 2.41 & 2.14 & 2.28 & @xmath195.17&@xmath193.6 & 5.41 + 15 & wn89h & 10.79 & 0.13 & 0.11 & 2.01 & 1.98 & 2.00 & @xmath195.71 & @xmath194.2 & 5.86 + 16 & wn89h & 10.46 & 0.13 & 0.11 & 2.17 & 2.24 & 2.15 & @xmath196.19&@xmath194.2 & 6.06 + 17 & wn5b & 11.43 & 0.36 & 0.26 & 2.40 & 2.09 & 2.24 & @xmath195.31&@xmath194.4 & 5.54 + 18 & p cyg - type oi & 9.55 & @xmath190.21 & @xmath190.10 & 3.37 & 3.35 & 3.36 & @xmath198.31 & @xmath193.7 & 6.70 + 19 & wc9 & 11.34 & 0.23 & 0.26 & 2.10 & 1.92 & 2.01 & @xmath195.17 & @xmath193.6 & 5.41 spectral type & o46i & o46i + cxogc j & 174531.4@xmath19285716 & 174537.9@xmath19290134 + r. a. ( j2000 ) & 266.38108 & 266.40829 + decl .
( j2000 ) & @xmath1928.95466 & @xmath1929.02625 + @xmath226 ( arcsec . ) & 0.5 & 0.3 + @xmath227 ( 0.52.0 kev ) & @xmath228 & @xmath229 + @xmath230 ( 2.08.0 kev ) & @xmath231 & @xmath232 + @xmath233 ( s@xmath34 @xmath53 ) & @xmath234 & @xmath235 + hr0 & @xmath236 & @xmath237 + hr2 & @xmath238 & @xmath239 + @xmath240/photon ( kev ) & 3.0 & 2.5 + @xmath71 ( kev ) & 0.61.0 & @xmath2410.2 + @xmath58 ( @xmath242 @xmath53 ) & 5.9 & @xmath243 + norm .
factor & & @xmath244 + @xmath245 ( erg s@xmath34 @xmath53 ) & @xmath67@xmath68 & @xmath246 + @xmath247 ( erg s@xmath34 ) & @xmath69@xmath70 & @xmath248 + @xmath249 & @xmath250 to @xmath73 & @xmath250 lccccccccc gcr isolated " & 3 & 15 & 0 & 10 & 28 & 0.56 & 1.00 & 0.36 & 1 , 2 , 3 , 4 , 5 , 6 , 12 + arches & 0 & 13 & 0 & 0 & 13 & n / a & 0.00 & 0.00 & 7 , 11 + quintuplet & 1 & 5 & 0 & 14 & 20 & 2.33 & 1.00 & 0.70 & 2 , 4 , 8 , 9 + central & 1 & 17 & 1 & 12 & 31 & 0.72 & 0.92 & 0.42 & 10 + * gcr total * & 5 & 50 & 1 & 36 & 92 & 0.67 & 0.97 & 0.40 & | we report the discovery of 19 hot , evolved , massive stars near the galactic center region ( gcr ) .
these objects were selected for spectroscopy owing to their detection as strong sources of paschen-@xmath0 emission - line excess , following a narrowband imaging survey of the central @xmath1 @xmath2 around sgr a@xmath3 with the _ hubble space telescope_. discoveries include six carbon - type ( wc ) and five nitrogen - type ( wn ) wolf - rayet stars , six o supergiants , and two b supergiants .
two of the o supergiants have x - ray counterparts having properties consistent with solitary o stars and colliding - wind binaries .
the infrared photometry of 17 stars is consistent with the galactic center distance , but two of them are located in the foreground .
several wc stars exhibit a relatively large infrared excess , which is possibly thermal emission from hot dust .
most of the stars appear scattered throughout the gcr , with no relation to the three known massive young clusters ; several others lie near the arches and quintuplet clusters and may have originated within one of these systems .
the results of this work bring the total sample of wolf - rayet stars in the gcr to 92 .
all sources of strong p@xmath0 excess have been identified in the area surveyed with _ hst _ , which implies that the sample of wn stars in this region is near completion , and is dominated by late ( wnl ) subtypes . the current wc sample ,
although probably not complete , is almost exclusively dominated by late ( wcl ) subtypes .
the observed wolf - rayet subtype distribution in the gcr is a reflection of the intrinsic rarity of early subtypes ( wne and wce ) in the inner galaxy , an effect that is driven by metallicity . |
molecular collisional excitation rate coefficients are important as they contribute to level excitation ( and quenching or deexcitation ) in competition with radiative processes . in cold environments such as
the ism , the dominant collision partners are often h@xmath3 and he , except in photodissociation regions ( pdrs ) and diffuse gas where collisions with electrons and
h can become important .
precise laboratory data including collisional ( de)-excitation rate coefficients are required , for a range of temperatures , to interpret the complicated interstellar spectra of molecular gas not in local thermodynamic equilibrium ( lte ) .
because of the complexity and difficulty in direct measurements , only limited state - to - state collision rate coefficients have been obtained for systems of astrophysical interest ( see , for example , * ? ? ?
therefore modeling astrophysical systems rely heavily on theoretical estimates @xcite . here
we consider the hydrogen halide hcl , which has been detected in the atmospheres of some planets , as well as in interstellar clouds @xcite .
it is an important tracer of chlorine and can be used to constrain the chlorine elemental abundance and isotope ratios .
@xcite observed it toward the sagittarious b2 complex and modeled its non - lte spectrum .
further , hcl was recently detected in the circumstellar envelope of irc+10216 by _ herschel _ @xcite .
it has also been surveyed throughout the interstellar medium ( ism ) via its rotational line emission by @xcite .
it appears that hcl is a particularly good tracer of molecular cloud cores of the highest density ; however , the density estimates are limited by the uncertainty in collision rates .
the hcl - he collision system has been investigated theoretically and experimentally in a variety of studies @xcite . the first reliable potential energy surface ( pes ) for the hcl - he complex
was presented by @xcite and used to theoretically fit experimental helium pressure broadening cross sections of dcl at very low temperature and at 300 k. @xcite computed hcl - he rate coefficients for temperatures between 10 k and 300 k on the potential by @xcite , but their calculations were limited to @xmath4 , where @xmath5 is the rotational quantum number . in their calculations , collision cross sections were computed using the molscat scattering code @xcite within the essentially exact close - coupling ( cc ) formalism . for total energies above 1200 k ,
the coupled - states ( cs ) approximation was used .
two newer pess for hcl - he have been reported @xcite .
the two - dimensional ( rigid monomer ) pes of @xcite was developed from ab initio calculations using symmetry - adapted perturbation theory ( sapt ) with an ( @xmath6 ) basis set plus midbond functions .
this sapt potential is in good agreement with the semiempirical pes of @xcite predicting that the global minimum is on the cl side .
the accuracy of the sapt potential was tested by performing rovibrational bound states calculations of the hcl - he complex , which show that the calculated transition frequencies were in excellent agreement with the measurements of @xcite .
@xcite also pointed out that the dissociation energy predicted by the sapt potential for the complex is probably more accurate than the experiment .
@xcite presented another two - dimensional hcl - he potential using the coupled - cluster with single , double , and triple excitations ( ccsdt ) method .
this potential is very similar to the sapt pes , but gives better agreement with the experimental dissociation energy and rotational energy levels of the hcl - he complex . very recently three - dimensional and two - dimensional pess were presented by @xcite and by @xcite , which used the supermolecular approach based on the ccsd(t ) method , but at different levels of theory .
these new potentials were also used in cc calculations of pure rotational excitation of hcl by he by these authors . cross sections for transitions among the first 11 rotational states of hcl
were calculated for total energy of up to 3000 @xmath2 using molscat with the log - derivative propagator of @xcite .
rate coefficients were presented from 5 to 300 k. in this work , we performed explicit quantum scattering calculations of rotational quenching of hcl by he to higher levels of rotational excitation using the sapt potential of @xcite .
hcl - he rate coefficients are presented for a large range of temperature ( 0.1 - 3000 k ) which will aid in modeling rotational spectra of hcl in various astrophysical and atmospheric environments .
we discuss the computational method in section 2 and the results in section 3 . in section 4 ,
an estimate of the uncertainty in the cross sections and rate coefficients is presented , while section 5 briefly discusses application of the current results to astrophysics .
the theory developed by @xcite for atom - diatom scattering is adopted .
hcl was treated as a rigid - rotor with bond length equal to the equilibrium distance @xmath7 .
the calculations presented here were performed by applying the cc method and the cs approximation ( see , for example , * ? ? ? * ) .
the hcl - he interaction potential is expressed by @xmath8 , where @xmath9 is the distance from the hcl center of mass to the he atom , and @xmath10 is the angle between @xmath11 and the hcl molecular axis , with @xmath12 defined for the collinear arrangement he - h - cl .
the potential @xmath13 was expanded in the form @xmath14 where @xmath15 are legendre polynomials and @xmath16 expansion coefficients of the potential . for a transition from an initial rotational state @xmath5 to a final rotational state @xmath17
, the integral cross section can be expressed in terms of the scattering matrix @xmath18 , within the cc formalism by @xmath19 where the total angular momentum @xmath20 , is composed of the rotational angular momentum @xmath21 of the hcl molecule and the orbital angular momentum @xmath22 of the collision complex .
@xmath23 is the wave vector for the incoming channel , @xmath24 the center - of - mass kinetic energy for the incoming channel corresponding to the initial rotational state @xmath5 of hcl , and @xmath25 the collision system reduced mass .
all reported scattering calculations were performed using the quantum - mechanical scattering code molscat @xcite .
the propagation was carried out from an intermolecular separation @xmath26 to @xmath27 . to ensure the accuracy of the state - to - state rate constants for temperatures from @xmath28 to 3000 k
, kinetic energies between @xmath29 @xmath2 and 15,000 @xmath2 were used in our state - to - state cross section calculations .
the angular dependence of the interaction potential was expanded in legendre polynomials shown in eq .
( 1 ) with @xmath30=22 .
24 points in @xmath10 from gauss - legendre quadrature were used to project out the potential expansion coefficients .
sufficient number of partial waves necessary for convergence of the cross sections were used ; in the higher collision energy region the maximum value of @xmath31 employed was 360 .
the cs approximation was adopted for collision energies greater than 2000 @xmath2 , while the cc method was used for all lower energies , the agreement between the cs and cc calculations is within 5% .
state - to - state quenching cross sections were computed for initial hcl rotational levels of @xmath32 .
rotational energy levels are presented in table 1 for the basis set adopted in the current calculation , which were obtained using rotational constant @xmath33=10.5933 @xmath2 @xcite and centrifugal distortion constant @xmath34=0.00053 @xmath2 @xcite . as examples ,
the state - to - state quenching cross sections from initial levels @xmath5=5 and 15 are presented in fig .
[ fig1](a ) and [ fig1](b ) , respectively .
cross sections using the cs approximation begin at 2000 @xmath2 and are seen to be in excellent agreement with those obtained with the cc method .
the cross sections display resonances in the intermediate energy region from @xmath350.1 @xmath2 to @xmath3510 @xmath2 due to quasibound levels supported by the attractive part of the interaction potential .
the @xmath36 transition dominants the quenching for all @xmath5 ( shown here for @xmath37 and @xmath38 ) , with the cross sections generally increasing with increasing @xmath17 with that for @xmath39 being the smallest .
the total deexcitation cross section from an initial state @xmath5 can be obtained by summing over all final states @xmath17 .
[ fig2 ] displays the total deexcitation cross section for quenching from selected initial levels @xmath5=2 , 4 , 6 , @xmath40 , 18 , and 20 .
generally , the total quenching cross sections have similar behavior , but differences result for small @xmath5 at high energy due to a limited number of final exit channels .
each of the cross sections exhibits the behavior predicted by wigner ( 1948 ) threshold laws at ultra - low collision energies below @xmath3510@xmath41 @xmath2 , where only @xmath42-wave scattering contributes and the cross sections vary inversely with the relative velocity . except for @xmath5=2
the total deexcitation cross sections decrease to a global minimum near 500 @xmath2 .
the state - to - state thermal rate coefficients can be calculated by thermally averaging the appropriate state - to - state cross sections over maxwell - boltzmann distribution of kinetic energy @xmath24 . to our knowledge , there has been no published experimental rate coefficients available for rotational transitions of hcl by collisions with he .
therefore , we compare our rate coefficients with the theoretical results of @xcite , @xcite , and @xcite , which were obtained over a limited temperature range of 10 to 300 k. as examples , figs .
[ fig3 ] and [ fig4 ] show selected transitions for state - to - state deexcitation rate coefficients from initial hcl levels @xmath5=1 , 3 , and 7 .
generally , our results show very good agreement with that of @xcite and @xcite , which were computed using different potentials . for initial state @xmath43 , fig .
[ fig3](a ) shows that the present rate coefficients are larger than those of @xcite , @xcite , and @xcite in the temperature range from 10 to 300 k , but their rate coefficients approach our results with increasing temperature .
[ fig3](b ) , which displays state - to - state deexcitation rate coefficients from initial level @xmath44 , shows that the results of @xcite , @xcite , and @xcite agree well with the present rate coefficients , except that for quenching to @xmath17=0 the rate coefficients of both @xcite and @xcite are larger than the present results .
state - to - state rate coefficients from initial state @xmath45 are given in fig .
[ fig4 ] where comparison of the present results with the rate coefficients of @xcite show excellent agreement . except for the transition @xmath46
, the rate coefficients of @xcite generally show good agreement with present results .
these differences are likely related to differences in the adopted pess ( see below ) . for illustration , in fig .
[ fig5 ] we also present the state - to - state deexcitation rate coefficients for temperatures from 0.1 k to 3000 k for initial levels @xmath5=10 and 20 . over the whole temperatures range considered , the rate coefficients generally increase with increasing temperature for all transitions .
further , one can clearly see that the rate coefficients decrease with increasing @xmath47 with the @xmath48 transitions dominant .
one essential prerequisite for accurate cross section and rate coefficient calculations is the availability of a pes with high accuracy .
however the quality of the pes depends not only on the methods used for the interaction energy calculation , but also the accuracy and reliability of the potential fitting and extension to long range .
convergence testing was performed in our calculations by using sufficiently large basis sets and adjustment of other calculation parameters .
consequently , the uncertainty associated with our cross section calculation is related primarily to the uncertainty in the adapted pes . in this work , the sapt pes of @xcite was applied . as discussed by @xcite
the main features of the hcl - he pes are a minimum in the linear configuration he - h - cl , global minimum at the he - cl - h configuration , and a saddle point close to the t - shaped configuration .
this also applies to pess of @xcite and @xcite .
however , the major difference between the three potentials is the well depth of the global minimum . to estimate the uncertainty of the cross sections and rate coefficients calculated from the pes of @xcite , a simple approach is to scale the sapt pes by the largest and smallest ratio of the well depths of the three pess
of course , the cross section and rate coefficients depend on the detailed structure of the potential including the shape , well depth , and position of the minumum which relies heavily on the methods applied in the hcl - he potential calculaltion . here , the well depth is used to illutrate the uncertainty in scattering calculations .
we calculated the deexcitation cross section and rate coefficient for the @xmath49 transition using the sapt pes scaled by factors of 1.0322 and 0.9688 , respectively , i.e. , a variation of @xmath50 . in fig .
[ fig6 ] we compare the @xmath49 quenching cross sections and rate coefficients obtained using the scaled sapt potential with the results obtained from the original sapt potential . due to differences in the pes well - depths , it can be seen that the cross sections and rate coefficients show evident disagreement at low collision energy and low temperature , @xmath51 k. in the van der waals interaction - dominated regime , the rate coefficients exhibit an oscillatory temperature dependence due to the presence of resonances .
the magnitudes , widths , and positions of the resonances are sensitive to the details of the pes .
however for high collision energy and high temperature , the cross sections and rate coefficients from the scaled and original sapt pes generally converge . in another words ,
the uncertainty due to the accuracy of the sapt pes is negligible at high temperatures .
we estimate the uncertainty in the computed rate coefficients to be 12% , 4% , and 1.3% at 1 , 10 , and 100 k , respectively .
rate coefficients for collisional excitation and deexcitation are of importance in describing the dynamics of energy transfer processes in interstellar objects .
in particular , accurate rotational and vibrational excitation rates are needed to interpret microwave and infrared observations of the interstellar gas for non - lte line formation .
further , the thermal balance of interstellar gas is partly determined by cooling processes involving molecular collisional excitation followed by radiative decay . despite some progress in laboratory measurements of state - to - state collisional rate coefficients and
cross sections , astrophysical models depend almost exclusively on theoretical data .
experiments do provide , when available , some confidence in the theoretical rate coefficients . as discussed in the introduction
, hcl has been observed in emission and absorption in a variety of astronomical environments .
agndez et al .
( 2011 ) reported observations of the @xmath5=1 - 0 , 2 - 1 , and 3 - 2 rotational lines of h@xmath52cl and h@xmath53cl in the carbon star envelope irc+10216 using the _ herschel_/hifi instrument .
it was inferred that hcl is produced in the inner layers of the envelope close to the asymptotic giant branch ( agb ) star .
recently , by observing the h@xmath53cl and h@xmath52cl 1 - 0 transitions with the hifi spectrometer , codella et al .
( 2012 ) presented the first detection of hcl towards protostellar shocks .
it is expected that more highly excited rotational lines may be observed with sofia in the future . however , except for the set of calculations done for temperatures between 10 and 300 k and @xmath54 by @xcite , @xcite , and @xcite , no other computations of the he excitation rate coefficients have appeared .
therefore , the current rate coefficient calculations , which extend from 0.1 to 3000 k , are the most comprehensive to date for he and can be utilized in a variety of applications augmenting the datasets developed for he , h@xmath3 and electrons ( see also , the lamda website , * ? ? ?
the rate coefficients for molecular scattering with para - h@xmath3 are often estimated using available rate coefficients with he by the application of a constant reduced - mass scaling factor of 1.4 @xcite . however , @xcite calculated the rate coefficients for sis scattering with para - h@xmath3 and compared their results to the sis - he rate coefficients @xcite scaled by the h@xmath3/he reduced - mass factor and found significant differences . in
much earlier work , schaefer ( 1990 ) found large deviations at temperatures of 100 k and below when comparing directly calculated rotational excitation rate coefficients for hd due to h@xmath3 and those obtained by reduced - mass - scaling of hd - he rates .
applying such a procedure to the current hcl - he rates coefficients to estimate rate coefficients for h@xmath3 is therefore not recommended .
the hyperfine structure splitting of hcl occurs due to its non - zero nuclear spin .
though hyperfine transitions are not considered here , the hyperfine excitation cross sections can be estimated from the current results using an infinite - order sudden approximation approach @xcite .
cross sections and rate coefficients for rotational quenching of hcl due to he collisions have been studied using the close - coupling method and the coupled - states approximation on the pes of @xcite for excited initial rotational levels of hcl up to @xmath5=20 .
state - to - state rate coefficients are obtained over a wide temperature range from 0.1 to 3000 k and available in tables formatted for astrophysical applications . the very good agreement with the results of @xcite and @xcite , computed on different
pess confirms the accuracy of the present calculations of the rate coefficients . the uncertainty in the current rate coefficient calculation , deduced by scaling the adopted potential , is less than 4% at temperatures of astrophysical interest , comparable to the divergence in the well depth of recently available pess | we report rotational quenching cross sections and rate coefficients of hcl due to collisions with he .
the close - coupling method and the coupled - states approximation are applied in quantum - mechanical scattering calculations of state - to - state cross sections for hcl with initial rotational levels up to @xmath0 for kinetic energies from 10@xmath1 to 15000 @xmath2 .
state - to - state rate coefficients for temperature between 0.1 and 3000 k are also presented .
comparison of the present rate coefficients with previous results reported in the literature for lowly - excited rotational levels shows reasonable agreement .
small differences are attributed to the differences in the interaction potential energy surfaces .
the uncertainty in the computed cross sections and rate coefficients is estimated by varying the potential well depth .
applications of current results to astrophysical systems are also briefly discussed . |
the ongoing experimental progress in the ability to efficiently couple polarizable emitters to waveguide structures opens many exciting possibilities in quantum optics research @xcite .
a central theme in this growing field is the possibility to design long - range and modified interactions between polarizable objects , mediated by the waveguide modes .
such enhanced interactions should prove useful for entanglement generation @xcite and for the study of novel many - body phenomena and quantum nonlinear optics with long - range interactions @xcite .
quantum vacuum fluctuations of waveguide modes may also drive and mediate novel casimir and van der waals forces , with drastically modified distance dependencies @xcite . of particular interest
are such vacuum - induced interactions mediated by transmission - line waveguides , which can become extremely long - range and strong @xcite , as opposed to e.g. exponentially decreasing interactions mediated by other waveguides @xcite .
the unique feature of electromagnetic phenomena in transmission - line ( tl ) waveguides is that it can be described by electrodynamics in one - dimension ( 1d ) , or equivalently , by tl circuit theory .
this is in contrast to electrodynamics in other types of waveguides , for which a 1d description of broadband and non - resonant phenomena , such as zero - point fluctuation forces , is not adequate .
transmission - line circuits in the quantum regime are now routinely used in the field of circuit quantum electrodynamics , where squeezed vacuum generation and the dynamical casimir effect were recently observed @xcite .
the aim of the present work is to consider a general theoretical framework for the description of quantum fluctuation - induced potentials inside tl circuits .
as explained below , we focus on the regime at which the 1d electrodynamic description is valid , and compare two equivalent approaches for such a theoretical framework .
the first approach considered here is a straightforward adaptation of qed to tl circuits , consisting of canonical quantization of propagating voltage and current fields @xcite .
quantum fluctuations and associated phenomena then originate in those quantized fields , and are described by an effective 1d qed theory , combined with a 1d scattering formalism @xcite . in an alternative approach
, quantization is performed only implicitly , by using the fluctuation - dissipation theorem ( fdt ) @xcite . here ,
quantum fluctuations in the form of noisy current sources , emerge from resistive elements within the circuit , in analogy to the lifshitz treatment of the casimir force between dispersive materials in free space @xcite .
the inclusion of these noisy currents within standard circuit theory , then provides an elegant route towards the calculation of fluctuation potentials in circuits , by simply using the kirchhoff circuit laws . moreover , much like in the lifshitz treatment , the advantage of the fdt description over that using canonical quantization , is the possibility to treat dissipative systems .
the paper is organized as follows . in sec .
ii we briefly review a few essential elements of transmission - line theory and present the casimir problem adapted to circuits . sec .
iii is devoted to the introduction of the qed and fdt approaches and the identification of the driving quantum fluctuations sources therein .
we proceed to the calculation of the casimir force between two circuit components within the qed and fdt approaches in sec .
iv and v , respectively , where in the latter the calculation is generalized to dissipative components and terminated lines . as an illustration
, we discuss in sec .
vi a few simple examples , where special care is taken to the possibility of repulsive forces .
we conclude by a discussion of a few formal aspects and possible experimentally relevant systems in sec .
a transmission line ( tl ) is formed by at least two conductors ( fig .
this guarantees the existence of the transverse - electromagnetic ( tem ) mode as the fundamental transverse mode of the tl @xcite .
the tem mode function , @xmath0 , where @xmath1 is the propagation direction and @xmath2 is independent of frequency , along with its linear dispersion , @xmath3 , with @xmath4 the wavenumber along @xmath1 and @xmath5 the speed of propagation , constitute 1d wave propagation at _ all _ frequencies .
this unique 1d propagation is in contrast to the guided modes of dielectric waveguides ( such as optical fibers ) or hollow metallic waveguides , where @xmath2 is frequency dependent @xcite .
all other guided modes supported by the tl possess a cutoff frequency which is higher than @xmath6 , @xmath7 being the separation between the conductors that form the tl ( fig . 1a ) .
therefore , the 1d description of electromagnetic wave propagation in tl circuits is valid for frequencies @xmath8 , where only the tem mode exists . for quantum fluctuation forces , where the driving fluctuations may exist in an immense frequency bandwidth , an electrodynamic description based only on
the tem mode is valid for @xmath9 , @xmath10 being the distance between the interacting objects @xcite . in principle , the tem mode possesses two orthogonal field polarizations , however , due to the anisotropy of circuit elements ( e.g. parallel - plate capacitors ) , we consider only a single polarization , resulting in the usual scalar tl description . considering only the tem mode ,
the electromagnetic field can be described by propagating voltage and current waves in 1d , obeying the so - called telegraphers equations @xcite @xmath11 with @xmath12 and @xmath13 the voltage and current waves at time @xmath14 and position @xmath1 along the tl . for a given section of the tl with capacitance and inductance per unit length , @xmath15 and @xmath16 , respectively , that are independent of @xmath1
, this leads to the 1d helmholtz equation in this section , @xmath17 for @xmath18 written in frequency domain .
the resulting forward and backward propagating voltage waves @xmath19 can then be reflected by an impedance @xmath20 that terminates the line , with a reflection coefficient @xcite @xmath21 and where @xmath22 is the line impedance ( real number ) . in turn , the forward / backward current amplitudes are related to their respective voltage amplitudes by @xmath23 . when the impedance @xmath24 is placed in between two semi - infinite tl sections ( fig .
1b ) , it acts as a mirror with reflection and transmission coefficients @xmath25 here we used the fact that the load of a semi - infinite tl seen by the circuit is its characteristic impedance @xmath22 ( connected in parallel to @xmath24 ) .
we note that @xmath24 may in general contain dissipative elements ( a real part ) , in which case @xmath26 ( see appendix a ) .
the electromagnetic energy of an infinitesimal section @xmath27 at position @xmath1 of the tl is given by that stored in its capacitance and inductance .
the electromagnetic energy @xmath28 of the tl is then @xmath29 with @xmath30 being the energy density ( energy per unit length ) . considering two lumped - element components with impedances @xmath31 and @xmath32 at positions @xmath33 and @xmath34 along the line ( fig .
1c ) , the force between the components mediated by the tl s tem modes is given by the discontinuity of the electromagnetic energy density around each component ( see appendix b ) @xmath35 here , the positive sign ( @xmath36 ) means an attractive force between the components . in the case of a casimir - like force ,
the task is therefore to find the voltage / current fields around the components ( e.g. at @xmath37 and @xmath38 ) , given the sources of quantum fluctuations that exist in the circuit , and insert them in eqs .
( [ h ] ) and ( [ f ] ) .
within qed , the casimir interaction is driven by the vacuum fluctuations of the electromagnetic field modes that are coupled to the interacting objects .
therefore , the first step is to perform a canonical quantization of the field modes of a `` free '' tl ( where @xmath15 and @xmath16 are independent of @xmath1 ) .
this has been done before by many authors @xcite and is also presented in appendix c. in short , we begin by writing the 1d helmholtz equation ( [ helm ] ) for the charge wave @xmath39 , find the lagrangian that yields it , and the canonical conjugate to @xmath40 , @xmath41 .
then , moving to hamiltonian formulation , quantization is performed by imposing the commutation relations @xmath42=i\hbar\delta(x - x')$ ] .
the resulting quantized voltage and current fields take the form ( appendix c ) @xmath43 , \nonumber \\
\hat{i}(x)&=&\frac{1}{z_0}\sum_{k>0}\left[\left(\hat{v}_{+k}e^{ikx}-\hat{v}_{-k}e^{-ikx}\right)+\mathrm{h.c.}\right ] .
\label{vi}\end{aligned}\ ] ] here , the terms in parentheses are the so - called positive - frequency components of the fields , and the forward / backward ( @xmath44 ) amplitudes of waves at frequency @xmath45 are given by @xmath46 where @xmath47 is a quantization length of the 1d plane wave modes @xmath48 , and where @xmath49 ( @xmath50 ) possess the commutation relations @xmath51=\delta_{\mu\mu'}\delta_{k k'}$ ] , @xmath52=0 $ ] .
the hamiltonian operator is given by eq .
( [ h ] ) with @xmath53 and @xmath54 now replaced by the operators from eq .
( [ vi ] ) , or , in terms of the operators @xmath55 by @xmath56 . in summary within the qed formulation ,
the source of casimir forces are the quantum fluctuations of the voltage and current fields ( [ vi ] ) , which propagate inside the tl with the amplitudes ( [ a ] ) . the way these fields propagate and scatter between different impedances determines the energy density and the force in eq .
( [ f ] ) . according to the fdt , the dissipation of electromagnetic energy in a resistor ( ohmic losses ) must be accompanied by current fluctuations @xmath57 inside it . at open circuit , when the resistor is disconnected , the spectrum of these fluctuations is given by ( appendix d ) @xcite @xmath58 with a corresponding voltage spectrum of @xmath59 , and where @xmath60 , @xmath61 is the duration of the experiment , and @xmath62 is the inverse temperature .
these fluctuations originate from the internal degrees of freedom of the resistor and hence should exist regardless of whether the resistor is connected to other components or not .
therefore , in order to account for these fluctuations within circuit theory , the resistor model is accompanied by a shunt current source with the spectrum ( [ in ] ) ( fig .
2a ; noting also the equivalence to a voltage source in series @xcite ) .
this guarantees the constant supply of the current noise as per the fdt . to conclude , the fdt approach for the calculation of quantum fluctuation forces in circuits goes as follows . in a given circuit , all resistors ( or equivalently , semi - infinite tls )
are modeled by their resistance @xmath63 in parallel to a current noise source with the spectrum ( [ in ] ) . by applying the kirchhoff laws in frequency domain ,
one can then find voltages and currents in the circuit and hence the energy density and force ( [ f ] ) . as already mentioned above
, the impedance of a semi - infinite tl seen by a circuit ( e.g. an electrical component ) , is equivalent to a resistance @xmath64 , with @xmath22 the characteristic impedance of the tl .
this is a manifestation of the fact that an infinite space , here 1d space , does not reflect incident waves , and is hence equivalent to dissipation , which is represented by a resistor in the case of circuits .
following the reasoning of the fdt , we would then expect that the noise inserted into a circuit from a semi - infinite tl with a characteristic impedance @xmath63 , is identical to that inserted by a resistor @xmath63 ( fig .
3 ) . this fluctuation - source equivalence can be demonstrated as follows .
first , consider the spectrum of voltage fluctuations inserted by a resistor @xmath63 into a circuit .
the input of any circuit is represented by its wires , characterized by the impedance @xmath22 of a semi - infinite tl ( fig .
the task is then to find the voltage fluctuations induced by @xmath63 that enter and propagate into this tl at @xmath33 .
the forward - propagating current @xmath65 at @xmath33 is easily found in frequency domain in terms of the current source @xmath66 , by effectively substituting the tl with an impedance @xmath22 and using kirchhoff s laws , obtaining @xmath67 .
the forward - propagating voltage is given by @xmath68 ( see sec .
iib ) , yielding the input fluctuation spectrum @xmath69 where the fdt fluctuation source of eq .
( [ in ] ) was used . in analogy , consider now a similar problem , where the resistor @xmath63 is replaced by a semi - infinite tl with characteristic impedance @xmath63 ( fig .
the forward - propagating voltage in the tl @xmath22 at @xmath37 , inserted by the tl @xmath63 , is given by @xmath70 with fluctuation spectrum @xmath71 here @xmath72 is the forward - propagating voltage fluctuation in the tl @xmath63 and @xmath73 is the transmission coefficient from left to right around @xmath33 [ eq . ( [ r ] ) for @xmath74 .
the remaining task is to find @xmath75 , which is the spectrum of the freely forward propagating fluctuations in a tl with characteristic impedance @xmath76 .
this can be done by considering the voltage operator of a free tl in the qed formulation , eqs .
( [ vi ] ) and ( [ a ] ) . in the heisenberg picture with respect to hamiltonian ( [ h ] ) ,
the operators evolve in time as @xmath77 so that @xmath78 .
\label{vp}\ ] ] assuming a thermal state for the voltage modes in tl @xmath63 at @xmath79 , we take @xmath80 with @xmath81 for @xmath82 . inserting eq .
( [ nk ] ) into eq .
( [ vp ] ) in the continuum limit @xmath83 , the voltage correlation at @xmath38 is obtained as @xmath84 . \label{rv}\end{aligned}\ ] ] by definition , the first term of the integrand in eq .
( [ rv ] ) is @xmath75 , which is inserted into eq .
( [ sqed ] ) with @xmath85 , to yield exactly the same spectrum as in eq .
( [ sfdt ] ) .
reconsidering the configuration from fig . 1c , we shall now turn to the calculation of the force between the electric components , eq .
( [ f ] ) .
the qed formalism is particulary useful for non - dissipative components for which the impedances @xmath86 are purely imaginary .
these components then form non - absorptive scatterers , or mirrors , with the reflection and transmission coefficients , eqs .
( [ rt ] ) , satisfying @xmath87 ( appendix a ) .
the expression for the force in eq .
( [ f ] ) is given in terms of the hamiltonian density , which in turn requires the knowledge of the voltage and current fields around the scatterers .
these are found below by a 1d scattering formulation . since
scattering problems are conveniently treated in frequency domain , we first write the hamiltonian density @xmath30 from eq .
( [ h ] ) using the voltage and current operators from eq .
( [ vi ] ) . taking the quantum mechanical average over @xmath30 , we assume a thermal state for the input field fluctuations arriving from both semi - infinite ends of the tl , so that @xmath88 , @xmath89 ( @xmath90 ) . the hamiltonian density then assumes the following spectral representation @xmath91 which depends on @xmath1 only through the tl section wherein @xmath1 is located . therefore , for the force ( [ f ] ) , we need to find the amplitudes @xmath92 of the forward and backward propagating voltages at the different tl sections , e.g. those of @xmath79 and @xmath93 around the scatterer @xmath31 at @xmath33 .
this defines the following scattering problem ( as in ref .
we divide the 1d space into three sections 1,2 and c for @xmath79 , @xmath94 and @xmath93 , respectively .
then , the field amplitudes of all sections are found as a function of input fields , consisting of the freely - propagating amplitudes , @xmath95 ( forward - propagating , section 1 ) and @xmath96 ( backward - propagating , section 2 ) , with @xmath97 from eq .
( [ a ] ) .
the solution is performed as usual by matching the fields at the boundaries @xmath33 and @xmath34 , obtaining @xmath98 with @xmath99 and @xmath100 the reflection / transmission coefficients , eqs .
( [ rt ] ) , for purely imaginary impedances @xmath101 . solving for the field
amplitudes in terms of the inputs , @xmath92 , we find @xmath102 with the reflection and scattering matrices , @xmath103 the force , calculated on the impedance @xmath31 , is given by @xmath104 , with @xmath105 the hamiltonian density in section @xmath106 .
beginning with @xmath107 we insert @xmath95 and @xmath108 from eqs .
( [ sqed ] ) and ( [ rs ] ) into eq .
( [ hqed ] ) .
taking the statistics of a thermal state , eq .
( [ nk ] ) , for the input fields @xmath92 from eq .
( [ a ] ) , and the continuum limit @xmath109 , we find @xmath110 with the notation @xmath111 for the scattering matrix from eq .
( [ rs ] ) . for the non - absorptive scatterers case
above , where @xmath86 are purely imaginary and @xmath112 , we verify that @xmath113 as it should , such that @xmath114 for @xmath115 we insert @xmath116 from eqs .
( [ sqed ] ) and ( [ rs ] ) inside eq .
( [ hqed ] ) and find @xmath117 where again @xmath112 was used . finally , by subtracting eq .
( [ hc ] ) from eq .
( [ h1 ] ) we obtain the force @xmath118 , \label{fqed}\ ] ] identical to that found for 1d scattering in ref .
@xcite .
an interesting point is that of _
renormalization_. both @xmath107 and @xmath119 are infinite , however only @xmath119 depends on the existence of the scatterers and their separation .
therefore , one may view @xmath119 as the `` bare '' force between the scatterers whereas @xmath107 as merely a reference term originating in the free tl .
indeed , @xmath107 is identical to the energy density of a free tl with @xmath120 modes traveling at velocity @xmath5 , each with a zero - point energy @xmath121 . as such
, it can be also associated with the free - tl value of section @xmath122 rather than that of section 1 .
using such an interpretation , the above renormalization is similar to that used in the lifshitz approach , where the reference energy subtracted from a diverging stress at a given point is that of the corresponding `` free '' homogenous system @xcite .
the casimir problem presented in fig . 1c and solved in the previous section presents a direct adaptation of the well - known free - space ( 3d ) problem to its tl counterpart , which is nevertheless not the most natural problem to consider in circuits .
in particular , the semi - infinite tl sections on the right and left sides of the scatterers @xmath32 and @xmath31 , respectively , that were introduced therein in analogy to mirrors placed in free space , can be readily replaced by general complex impedances @xmath123 ( @xmath124 ) that _ terminate _ the line , as in fig .
we can decompose any impedance @xmath123 into a pair of real and imaginary impedances , @xmath125 and @xmath126 , respectively , connected in parallel ( fig .
4b ) and satisfying @xmath127=x_n\frac{r_nx_n}{r_n^2+x_n^2 } , \quad \mathrm{im}[z_n]=r_n\frac{r_nx_n}{r_n^2+x_n^2}. \label{zn}\ ] ] the lossless - mirrors problem of the previous section then corresponds to @xmath128 , whereas the general @xmath129 problem is equivalent , in the scattering formulation , to two lossless mirrors @xmath130 `` sandwiched '' between two different `` media '' with characteristic impedances @xmath131 and @xmath132 and which are separated by a medium with characteristic impedance @xmath22 .
the general casimir problem of fig .
4a is however naturally understood without the need of such scattering interpretation : the task is to find the zero - point force induced between two electric components due to the fact that they are connected by a wire ( tl ) .
the fdt approach used below , allows to solve this problem using the kirchhoff circuit laws , naturally applied to circuits in the frequency domain . in the following
, it will be therefore useful to write the hamiltonian density of the tl , eq .
( [ h ] ) , in frequency representation @xmath133 \nonumber\\ & = & \frac{1}{2}l'\int_{-\infty}^{\infty}\frac{d\omega}{2\pi}\left[s_{i_+(x)}(\omega)+s_{i_-(x)}(\omega)\right ] .
\label{hfdt}\end{aligned}\ ] ] here the spectrum @xmath134 of a signal @xmath135 is the same as that defined in eq .
( [ in ] ) , and we recall the forward / backward waves at frequency @xmath45 which are defined by @xmath136 with @xmath137 and the relation @xmath138 ( see eq .
[ helm ] ) . in analogy to sec .
iv a , the idea is to find the hamiltonian density between the scatterers , e.g. at @xmath139 , from which the force is deduced by a proper renormalization ( subtraction ) . noting eq .
( [ hfdt ] ) , we then need to find the forward and backward current amplitudes at @xmath33 , @xmath140 .
using the circuit diagram of fig .
4b at frequency @xmath141 , the current @xmath142 on @xmath31 is given by @xmath143=\frac{z_0}{z_1}[i_+(0)-i_-(0 ) ] .
\label{i1}\ ] ] combining it with the kirchhoff current law at @xmath33 , @xmath144 , with @xmath145 the noise current eq .
( [ in ] ) with resistance @xmath131 ( fig .
4 ) , we obtain @xmath146 similar considerations at @xmath34 lead to @xmath147 considering the definition of the reflection coefficient @xmath148 of a line terminated by impedance @xmath149 , eq .
( [ r ] ) , along with the definition @xmath150 for @xmath124 , the solution of eqs .
( [ k1 ] ) and ( [ k2 ] ) becomes @xmath151 , \nonumber\\ i_-(0)&=&\frac{-e^{ikl}}{2(1-r_1r_2e^{i2kl})}\left[t_2i_{n_2}+ r_2t_1e^{ikl}i_{n_1}\right ] , \label{ipm}\end{aligned}\ ] ] where the relation between the propagating current waves , @xmath152 , was used .
assuming that the noise input from the two resistances @xmath129 ( fig .
4 ) are uncorrelated , and considering their corresponding noise spectra from eq .
( [ in ] ) , the spectra @xmath153 of the currents @xmath140 from eq .
( [ ipm ] ) are found .
inserting @xmath153 into eq .
( [ hfdt ] ) , the energy density at @xmath33 becomes @xmath154 using @xmath155 for @xmath124 ( appendix a ) , @xmath156 , and the expressions for @xmath5 and @xmath22 [ eqs .
( [ helm ] ) and ( [ r ] ) ] , we obtain @xmath157 in order to convert the integral to positive @xmath4-values we have used the fact that the absolute values of the linear responses @xmath99 and @xmath158 are even functions of @xmath4 , since they result from fourier transforms of real functions .
the above result is identical to @xmath159 calculated within the qed formalism , eq .
( [ hc ] ) .
the physical picture provided by the qed scattering problem was that of non - absorptive mirrors and fields transmitted to infinity , equivalent to the specific case @xmath160 in fig .
4 . the circuit problem of fig .
4 is however more general , so that the above expression for @xmath161 is valid for reflection coefficients of _ any _ complex impedances that effectively _ terminate _ the line . since the fdt result for @xmath161 , eq .
( [ h0 ] ) , is identical to that from eq .
( [ hc ] ) , it can also be renormalized in the same way , i.e. by subtraction of the free - tl result of eq .
( [ h1 ] ) . whereas in the qed approach the physical justification for this subtraction could be attributed to a counter force acting at @xmath38 [ eq .
( [ f ] ) ] , a more constructive point - of - view is the latter lifshitz - theory - like interpretation given above ; namely , the stress @xmath161 at @xmath37 is renormalized by the subtraction of that of a free tl with the same line parameters @xmath15 and @xmath16 as in @xmath37 . to conclude ,
the tl circuit - mediated casimir force between two general complex impedances @xmath86 which terminate a tl with characteristic impedance @xmath22 is given by @xmath118 , \label{ffdt}\ ] ] with @xmath99 the reflection coefficients from eq .
( [ r ] ) .
this result is in contrast to that obtained by the qed approach , eq . ( [ fqed ] ) , where the reflection coefficients are those from eq . ( [ rt ] ) .
therefore , the above fdt - obtained result ( eq . [ ffdt ] ) extends its validity of its qed - obtained counterpart ( eq . [ fqed ] ) to include dissipative scatterers / impedances that terminate the line .
we shall now illustrate the application of the general formula , eq .
( [ ffdt ] ) , with a few simple examples at zero temperature , where we take the two impedances to terminate the tl ( as in fig .
4 ) . in particular
, we wish to focus on the sign of the force , either attractive or repulsive , which is facilitated by performing the integral ( [ ffdt ] ) in imaginary frequencies , the so - called wick rotation @xcite . out of the four examples considered below , the first two are of pedagogical nature , whereas the latter two may also have practical implications .
we begin by rewriting eq .
( [ ffdt ] ) at zero temperature ( @xmath162 ) as @xmath163 , \quad i=\int_0^{\infty}dv v\frac{r_1(v)r_2(v)e^{i2v}}{1-r_1(v)r_2(v)e^{i2v } } , \label{i}\ ] ] with @xmath164 being a dimensionless frequency .
considering that the reflections @xmath99 and the fabry - prot cavity response @xmath158 are causal linear response functions , they are analytic in the upper half complex plane as a function of @xmath165 ( dimensionless frequency @xmath141 , extended to be a complex variable ) @xcite ( this means that any pole / resonance that appears on the real axis is slightly shifted below the real axis , i.e. includes dissipation , to account for realistic causal systems ) .
therefore , the integrand of @xmath166 in ( [ i ] ) must also be analytic in the upper half plane , where it vanishes exponentially for @xmath167 .
this allows to deform the contour of integration from the real axis of @xmath165 into the imaginary axis @xcite .
the resulting integral in terms of the dimensionless imaginary frequency @xmath168 becomes @xmath169 where for the evaluation of @xmath170 we recall that @xmath171 .
we also note that this integral is guaranteed to take real values since linear response functions are real and monotonically decreasing in the positive imaginary axis @xcite .
this also means that @xmath170 take only real values smaller than @xmath172 , which is extremely useful in determining the sign of the force .
the denominator must therefore be positive such that the sign of the integrand is exclusively determined by the product @xmath173 in the numerator .
considering impedances @xmath86 that terminate the line , with the reflection coefficients of eq .
( [ r ] ) , repulsion ( @xmath174 ) is then obtained ( per frequency ) for @xmath175 where @xmath176 and @xmath177 are interchangeable .
this is in analogy to the known condition in 3d @xcite , which is nevertheless typically stated in terms of electric permittivity instead of impedance , hence ignoring the magnetic response .
here we consider that both impedances are either shorted , i.e. replaced by a wire so that @xmath178 , or disconnected @xmath179 ( `` open circuit '' , equivalent to vanishing capacitance or infinite inductance ) . whereas the first case gives @xmath180 and the latter @xmath181 [ eq .
( [ r ] ) ] , both cases yield @xmath182 at all frequencies @xmath183 , and hence to an attractive force . this case constitutes the 1d analogue to the original situation considered by casimir @xcite . performing the integral in eq .
( [ fu ] ) with @xmath182 we obtain @xmath184 as expected in 1d @xcite .
consider now a short circuit , @xmath185 , interacting with an open circuit , @xmath186 , satisfying the condition ( [ rep ] ) for a repulsive force .
inserting @xmath187 into eq .
( [ fu ] ) for all frequencies @xmath183 , we find @xmath188 .
this result was obtained in ref .
@xcite by similar considerations to those of the original casimir treatment , adopted to 1d field quantization with unlike boundary conditions .
here however , a concrete physical system of impedances inside circuits is considered , which allows an interpretation in analogy to the result by boyer in ref .
@xcite , where the free - space casimir force between a perfect magnetic conductor @xmath189 and a perfect electric conductor @xmath190 was found to be repulsive .
therefore , the impedance @xmath185 can be viewed as that of a capacitor @xmath191 in the limit @xmath192 in analogy to @xmath190 , whereas @xmath186 can be seen as an inductor @xmath193 in the limit of infinite inductance analogous to @xmath189 .
the former two examples can be made more realistic by considering circuit elements such as a capacitor @xmath122 and an inductor @xmath47 .
the frequency dependencies of their respective impedances , @xmath194 and @xmath195 , exhibit opposite limiting behaviors at @xmath196 and @xmath197 , namely , when @xmath31 appears as open circuit @xmath32 appears as short circuit and vise versa .
therefore , both behaviors found in examples 1 and 2 above should appear in this case . more explicitly ,
consider their corresponding reflection coefficients in imaginary frequency @xmath198 where @xmath199 and @xmath200 are the scaled frequencies associated with the capacitor and inductor , respectively .
figure 5b displays the force between the components , which is calculated by numerically integrating eq .
( [ fu ] ) as a function of the parameters @xmath199 and @xmath200 .
indeed , both of the results from examples 1 ( attractive force ) and 2 ( repulsive force ) emerge at the following limits : when the capacitance is small ( @xmath201 ) @xmath31 acts as open circuit and vise versa , whereas when the inductance is small ( @xmath202 ) @xmath32 acts as short circuit and vise versa .
all previous examples included non - dissipative impedances .
we shall now exploit the general formula , eq .
( [ ffdt ] ) , derived using the fdt approach to consider a dissipative case that goes beyond that treated using the qed approach .
specifically , we consider an impedance @xmath203 of a capacitor in series with a resistor . the corresponding reflection coefficient in imaginary frequency is given by @xmath204 which changes sign as a function of @xmath183 .
figure 5c presents the resulting force for @xmath205 ( short circuit ) , plotted as a function of the parameters @xmath199 and @xmath206 .
again , for small capacitance , where @xmath207 , the capacitor , and hence @xmath31 , behave as open circuit , so that the repulsion of example 2 is recovered .
we have considered the fluctuation force between two neutral objects connected by wires .
the objects are characterized by their impedances and , within the fdt approach , the fluctuations emerge from their dissipative part .
this approach is simple to generalize for any dissipative circuit , and in particular to lossy tls , by modelling any resistance with a current source as in fig .
moreover , it is also natural to apply to any general circuit configuration such as lumped circuits , since it does not rely on the quantization of electromagnetic waves , as in the qed approach . as such
, it can become useful in the exploration of a variety of fluctuation phenomena in circuits , such as those studied in refs .
@xcite , by merely solving a set of kirchhoff - law equations . in this respect
, the study of fluctuation forces in circuits could provide a conceptually simple system wherein casimir physics can be explored in the absence of technical complications which may not add to its essence .
we note that in principal , the general configuration beyond that of fig .
1c can also be calculated within the qed scattering formalism .
that would require to replace any resistive element @xmath63 by a semi - infinite tl with characteristic impedance @xmath63 and corresponding quantum field fluctuations .
this would result in a more complicated scattering problem with multiple channels and inputs which is nevertheless fully equivalent to the more elegant circuit solution of the fdt formalism . in this respect ,
let us also refer back to the fact that the result for the force in the qed approach , eq .
( [ fqed ] ) , is identical to that found using the fdt approach , eq .
( [ ffdt ] ) , even though the latter was derived for the general case whereas the former for the specific case of non - dissipative impedances embedded in a uniform tl .
in fact , a similar situation exists also in 3d derivations of the casimir force between dielectric slabs or mirrors .
there , the lifshitz treatment , which is performed for generally lossy dielectrics @xcite , formally leads to the same result obtained by derivations which assume lossless dielectrics and mirrors @xcite .
although the main focus of this paper is on formalism , we shall conclude with a brief account on possible experimental realizations . a straightforward realization where the force between the circuit elements would in principle be observable
is a systems wherein the motion of the center of mass of one of the elements is allowed and is measurable .
a different approach would be to consider the interaction potential built between the two elements , which is associated with the force .
then , if the energy spectrum describing the internal degrees of freedom of one of the elements is quantized , such as in a super - conducting qubit , the interaction potential could possibly be measured via the shift it induces on these energy levels @xcite . alternatively , in case one of the interacting elements possesses a movable internal coordinate , such as in a variable capacitor or inductor , the interaction potential may be inferred from the force induced on this coordinate @xcite .
i would like to thank ulf leonhardt for valuable discussions and itay griniasty for careful reading of the manuscript .
the financial support of the european research council ( erc ) , the israel science foundation ( nsf ) and the mit - harvard center for ultracold atoms is acknowledged .
by energy conservation , it is clear that the relation between the power reflection and transmission coefficients of a non - dissipative 1d scatterer satisfies @xmath87 .
similar energy conservation reasoning lead to the modification of this relation when the scatterer is situated at the boundary between media with different impedances .
we will now show how these relations can be obtained using simple circuit analysis for the reflection of an impedance @xmath24 that _ terminates _ a tl , eq .
( [ r ] ) , when the transmission ( absorption ) is defined as @xmath208 . denoting @xmath209 $ ] and @xmath210 $ ] , we obtain @xmath211 ^ 2 } , \nonumber\\ & & |t|^2=\frac{4|z|^2\left(|z|^2 + 2z_0z'+z_0 ^ 2\right)}{\left[|z|^2+z_0(2z'+z_0)\right]^2}. \label{a1}\end{aligned}\ ] ] then , representing the impedance by a parallel resistor @xmath63 and reactance @xmath212 as in fig .
4b , we insert the relations ( [ zn ] ) into eqs .
( [ a1 ] ) and obtain @xmath213 the relation to energy conservation in 1d scattering becomes clear if we think of the resistor @xmath63 in fig .
4b as a semi - infinite tl extending to left side , with a characteristic impedance @xmath63
. then , eq . ( [ a2 ] ) describes the energy transfer between a medium with impedance @xmath63 to a medium with impedance @xmath22 through a lossless scatterer @xmath214 , and therefore the ratio of impedances appears . for a lossless mirror between identical media , @xmath64 , we obtain the usual relation @xmath87 .
consider a lumped - circuit component embedded in a tl at position @xmath215 , as in fig .
the force acting on the component is that exerted by the electromagnetic fields on a volume @xmath216 which surrounds the component ( blue dashed line in fig .
6 ) @xcite , @xmath217 here @xmath218 is a vector of an area element directed normal to the surface @xmath219 confining the volume @xmath216 , and @xmath220 is the maxwell stress tensor ( outside of the component ) @xmath221 since we are interested in forces mediated ( exerted ) by the tem field mode of the tl , we consider @xmath222 for electric and magnetic fields perpendicular to each other and to the propagation axis @xmath1 , @xmath223 and @xmath224 , yielding a diagonal stress tensor @xmath225 with @xmath226 . the force along the positive @xmath1 direction , @xmath227 is then found to be @xmath228 \nonumber\\ & = & h'(x_0 ^ -)-h'(x_0^+ ) .
\label{fff}\end{aligned}\ ] ] here we used the fact that the integral of the electromagnetic energy density @xmath229 over @xmath230 and @xmath231 is equivalent to the energy per unit length @xmath30 .
considering now the impedance 1 from the casimir configuration of fig .
1c , the force in the positive @xmath1 direction , @xmath232 , is the attractive force acting on this circuit element . for the impedance 2 ,
the attractive force is that in the negative @xmath1 direction and is hence given by @xmath233 .
finally , since the energy density is constant at each section of the tl [ see eq .
( [ hqed ] ) ] , the two forces are identical as expected , @xmath234 .
in the following , we provide a brief account of the quantization of the tl .
the voltage , current , and hence the charge fields in a homogenous tl satisfy the 1d helmholtz equation , eq .
( [ helm ] ) .
it can be easily shown that this equation , e.g. for the charge @xmath235 , can be derived from the following lagrangian , @xmath236 .
\label{b1}\ ] ] the conjugate variable to @xmath40 is found to be the flux per unit length , @xmath237 , resulting in the hamiltonian , @xmath238 .
\label{b2}\ ] ] canonical quantization is performed by demanding the commutation relations @xmath239=0 $ ] , @xmath240=0 $ ] and @xmath241=i\hbar \delta(x - x')$ ] . using the expansion @xmath242 and
hence @xmath243 with @xmath244=0 $ ] and @xmath245=\delta_{kk'}$ ] , we find the coefficients @xmath246 .
finally , by using the telegraphers equation @xmath247 and the definition of the flux @xmath248 , we find the voltage and current operators from eqs .
( [ vi ] ) and ( [ a ] ) .
a simple way to derive eq .
( [ in ] ) goes as follows . treating the charge @xmath249 on a lumped circuit element as its relevant dynamical variable , and considering an applied voltage @xmath250
, we obtain the interaction hamiltonian in a linear form @xmath251 . considering the impedance of the circuit element @xmath20 , and using @xmath252 and @xmath253 , we find @xmath254 with the linear response @xmath255 .
then , inserting @xmath256=i\mathrm{re}[z]/(\omega|z|^2)$ ] into the fdt @xcite we find the charge fluctuations @xmath257}{\omega |z|^2}\frac{1}{1-e^{-\beta\hbar\omega}}. \label{c1}\ ] ] finally , the current fluctuations are obtained from @xmath258 , where eq .
( [ in ] ) corresponds to the case @xmath259 .
e. vetsch , d. reitz , g. sagu , r. schmidt , s. t. dawkins and a. rauschenbeutel , phys .
* 104 * , 203603 ( 2010 ) . c. sayrin , c. junge , r. mitsch , b. albrecht , d. oshea , p. schneeweiss , j. volz and a. rauschenbeutel , phys .
x * 5 * , 041036 ( 2015 ) .
a. goban , c .-
l . hung , s .-
yu , j. d. hood , j. a. muniz , j. h. lee , m. j. martin , a. c. mcclung , k. s. choi , d. e. chang , o. painter and h. j. kimble , nat . commun . * 5 * , 3808 ( 2014 ) .
a. goban , c .-
. hung , j. d. hood , s .-
yu , j. a. muniz , o. painter and h. j. kimble , phys .
* 115 * , 063601 ( 2015 ) .
m. arcari , i. sllner , a. javadi , s. lindskov hansen , s. mahmoodian , j. liu , h. thyrrestrup , e. h. lee , j. d. song , s. stobbe , and p. lodahl , phys .
lett . * 113 * , 093603 ( 2014 ) .
a. sipahigil , r. e. evans , d. d. sukachev , m. j. burek , j. borregaard , m. k. bhaskar , c. t. nguyen , j. l. pacheco , h. a. atikian , c. meuwly , r. m. camacho , f. jelezko , e. bielejec , h. park , m. loncar , m. d. lukin , science * 354 * , 847 ( 2016 ) .
a. f. van loo , a. fedorov , k. lalumire , b. c. sanders , a. blais , and a. wallraff , science * 342 * , 1494 ( 2013 ) . j. a. mlynek , a. a. abdumalikov , c. eichler and a. wallraff , nat
. commun . * 5 * , 5186 ( 2014 ) .
a. gonzalez - tudela , d. martin - cano , e. moreno , l. martin - moreno , c. tejedor and f. j. garcia - vidal , phys .
lett * 106 * , 020501 ( 2011 ) .
e. shahmoon and g. kurizki , phys .
a * 87 * , 033831 ( 2013 ) .
d. e. chang , j. i. cirac and h. j. kimble , phys .
lett . * 110 * , 113606 ( 2013 ) .
t. grieer and h. ritsch , phys .
lett . * 111 * , 055702 ( 2013 ) .
z. eldredge , p. solano , d. chang and a. v. gorshkov , phys .
a * 94 * , 053855 ( 2016 ) .
g. calaj , f. ciccarello , d. chang and p. rabl , phys .
a * 93 * , 033833 ( 2016 ) .
e. shahmoon , i. mazets and g. kurizki , opt
. lett . * 39 * , 3674 ( 2014 ) .
j. s. douglas , h. habibian , c .-
. hung , a. v. gorshkov , h. j. kimble and d. e. chang , nat .
* 9 * , 326 ( 2015 ) .
e. shahmoon , p. grisins , h. p. stimming , i. mazets and g. kurizki , optica * 3 * , 725 ( 2016 ) .
j. s. douglas , t. caneva and d. e. chang , phys .
x * 6 * , 031017 ( 2016 ) .
e. shahmoon , i. mazets and g. kurizki , proc .
usa * 111 * , 10485 ( 2014 ) .
e. lvarez and f. d. mazzitelli , phys .
d * 79 * , 045019 ( 2009 ) .
s. i. maslovski and m. g. silveirinha , phys .
a * 82 * , ( 2010 ) .
e. shahmoon and g. kurizki , phys .
a * 87 * , 062105 ( 2013 ) .
h. r. haakh and s. scheel , phys .
rev . a * 91 * , 052707 ( 2015 ) .
r. de melo e souza , w. j. m. kort - kamp , f. s. s. rosa and c. farina , phys .
rev . a * 91 * , 052708 ( 2015 ) .
s. scheel , s. y. buhmann , c. clausen , and p. schneeweiss , phys .
a * 92 * , 043819 ( 2015 ) . c. eichler , d. bozyigit , c. lang , m. baur , l. steffen , j. m. fink , s. filipp , and a. wallraff , phys .
lett . * 107 * , 113601 ( 2011 ) .
k. w. murch , s. j. weber , k. m. beck , e. ginossar , i. siddiqi , nature , * 499 * , 62 ( 2013 ) .
c. m. wilson , g. johansson , a. pourkabirian , m. simoen , j. r. johansson , t. duty , f. nori and p. delsing , nature * 479 * , 376 - 379 ( 2011 ) .
p. lhteenmki , g. s. paraoanu , j. hassel and j. hakonen , proc .
sci . * 110 * , 4234 ( 2013 ) .
b. yurke and j. s. denker , phys .
a * 29 * , 1419 ( 1984 ) .
m. h. devoret , _ quantum fluctuations in electrical circuits _ , in _ quantum fluctuations _
, les houches session lxiii , pp . 351 ( elsevier 1997 ) . m. t. jaekel and s. reynaud , journal de physique i * 1 * , 1395 ( 1991 ) .
l. d. landau and e. m. lifshitz , _ statistical physics , part 1 _ ( pergamon press , 3rd edtion 1980 ) .
i. e. dzyaloshinskii , e. m. lifshitz and l. p. pitaevskii , advances in physics , * 10 * , 165 ( 1961 ) . y. s. barash and v. l. ginzburg , sov .
usp . * 27 * , 467 ( 1984 ) . p.
w. milonni , _ the quantum vacuum : an introduction to quantum electrodynamics _ ( academic , 1993 ) .
s. scheel , in _ forces of the quantum vacuum _
( w. m .r .
simpson and u. leonhardt , eds . ) , ch . 3 ( world scientific , 2015 )
. j. a. kong , _ electromagnetic wave theory _ , ( john wiley and sons , inc . , 1986 ) .
d. m. pozar _ microwave engineering _ ( john wiley and sons , 2005 ) .
a. blais , r. s. huang , a. wallraff , s. m. girvin and r. j. schoelkopf , hys .
a * 69 * , 062320 ( 2004 ) .
j. r. zurita - snchez and c. henkel , phys .
rev . a * 73 * , 063825 ( 2006 ) .
h. nyquist , phys . rev . * 32 * , 110 ( 1928 ) .
e. shahmoon , in _ forces of the quantum vacuum _ ( w. m .r .
simpson and u. leonhardt , eds . ) , ch .
3 ( world scientific , 2015 ) .
j. munday , f. capasso ans v. a. parsegian , nature * 457 * , 170 ( 2009 ) .
h. b. g. casimir , proc .
akad . wet . *
51 * , 793 ( 1948 ) .
t. h. boyer , am .
* 71 * , 990 ( 2003 ) .
t. h. boyer , phys .
rev . a * 9 * , 2078 ( 1974 ) .
e. shahmoon and u. leonhardt , arxiv:1612.03250 .
d. drosdoff , i. v. bondarev , a. widom , r. podgornik and l. m. woods , phys .
x * 6 * , 011004 ( 2016 ) .
a. fragner , m. gppl , j. m. fink , m. baur , r. bianchetti , p. j. leek , a. blais and a. wallraff , science * 322 * , 1357 ( 2008 ) .
y. tabuchi , s. ishino , a. noguchi , t. ishikawa , r. yamazaki , k. usami and y. nakamura , science * 349 * , 405 ( 2015 ) .
j. d. teufel , t. donner , dale li , j. w. harlow , m. s. allman , k. cicak , a. j. sirois , j. d. whittaker , k. w. lehnert and r. w. simmonds , nature * 475 * , 359 ( 2011 ) .
l. novotny and b. hecht , _ principles of nano - optics _ ( cambridge university press , 2006 ) . | transmission - line waveguides can mediate long - range fluctuation - induced forces between neutral objects .
we present two approaches for the description of these forces between electric components embedded in transmission - line circuits .
the first , following ordinary quantum electrodynamics ( qed ) , consists of the quantization and scattering theory of voltage and current waves inside transmission lines .
the second approach relies on a simple circuit analysis with additional noisy current sources due to resistors in the circuit , as per the fluctuation - dissipation theorem ( fdt ) .
we apply the latter approach to derive a general formula for the casimir force induced by circuit fluctuations between any two impedances .
the application of this formula , considering the sign of the resulting force , is discussed .
while both qed and fdt approaches are equivalent , we conclude that the latter is simpler to generalize and solve . |
polymers and polymer solutions belong to the most intensively studied objects in condensed matter physics @xcite .
the adsorption of polymers on surfaces and interfaces is of special importance @xcite .
adsorption of free polymers in solution on the container wall or other boundaries occurs in the presence of attractive interactions between the surface and the chain monomers .
examples of such interactions include coulomb and van der waals forces , and more specific molecular interactions .
coulomb forces are screened by counter - ions in the solution and can be tuned to some extent by adding salt to the solvent . thus on changing the properties of the solvent
an individual polymer chain can undergo a reversible transition from a freely floating , desorbed state to an adsorbed state in which the chain monomers are close to the surface on average .
the adsorption of polymers on flat surfaces has been studied theoretically and experimentally , and is by now well understood @xcite . due to the importance to colloidal dispersions the interaction of polymers with
spherical and rodlike particles has been studied as well @xcite .
the adsorption of flexible polymers on rodlike particles is relevant , for example , in gels @xcite , and for the binding of flexible polymers to relatively stiff biomolecules such as dna @xcite .
another class of polymer adsorption transitions involve two flexible self - avoiding but mutually attracting polymers which can form a bound , double - stranded , so - called `` zipped '' state .
a prominent example of this kind of transition is the denaturation transition of double - stranded dna @xcite which recently attracted considerable attention regarding its statistical - mechanical properties and thermodynamic order @xcite ( for a recent review on the biophysics related to dna topology , see @xcite ) .
the dna denaturation transition is usually modelled in such a way that monomer @xmath1 of one strand can only interact with monomer @xmath1 of the other strand , reflecting the key - lock principle of natural , inhomogeneous dna with its specific sequence of base pairs .
two - chain systems in which any monomer of one chain can interact with any monomer of the other chain include diblock copolymers , which consist of linear chains of @xmath2 monomers of type @xmath3 followed by @xmath4 monomers of type @xmath5 , with different @xmath6 , @xmath7 , and @xmath8 interactions ; systems of this kind have been intensively studied as well @xcite .
recently it was found that self - avoiding mutually attracting diblock copolymers can adopt a zipped state in which the two components form a bound , double - stranded structure which however remains swollen and does not assume compact configurations .
the zipped state is located between a swollen , unbound high - temperature state and a compact low - temperature state , separated by second - order and first - order phase transitions , respectively @xcite .
( 10,8 ) ( -0.5,-30.5 ) in this work we focus on the adsorption transition of a long flexible self - avoiding polymer chain onto a rigid thin elongated rod , as shown in figure [ fig_rod ] .
we assume that the rod is endowed with a short - ranged surface potential , or adsorption energy , @xmath9 , for the chain monomers ; the more positive @xmath9 , the more attractive the interaction .
thus , on increasing @xmath9 from a low value , at some threshold value @xmath10 the chain is expected to undergo a transition from an unbound , free state to a bound state in which the overall gain in binding energy compensates the loss of configurational entropy . an interesting feature of this transition is the fact that it represents a true phase transition in the thermodynamic sense ; the adsorbed state forming an elongated , aligned structure , see figure [ fig_rod ] .
in contrast , for a particle of finite size , a self - avoiding polymer chain can not undergo a true adsorption phase transition due to steric constraints .
the focus of the present work is on the former case . since the polymer adsorption transition is characterized by large fluctuations we expect scaling and universal behaviour .
we thus use the renormalization group to obtain general scaling results for the chain partition function .
we also obtain estimates for relevant critical exponents by geometric arguments . before we introduce and study our model in section [ sec_rod ] we recall some general ideas and concepts for later reference . in section [ subsec_general ]
we discuss typical scaling arguments related to the polymer adsorption transition .
since our work strongly relies on field - theoretical methods , in section [ subsec_map ] and [ sec_app ] we sketch the mapping of the polymer system on the ginzburg - landau model .
consider the adsorption of a long flexible polymer chain onto an object @xmath11 .
for the time being , this object can be a surface , a thin rod , another flexible polymer chain , or any other extended manifold that allows for a thermodynamic adsorption phase transition .
the quantity of interest is the partition function @xmath12 of the chain with one end fixed close to the object @xmath11 and the other end moving freely . upon adjusting the system
s thermodynamic variables close to the adsorption transition point , only the number of chain monomers @xmath13 and the adsorption energy @xmath14 remain as relevant parameters , where @xmath10 is the adsorption energy at the transition point .
the partition function @xmath12 is expected to obey the scaling @xmath15\ ] ] where @xmath16 is the lattice - dependent connectivity constant and @xmath17 , @xmath18 are critical exponents .
the scaling function @xmath19 is regular at @xmath20 since @xmath21 has no singularity for finite @xmath2 and @xmath14 .
the exponent @xmath17 thus characterizes the scaling of @xmath12 right at the transition point : @xmath22 .
note that @xmath17 is not necessarily equal to the critical exponent @xmath23 for an unbounded , free chain , for which @xmath24 ( compare equation ( [ scalbehz ] ) in section [ subsec_ren ] , with @xmath25 introduced in equation ( [ scalbeh ] ) and @xmath26 ) .
the exponent @xmath18 is referred to as the crossover - exponent .
since @xmath27 acts as a chemical potential for monomers close to @xmath11 , the number @xmath28 of adsorbed monomers scales as @xmath29 equation ( [ scale ] ) implies three distinct scaling regimes for @xmath28 .
\(i ) @xmath30 .
equations ( [ scale ] ) and ( [ ns ] ) yield @xmath31 for @xmath32 this implies that @xmath28 grows with @xmath2 but the _ fraction _ of adsorbed monomers , @xmath33 , vanishes for @xmath34 . for @xmath35 ,
the behaviour @xmath36 at @xmath30 indicates that the adsorption transition is of first order .
\(ii ) @xmath37 . equation ( [ scale ] ) implies that the scaling behaviour of @xmath12 for @xmath34 is governed by the behaviour of @xmath19 for @xmath38 , regardless the precise value of @xmath9 . in this case
@xmath11 is repulsive for the chain monomers and @xmath28 stays finite even for @xmath34 .
\(iii ) @xmath39 .
the chain adopts an adsorbed state and stays close to @xmath11 on average .
thus , @xmath36 , which implies a _
finite _ fraction of adsorbed monomers for @xmath34 : @xmath40 to analyze the behaviour of @xmath41 it is useful to consider the grand canonical ensemble .
the partition function in the grand canonical ensemble , @xmath42 , is related to @xmath21 by a laplace transform : @xmath43 where @xmath44 is the chemical potential conjugate to @xmath2 .
equation ( [ grand ] ) is valid for @xmath45 with @xmath46 .
one is allowed to set @xmath47 for simplicity , so that @xmath48 . equation ( [ scale ] ) then implies the scaling behaviour @xmath49 \ , \ , \
, , \qquad \mu > 0 \ , \ , .\ ] ] by reasoning similar to that below equation ( [ scale ] ) one finds that the scaling function @xmath50 is regular at @xmath51 and @xmath17 characterizes the scaling of @xmath52 right at the transition point : @xmath53 . on the other hand ,
we know that for @xmath39 the chain takes an adsorbed state . for the grand canonical ensemble this implies , for given @xmath39 , that @xmath54 for @xmath55 with some @xmath56 . in this limit
we thus expect the scaling behaviour @xmath57 where @xmath58 is characteristic for the _ adsorbed _ state and _ different _ from @xmath17 .
for example , if @xmath11 is another flexible polymer chain , the adsorbed state forms a double - stranded structure which , as a whole , behaves like an unbounded , free , self - avoiding chain , which implies @xmath59 in this case @xcite .
using equation ( [ scalex ] ) it follows that the scaling function @xmath50 must have a singularity at some @xmath60 of the form @xmath61 the relation @xmath62 determines the shape of the line @xmath63 as @xmath64 figure [ fig_pd ] shows typical phase diagrams for polymer adsorption in the grand canonical ensemble ( fixed @xmath44 ) and canonical ensemble ( fixed @xmath2 ) .
( 10,5.2 ) ( -2.7,-36.2 ) according to the above , for @xmath39 , @xmath42 in equation ( [ grand ] ) has a singularity for @xmath65 , and @xmath2 @xmath66 diverges in this limit .
this , in turn , implies @xmath67 \ , \ , , \qquad { \cal e } > { \cal e}^ * \ , \ , , \ , \ , n \to \infty \ , \ , .\ ] ] using equation ( [ ns ] ) one finds @xmath68 and thus , using equations ( [ fraction ] ) and ( [ s ] ) , @xmath69 where the exponent @xmath70 is related to the crossover - exponent @xmath18 in equation ( [ nscale ] ) by @xmath71 in particular , for @xmath35 the fraction @xmath72 jumps from zero for @xmath37 to a finite value for @xmath39 , which is then independent of @xmath9 ; this corresponds to a first - order transition ( compare case ( i ) above ) .
the scaling behaviours ( [ nscale ] ) and ( [ resfracintro ] ) , ( [ relation ] ) have been proven rigorously for the polymer adsorption transition on a _ flat _
surface @xcite .
this system is closely related to the semi - infinite ginzburg - landau model , see @xcite for reviews ; the mapping of the polymer system on the ginzburg - landau model is discussed below .
the scaling behaviours ( [ nscale ] ) and ( [ resfracintro ] ) , ( [ relation ] ) also hold reasonably well in a recent numerical study of the dna denaturation transition @xcite . according to edwards continuous chain model
we represent the configuration of a linear chain of length @xmath73 by a curve @xmath74 , parameterized by its arc length @xmath75 , in @xmath76-dimensional space .
the chain length @xmath73 is proportional to the number of chain monomers @xmath2 . in the presence of an external potential @xmath77
the partition function of the chain is given by @xmath78 & & \times \ , \exp \left\{- \int d^dr \left[v({\bf r } ) \rho({\bf r } ) + \frac{u}{6 } \ , \rho^2({\bf r } ) \right ] \right\ } \nonumber \end{aligned}\ ] ] with the monomer density @xmath79 the symbol @xmath80 denotes functional integration over all chain configurations with the chain ends fixed at @xmath81 and @xmath82 .
the superscript `` @xmath83 '' on @xmath12 indicates that the chain is fixed with both ends .
the coupling constant @xmath84 of the @xmath85 interaction characterizes the strength of the contact interaction between chain monomers : @xmath86 describes a gaussian random walk whereas @xmath87 describes a self - avoiding chain .
the case @xmath88 is related to the polymer collapse transition to a compact state in a poor solvent @xcite ; in this work we do not consider this collapse transition , and therefore exclude the case @xmath88 .
as first noticed by de gennes @xcite , the polymer system can be mapped on the ginzburg - landau model of an @xmath89-component order parameter field @xmath90 in the limit @xmath91 .
it is worthwhile to note that this mapping not only works in perturbation theory but already on the level of the hamiltonian in the ginzburg - landau model .
the derivation , using a gaussian transformation to linearize the @xmath85 interaction in equation ( [ pf ] ) @xcite , is left to [ sec_app ] .
the result is @xmath92 where @xmath93 is the two - point correlation function in the ginzburg - landau model with the standard hamiltonian @xcite @xmath94 \ , \ , .\ ] ] the operation @xmath95 acting on the correlation function in equation ( [ res ] ) is an inverse laplace transform in which the integration path @xmath96 in the complex @xmath97-plane is a parallel to the imaginary axis to the right of all singularities .
equations ( [ res ] ) - ( [ laplace ] ) describe the statistics of the polymer chain in terms of properties of the near - critical ferromagnetic @xmath89-vector model in the limit @xmath91 . in the context of the @xmath89-vector model
, the parameter @xmath98 describes the deviation of the temperature @xmath99 from the critical temperature @xmath100 .
the form of the interaction involving the potential @xmath77 in equation ( [ action ] ) shows that the @xmath101-invariant scalar @xmath102 is related to the monomer density @xmath103 of the polymer chain in equation ( [ pf ] ) .
thus , translated to the polymer system , the term @xmath104 is related to the contact interaction between chain monomers and @xmath97 plays the role of a chemical potential for chain monomers in the bulk .
the objective of this work is the study of the adsorption of a long flexible polymer chain onto a rigid thin rod .
however , for the time being we model the rod by an infinitely elongated cylinder with small but finite radius @xmath0 .
the introduction of a finite cylinder radius @xmath0 is necessary since the limit of a rod with zero radius turns out to be too singular for the present field - theoretical treatment ; see figure [ fig_curved ] and the discussion below equation ( [ fracd2 ] ) . on the other hand ,
the fact that the adsorption transition now takes place on a _ surface _ , albeit a curved one , allows us to take advantage of available results for field theories with curved boundaries @xcite ; see section [ subsec_ren ] below .
the chain of total length @xmath105 , where @xmath2 is the number of chain monomers , is fixed with one end at the point @xmath106 on the cylinder surface @xmath107 while the other end is moving freely . the cylinder surface is endowed with a short - ranged surface potential @xmath108 acting on the chain monomers , and it is understood that the chain monomers are excluded from the interior of the cylinder .
the potential v(*r * ) in equation ( [ pf ] ) is thus given by ( 10,9 ) ( 1,-25 ) @xmath109 and @xmath110 if @xmath81 is located in the interior of the cylinder . by virtue of equation ( [ res ] )
, the chain partition function is given by @xmath111 the integration volume @xmath112 is the outer space of the cylinder bounded by the cylinder surface @xmath107 . on the right hand side , @xmath113 , where the _ susceptibility _ @xmath114 in the ginzburg - landau model is obtained by integrating the two - point correlation function , i.e. , @xmath115 in a ginzburg - landau type field theory with hamiltonian @xcite @xmath116 + \int\limits_s ds \ , \frac{c_0}{2 } \ , \vec{\phi}^2\ ] ] of an @xmath89-component order parameter field @xmath90 . in the following
we will understand @xmath114 as the chain partition function in the grand canonical ensemble , where @xmath117 is conjugate to @xmath118 ( compare section [ subsec_general ] with @xmath119 , @xmath105 , and @xmath120 ) .
the subscript `` 0 '' on @xmath108 , @xmath117 , @xmath118 , @xmath121 is used to distinguish these parameters from their renormalized counterparts that will be introduced below .
if @xmath122 in @xmath114 we suppress the argument @xmath106 since @xmath123 does not depend of the choice of @xmath106 by symmetry ; compare equation ( [ intz ] ) . before we proceed with the renormalization of the model defined by equation ( [ ham ] ) we review some cases in which results are available .
to this end it is useful to consider not only a cylinder in @xmath124 dimensions but bodies of more general shape .
the `` generalized cylinders '' have an infinitely extended axis of dimension @xmath125 and a curved surface with constant curvature radius @xmath0 in the subspace of co - dimension @xmath126 . in a space of dimension @xmath76
is given by @xmath127 .
] the axis can be the axis of an ordinary cylinder in three dimensions , for which @xmath128 , the midplane of a plate @xmath129 , or the centre of a sphere @xmath130 .
the explicit form of a `` generalized cylinder '' is given by the set @xmath131 .
figure [ fig_cyl ] shows some examples in the @xmath132-plane ; compare also references @xcite .
( 10,15 ) ( -2,-18.5 ) first , consider the line @xmath133 with @xmath76 arbitrary in figure [ fig_cyl ] .
then , of course , equation ( [ ham ] ) corresponds to the semi - infinite @xmath89-vector model bounded by a _ flat
_ surface for which many results are available @xcite .
in particular , for the polymer case @xmath134 , the considerations of section [ subsec_general ] can be made explicit , proving equations ( [ nscale ] ) and ( [ resfracintro ] ) , ( [ relation ] ) @xcite . next , consider the line @xmath135 with @xmath126 arbitrary in figure [ fig_cyl ] .
since @xmath135 is the upper critical dimension of the model in equation ( [ ham ] ) , this case corresponds to the gaussian field theory and to gaussian random walks , respectively .
both cases are described by @xmath136 in equation ( [ ham ] ) and can be solved exactly by standard methods @xcite .
thus , the susceptibility defined in equation ( [ intc ] ) , corresponding to the chain partition function in the grand canonical ensemble , is given by @xcite @xmath137\ ] ] where @xmath138 ( so that @xmath139 for @xmath122 ) , @xmath140 and @xmath141 .
the functions @xmath142 and @xmath143 are modified bessel functions @xcite with @xmath144 . from equation ( [ suscept ] )
one obtains the asymptotic behaviour right at the transition point : @xmath145 with @xmath146 the crossover - exponent @xmath18 in equation ( [ nsgauss ] ) is given by @xcite @xmath147 @xmath148 for @xmath149 , @xmath150 for @xmath151 , and for @xmath152 one finds @xmath35 corresponding to a first - order transition . for @xmath153 the finite fraction of adsorbed monomers scales like @xmath154 with the exponent @xcite @xmath155 for @xmath149 one finds @xmath156 which formally corresponds to @xmath157 . for @xmath151
one finds @xmath158 while for @xmath152 the fraction tends to a finite value , which reflects the fact that in this case the transition is of first order .
equations ( [ nsgauss ] ) - ( [ fracd2 ] ) are to be compared with equations ( [ nscale ] ) - ( [ relation ] ) in section [ subsec_general ] , where the variable @xmath159 here corresponds to @xmath160 there .
in particular , for given @xmath153 the chain partition function @xmath161 exhibits a singularity for @xmath162 where the function @xmath163 is determined by an analysis of the zero of the denominator in equation ( [ suscept ] ) @xcite .
the values of the exponents @xmath18 and @xmath70 in equations ( [ co ] ) and ( [ ka ] ) obey the scaling relation @xmath164 from equation ( [ relation ] ) .
note that for @xmath165 the limit @xmath166 yields merely the trivial bulk result @xmath167 , and hence no phase transition for @xmath165 .
thus , in the present treatment it is necessary to keep the cylinder radius @xmath0 finite even though the results for @xmath18 and @xmath70 do not depend on @xmath0 .
finally , we note that the adsorption of a gaussian chain onto a rigid rod is equivalent to the denaturation transition of two _ flexible _ gaussian chains @xmath3 and @xmath5 if monomer @xmath75 of chain @xmath3 can only interact with monomer @xmath75 of chain @xmath5 .
this corresponds to an interaction of the form @xmath168 $ ] in the partition function ( [ pf ] ) .
it is easy to see that the system of two flexible gaussian chains with the above interaction can be mapped on the system of one flexible gaussian chain interacting with a rigid rod , using the transformation @xcite
@xmath169 \ , \ , .\ ] ] for gaussian chains ( and only for gaussian chains ) the degrees of freedom described by the centre of mass ( cm ) coordinates @xmath170 and the relative coordinates @xmath74 decouple from each other .
the relative coordinates @xmath74 describe a gaussian chain which interacts with the origin at @xmath171 in the @xmath126-dimensional subspace , while the degrees of freedom of the remaining @xmath172 dimensions are unbounded and independent from the degrees of freedom of the @xmath126-dimensional subspace . by virtue of this mapping ,
the above results ( [ nsgauss ] ) - ( [ fracd2 ] ) have also been obtained in @xcite .
we now turn to the renormalization of the @xmath89-vector model defined by equation ( [ ham ] ) .
the objective is to determine the scaling behaviour of the renormalized chain partition function @xmath173 in the grand canonical ensemble in terms of the cylinder radius @xmath0 and renormalized parameters @xmath97 ( conjugate to the renormalized chain length @xmath73 ) and @xmath174 . the @xmath89-vector model in equation ( [ ham ] )
can be dimensionally regularized and renormalized by minimal subtraction of poles in @xmath175 .
the renormalizations of the bulk field @xmath176 , @xmath177 , and the bulk parameters @xmath117 , @xmath121 have the same form as in the unbounded case , and are given by @xcite ( we follow the convention of @xcite ) @xmath178 t_0 & = & \mu^2 \
, z_t \ , t \ , + \ , t_b \\[2 mm ] u_0 & = & \mu^{\varepsilon } \ , 2^d \ , \pi^{d/2 } \
, z_u \ , u \ , \ , \ , \ , .\end{aligned}\ ] ] the parameters @xmath179 , @xmath97 , @xmath84 are renormalized counterparts of the bare parameters @xmath180 , @xmath117 , @xmath121 in equation ( [ ham ] ) . in a regularization scheme using a large momentum - cutoff @xmath181 ,
the bulk renormalization factors @xmath182 , @xmath183 , @xmath184 absorb divergencies logarithmic in @xmath181 , corresponding to poles in @xmath185 in dimensional regularization .
the parameter @xmath186 absorbs divergencies quadratic in @xmath181 and describes the shift of the critical temperature @xmath100 of the @xmath89-vector model due to fluctuations . in dimensional regularization , @xmath187 , and @xcite @xmath188 z_t \ , z_{\phi } & = & 1 + \frac{n+2}{3 \
, \varepsilon } \ , u \ ,
+ \ , \left [ \frac{(n+2)(n+5)}{9 \ , \varepsilon^2 } - \frac{n+2}{6 \ , \varepsilon } \right ]
u^2 \ , + \ , \or(u^3 ) \\[2 mm ] z_u & = & 1 + \frac{n+8}{3 \ ,
\varepsilon } \ , u \
, + \ , \left [ \frac{(n+8)^2}{9 \ , \varepsilon^2 } - \frac{3n+14}{6 \ , \varepsilon } \right ] u^2 \ , + \ , \or(u^3 ) \ , \ , \ , .\end{aligned}\ ] ] the presence of the cylinder surface @xmath107 requires , in addition , renormalization of the surface field @xmath189 , @xmath190 , and of the surface parameter @xmath108 in equation ( [ ham ] ) . for a
_ flat _ surface these additional renormalizations are given by @xcite @xmath191 c_0 & = & \mu \ , z_c \ , c \
, + \ , c_{sp } \label{rbsurface2}\end{aligned}\ ] ] defining the renormalized surface field @xmath192 and surface parameter @xmath174 . the new renormalization factors @xmath193 and @xmath194 absorb divergencies logarithmic in @xmath181 which occur at a flat surface , corresponding to poles in @xmath185 in dimensional regularization .
the parameter @xmath195 absorbs divergencies linear in @xmath181 and describes the shift of the multicritical point due to fluctuations ( compare figure [ fig_pd ] , @xmath196 ) . in dimensional regularization , @xmath197 , and @xmath193 and @xmath194
are given by @xcite @xmath198 u^2 \ , + \ , \or(u^3 ) \\[2 mm ] z_c & = & 1 + \frac{n+2}{3 \
, \varepsilon } \ , u \ , + \ , \left [ \frac{(n+2)(n+5)}{9 \ , \varepsilon^2 } + \frac{n+2}{36 \ ,
\varepsilon } \ , ( 1 - 4 \pi^2 ) \right ] u^2 \ , + \ , \or(u^3 ) \ , \ , \ , .
\label{zs2}\end{aligned}\ ] ] equations ( [ rbsurface ] ) and ( [ rbsurface2 ] ) hold for a flat surface . as shown by mcavity and osborn @xcite ,
the required renormalizations become modified if the surface @xmath107 is curved , like in the present case .
while the renormalization of the surface field @xmath192 remains unchanged , the renormalization of the surface parameter @xmath108 requires an additional , additive term that depends on the mean curvature @xcite : @xmath199 or , with @xmath140 and @xmath200 , @xmath201 @xmath194 is the same renormalization factor as in equation ( [ rbsurface2 ] ) for a flat surface and @xmath202 to second order in @xmath84 can be deduced from reference @xcite : @xmath203 u^2 \right\ } \ , + \ , \or(u^3 ) \ , \ , \ , .\ ] ] to proceed , we consider the two - point correlation function @xmath204 where @xmath106 is located on the cylinder surface @xmath107 and @xmath205 . from @xmath206 , the chain partition function @xmath207 follows by an integration over @xmath82 , see equation ( [ intc ] ) ; by symmetry , @xmath123 does not depend on @xmath106 .
the renormalization group ( rg ) equation for the renormalized counterpart @xmath208 of @xmath206 follows in the standard way , using the relation @xmath209 and the fact that @xmath206 does not depend on @xmath44 : @xmath210 .
this leads to the rg equation @xmath211 g_{ren } \
, = \ , 0\ ] ] where we have used the abbreviation @xmath212 with @xmath213 and the exponent functions @xmath214 with @xmath215 .
the new feature of the rg equation ( [ rg ] ) generated by the surface curvature is the function @xmath216\ ] ] where @xmath217 & = & - \ , \frac{n+2}{3 } \ , u \ , + \ , \frac{n+2}{18 } \ , ( 4 \pi^2 - 1 ) u^2 \ , + \ , \or(u^3 ) \nonumber\end{aligned}\ ] ] is the exponent function for @xmath174 corresponding to a _ flat _
surface , with @xmath194 from equation ( [ zs2 ] ) @xcite .
a necessary condition for the two - point correlation function @xmath206 to be scale invariant ( si ) is that the right hand side of equation ( [ zeta ] ) vanishes , which is either the case for @xmath86 ( gaussian model ) or if @xmath218 takes the value for which the square bracket in equation ( [ zeta ] ) vanishes : @xmath219 note that the value of @xmath220 to leading order in @xmath84 , @xmath221 , is _ different _ from the value @xmath222 in equation ( [ zetastar ] ) corresponding to the onset of the adsorption transition in the gaussian model .
it is also interesting to note that for a sphere , corresponding to @xmath223 , the value of @xmath220 to leading order in @xmath84 , @xmath224 , coincides with the value @xmath225 for which the gaussian two - point correlation function at @xmath226 is conformal invariant ( ci ) @xcite .
however , within the @xmath227-model the special value @xmath220 is already fixed by scale invariance , whereas in the gaussian model the value @xmath228 is only fixed if one requires conformal invariance . by solving the rg equation ( [ rg ] ) for the two - point function in the standard way using characteristics and integrating over @xmath82 , one derives the general scaling behaviour of the chain partition function in the grand canonical ensemble : @xmath229 where @xmath230 with @xmath220 from equation ( [ van ] ) .
the exponent @xmath231 is a bulk critical exponent , while @xmath232 ( the crossover exponent ) and @xmath233 are critical exponents associated with a _ flat _
surface @xcite .
the exponent functions @xmath234 for @xmath235 are defined below equation ( [ beta ] ) and in equation ( [ etac ] ) .
the values @xmath236 are the values of the exponent functions at the fixed point @xmath237 .
the constants @xmath238 and @xmath239 in equation ( [ scalbeh ] ) are nonuniversal prefactors , while the function @xmath240 is a universal scaling function .
equation ( [ scalbeh ] ) is in line with the gaussian model ( where @xmath241 and @xmath242 ) : @xmath243 compare equation ( [ suscept ] ) .
finally , equation ( [ scalbeh ] ) can be compared with the corresponding behaviour for a flat surface : @xmath244 note that this limit can be obtained from equation ( [ scalbeh ] ) by rewriting the second scaling argument as @xmath245 .
equation ( [ scalbehplanar ] ) then follows as the limit @xmath246 of the scaling form @xmath247 , where @xmath248 .
let us come back to equation ( [ scalbeh ] ) .
according to recent estimates one has @xmath249 @xcite and @xmath250 @xcite for @xmath251 in @xmath252 so that the exponent @xmath253 in equation ( [ scalbeh ] ) is small but positive . from a naive point of view
this would imply that the scaling variable @xmath254 in the second scaling argument of @xmath255 is _ irrelevant _ and could be omitted from the outset ; however , one should keep in mind that the radius @xmath0 in the first scaling argument is also irrelevant in principle .
now , the relevant question is whether the scaling function @xmath240 exhibits a singularity on a certain subset of @xmath256 , corresponding to the polymer adsorption transition ; compare the related discussion below equation ( [ scalex ] ) and below equation ( [ fracd2 ] ) .
in fact , we expect this singularity to occur for @xmath257 , where now @xmath258 is a function of the first scaling variable @xmath259 .
note that the scaling function @xmath260 in equation ( [ scalbehgauss ] ) , corresponding to the gaussian model , exhibits this kind of singularity indeed ; compare section [ subsec_gauss ] .
thus , in the present description , the adsorption transition is characterized by a balance of the two irrelevant variables @xmath254 and @xmath0 . in this sense
the scaling variable @xmath254 can be considered as a dangerously irrelevant variable .
finally we note that equation ( [ scalbeh ] ) implies a corresponding scaling form of the chain partition function with fixed chain length @xmath73 : @xmath261 with ( different ) nonuniversal prefactors @xmath262 and @xmath263 and a universal scaling function @xmath264 . from equations
( [ scalbeh ] ) and ( [ scalbehz ] ) the number of adsorbed monomers @xmath28 for @xmath265 and the finite fraction of adsorbed monomers @xmath33 for @xmath266 can be derived as outlined in section [ subsec_general ] in principle
. however , the thin rod limit @xmath166 corresponds to singular limits of the scaling functions @xmath255 and @xmath264 which are rather difficult to obtain .
at least it is easy to see that the exponents @xmath18 and @xmath70 defined in equations ( [ nscale ] ) and ( [ resfracintro ] ) are _ universal _ , using the fact that @xmath255 and @xmath264 are universal scaling functions . to proceed , in the next section we use a different method to obtain estimates for the exponents @xmath18 and @xmath70 for a cylinder in @xmath124 . in this section
we obtain estimates for the exponents @xmath18 and @xmath70 for a cylinder in @xmath252 introduced in equations ( [ nscale ] ) and ( [ resfracintro ] ) by means of an interpolation procedure between two known cases .
firstly , in the gaussian model one has @xmath267 for @xmath268 , see equation ( [ co ] ) , corresponding to the point @xmath269 in figure [ fig_cyl ] .
likewise , @xmath270 from the scaling relation @xmath164 , or from equation ( [ ka ] ) , for @xmath269 .
secondly , one has @xmath271 on the whole line @xmath272 in figure [ fig_cyl ] .
this result can be obtained by using the _ co - dimension additivity theorem _ , stating that the co - dimension of the intersection points of two objects of dimensions @xmath273 and @xmath274 is given by the sum of their co - dimensions : @xmath275 , i.e. , @xmath276 for example , two - dimensional surfaces generically intersect along curves in @xmath124 ( @xmath277 ) and only at isolated points in @xmath135 ( @xmath278 ) . equation ( [ codim ] ) is also expected to hold if one or both objects are fractal . in the present case ,
one object is a self - avoiding random walk with fractal ( hausdorff ) dimension @xmath279 and the other one is a `` generalized cylinder '' with co - dimension @xmath126 ( see figure [ fig_cyl ] ) ; the dimension of intersection points of these two objects is thus given by @xmath280 in figure [ fig_cyl ] , the line @xmath278 as a function of @xmath126 and @xmath76 is shown as the blue dashed line .
an unbounded , free , self - avoiding random walk does not intersect with `` generalized cylinders '' located above the dashed line , apart from exceptional cases . in this sense , `` generalized cylinders '' above the dashed line are irrelevant perturbations for a free , self - avoiding random walk .
now , `` generalized cylinders '' located right on the line @xmath278 correspond to marginal cases : an unbounded , free , self - avoiding random walk _ does _ intersect with `` generalized cylinders '' located on the dashed line , but only at isolated points .
we thus expect that the number of intersecting monomers @xmath28 grows with @xmath2 for @xmath34 , but only logarithmically , i.e. , @xmath148 , which implies @xmath271 ; compare the case @xmath149 for the gaussian model discussed in section [ subsec_gauss ] , and compare the case @xmath30 with @xmath271 in section [ subsec_general ] .
it should be noted that this argument only applies to _ unperturbed _ random walks , and does not make any statement for walks that interact with the body .
thus , the values of the exponent @xmath18 at the end points of the green line in figure [ fig_cyl ] are available .
this can be used to obtain an estimate for @xmath18 for an ordinary cylinder in @xmath124 as follows .
the shape of the dashed line in figure [ fig_cyl ] is known quite accurately by means of the @xmath185-expansion of @xmath281 in conjunction with the exact value @xmath282 for @xmath134 in @xmath283 @xcite .
thus , one may estimate @xmath18 for a cylinder in @xmath252 , located at the point @xmath128 in figure [ fig_cyl ] , by means of a linear interpolation between the known values of @xmath18 at the end points of the green line ( compare references @xcite ) . in this way we find for a cylinder in @xmath124 the estimates , using equation ( [ relation ] ) , @xmath284
since these exponents are universal and do not depend on the cylinder radius @xmath0 , they are also expected to hold for a rigid rod with vanishing radius , or for a line of lattice sites in a numerical simulation of this system .
we have investigated the adsorption transition of a long flexible self - avoiding polymer chain onto a rigid thin rod by field - theoretical methods .
the rod is endowed with a short - ranged adsorption energy @xmath9 for the chain monomers so that , on increasing @xmath9 , at some threshold value @xmath10 the chain undergoes a transition from an unbound state to a bound state , as shown in figure [ fig_rod ] .
the main results and remaining questions are summarized below .
* by means of general scaling arguments we obtained the scaling relation ( [ relation ] ) for the exponents @xmath18 and @xmath70 defined in equations ( [ nscale ] ) and ( [ resfracintro ] ) , and the phase diagrams shown in figure [ fig_pd ] in terms of the number of chain monomers @xmath2 , the chemical potential @xmath44 conjugate to @xmath2 , and the adsorption energy @xmath9 . * by representing the rod by a cylinder of finite radius @xmath0 we could use available results for field theories with curved boundaries @xcite ; see figure [ fig_curved ] . by using renormalization group arguments , we derived the scaling behaviour of the chain partition function in the grand canonical ensemble , equation ( [ scalbeh ] ) , and in the canonical ensemble , equation ( [ scalbehz ] ) , where @xmath26 and @xmath285 .
notable features of the scaling results are the distinct form of the scaling variable @xmath286 , where the parameter @xmath174 is related to the surface potential for chain monomers , and the curvature - induced shift of @xmath218 in equation ( [ variable ] ) with @xmath220 from equation ( [ van ] ) .
it also follows that the exponents @xmath18 and @xmath70 introduced in equations ( [ nscale ] ) and ( [ resfracintro ] ) are universal . * because the cylinder radius @xmath0 enters the scaling functions @xmath255 and @xmath264 in equations ( [ scalbeh ] ) and ( [ scalbehz ] ) explicitly it is difficult to obtain the universal exponents @xmath18 and @xmath70 directly from them .
therefore we used the co - dimension additivity theorem in conjunction with an interpolation procedure , as shown in figure [ fig_cyl ] , to obtain the estimates for @xmath18 and @xmath70 in equation ( [ exp ] ) .
the check of these exponents and the scaling relation ( [ relation ] ) is a possible starting point for numerical simulations of this system .
* it would be interesting to introduce new methods to derive the exponents @xmath18 and @xmath70 , possibly avoiding the introduction of a finite cylinder radius @xmath0 from the outset .
* it would also be interesting to explain the relation between @xmath222 and @xmath220 discussed below equation ( [ van ] ) .
i would like to thank c v ferber for useful correspondence .
[ sec_app ] the @xmath85 interaction in equation ( [ pf ] ) can be linearized by means of a gaussian transformation @xcite .
this procedure makes use of the gaussian integral @xmath287 = \left ( \det \frac{a}{2 \pi } \right)^{-1/2 } \ , \exp\left[\frac{1}{2 } \ , b^t a^{-1 } \ , b \right]\ ] ] where @xmath288 is a vector with discrete or continuous indices and the symmetric matrix @xmath3 must have a positive definite real part .
using @xmath289 with purely imaginary @xmath290 , the matrix @xmath291 , and @xmath292 , one finds @xmath293 & & = \int { \cal d } \sigma \ , \exp\left [ \frac{3}{2 u } \int d^d r \ , \sigma^2({\bf r } ) \ , - \int d^dr \ , \sigma({\bf r } ) \rho({\bf r } ) \right ] \ , \ , .
\nonumber\end{aligned}\ ] ] note that @xmath3 is positive definite due to our assumption @xmath87 . inserting ( [ aux ] ) in ( [ pf ] )
yields @xmath294 \\[2 mm ] & & \ , \ , \times \int_{\bf r}^{{\bf r } ' } { \cal d } { \bf r }
\ , \exp \left\ { - \frac{1}{4 } \int_{0}^{l } ds \left(\frac{d { \bf r}}{ds}\right)^2 - \int d^dr \ , \left [ v({\bf r } ) + \sigma({\bf r } ) \right ] \rho({\bf r } ) \right\ } \nonumber\end{aligned}\ ] ] the @xmath85 interaction in equation ( [ pf ] ) has been replaced by the interaction of @xmath103 with an external , fluctuating potential @xmath290 .
the second line of equation ( [ pf2 ] ) can be interpreted as the path integral representation of the evolution operator @xmath295 in imaginary time @xmath75 of a quantum - mechanical particle with hamiltonian @xmath296 .
the laplace transform of this evolution operator with respect to @xmath73 yields the resolvent @xmath297 the resolvent can be represented in the standard way by the two - point correlation function of an @xmath89-component field @xmath90 in the limit @xmath91 .
the result is @xmath298 \\[2 mm ] & \times & { \cal l}_{t \to l } \ , \lim\limits_{n \to 0 } \ , \int { \cal d } \vec{\phi } \ , \phi_1({\bf r } ) \ , \phi_1({\bf r } ' ) \ , e^{- s\{\vec{\phi}\ } } \nonumber\end{aligned}\ ] ] with the action @xmath299
\vec{\phi}^2 \right ] \ , \ , \ , .\ ] ] the gaussian integration in equation ( [ ev ] ) can be carried out using equation ( [ gauss ] ) with the same @xmath289 and matrix @xmath291 as before , and now @xmath300 .
this leads to equations ( [ res ] ) - ( [ action ] ) .
metzler r and hanke a 2005 _ knots , bubbles , unwinding , and breathing : probing the topology of dna and other biomolecules _ in _ handbook of theoretical and computational nanotechnology _
rieth m and schommers w editors ( in press ) | the subject of this work is the adsorption transition of a long flexible self - avoiding polymer chain onto a rigid thin rod .
the rod is represented by a cylinder of radius @xmath0 with a short - ranged attractive surface potential for the chain monomers .
general scaling results are obtained by using renormalization group arguments in conjunction with available results for quantum field theories with curved boundaries [ mcavity and osborn 1993 _ nucl .
phys .
b _ * 394 * 728 ] .
relevant critical exponents are identified and estimated using geometric arguments .
it is a pleasure to dedicate this work to l. schfer on the occasion of his 60th birthday . |
in recent years there has been considerable interest in the study of nearby groupings of young stars that do not appear to be associated with any molecular clouds .
the prototypical example is perhaps the tw hydrae association @xcite , with an estimated age of 810 myr and some two dozen members .
searches for new members of this group have relied on x - ray properties , kinematics , infrared ( 2mass ) colors , and spectral features ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and have produced many candidates and motivated detailed follow - up studies to confirm them .
one of those searches , by @xcite , used proper motions from the tycho-2 catalog @xcite for x - ray sources listed in the rosat bright source catalog @xcite .
the authors considered a large area surrounding the previously known members of the tw hya association , and proposed a list of 23 candidate members that were subsequently followed up spectroscopically by @xcite , @xcite , and @xcite .
the present paper discusses one of those objects , ( also known as bd@xmath509 3055 , gsc 05493 00324 , 1rxs j102219.2@xmath5103302 , tyc 54933241 , @xmath6 , @xmath7 , j2000 , @xmath8 , spt = k ) .
the kinematic model of the association constructed by @xcite made specific predictions as to the radial velocity expected for each candidate . although the measurements for by @xcite happened to agree perfectly with those predictions ( @xmath9 ) , @xcite found that the star shows no detectable li @xmath106708 absorption in its spectrum ( equivalent width @xmath11 m ) , which indicates it is not very young and therefore essentially rules it out as a true member .
they also measured a radial velocity of @xmath12 that is nearly 10 different from the previous measurements , and proposed that the object is a spectroscopic binary .
further measurements by @xcite confirmed the velocity variations and indicated an orbital period of perhaps 1 or 2 years .
even though it was clear that is not related the tw hya association , the object was kept on the observing list at the harvard - smithsonian center for astrophysics ( cfa ) for the purpose of establishing the orbit .
preliminary solutions appeared to imply a mass for the companion that is larger than the primary , which was puzzling given that the spectrum is single - lined .
additionally , the hipparcos satellite @xcite detected curvature in the proper motion further supporting the binary nature of the object .
we present here the complete analysis of that combines the astrometry and the spectroscopy and allows us to investigate in more detail the nature of the overmassive secondary star .
was observed spectroscopically at the cfa with an echelle spectrograph on the 1.5-m tillinghast reflector at the f. l. whipple observatory on mt .
hopkins ( arizona ) .
a single echelle order was recorded using a photon - counting intensified reticon detector at a central wavelength of 5187 , giving a spectral coverage of 45 .
the resolving power is @xmath13 .
a total of 34 spectra were obtained from april 2002 until may 2005 , and the signal - to - noise ( s / n ) ratios achieved range from 5 to about 30 per resolution element of 8.5 .
four archival spectra ( s / n of 930 ) were also used , and were obtained for another program in 19851986 with nearly identical instrumentation on the multiple mirror telescope ( also on mt .
hopkins , arizona ) , prior to its conversion to a monolithic mirror .
radial velocities were derived using xcsao @xcite , a one - dimensional cross - correlation program well suited to our relatively low s / n spectra that runs under iraf . for the template we used a synthetic spectrum selected from an extensive library of spectra based on model atmospheres by r. l.kurucz ( see * ? ? ?
* ; * ? ? ?
these calculated spectra are available for a wide range of effective temperatures ( @xmath14 ) , projected rotational velocities ( @xmath15 ) , surface gravities ( @xmath16 ) and metallicities .
the optimum template was determined from grids of cross - correlations over broad ranges in @xmath14 and @xmath17 , seeking to maximize the average correlation weighted by the strength of each exposure .
solar metallicity was assumed to begin with .
the above calculations were repeated for a range of @xmath18 values from 2.0 to 5.0 , which allowed us to derive a rough estimate of the surface gravity for .
we obtained @xmath19 and @xmath20 k. the entire procedure was repeated for metallicities different from solar , but the best match was achieved for templates with the solar composition .
the temperature we derive corresponds to a main - sequence star with spectral type mid - k @xcite .
the rotational broadening was found to be negligible ( formally @xmath21 ) .
@xcite reported a considerably larger value of 8 .
the stability of the zero - point of our velocity system was monitored by means of exposures of the dusk and dawn sky , and small run - to - run corrections were applied in the manner described by @xcite .
the radial velocities in the heliocentric frame including these corrections are listed in table [ tab : rvs ] .
the typical precision of a single measurement is 0.5 .
variations over the 19.4 years of coverage are obvious and show a peak - to - peak amplitude of about 40 .
the period is 570 days , and the eccentricity ( @xmath0 ) is quite significant .
an orbital solution based on these velocities is presented in the second column of table [ tab : elements ] .
although it is no longer relevant in connection with the possible membership to the tw hya association , we mention in passing that the center - of - mass velocity @xmath22 of is far from the value of @xmath23 that had been predicted by @xcite .
the most significant finding , however , is the large mass function of the binary : @xmath24 . for a typical primary mass corresponding to the temperature
we derive , the mass function implies a secondary that is more massive than the primary , yet no signs of another star are obvious in our spectra .
we discuss this further below .
was observed by the hipparcos satellite from december 1989 to november 1992 , during which 53 astrometric measurements were made .
the trigonometric parallax was determined to be @xmath25 mas .
the uncertainty is somewhat larger than usual , but perhaps understandably so given that the star is relatively faint ( @xmath26 ) . additionally , the hipparcos reductions revealed significant curvature in the proper motion in the amount of @xmath27 mas yr@xmath28 and @xmath29 mas yr@xmath28 , most likely a reflection of the orbital motion of the object
. given its period of 570 days , however , nearly two cycles of the orbit were covered during the hipparcos campaign ( which lasted 1058 days ) so that the meaning of those coefficients is obscured . the proper motions and the parallax as reported in the hipparcos catalog are also likely to be compromised .
since the satellite observations are sensitive enough to have detected the orbital motion on the plane of the sky , the best course of action is therefore to make use of the individual astrometric measurements together with the velocities to constrain the orbit , and at the same time re - derive the position , proper motion , and parallax .
those hipparcos measurements are listed in table [ tab : hip ] .
we describe the global solution below .
the radial velocities allow for the determination of the usual spectroscopic elements , which are the period ( @xmath30 ) , center - of - mass velocity ( @xmath31 ) , semi - amplitude of the velocity variation ( @xmath32 ) , eccentricity ( @xmath33 ) , longitude of periastron for the primary ( @xmath34 ) , and time of periastron passage ( @xmath35 ) . on the other hand ,
the visual elements constrained by the astrometry are the angular semimajor axis ( @xmath36 ) , the inclination angle ( @xmath37 ) , the position angle of the ascending node ( @xmath38 ) , as well as @xmath30 , @xmath33 , @xmath34 , and @xmath35 . the combination of the two kinds of measurements thus provides redundancy in the four latter orbital elements that strengthens the solution .
furthermore , the fact that we see no indication of the secondary in the spectrum of means that the motion seen by hipparcos should essentially be that of the primary star around the center of mass of the binary . this introduces a relation between the angular semimajor axis of the primary , @xmath39 , and its velocity amplitude , given by @xmath40 where @xmath39 and the trigonometric parallax @xmath41 are expressed in milli - arc seconds , @xmath32 in @xmath42 , and @xmath30 in days .
we use this relation to eliminate the angular semimajor axis ( @xmath39 ) as an unknown . since the hipparcos observations are made in an absolute frame of reference , the abscissae residuals contain information on the parallax as well as the position and proper motion of the barycenter of the binary . five additional variables
are thus introduced ( @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 ) and @xmath48 . ] , which represent corrections to a fiducial point of reference that yield improved estimates of those quantities over the catalog values ( see * ? ? ?
a total of 13 variables were adjusted in the solution , using standard non - linear least - squares techniques @xcite .
the formalism used for incorporating the abscissae residuals from hipparcos into the fit follows closely that described by @xcite and @xcite , and is described in more detail in the appendix .
the relative weights of the spectroscopic and astrometric observations were adjusted to yield separate reduced @xmath49 values near unity . in this way we determined a scale factor for the original uncertainties of the hipparcos abscissae residuals of 1.14 , and a factor of 1.43 for the original velocity errors .
due to the faintness of the target the median error of a single hipparcos measurement is 5.7 mas , so that the constraint provided by the astrometry is relatively weak . nevertheless , the results of this initial fit yielded improvements in all of the orbital elements in common with the spectroscopy .
the value derived for the angular semimajor axis of the primary was @xmath50 mas .
as a test to see if the astrometry is able to detect the light contribution from the companion , we ran another solution in which we left the semimajor axis of the apparent orbit as a free parameter . in this case
that semimajor axis ( @xmath51 ) corresponds to the motion of the center of light of the binary , or photocenter , rather than that of the primary , and it should in principle be smaller if the light from the secondary is significant .
this fit gave a value of @xmath52 mas , which is not only not smaller but is also much more uncertain ( because it is no longer constrained by the spectroscopy through eq.[[eq : a1 ] ] ) .
nevertheless , it is still consistent with the previous result , within the errors .
we conclude from this that the companion is not bright enough to produce a measurable astrometric effect , and we proceed under the assumption that that star is invisible ( but see [ sec : orbit2 ] ) .
slight differences were found in the proper motion components ( @xmath53 , @xmath54 ) compared to the values reported in the hipparcos catalog , which was not unexpected given that the catalog values do not account for orbital motion .
an external check on these motions is available from the tycho-2 catalog @xcite , in which @xmath53 and @xmath54 are derived by combining the hipparcos position ( mean epoch 1991.25 ) with positions from ground - based meridian - circle and photographic catalogs going back several decades , and up to a century in some cases . given that the period of is only 570 days , the orbital motion over a baseline of decades should tend to average out in the tycho-2 analysis , resulting in a more accurate measure of the proper motion than that reported by hipparcos .
indeed , as shown in table [ tab : pm ] the values of @xmath53 and @xmath54 from our initial fit are much closer to the tycho-2 values than the hipparcos values .
this also suggests that the solution might benefit if we made use of the constraint provided by tycho-2 .
consequently , we incorporated the tycho-2 proper motions into the fit as measurements , along with their uncertainties .
the resulting proper motions from this combined solution are listed for comparison in the fourth row of table [ tab : pm ] .
the remaining orbital elements are hardly affected , except for the slightly smaller uncertainties .
the complete results of this combined fit are given in the third column of table [ tab : elements ] .
the semimajor axis of the relative orbit , @xmath36 , is inferred to be 30.8 mas ( 1.58 au in linear units ) .
the orbital period is determined to better than 0.1% by virtue of the 19.4-yr baseline provided by the velocity measurements ( more than 12 cycles ) .
perhaps one of the most significant improvements coming from the combination of the astrometry and spectroscopy is in the trigonometric parallax . compared to the value from the hipparcos catalog ,
our parallax that takes full account of the orbital motion is nearly 10 mas smaller , which corresponds to a 50% change in the distance to ( see table [ tab : pm ] ) , and is considerably more precise .
this has important consequences for the luminosity estimates discussed below . according to our solution the orbit of the binary
is seen nearly edge - on : the inclination angle is @xmath55 .
although in principle this would allow for eclipses , these are very unlikely given the 570-day period .
inspection of the hipparcos photometry for shows no evidence of any eclipse events ( four conjunctions occurred during the mission ) , and the peak - to - peak brightness variation recorded is @xmath20.1 mag , which is probably at the level of the uncertainties for a star as faint as @xmath8 . nevertheless , carefully scheduled observations may perhaps be of interest since the satellite measurements do not have the optimal sampling .
the information provided by our combined solution allows a direct determination of the mass of the unseen companion of , given an estimate of the primary mass . for this
we use our effective temperature determination in [ sec : spectroscopy ] ( @xmath56 k ) , and the brightness of the object .
the latter was measured during the hipparcos mission as @xmath57 , in the passband of the satellite .
conversion to the johnson system @xcite yields @xmath58 , where the uncertainty is our conservative estimate that includes the transformation . with our parallax of @xmath59 mas ( corresponding to a distance of @xmath60 pc )
the absolute magnitude in the visual band is @xmath61 , ignoring extinction , in which essentially all of the uncertainty comes from the parallax error .
the location of the primary star in the h - r diagram is shown in figure [ fig : hr ] .
model isochrones from the series by @xcite for solar metallicity agree well with the measured properties of , quite independently of age ( ranging from 1 gyr to 5 gyr in the figure ) . from these models we estimate the primary mass to be @xmath62 m@xmath1 ,
in which the uncertainty accounts for observational errors as well as age .
the surface gravity inferred from the models is @xmath63 , in agreement with our crude estimate from [ sec : spectroscopy ] .
although the precise metallicity of the star is unknown , our tests in [ sec : spectroscopy ] suggested a composition near solar .
additionally , the space motion of in the galactic frame based on our parallax , proper motion , and center - of - mass velocity is quite small ( @xmath64 , @xmath65 , @xmath66 , relative to the local standard of rest ) , which supports a population i origin and a heavy element abundance probably close to that of the sun , as we have assumed .
the adopted mass for the primary leads to a companion mass of @xmath67 m@xmath1 .
the uncertainty in the primary mass contributes only about 25% to this error .
the secondary star is thus 20% more massive than the primary , which implies it can not be a single main - sequence star .
a more careful examination of our spectra was carried out using the two - dimensional cross - correlation algorithm todcor @xcite , to place limits on the brightness of the companion .
this technique often allows the detection and measurement of faint secondaries that are difficult to see in standard one - dimensional cross - correlation diagrams .
no evidence of another set of lines was found down to the level of the noise , at about 10% of the brightness of the primary .
additionally , @xcite reported an observation of by speckle interferometry in which no companions were apparently seen in the range from 30 mas to 07 .
details of this measurement are unavailable .
one possibility is that the companion is a white dwarf .
depending on its temperature , it may be possible to detect it in the ultraviolet , although no such observations are available to our knowledge . given the present mass of the object , initial - final mass relations for white dwarfs ( e.g. , * ? ? ?
* and references therein ) suggest a progenitor of approximately 45 m@xmath1 , corresponding to a mid - b star . alternatively , the companion of could itself be a ( closer ) binary composed of lower main - sequence stars , making the system a hierarchical triple . in this case , however , the mass ratio of the binary can not be too small or we would have seen the brighter of the two stars in our spectra .
possible signs of this type of configuration might be seen in the form of an infrared excess .
@xmath68 measurements for are available from the 2mass catalog . with the visual magnitude listed earlier , and after conversion to the standard johnson system as defined by @xcite using transformations by @xcite , we obtain the observed colors @xmath69 , @xmath70 , and @xmath71 listed in the first line of table [ tab : photometry ] .
the predicted colors for a single star of the assumed mass ( @xmath72 m@xmath1 ) according to the models by @xcite for a representative age of 2 gyr are indeed bluer , by as much as 0.4 mag .
however , there are a number of reasons to doubt the theoretical colors in this case , including indications of missing opacity sources and other limitations in the physics ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) , and especially disagreements with empirical mass - luminosity relations , particularly in the visual band
. therefore , we have chosen here to rely on the empirical relations .
those by @xcite give redder colors than @xcite for the primary ( see third line of table [ tab : photometry ] ) , but are still considerably bluer than observed , by 0.17 mag , 0.29 mag , and 0.25 mag for @xmath69 , @xmath70 , and @xmath71 , respectively .
these differences are 67 times larger than the observational errors , suggesting the infrared excess may be real . with the aid of the @xcite relations , we have modeled this excess by computing the visual and @xmath73 magnitudes of the components of the unseen companion ( hereafter star 2 and star 3 ) over a range of possible masses for each star , subject to the constraint that their sum be the value we determined above , @xmath74 m@xmath1 .
we then added the light of these stars to that of the primary ( star 1 ) in each passband , and calculated the combined colors for the triple system .
for convenience we parameterized the problem in terms of the mass ratio of the binary ( @xmath75 , where star 2 is the more massive component ) . additionally , we allowed for changes in the mass of the main star , and sought to produce the best match to the observed colors as well as the combined absolute visual magnitude , in a @xmath49 sense .
a solution was found for a mass @xmath76 m@xmath1 and a binary mass ratio @xmath77 , which reproduces the observed visual brightness and all infrared colors within the uncertainties ( see last row of table [ tab : photometry ] ) .
contours of the @xmath49 surface as a function of @xmath78 and @xmath79 are shown in figure [ fig : contours ] , where the best fit is indicated by a dot .
the masses of the binary components from this model are @xmath80 m@xmath1 and @xmath81 m@xmath1 .
the predicted magnitude difference in the visual band between the brighter of these stars and star 1 is 3.0 mag , which is below our threshold for spectroscopic detection ( @xmath22.5 mag ; see above ) and is thus consistent with all observational constraints .
although this is formally the best fit to the observations , we note that the @xmath49 surface is nearly flat in the @xmath79 direction for @xmath79 larger than about 0.70.8 at a fixed value of @xmath78 , so that larger mass ratios up to and including unity are equally acceptable and reproduce the observed colors just as well ( see figure [ fig : contours ] ) .
the mass of star 1 , on the other hand , is very tightly constrained . at @xmath82 stars 2 and 3
would be at their faintest , each about 3.3 mag fainter than star 1 .
even though , according to the model above , both stars in the binary are too faint to be seen individually in our spectra , their combined light is not negligible compared to the brightness of star 1 . the difference is approximately 2.5 mag in the visual band .
this suggests that our earlier assumption that the astrometric motion detected by hipparcos traces only the motion of star 1 may not be entirely correct . the extra light ( even with @xmath83 ) must affect the astrometry at some level , despite our failure to detect the effect directly , as described in [ sec : orbit1 ] .
this will cause the semimajor axis of the photocenter , @xmath51 , to be slightly smaller than that of the primary alone by a factor @xmath84 , where @xmath85 is the mass fraction of the companion and @xmath86 is its fractional light .
we therefore adjusted our orbital solution to allow for this effect , and recomputed all orbital elements .
the results of this constrained " solution are seen in the last column of table [ tab : elements ] .
the semimajor axis of the photocenter is reduced from @xmath87 mas to @xmath88 mas , a small effect .
the residuals of the radial - velocity observations from this fit are listed in table [ tab : rvs ] , and those of the hipparcos measurements are given in table [ tab : hip ] .
figure [ fig : rvs ] displays the spectroscopic measurements along with the velocity curve as a function of orbital phase .
the astrometric observations on the plane of the sky are illustrated in figure [ fig : hip1 ] and figure [ fig : hip2 ] , in which the axes are parallel to the right ascension and declination directions .
the curious pattern in figure [ fig : hip1 ] is the result of the combined effects of annual parallax , proper motion , and orbital motion .
the dominant contribution is from the proper motion ( 80 mas yr@xmath89 ) , which is indicated with an arrow .
parallax and orbital motion are smaller and comparable effects . in figure
[ fig : hip2 ] we have subtracted the proper motion and parallactic contributions , leaving only the orbital motion with the 570-day period and a semimajor axis of @xmath90 mas .
the direction of motion is direct ( counterclockwise ) .
the individual hipparcos observations are represented schematically in both of these figures , but are seen more clearly in figure [ fig : hip2 ] . because they are one - dimensional in nature @xcite , the exact location of each measurement on the plane of the sky can not be shown graphically .
the filled circles represent the predicted location on the computed orbit .
the dotted lines connecting to each filled circle indicate the scanning direction of the hipparcos satellite for each measurement , and show which side of the orbit the residual is on .
the short line segments at the end of and perpendicular to the dotted lines indicate the direction along which the actual observation lies , although the precise location is undetermined .
occasionally more than one measurement was taken along the same scanning direction , in which case two or more short line segments appear on the same dotted lines . compared to our previous solution ,
the parallax is increased slightly to @xmath4 mas , which corresponds to a distance of @xmath91 pc .
the proper motion components change very little ( table [ tab : pm ] ) .
the binary mass derived with @xmath92 m@xmath1 is marginally larger than before ( @xmath93 m@xmath1 ) .
the change in the parallax leads to a revised absolute magnitude for the system of @xmath94 .
with these updated values of @xmath95 and @xmath96 we repeated the photometric modeling using the @xcite mass - luminosity relations , and obtained a solution not very different from the previous one ( the same value of @xmath78 , and @xmath97 , implying @xmath98 m@xmath1 and @xmath99 m@xmath1 ) .
the brightness difference between star 1 and the sum of stars 2 and 3 is unchanged , making another iteration of our orbital solution unnecessary .
is an interesting illustration of the complementarity of spectroscopic and astrometric observations , and in particular of the utility of the hipparcos intermediate data ( see also * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . while the radial - velocity measurements clearly reveal this object to be a binary and provide the ( single - lined ) spectroscopic orbit with a period of 570 days , the combination with the astrometry has allowed us to derive the dynamical mass of the companion ( @xmath100 m@xmath1 ) and also to correct the seriously biased parallax value from the original hipparcos reductions .
no indications of the companion are detected in our spectra , yet the orbital solution shows that it is clearly more massive than the visible star , which is a normal k dwarf with @xmath101 m@xmath1 . while we can not completely rule out that the secondary is a massive white dwarf , all the evidence points quite convincingly toward the conclusion that the companion is itself a closer binary composed of m dwarfs , making the system a hierarchical triple .
the infrared excess that might be expected from such a configuration appears indeed to be present , as indicated by our modeling of the measured visual and near - infrared ( 2mass ) photometry using empirical mass - luminosity relations by @xcite , along with the constraint on the total mass of the unseen companion .
this modeling can not determine the precise mass ratio of the close binary , but is able to place a lower limit of about @xmath102 .
smaller values would make one of the stars bright enough that it would be seen in our spectra .
we infer masses for these stars of approximately @xmath103 0.440.48 m@xmath1 and @xmath104 0.410.44 m@xmath1 .
the period of the close binary is unknown .
we note also that is listed as an x - ray source in the rosat catalog , with @xmath105 @xcite and @xmath106 @xcite .
this x - ray emission might arise naturally from the presence of the m dwarfs , which are frequently active , particularly if they are in a short - period configuration so that tidal forces compel the stars to be in synchronous rotation with the orbital motion . finally , the relative orbit of and its unseen companion has an angular semimajor axis of about 33 mas . given the eccentricity of the orbit , separations up to about 50 mas are possible at times , which may permit a direct detection of the secondary with high - resolution techniques .
alternatively , infrared spectroscopy should be able to reveal the presence of at least the brighter of the m dwarfs directly , given the more favorable contrast with the main star at those wavelengths .
the author is grateful to p. berlind , m. calkins , d. w. latham , and r. p. stefanik for obtaining the spectroscopic observations used in this work , and to r. j. davis for maintaining the cfa echelle database .
b. mason and p. hemenway are thanked for information on speckle measurements .
this paper benefited also from helpful comments by an anonymous referee .
partial support for this work from nsf grant ast-0406183 and nasa s massif sim key project ( blf57 - 04 ) is acknowledged .
this research has made use of the simbad database , operated at cds , strasbourg , france , of nasa s astrophysics data system abstract service , and 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 nasa and the nsf .
the intermediate data provided with the hipparcos catalog are the abscissae residuals " , @xmath107 , which are the difference between the satellite measurements ( abscissae ) along great circles and the abscissae computed from the 5 standard astrometric parameters .
the standard parameters are the position of the object ( @xmath108 , @xmath109 ) at the reference epoch @xmath110 , the proper motion components ( @xmath53 , @xmath54 ) , and the parallax ( @xmath41 ) .
we follow here the notation in the hipparcos catalog and define @xmath111 and @xmath112 , to incorporate the projection factors .
the goal of an orbital solution making use of the hipparcos data is to reduce the abscissae residuals below the values obtained from the 5-parameter solution by taking the orbital motion into account .
following @xcite the @xmath49 sum for the abscissae residuals is @xmath113 , where @xmath114 and is the transpose of . in this expression
is the array of n abscissae residuals provided by hipparcos , and @xmath115 is the array of partial derivatives of the abscissae with respect to the @xmath116-th fitted parameter .
the number @xmath117 of parameters fitted to the astrometry in the general case is 12 : the 5 standard hipparcos parameters ( @xmath118 , @xmath119 , @xmath120 , @xmath121 , @xmath122 ) and 7 orbital elements ( @xmath39 , @xmath30 , @xmath33 , @xmath37 , @xmath34 , @xmath38 , @xmath35 , represented as @xmath123 with @xmath124 ) .
is the inverse of the covariance matrix of the observations , containing the abscissae uncertainties and correlation coefficients ( * ? ? ?
* vol . 3 , eqs .
17 - 10 and 17 - 11 ) also provided with the hipparcos catalog .
correlations arise because the same original data were reduced independently by two data reduction consortia ( ndac and fast ; see * ? ? ?
* ) , and both results are included in the solution .
the partial derivatives @xmath125 for @xmath126 to 5 are given in the hipparcos catalog along with the abscissae residuals .
the remaining derivatives can be expressed in terms of the partial derivatives of with respect to @xmath108 and @xmath109 .
these are ( * ? ? ?
* vol . 3 , eq.17 - 15 )
@xmath127 in which @xmath128 and @xmath129 are in our case the rectangular coordinates of the photocenter relative to the center of mass of the binary on the plane tangent to the sky at ( @xmath108 , @xmath109 ) , which are given by @xmath130 @xmath131 and @xmath132 are the parallactic factors , and the terms @xmath133 and @xmath134 represent the orbital motion components , where @xmath135 , @xmath136 , @xmath137 , and @xmath138 are the classical thiele - innes constants .
these depend only on the orbital elements @xmath39 , @xmath37 , @xmath34 , and @xmath38 ( see , e.g. , * ? ? ?
@xmath139 and @xmath140 are the rectangular coordinates in the unit orbit given by @xmath141 and @xmath142 , with @xmath143 being the eccentric anomaly .
as described by @xcite , the nature of the orbital solution is such that only the derivative corresponding to the semimajor axis in eq.([eq : derivative ] ) needs to be considered .
the expression for in eq.([eq : xi ] ) then reduces to @xmath144 that the semimajor axis @xmath39 has actually been eliminated as a formal adjustable parameter in our case is irrelevant , since it still appears in the thiele - innes constant but is computed from other elements using eq.([eq : a1 ] ) .
hemenway , p. d. , duncombe , r. l. , bozyan , e. p. , lalich , a. m. , argue , a. n. , franz , o. g. , mcarthur , b. , nelan , e. , taylor , d. , white , g. , benedict , g. f. , crifo , f. , fredrick , l. w. , jefferys , w. h. , johnston , k. , kovalevsky , j. , kristian , j. , perryman , m. a.c . ,
preston , r. , shelus , p. j. , turon , c. , & van altena , w. 1997 , , 114 , 2796 jorissen , a. , jancart , s. , & pourbaix , d. 2004 , in spectroscopically and spatially resolving the components of close binary stars , asp conf .
318 , eds .
r. w. hilditch , h. hensberge and k.pavlovski ( san francisco : asp ) , 141 pourbaix , d. , jancart , s. , & jorissen , a. 2004 , in spectroscopically and spatially resolving the components of close binary stars , asp conf .
318 , eds .
r. w. hilditch , h. hensberge and k.pavlovski ( san francisco : asp ) , 144 voges , w. , aschenbach , b. , boller , th . , bruninger , h. , briel , u. , burkert , w. , dennerl , k. , englhauser , j. , gruber , r. , haberl , f. , hartner , g. , hasinger , g. , krster , m. , pfeffermann , e. , pietsch , w. , predehl , p. , rosso , c. , schmitt , j. h. m. m. , trmper , j. , & zimmermann , h. u. 1999 , , 349 , 389 cccccc 46422.9750 & 1985.9767 & @xmath14516.31 & 0.31 & @xmath50.06 & 0.605 + 46537.6905 & 1986.2907 & @xmath14514.05 & 0.57 & @xmath1451.11 & 0.806 + 46540.7560 & 1986.2991 & @xmath14512.68 & 0.41 & @xmath50.25 & 0.812 + 46569.7043 & 1986.3784 & @xmath14512.01 & 0.73 & @xmath51.43 & 0.862 + 52391.6615 & 2002.3180 & @xmath14547.99 & 0.86 & @xmath51.38 & 0.064 + 52395.7197 & 2002.3291 & @xmath14548.10 & 0.59 & @xmath1450.03 & 0.071 + 52419.6668 & 2002.3947 & @xmath14540.40 & 0.70 & @xmath50.96 & 0.113 + 52421.6565 & 2002.4002 & @xmath14540.59 & 0.50 & @xmath50.30 & 0.116 + 52423.6634 & 2002.4056 & @xmath14540.47 & 0.60 & @xmath1450.05 & 0.120 + 52424.6403 & 2002.4083 & @xmath14541.03 & 0.70 & @xmath1450.83 & 0.122 + 52655.0264 & 2003.0391 & @xmath14518.00 & 0.71 & @xmath50.54 & 0.525 + 52656.9803 & 2003.0444 & @xmath14517.85 & 0.50 & @xmath50.60 & 0.529 + 52663.9288 & 2003.0635 & @xmath14518.03 & 0.56 & @xmath50.09 & 0.541 + 52691.9163 & 2003.1401 & @xmath14516.96 & 0.54 & @xmath1450.09 & 0.590 + 52712.8028 & 2003.1973 & @xmath14514.63 & 0.97 & @xmath51.37 & 0.627 + 52752.7176 & 2003.3066 & @xmath14514.95 & 0.60 & @xmath1450.44 & 0.697 + 52772.7344 & 2003.3614 & @xmath14514.13 & 0.71 & @xmath1450.26 & 0.732 + 52985.0078 & 2003.9425 & @xmath14543.29 & 0.51 & @xmath1450.53 & 0.104 + 53013.9323 & 2004.0217 & @xmath14536.91 & 0.44 & @xmath1450.36 & 0.154 + 53035.8873 & 2004.0818 & @xmath14532.92 & 0.51 & @xmath50.19 & 0.193 + 53044.9091 & 2004.1065 & @xmath14531.77 & 0.47 & @xmath50.14 & 0.209 + 53046.9420 & 2004.1121 & @xmath14531.34 & 0.47 & @xmath50.32 & 0.212 + 53073.7601 & 2004.1855 & @xmath14528.56 & 0.44 & @xmath50.12 & 0.259 + 53125.7468 & 2004.3278 & @xmath14525.00 & 0.44 & @xmath1450.67 & 0.350 + 53132.7248 & 2004.3470 & @xmath14524.29 & 0.44 & @xmath1450.45 & 0.362 + 53337.0326 & 2004.9063 & @xmath14513.83 & 0.46 & @xmath50.25 & 0.720 + 53362.0644 & 2004.9749 & @xmath14513.70 & 0.46 & @xmath1450.32 & 0.764 + 53371.9950 & 2005.0020 & @xmath14512.45 & 0.39 & @xmath50.72 & 0.782 + 53393.8773 & 2005.0620 & @xmath14513.90 & 0.39 & @xmath1450.97 & 0.820 + 53420.9096 & 2005.1360 & @xmath14513.88 & 1.13 & @xmath1450.47 & 0.867 + 53426.8509 & 2005.1522 & @xmath14513.46 & 0.51 & @xmath50.27 & 0.878 + 53428.8623 & 2005.1577 & @xmath14513.35 & 0.51 & @xmath50.52 & 0.881 + 53451.7621 & 2005.2204 & @xmath14517.17 & 0.36 & @xmath1450.11 & 0.921 + 53457.7802 & 2005.2369 & @xmath14519.09 & 0.50 & @xmath1450.37 & 0.932 + 53480.8179 & 2005.3000 & @xmath14532.11 & 0.41 & @xmath1450.03 & 0.972 + 53482.6947 & 2005.3051 & @xmath14533.57 & 0.44 & @xmath50.26 & 0.976 + 53487.7003 & 2005.3188 & @xmath14538.96 & 0.44 & @xmath1450.07 & 0.984 + 53510.6947 & 2005.3818 & @xmath14554.52 & 0.43 & @xmath1450.13 & 0.025 + lccc + @xmath30 ( days ) & 570.70 @xmath146 0.73 & 570.95 @xmath146 0.52 & 570.98 @xmath146 0.52 + @xmath31 ( ) & @xmath147 @xmath146 0.14 & @xmath148 @xmath146 0.11 & @xmath149 @xmath146 0.11 + @xmath32 ( ) & 20.62 @xmath146 0.28 & 20.75 @xmath146 0.20 & 20.76 @xmath146 0.20 + @xmath33 & 0.6103 @xmath146 0.0071 & 0.6110 @xmath146 0.0052 & 0.6110 @xmath146 0.0051 + @xmath34 ( deg ) & 313.75 @xmath146 0.95 & 314.01 @xmath146 0.76 & 314.02 @xmath146 0.76 + @xmath35 ( hjd@xmath52,400,000 ) & 52355.2 @xmath146 1.4 & 52355.0 @xmath146 1.1 & 52355.0 @xmath146 1.1 + @xmath37 ( deg ) & & 85 @xmath146 13 & 83 @xmath146 15 + @xmath38 ( deg ) & & 175.0 @xmath146 8.8 & 179 @xmath146 10 + @xmath43 ( mas ) & & @xmath150 @xmath146 4.0 & @xmath151 @xmath146 4.2 + @xmath44 ( mas ) & & @xmath152 @xmath146 1.8 & @xmath153 @xmath146 1.7 + @xmath45 ( mas yr@xmath89 ) & & @xmath154 @xmath146 1.4 & @xmath155 @xmath146 1.4 + @xmath46 ( mas yr@xmath89 ) & & @xmath156 @xmath146 1.3 & @xmath157 @xmath146 1.3 + @xmath47 ( mas ) & & @xmath158 @xmath146 1.8 & @xmath159 @xmath146 1.9 + + @xmath160 ( m@xmath1 ) & 0.258 @xmath146 0.011 & & + @xmath161 ( m@xmath1 ) & 0.6364 @xmath146 0.0092 & & + @xmath162 ( @xmath163 km ) & 128.2 @xmath146 1.9 & & + @xmath39 ( mas ) & & 16.9 @xmath146 1.5 & 17.9 @xmath146 1.6 + @xmath51 ( mas ) & & 16.9 @xmath146 1.5 & 14.9 @xmath146 1.3 + @xmath36 ( au ) & & 1.580 @xmath146 0.029 & 1.582 @xmath146
0.032 + @xmath36 ( mas ) & & 30.8 @xmath146 3.2 & 32.5 @xmath146 3.3 + @xmath53 ( mas yr@xmath89 ) & & @xmath164 @xmath146 1.4 & @xmath165 @xmath146 1.4 + @xmath54 ( mas yr@xmath89 ) & & @xmath166 @xmath146 1.3 & @xmath167 @xmath146 1.3 + @xmath41 ( mas ) & & 19.5 @xmath146 1.8 & 20.6 @xmath146 1.9 + cccccccc 47884.3184 & 1989.9776 & @xmath56.60 & 4.60 & f & 0.762 & @xmath55.24 & 0.170 + 47884.1723 & 1989.9772 & @xmath58.65 & 5.40 & n & 0.762 & @xmath57.30 & 0.170 + 47901.6313 & 1990.0250 & + 1.61 & 5.67 & f & 0.531 & @xmath56.77 & 0.200 + 47901.5947 & 1990.0249 & + 5.55 & 4.55 & n & 0.531 & @xmath52.84 & 0.200 + 48020.3010 & 1990.3499 & + 3.81 & 4.31 & f & 0.736 & + 0.01 & 0.408 + 48020.1914 & 1990.3496 & @xmath52.70 & 5.51 & n & 0.736 & @xmath56.49 & 0.408 + 48046.1241 & 1990.4206 & + 20.01 & 8.67 & f & 0.868 & + 12.78 & 0.453 + 48046.0146 & 1990.4203 & + 20.01 & 10.75 & n & 0.868 & + 12.78 & 0.453 + 48046.3068 & 1990.4211 & + 1.86 & 4.72 & f & 0.773 & @xmath55.37 & 0.454 + 48046.3798 & 1990.4213 & + 8.67 & 5.82 & n & 0.773 & + 1.44 & 0.454 + 48069.0619 & 1990.4834 & + 6.63 & 8.84 & f & 0.878 & + 0.25 & 0.494 + 48069.0619 & 1990.4834 & @xmath50.14 & 9.83 & n & 0.878 & @xmath56.53 & 0.494 + 48069.2445 & 1990.4839 & @xmath52.29 & 6.59 & f & 0.000 & @xmath58.62 & 0.494 + 48203.1086 & 1990.8504 & @xmath51.80 & 3.97 & f & 0.789 & + 2.00 & 0.728 + 48203.0721 & 1990.8503 & @xmath54.01 & 4.66 & n & 0.789 & @xmath50.22 & 0.728 + 48249.2397 & 1990.9767 & @xmath53.38 & 11.29 & f & 0.950 & @xmath53.45 & 0.809 + 48249.2397 & 1990.9767 & @xmath51.29 & 12.87 & n & 0.950 & @xmath51.37 & 0.809 + 48367.3615 & 1991.3001 & @xmath517.97 & 5.69 & f & 0.798 & @xmath51.14 & 0.016 + 48367.2520 & 1991.2998 & @xmath524.12 & 7.29 & n & 0.798 & @xmath57.32 & 0.016 + 48383.7978 & 1991.3451 & @xmath57.19 & 6.53 & f & 0.853 & @xmath55.25 & 0.045 + 48383.7978 & 1991.3451 & @xmath58.67 & 7.31 & n & 0.853 & @xmath56.71 & 0.045 + 48454.3275 & 1991.5382 & @xmath51.37 & 6.17 & f & 0.853 & @xmath50.54 & 0.168 + 48454.2910 & 1991.5381 & + 1.18 & 7.44 & n & 0.853 & + 2.00 & 0.168 + 48454.8024 & 1991.5395 & + 4.31 & 3.93 & f & 0.768 & + 5.72 & 0.169 + 48454.7658 & 1991.5394 & + 2.88 & 4.82 &
n & 0.768 & + 4.26 & 0.169 + 48458.3088 & 1991.5491 & @xmath517.61 & 4.83 & f & 0.825 & @xmath512.14 & 0.175 + 48458.3453 & 1991.5492 & @xmath519.79 & 5.64 & n & 0.825 & @xmath514.34 & 0.175 + 48458.7471 & 1991.5503 & @xmath510.35 & 4.28 & f & 0.753 & @xmath54.42 & 0.176 + 48458.7471 & 1991.5503 & @xmath513.81 & 4.59 & n & 0.753 & @xmath57.92 & 0.176 + 48551.9954 & 1991.8056 & @xmath56.38 & 5.31 & f & 0.799 & @xmath57.26 & 0.339 + 48551.9223 & 1991.8054 & @xmath55.40 & 5.89 & n & 0.799 & @xmath56.29 & 0.339 + 48552.2145 & 1991.8062 & @xmath52.14 & 6.75 & f & 0.683 & @xmath52.89 & 0.340 + 48552.3606 & 1991.8066 & @xmath511.07 & 10.06 & n & 0.683 & @xmath511.80 & 0.340 + 48562.6972 & 1991.8349 & + 5.40 & 6.78 & f & 0.779 & + 9.22 & 0.358 + 48562.5146 & 1991.8344 & + 8.30 & 9.05 & n & 0.779 & + 12.09 & 0.358 + 48562.9164 & 1991.8355 & @xmath51.35 & 5.10 & f & 0.730 & + 2.59 & 0.358 + 48562.9894 & 1991.8357 & @xmath54.05 & 6.93 & n & 0.730 & @xmath50.10 & 0.359 + 48595.8984 & 1991.9258 & @xmath59.20 & 9.97 & f & 0.894 & @xmath510.50 & 0.416 + 48595.7889 & 1991.9255 & @xmath511.43 & 11.07 & n & 0.894 & @xmath512.75 & 0.416 + 48629.5380 & 1992.0179 & @xmath517.27 & 4.36 & f & 0.778 & @xmath57.63 & 0.475 + 48629.7206 & 1992.0184 & @xmath523.85 & 4.97 & n & 0.778 & @xmath514.20 & 0.475 + 48638.9614 & 1992.0437 & + 5.21 & 4.58 & f & 0.735 & + 11.59 & 0.492 + 48638.8519 & 1992.0434 & @xmath50.26 & 4.85 &
n & 0.735 & + 6.12 & 0.491 + 48639.2901 & 1992.0446 & @xmath55.36 & 4.10 & f & 0.686 & + 0.77 & 0.492 + 48639.3632 & 1992.0448 & @xmath57.20 & 4.02 & n & 0.686 & @xmath51.06 & 0.492 + 48760.4436 & 1992.3763 & + 9.38 & 9.65 & f & 0.751 & + 10.30 & 0.704 + 48760.4070 & 1992.3762 & @xmath52.08 & 13.41 & n & 0.751 & @xmath51.18 & 0.704 + 48808.2183 & 1992.5071 & + 10.67 & 11.83 & f & 0.875 & + 8.18 & 0.788 + 48809.0218 & 1992.5093 & + 4.26 & 13.06 &
n & 0.875 & + 1.84 & 0.789 + 48809.2410 & 1992.5099 & + 3.03 & 5.48 & f & 0.505 & + 0.89 & 0.790 + 48809.0949 & 1992.5095 & + 1.71 & 9.46 & n & 0.505 & @xmath50.42 & 0.790 + 48942.7033 & 1992.8753 & @xmath52.01 & 6.43 & f & 0.777 & @xmath59.08 & 0.024 + 48942.7033 & 1992.8753 & + 2.19 & 7.92 & n & 0.777 & @xmath54.88 & 0.024 + lccc hipparcos catalog @xcite & @xmath580.0 @xmath146 2.4 & @xmath521.6 @xmath146 1.9 & 29.4 @xmath146 2.7 + initial fit ( this paper ) & @xmath576.1 @xmath146 2.2 & @xmath524.6 @xmath146 1.8 & 19.4 @xmath146 1.9 + tycho-2
catalog @xcite & @xmath574.9 @xmath146 1.8 & @xmath525.3 @xmath146 1.8 & + combined fit ( this paper ) & @xmath575.4 @xmath146 1.4 & @xmath525.0 @xmath146 1.3 & 19.5 @xmath146 1.8 + constrained fit ( this paper ) & @xmath575.5 @xmath146 1.4 & @xmath524.8 @xmath146 1.3 & 20.6 @xmath146 1.9 + lcccc observed ( @xmath41 from combined solution ) & 7.25 @xmath146 0.20 & 2.32 @xmath146 0.03 & 2.99 @xmath146 0.04 & 3.10 @xmath146 0.04 + @xcite isochrones & 7.25 & 1.96 & 2.60 & 2.67 + @xcite & 7.22 & 2.15 & 2.70 & 2.85 + observed ( @xmath41 from constrained solution ) & 7.37 @xmath146 0.20 & 2.32 @xmath146 0.03 & 2.99 @xmath146 0.04 & 3.10 @xmath146 0.04 + model for triple system & 7.14 & 2.34 & 2.95 & 3.12 + | we report spectroscopic observations of the star , previously considered ( but later rejected ) as a candidate member of the tw hya association .
our measurements reveal it to be a single - lined binary with an orbital period of 570 days and an eccentricity of @xmath0 .
the astrometric signature of this orbit was previously detected by the hipparcos satellite in the form of curvature in the proper motion components , although the period was unknown at the time . by combining our radial velocity measurements with the hipparcos intermediate data ( abscissae residuals ) we are able to derive the full three - dimensional orbit , and determine the dynamical mass of the unseen companion as well as a revised trigonometric parallax that accounts for the orbital motion .
given our primary mass estimate of 0.73 m@xmath1 ( mid - k dwarf ) , the companion mass is determined to be 0.89 m@xmath1 , or @xmath220% _ larger _ than the primary .
the likely explanation for the larger mass without any apparent contribution to the light is that the companion is itself a closer binary composed of m dwarfs .
the near - infrared excess and x - ray emission displayed by support this .
our photometric modeling of the excess leads to a lower limit to the mass ratio of the close binary of @xmath3 , and individual masses of 0.440.48 m@xmath1 and 0.410.44 m@xmath1 .
the new parallax ( @xmath4 mas ) is significantly smaller than the original hipparcos value , and more precise . |
there has been much attention to the molecular bose - einstein condensate ( bec ) which is a versatile playground not only for cold atomic physics experiments but also for other research areas , such as condensed matter theory as well as quantum field theory .
it is no longer a pure bose system as the molecular bec can be a system of bose - fermi mixtures .
this makes the theoretical structure rich and colorful . to realize the molecular bec , one must create stable molecules with long lifetime . in the recent experiments @xcite , the technique of feshbach resonance plays an important role in the creation of molecules . since the molecules created through such a technique usually suffer from fast decay due to the vibrational excitation , the stimulated raman adiabatic passage ( stirap ) in photoassociation @xcite has been regarded as an effective approach to create ground - state molecules .
the success of stirap technique requires the existence of a coherent population trapping ( cpt ) state which can be followed adiabatically @xcite .
such a condition can be fulfilled for linear @xmath2 systems by appropriately choosing laser frequencies . however , for the system with inter - particle interactions , the two - photon resonance condition dynamically changes when population is transferred from atomic states to molecular states .
this makes the cpt state more difficult to be followed adiabatically .
the adiabatic property was studied by means of adiabatic fidelity in a recent theoretical work @xcite for a simplified model of monoatomic system without inter - particle interaction .
it is worthwhile to appropriately define the adiabatic fidelity to study the adiabatic property for a more realistic system , such as the diatomic system which has been realized in serval experiments @xcite . in this paper , we consider systems consisting of fermionic atoms in different hyperfine states and their compounded molecules coupled through the stirap technique aided by feshbach resonance .
we discuss systems of @xmath0li and @xmath1k as concrete examples . in section [ sec : general ] , we model the systems with inter - particle interaction and derive the dynamical equations for them .
we make a general consideration on the symmetry and the relevant conservation law , and then introduce the definition of fidelity for our system . in section [ sec
: cpt - state ] , we look for solutions of the cpt state . in section [ sec : efficiency ] , we calculate the conversion efficiency for @xmath1k atoms and @xmath0li atoms , respectively , and discuss the corresponding features . in section [ sec : fidelity ] , we study the relation between the atom - to - molecule conversion efficiency and the adiabatic fidelity for cpt state .
we also study the effect of the decay of quansibound molecules and compare the difference between @xmath1k and @xmath0li systems .
our results are briefly summarized in section [ sec : summary ] .
we consider that two species of fermionic atoms are converted into stable molecules via the stirap technique aided by feshbach resonance . here
we use @xmath3 and @xmath4 stand for the states of the free atoms in the open channel , and @xmath5 and @xmath6 for the quasibound and ground molecular states in the close channel , respectively .
a pair of ferimonic atoms are coupled with the quasibound molecular state @xmath5 through a feshbach resonance , where the coupling strength is denoted by @xmath7 and the detuning by @xmath8 .
additionally , the sates @xmath5 and @xmath6 are coupled with each other through a laser field with the coupling strength @xmath9 and detuning @xmath10 .
since the system is a fermi - bose mixture whose hilbert space actually carries out the representation of a graded unitary group su@xmath11 , it will be an arduous work to study the dynamics of the system by means of the heisenberg equation of motion . whereas , with the help of mean - field langrange density @xcite , one can investigate the dynamics conveniently , @xmath12 where @xmath13 denotes the kinetic energy term of the @xmath14th component and @xmath15 . here
the coefficients @xmath16 are the interaction strengths between particles with @xmath17 being the reduced mass and @xmath18 the @xmath19-wave scattering length .
the lagrangian ( [ eq : lag ] ) is no more invariant under a simultaneous global phase transformation due to the presence of the atom - to - molecule conversion term . as the lagrangian does not include the term flipping the two fermionic components into each other , their corresponding phase parameters are not necessarily the same in general .
one can find that the lagrangian eq .
( [ eq : lag ] ) is invariant under the phase transformation , @xmath20 with @xmath21 following the formulism of noether theorem , we evaluate the variation of the action caused by the infinitesimal phase transformation @xmath22 , @xmath23 for simplicity in the above expression , we omitted the subscripts of @xmath24 labelling different components , abbreviated @xmath25 for @xmath26 and @xmath27 for @xmath28 with @xmath29 . because of , explicitly , @xmath30 , @xmath31 , @xmath32 , and their complex conjugations , we derive the following result , @xmath33 where @xmath34 , @xmath35 with @xmath36 the system is invariant under the transformation eq .
( [ eq : trans ] ) such that @xmath37 , which gives rise to two conserved currents .
@xmath38 in the present paper , we focus on uniform system ( @xmath39 ) and hence neglect the kinetic and trapping potential terms
. then the conservation law eq .
( [ eq : conserved - currents ] ) reads @xmath40 , @xmath41 , which implies that @xmath42 and @xmath43 are determined by @xmath44 , the volume of the system , together with @xmath45 and @xmath46 , the initial numbers of species @xmath47 and @xmath48 , _ i.e. _ , @xmath49 , @xmath50 .
here we assume there are no molecules at the initial time in the system . to guarantee the compatibility with the constraints given by eq .
( [ eq : constraints ] ) , we introduce two lagrange multipliers @xmath51 and @xmath52 into the mean - field lagrange density eq .
( [ eq : lag ] ) , @xmath53 here the real parameters @xmath54 and @xmath55 can be identified as the chemical potentials of the corresponding components .
owning to eq .
( [ eq : constraints ] ) , we can conveniently introduce new notations : @xmath56 , @xmath57 , @xmath58 , @xmath59 , @xmath60 , @xmath61 , and @xmath62 .
reexpressing equation ( [ eq : lc ] ) in terms of @xmath63 s and substituting it into the euler - lagrange equation , we obtain a set of equations @xmath64 where a phenomenological parameter @xmath65 is introduced to characterize the decay of quasibound molecules . in terms of @xmath66 s , the conservation relations ( [ eq : constraints ] ) turn to be @xmath67 characterizing the population imbalance between fermionic atoms in different states .
now we are in the position to introduce a proper definition of fidelity for our system .
as our system is a four - component system of bose - fermi mixture which is related to a graded unitary group su@xmath11 , we need to define the fidelity carefully as it must obey several basic properties @xcite . equation ( [ eq : constraint - redef ] ) actually provides us the normalization condition @xmath68 which can be expressed as the following form @xmath69 where @xmath70 denotes @xmath71 .
one might think of a naive expression for the @xmath72-matrix , @xmath73 .
however , because the relation of eq .
( [ eq : normalization ] ) should be invariant under the transformation given in eq .
( [ eq : trans ] ) , just like that the conventional inner product in quantum mechanics is invariant under the u(1 ) transformation , the simplest correct expression of the @xmath72-matrix ought to be @xmath74 as a result , a nature definition of fidelity of a state labelled by @xmath63 with that labelled by @xmath75 is given by @xmath76 where the @xmath72-matrix was given in eq .
( [ eq : f ] ) .
clearly , such a definition fulfills @xmath77 , which means the phase transformation given in eq .
( [ eq : trans ] ) does not vary the magnitude of fidelity ; and the fidelity of a state with itself is always the unit @xmath78 which is just the normalization condition .
now we consider the stationary states where we neglect the decay of quasibound molecules ( _ i.e. _ , @xmath79 ) .
we know that the existence of stationary solutions of eq .
( [ eq : de ] ) requires the system satisfies the adiabatic approximation condition . once the adiabatic approximation is valid , _
i.e. _ , @xmath80 , the time - evolution equations ( [ eq : de ] ) become algebraic ones for @xmath63 s , namely , @xmath81 although it is difficult to find the exact solutions of the above equations , one can easily obtain a set of steady state solutions for eq .
( [ eq : station ] ) by taking @xmath82 .
such a state is called coherent population trapping ( cpt ) state which yields , @xmath83 where @xmath84 .
the resonance condition corresponding to this solution is @xmath85 we consider a laser pulse with @xmath86 for @xmath87 and @xmath88 for @xmath89 .
for such a laser field , at the initial time ( @xmath90 ) , we have @xmath91 , @xmath92 and @xmath93 , which implies there are no molecules in the system at the initial time . at the final time ( @xmath89 ) , @xmath94 , @xmath95 and @xmath96 for @xmath97 ; whereas @xmath98 , @xmath99 and @xmath100 for @xmath101 . with the help of initial values of @xmath102 and their asymptotic values at final time ,
it is easy to find that those fermionic atoms , in the presence of their counterparts , can be converted into molecules if the cpt states can be followed adiabatically .
the residual atoms can not be converted into molecules due to the lack of counterpart atoms .
after the numerical calculation in next section , we will go back to study whether the cpt state can be followed adiabatically with the help of the useful concept , adiabatic fidelity .
we know that there are two sorts of fermionic atoms @xmath0li and @xmath1k in the group of alkali - metal atoms .
they have been converted into molecules in experiments successfully through the bare feshbach resonance @xcite . in those experiments , 60% to 80% of @xmath1k atoms and
no more than 85% of @xmath0li atoms can be converted into molecules . for @xmath0li atoms , the atom - to - molecule conversion efficiency via
bare feshbach resonance is lower than that via stirap technique aided by feshbach resonance , which is in contrast to the case for @xmath1k atoms . due to the difference in atomic properties , the atom - to - molecule conversion efficiency differs for different atoms even if the same technique is applied .
now we evaluate the atom - to - molecule conversion efficiency , respectively , for @xmath0li and @xmath1k atoms with concrete magnetic and laser fields .
for these two sorts of atoms , we adopt the same time - dependent rabi frequency , @xmath103,\ ] ] where the parameters @xmath104 , @xmath105 and @xmath106 are determined by the applied laser field that couples the two molecular states . in the numerical calculation ,
the detuning strength @xmath107 is given by eq .
( [ eq : rc ] ) .
we assume there are no molecules in the system at the initial time _
i.e. _ , @xmath108 at @xmath90 .
( color online ) the time dependence of the population of particles for @xmath1k system for different @xmath65 .
the parameters are @xmath109 , @xmath110 , @xmath111 , @xmath112 , @xmath113 , @xmath114 , @xmath115 , and the other parameters are zero .
time is in unit of @xmath116 and all other coefficients are in units of @xmath117 , where @xmath118j.,width=283 ] for @xmath1k atoms , we know that the feshbach resonance occurs at a magnetic field strength of 202.1 g and the resonance width is about 7.8 g @xcite
. then we can get the atom - to - molecule coupling strength @xmath119 according to ref .
@xcite . here
we choose that the magnetic field is @xmath120 g and particle density @xmath121 is about @xmath122m@xmath123 .
it is easy to obtain @xmath124 , @xmath125 and @xmath126 .
the time evolution of the corresponding population can be obtained by solving eq .
( [ eq : de ] ) .
we plot the numerical results in fig .
[ fig : k ] . from this figure
, we can see that the conversion efficiency @xmath127 for @xmath1k atoms is less than 60% , lower than that via the technique of bare feshbach resonance @xcite .
the low conversion efficiency implies that the cpt state can not be followed adiabatically , which will be confirmed confidently by evaluating the adiabatic fidelity in next section .
comparing the two panels in fig.[fig : k ] , we can know that the influence of @xmath65 on the atom - to - molecule conversion efficiency is very small .
this is due to that @xmath128 is close to zero at any time in contrast to the case for @xmath0li atoms .
( color online ) the time dependence of the population of particles for @xmath0li system for different @xmath65 .
the parameters are @xmath109 , @xmath129 , @xmath130 , @xmath131 , @xmath113 , @xmath114 , @xmath115 , and the other parameters are zero .
time is in unit of @xmath116 and all other coefficients are in units of @xmath117 , where @xmath132j.,width=283 ] for @xmath0li atoms , the feshbach resonance occurs at two distinct strengths of magnetic field ( the called narrow and broaden feshbach resonance respectively ) . in our calculation , we focus on the narrow feshbach resonance for which the atom - to - molecule coupling strength @xmath7 is about 3.29@xmath133j .
if the particle density @xmath121 is about @xmath122m@xmath123 and the magnetic field is about 543.6 g , one can get @xmath134 , @xmath135 and @xmath136 .
figure [ fig : li ] shows the time evolution of particle populations for the conversion of @xmath0li atoms into molecules . from the top panel ,
we find that almost all of @xmath0li atoms can be converted into molecules . for @xmath0li atoms , the atom - to - molecule conversion efficiency via the stirap technique aided by feshbach resonance is higher than that via bare feshbach resonance @xcite , which is in contrast to the case for @xmath1k atoms . from the bottom panel , we can see that the atom - to - molecule conversion efficiency decreases distinctly due to the existence of the quasibound molecular decay . the dependence of the conversion efficiency on the decay rate @xmath65 of quasibound molecules for both @xmath1k and @xmath0li is plotted in fig .
[ fig : gama ] .
clearly , the influence of the decay on the conversion efficiency for @xmath0li atoms is more distinct than that for @xmath1k atoms , which confirms the previous interpretation . the atom - to - molecule conversion efficiency versus the decay rate of the quansibound molecules for @xmath1k ( left panel ) and for @xmath0li ( right panel ) system .
the parameter choice for left ( right ) panel is the same as fig .
[ fig : k ] ( fig .
[ fig : li ] ) .
, width=283 ]
from the above numerical results , we can find that the atom - to - molecule conversion efficiencies of stirap technique aided by feshbach resonance for @xmath1k and @xmath0li are distinctly different although the cpt states are assumed to exist for the two systems .
as we know that most of the atoms can be converted into molecules only when the cpt state is followed adiabatically , and the existence of the cpt state does not guarantee the state can be followed adiabatically @xcite .
the adiabatic properties for the atom - to - molecule conversion have been studied in ref .
whereas , we can not apply this method to our system since all the possible solutions of the stationary equation ( [ eq : de ] ) must be known in the approach @xcite and it is difficult to obtain the other solutions beyond the cpt - state solution in our system . the time dependence of adiabatic fidelity for cpt state : ( * a * ) for @xmath1k , the same parameter choice as in fig.[fig : k ] ; ( * b * ) for @xmath137li , the same parameter choice as in fig.[fig : li ] .
, width=321 ] recently , the fidelity was employed @xcite to characterize the adiabatic condition for systems with atom - to - molecule conversion .
the key point is that the value of the fidelity should close to unity if the system can adiabatically evolve in the cpt state .
now we evaluate the adiabatic fidelity @xmath138 for our model , in which @xmath139 is the wave function obtained by solving the dynamical eq .
( [ eq : de ] ) with @xmath79 , and @xmath140 is the wave function corresponding to the cpt state which is given in eq .
( [ eq : cpt ] ) .
the time evolution of the adiabatic fidelity for cpt state is plotted in fig .
[ fig : fidelity](a ) for @xmath1k system .
one can see that the magnitude of fidelity is about @xmath141 at the initial time , which implies the system adiabatically evolves along cpt state . at the time @xmath142
, the fidelity diminishes to the minimal value @xmath143 , which implies the system deviates away from the cpt state distinctly at that time .
although the fidelity begins to fluctuate later on , its final value is still no more than @xmath144 .
the above analysis implies that the cpt state can not be followed adiabatically , so the atom - to - molecule conversion efficiency is not high for @xmath1k atoms .
[ fig : fidelity](b ) shows the time evolution of the adiabatic fidelity for cpt state corresponding to the conversion of @xmath0li atoms into molecules .
obviously , the fidelity for this system is very close to @xmath141 at all the time , which implies that the system adiabatically evolves along the cpt state .
this result is consistent with the fact that almost all of the @xmath0li atoms can be converted into molecules .
since the fidelity for cpt state can well reflect the atom - to - molecule conversion efficiency , one can improve the conversion efficiency by optimizing the parameters of the system to achieve a higher adiabatic fidelity .
with the help of mean - field lagrangian density , we studied the conversion of two - species fermionic atoms into bosonic molecules via stirap technique aided by fechbach resonance .
we calculated conversion efficiencies for @xmath1k and @xmath0li systems respectively , and found that almost all of the @xmath0li atoms can be converted into molecules , which implies that the stirap technique aided by feshbach resonance is more effective than the bare fechbach resonance for @xmath0li atoms rather than @xmath1k atoms .
we also compared the influence of the decay rate of quansibound molecules on the conversion efficiency for @xmath1k and @xmath0li systems respectively , and found that there is a big difference between them .
the success of stirap technique not only requires the existence of the cpt state but also requires that the system can adiabatically evolve within such a state .
the adiabatic fidelity was recently recognized to be a useful measurement for characterizing the adiabatic properties .
our analysis of the symmetry and the corresponding conservation law for the systems under consideration helped us to introduce an appropriate definition of adiabatic fidelity for cpt state .
for @xmath1k system , the cpt state can not be adiabatically followed since the fidelity was found to be less than @xmath144 at final time , which is well consistent with the corresponding low conversion efficiency . in order to improve the conversion efficiency in @xmath1k system , one should achieve a higher fidelity through optimizing the parameters of the system . whereas , for @xmath0li system , the fidelity is very close to @xmath141 , hence the cpt state can be followed adiabatically , which is the reason for a high conversion efficiency .
our evaluation of the adiabatic fidelity enable us to understand why the conversion efficiencies for @xmath1k and @xmath0li are distinctly different .
j. cubizolles , t. bourdel , s. j. j. m. f. kokkelmans , g. v. shlyapnikov , and c. salomon , phys .
* 91 * , 240401 ( 2003 ) . c. a. regal , m. greiner , and d. s. jin , phys .
lett . * 92 * , 040403 ( 2004 ) . c. a. stan , m.w .
zwierlein , c. h. schunck , s.m .
f. raupach , and w. ketterle , phys .
lett . * 93 * , 143001 ( 2004 ) . | the efficiency of converting two - species fermionic atoms into bosonic molecules is investigated in terms of mean - field lagrangian density .
we find that the stirap technique aided by feshbach resonance is more effective than the bare fechbach resonance for @xmath0li atoms rather than @xmath1k atoms .
we also make general consideration on the symmetry and its relevant conservation law , which enable us to introduce a natural definition of adiabatic fidelity for cpt state .
the calculated values of the fidelity then provide an interpretation on why the conversion efficiencies for @xmath1k and @xmath0li are distinctly different . |
jupiter - family comet 103p / hartley 2 became one of the best - studied comets of all time when , in late-2010 , it simultaneously passed very close to the earth and was the subject of a flyby by the _ epoxi _ spacecraft ( e.g. , @xcite ) .
first results have already been published for _ epoxi _ spacecraft data @xcite , radar observations of the nucleus @xcite , composition of the coma @xcite , and morphology of the coma @xcite . these and earlier results will undoubtedly be discussed in more detail elsewhere in this special issue ; in the interest of brevity , we restrict our discusion to those results having bearing on our current work .
we obtained both photometry and imaging of hartley 2 , and discuss the observations and reductions in section [ sec : obs ] . in section [ sec : photometry ]
we discuss our photoelectric photometer measurements acquired during the 1991 , 1997/98 , and 2010/11 apparitions . due to unfavorable viewing geometry
, we did not observe hartley 2 on its two other known apparition , 1986 ( when it was discovered ) or 2004 .
previous investigators @xcite found a typical composition and a low dust - to - gas ratio . assuming that hartley 2 had a comparable active fraction to other comets ( @xmath010% ) , it was inferred to have a rather large nucleus ( of order a few km ) .
the flyby conclusively showed that hartley 2 has a small nucleus ( effective radius of [email protected] km ; @xcite ) , confirming the conclusions of remote studies of the nucleus @xcite
. also revealed that the active fraction of the nucleus is relatively low and thus its `` hyperactivity '' necessitates an extended source of volatiles .
numerous new water production measurements have recently been published in support of and are compared with our measurements .
we also follow up our previous investigation of the cn coma morphology in the 2010/2011 apparition .
we discovered the existence of time - varying cn coma structures and used the repetition of these structures to measure a rotation period @xcite . in @xcite , henceforth `` paper 1 '' , we interpreted the morphology as likely being due to two `` jets '' whose appearance varied as the nucleus rotated , and evolved during the apparition .
we used repetitions in the morphology to measure rotation periods monthly from 2010 august through 2010 november and determined that the rotation period was increasing during this time from @xmath016.7 hr in august to @xmath018.7 hr in november .
we noted that the morphology did not exactly repeat from cycle to cycle , but repeated better after 3 , 6 , 9 ... cycles and inferred that it was in non - principal axis rotation with a component period roughly three times as long as the measured rotation period .
similar coma morphology and rotation periods were reported by @xcite , whose images covered a comparable range of dates ( 2010 september 1 through december 15 ) .
@xcite acquired snapshot imaging near perigee , reporting similar cn morphology and showing that the c@xmath2 coma morphology also resembled the cn coma morphology .
et al . _ also reported the existence of a much smaller - scale dust jet in the sunward direction .
a similar feature is visible in movies made from hri images posted on the website . @xcite
obtained cn and c@xmath2 images near the time of the flyby , finding a similar periodicity as we reported and noting the agreement of the cn and c@xmath2 morphology . in section [ sec : gas_morphology ] we intercompare the coma morphology exhibited by various gas species ( cn , c@xmath2 , c@xmath1 , oh , and nh ) in 2010 october and november , and discuss the dust morphology , in particular looking for evidence of the dust jet .
finally , in section [ sec : disc ] we discuss our results and how they fit into the larger body of knowledge about hartley 2 acquired by and numerous remote observers .
we summarize our findings in section [ sec : summary ] and briefly discuss their application to future modeling efforts .
the goals of our studies of hartley 2 have varied , with composition and production rate studies the emphasis of the 1991 and 1997/98 apparitions and multiwavelength imaging the emphasis in 2010/11 ( with production rate and compositional monitoring a secondary emphasis ) .
thus , our techniques and instrumentation have varied .
the production rate and composition studies continue to be made with a traditional photoelectric photometer .
this provides superior signal - to - noise for bulk coma measurements as compared to a ccd and also ensures continuity with earlier data sets .
the imaging necessarily requires use of a ccd .
all photometer observations were obtained with the hall 42-in ( 1.1-m ) telescope at lowell observatory except for one night in 1997 which was made using lowell s 0.9-m telescope at perth observatory .
imaging observations were acquired with either the hall 42-in or the 31-in ( 0.8-m ) telescopes at lowell observatory . in total , we acquired two nights of photometry in 1991 , four nights of photometry in 1997/98 , 13 nights of photometry in 2010/11 , and 39 nights of imaging in 2010/11 .
the observing circumstances for all nights of imaging and photometry are given in tables [ t : imaging_circ ] and [ t : phot_circ ] , respectively .
the photometers used had emi 6256 photomultiplier tubes and pulse counting systems except for the sole night ( 1997 december 1 ) at perth observatory where a dc amplifier was employed . the hale - bopp ( hb ) narrowband filter set @xcite was utilized in 1997 and later , superceding the international halley watch ( ihw ) set @xcite used in 1991 .
hb filters were also used for the ccd observations , along with a broadband cousins r filter for increased signal - to - noise for dust measurements .
the narrowband comet filters used for both our photometry and imaging isolate emission from daughter gas species ( oh , nh , cn , c@xmath2 , c@xmath1 ) or reflected solar continuum from dust ( continuum filters in the uv ( uc ) , blue ( bc ) , green ( gc ) , and red ( rc ) ) .
the comet filters are regularly used to study dust and coma in comets ( e.g. , @xcite ) . using the method described in @xcite and @xcite ,
we `` decontaminate '' the data by removing the underlying signal due to reflected solar continuum ( and in the case of cn and nh , contamination from c@xmath2 as well ) from gas bandpasses to yield pure gas and pure dust data .
the 1.1-m images were obtained using an e2v ccd231 - 84 chip with 4k@xmath44k pixels which was binned 2@xmath42 at the telescope , resulting in a pixel scale of 0.74 arcsec .
the 0.8-m images were obtained with an e2v ccd42 - 40 chip with 2k@xmath42k pixels and a pixel scale of 0.46 arcsec . at times images
were rebinned by an additional factor of 2 during processing to improve the effective signal - to - noise ; the final pixel scales of these images were 1.48 arcse for 1.1-m images and 0.92 arcsec for 0.8-m images .
earlier studies of hartley 2 had concluded that it has a small , `` hyperactive '' nucleus ( e.g. , @xcite ) , so our imaging program focused on coma morphology rather than direct nucleus studies . throughout the apparition , we regularly monitored the comet in broadband r and narrowband cn filters , with occasional narrowband blue continuum ( bc ) images as well .
on photometric nights we used additional hb narrowband filters , increasing the variety and frequency as the comet brightened and shorter exposure times were possible .
images using an h@xmath1o@xmath5 filter were only obtained on 2010 november 4 , in support of concurrent _ chandra _ observations being conducted by @xcite ; they are not discussed further in this paper .
see table [ t : imaging_circ ] for a list of the filters used on a given night .
exposure times varied by filter and with the comet s brightness , but ranged from as long as 900 s ( oh in september ) to as short as 30 s ( r in november ) .
most r , oh , and cn images were obtained in sets of 310 images while most images in other filters were obtained as single images . all 1.1-m images were guided at the comet s rate of motion while all 0.8-m images were tracked at the comet s rate of motion ( the 0.8-m does not have guiding capability ) .
the hb filters are parfocal and since the primary goal of the imaging campaign was gas coma morphology , we focused for these filters rather than for the r - band .
our entire hartley 2 campaign coincided with a separate effort to measure the nucleus lightcurve of 10p / tempel 2 @xcite .
as tempel 2 s period is known precisely , and is near 9 hr and changing by @xmath0 + 0.005 hr per orbit @xcite , we required regular monitoring of its lightcurve over many months to detect a change in the period .
fortunately , tempel 2 generally set as hartley 2 was rising , but occasionally we split time on both objects .
thus , our temporal coverage of hartley 2 was slightly less complete than it might have otherwise been , but this did not significantly affect any science results on hartley 2 .
we removed the bias and applied a flat field following standard reduction techniques .
we observed hb narrowband standard stars @xcite on all photometric nights .
for these nights , we followed our standard photometric procedures @xcite to determine flux calibrations and process the narrowband images into pure gas and pure dust images .
as discussed in paper 1 , there was not significant contamination from the dust continuum in cn images .
this allowed us to make morphological assessments of cn images on non - photometric nights ( when we could not generate fully `` decontaminated '' images ) .
we normally only observe additional species on photometric nights , but obtained images in other filters despite non - photometric conditions on november 4 in support of the flyby
. we centroided by fitting a two - dimensional gaussian to the apparent photocenter of each image .
as discussed in paper 1 , the jets observed in hartley 2 can pull the centroid away from the nucleus .
however , the central condensation generally dominated over any coma features and we estimate that our centroids are accurate to better than 1.5 arcsec .
we investigated various enhancement techniques ( e.g. , @xcite ) but utilized division by an azimuthal median profile for all images shown in this paper .
an observational set for photometry consisted of five gas filters ( oh , nh , cn , c@xmath2 , and c@xmath1 ) along with two ( 3650 and 4845 ) or three ( 3448 , 4450 , and 5260 ) continuum filters associated with the ihw or hb filter sets , respectively .
three observational sets were obtained in 1991 , four sets in 1997/98 , and 55 sets in 2010/11 in support of , with multiple sets often obtained on a single night .
photometer entrance aperture diameters ranged between 24 and 156 arcsec while projected aperture radii ranged from as small as 1950 km near perigee in 2010 october to as large as 47,900 km , with a median radius of 10,500 km .
our standard procedures were used in data acquisition and basic reductions ( see @xcite ) , with appropriate changes associated with the newer hb filters ( cf .
@xcite ) ; and improved decontamination for the ihw filters ( cf .
gas fluorescence efficiencies that vary with heliocentric velocity are listed in table [ t : phot_circ ] ; sources are given in @xcite . to extrapolate the resulting column abundances to total coma abundances ,
a standard haser model was applied , using the scalelengths given in @xcite , and then gas production rates ( @xmath6 ) were computed by dividing the total abundances by the assumed daughter lifetimes .
finally , water production rates were computed from the oh results using our empirical conversion ( @xcite ; also see @xcite ) . to quantify the dust
, we continue to use the quantity and , as appropriate , we have now applied a phase correction to the phase angle , @xmath7 , listed in table [ t : phot_circ ] ( see @xcite ) to obtain @xmath8(0)@xmath9 .
one - sigma uncertainties based on the photon statistics associated with each data point were computed for the resulting @xmath6s and s. these were generally fairly small except for early and late in the 2010/11 apparition when we were attempting to extend the temporal coverage as much as possible and hartley 2 was quite faint .
we first present our photometric results for the three apparitions for which we have data . in table
[ t : phot_flux ] we list emission band and continuum fluxes ( as logarithms ) along with the resulting gas column abundances , m(@xmath10 ) , while the derived gas production rates and values for each continuum point are given in table [ t : phot_rates ] as well as the vectorial - equivalent water production rate in the rightmost column . note that although the 1-@xmath11 uncertainties are unbalanced in log - space , we only list the `` + '' log value for clarity and to save space ; the `` @xmath12 '' values can be computed knowing that the percent uncertainties are balanced .
the derived production rates for each gas species and for the green continuum are plotted as logarithms in figure [ fig : phot1 ] as a function of time from perihelion .
the two most obvious characteristics are that the earlier apparitions generally have higher production rates than the most recent apparition and that all species exhibit a bulk asymmetry about perihelion , with values higher after but by varying amounts . as is usual for jupiter - family comets ( cf .
@xcite ) , the heliocentric distance dependancies for the gas species ( log @xmath6 vs log ) are significantly steeper than a canonical @xmath13 ( or @xmath14 if adjusting for gas outflow velocities ) . based solely on the 2010/11 apparition , the slopes for the carbon - bearing species range from @xmath123.4 to @xmath124.1 before perihelion and from @xmath123.2 to @xmath123.5 after , while oh is @xmath124.6 prior to and @xmath124.0 after , and nh has an uncertain value of @xmath127.1 before and @xmath124.9 after perihelion ( see table [ t : table5 ] ) .
the -dependancy for the dust , @xmath15 , before perihelion is quite similar to those exhibited by the carbon - bearing species , especially if one normalizes to 0 phase angle . specifically , the three continuum points have values for the slopes from @xmath122.8 to @xmath123.6 ( self - consistent given the large uncertainties ) when using while somewhat higher values from @xmath123.4 to @xmath124.1 using @xmath8(0)@xmath16@xmath10 .
although the adjustment for phase angle has a larger effect after perihelion going from @xmath121.0 to @xmath121.2 without adjustment to @xmath121.7 to @xmath121.9 when normalized as seen in other comets , including 19p / borrelly @xcite and 67p / churyumov - gerasimenko @xcite , the -dependancies for dust are much shallower than that of the gas species following perihelion .
as we suggested in borrelly s case , the very shallow slope after perihelion is likely due to the release of larger , very slow moving grains near perihelion and peak water production , and these grains remain in the inner coma far longer than the more typical micron - sized dust particles . note that @xmath15 also exhibits the usual trend with aperture size , where larger apertures yield smaller values , indicating that the dust spatial distribution with projected distance from the nucleus is steeper than the canonical 1/@xmath10 . since our projected aperture sizes were generally larger when the comet was at larger heliocentric distances , the -dependences just - discussed would have been even shallower had an adjustment for aperture size been performed , thereby magnifying the post - perihelion effect .
returning to the earlier apparitions of the 1990s , we first note that all observational sets from 1997/98 are higher than corresponding data from 2010 except for the sole continuum point after perihelion that can be explained by a combination of aperture and phase angle effects .
the measured offset between these two apparitions ranges between about a factor of 1.5 and 2.0 among all species , far more than can be explained by the change in solar flux caused by the increase in perihelion distance from 1.032 au in 1997 to 1.059 au in 2010 . while our data are even sparser in 1991 , with only three data sets on two nights , the measured offset is much greater , corresponding to a factor of 34 as compared to 2010 .
again , this is far greater than expected due to solar illumination
perihelion was 0.953 au in 1991 , resulting in only a 25% difference . to examine this further
, we next intercompare a variety of water measurements from all three apparitions . as indicated in section [ sec : phot_obs ] , we can compute vectorial - equivalent water production rates from our haser - model oh production rates , and we plot these in figure [ fig : phot2 ] .
we also overlay all other published water production values , including when available the uncertainties .
we were surprised at the large amount of dispersion , even when only examining one apparition at a time .
while some of this dispersion is likely due to differences in modeling parameters , some due to species - related issues ( data are from direct water measurements in the ir , oh in the near - uv and radio , and h in the far - uv ) , and some due to varying dilutions by older material associated with an extremely large range of effective aperture sizes for the measurements , there remains a great deal of variability within some individual data sets , especially as compared to the apparent smooth curve with time evident in our own data .
looking in detail first at the 2010/11 apparition , the variability seen in the most comprehensive data set from _
soho_/swan @xcite is sometimes very small ( @xmath17 t = @xmath1227 to @xmath1210 day ) but other times more than a factor of two in only a few days . and
while it first appears that our data are systematically higher than that from _ soho_/swan , a closer examination reveals excellent agreement when data were obtained near - concurrently , such as at @xmath17 t of @xmath1228/@xmath1227 day , + 19 day , and + 46 day ( where the _
soho_/swan point is nearly invisible due to the overlap ) .
other data sets , such as keck @xcite , also suggest considerable short - term variability during some intervals in the apparition .
based on the ensemble of data , we conclude that our own smooth curve during 2010/11 was partly caused by having only one successful night of photometry on the majority of our observing runs , due to a combination of relatively poor weather and our focus towards obtaining imaging data ( section 4 ) .
another factor is simple random chance ; for instance our data sets from nov 16 ( + 19 days ) and dec 13 ( + 46 days ) are in complete agreement with other data on those nights .
however , in neither case are these data similar in value to data obtained during the prior @xmath010 days . turning next to 1997/98 , _
soho_/swan @xcite again provides the most complete record , with one interval after perihelion showing a long steady decline while at other times much larger variability is evident . also , in 1997 , _
soho_/swan measurements are systematically higher than ours , opposite of the case in 2010 , but our two pre - perihelion data points each occur at minima of the _ soho_/swan data set .
when looked at together , it becomes clear that there was a much greater drop in production rates between 1997 and 2010 , by about a factor of 3 , than our data alone suggested .
by extension , this would also indicate a smaller change took place between 1991 and 1997 , and this is consistent with the only other published data point from 1991 ( _ hst _ ; @xcite ) , along with an unpublished _ iue _ measurement taken within the week that was stated by weaver et al . to be similar in value . allowing for the change in perihelion distances by comparing the 1991 data with an extrapolated heliocentric distance trend post - perihelion , our 1991 data are about 50% higher while the _ hst _ ( and _ iue _ ) points are nearly 90% higher than 1997 .
while the sparseness of the 1991 data and the large variability that is expected to be present make this result somewhat uncertain , the data are suggestive that water production rates were @xmath070% greater in 1991 than in 1997 . from these water data ,
we conclude that there was a large ( @xmath03@xmath4 ) decrease in water vaporization between 1997 and 2010 ( an interval of two orbits ) , and a smaller ( @xmath01.7@xmath4 ) decrease between 1991 and 1997 ( a single orbit ) , which implies a relatively consistent drop of about 40% from one apparition to the next .
there is also considerable evidence for a stochastic amount of variability in water production as a function of time .
there were intervals when variability was small , smooth , and exhibited trends consistent with the change in distance from the sun , and other times when variability was far larger and apparently sporadic .
another striking comparison is the opposite trends evident in the soho data sets in the 3 - 4 weeks immediately after perihelion in 1997 versus 2010 .
given the evidence from the mission that there were several isolated source regions on the nucleus , coupled with evidence for a complex rotational state and a changing rotational period of at least one of the rotational components , we conclude that at least one source region experiences a very complicated solar illumination from a combination of these rotational effects .
we further suggest that the intervals of smooth trends interrupted by intervals of much larger variability may be directly associated with the beat between the two periods , and how this beat changed with the changing period(s ) .
the absolute values of the water production can also yield a nominal surface area of vaporization .
our values in 2010 near perihelion and the encounter imply an active area of 34 km@xmath18 , just below the measured nucleus surface area of @xmath04.2 km@xmath18 ( based on a mean radius of [email protected] km from @xcite ) .
note that far more surface area than exists on the nucleus would have been required to produce the amount of water detected in the 1990s .
the significance of this will be returned to in section [ sec : disc ] .
the composition of hartley 2 , based on the abundance ratios of the minor gas species computed using the ratios of their respective production rates with respect to oh , are given in table 5 .
here we also provide our standard uncertainties , the @xmath11 of the data , which we have tabulated in our previous single comet papers , along with the @xmath11 of the mean .
the former describes the dispersion of the data , due to observational errors , modeling effects , and intrinsic variations in the comet itself , while the latter better describes how well each mean ratio is determined .
hartley 2 s composition clearly places it in the middle of the `` typical '' classification from @xcite , consistent with our result from the 1991 apparition that was included in ahearn et al . using this same classification scheme , @xcite find the same result from their spectroscopic observations .
the comet is also classified as `` typical '' by @xcite using his own definition of compositional classes . determining a single mean value for
the dust - to - gas ratio is more problematic for several reasons , including the trend in @xmath15 values with aperture size , the much shallower -dependence following perihelion , and phase angle effects .
since the first two issues are incorporated in our calculation of the uncertainties of the data for the dust - to - gas ratio , we make no further adjustments . regarding phase angle effects
, we compute both an unadjusted value , i.e. /@xmath6(oh ) , for direct comparison to other comets in @xcite , along with a normalized value to 0 phase angle . our results for the green continuum were @xmath1225.84 and @xmath1225.44 , respectively , corresponding to a relatively low dust - to - gas ratio , especially as compared to other jupiter - family comets .
finally , based on the three continuum points we measure , the dust is nearly grey in color , with reddening of less than 10% per 1000 through most of the apparition .
however , at the end of october , when our projected apertures were smallest , the color in the uv was as high as 30% per 1000 .
this result is consistent with that by @xcite who detect a stronger reddening in the inner - most coma , but note that the reddening decreases with distance and also varies with direction in the coma and that the dust is blue in some locations .
paper 1 focused on hartley 2 s cn coma morphology .
we summarize the relevant results here to place the morphology observed in oh , nh , c@xmath2 , c@xmath1 , and dust into context .
we observed two cn features , generally towards the north and south . in august and
september only the northern feature was seen , and had the appearance of a nearly face - on spiral . in october ,
november , and december the morphology was roughly an hourglass shape , with outward motion visible and the relative brightnesses of the northern and southern features varying . in january the signal - to - noise was poor and only the southern feature was detected .
the cn morphology varied smoothly during a night and similar , but not identical , morphology was seen on subsequent nights .
the similar morphology allowed us to estimate a rotation period during the august to november runs , and the period increased by @xmath02 hr ( from @xmath016.7 hr to @xmath018.7 hr ) during this time .
the differences in morphology implied the existence of small deviations from a principal axis rotation .
the morphology repeated much better 3 , 6 , 9 ... cycles apart than it did 1 , 2 , 4 , 5 ... cycles apart , implying a component rotation period @xmath03@xmath4 the `` primary '' rotation period .
similar morphology was reported by @xcite , @xcite , and @xcite , and the changing rotation period and non - principal axis rotation were confirmed by other observers ( cf .
@xcite ) .
cn is generally the gas species of choice for investigations of gas coma features with narrowband filters because it is vastly brighter than oh or nh and has a better contrast relative to the dust than c@xmath1 , c@xmath2 , or nh . cn jets were first detected in the coma of 1p / halley @xcite and jets and/or fans have now been detected in numerous other comets , e.g. , c/1995 o1 hale - bopp @xcite , c/2004 q2 machholz @xcite , 8p / tuttle @xcite , and c/2007 n3 lulin @xcite .
the coma morphology of the other gas species has only been published for a few very bright comets , e.g. , 1p / halley @xcite and c/1995 o1 hale - bopp @xcite .
hartley 2 s very small geocentric distance during the 2010/11 apparition allowed us to observe it with unusually high spatial resolution , making a multiwavelength study of coma morphology highly desirable .
as listed in table [ t : imaging_circ ] , we obtained c@xmath1 and c@xmath2 images on photometric nights from august through november , oh images on photometric nights from september through november , and nh images on photometric nights in october and november ( although the lone october nh image is unusable due to a tracking error ) .
the comet was easily detectable in all of these raw images and after removal of dust contamination .
however , sufficient signal - to - noise to allow meaningful analysis of the enhanced images was only possible in october and november for species other than cn .
we show cn , c@xmath2 , c@xmath1 , oh , nh , and dust on november 2 and 3 in figure [ fig : gas_morph1 ] ; comparable data were obtained on october 16 and november 7 but are not shown .
as shown in figure [ fig : gas_morph1 ] , the c@xmath2 and c@xmath1 morphology looked generally similar to the cn morphology . on november 2
, all three species exhibited an hourglass - like shape , with a stronger feature to the north and a fainter feature to the south ; the southern feature can be seen leaving the nucleus towards the southwest . on november 3 ,
the cn , c@xmath2 , and c@xmath1 showed a somewhat different hourglass shape , with the the southern feature much brighter relative to the northern feature than on november 2 .
all three species were much fainter along the sun - tail line ( roughly east - west ) than along the axis of the hourglass feature .
the cn signal was typically lowest in the tailward direction , whereas the c@xmath2 and c@xmath1 signals were typically slightly fainter in the sunward direction .
the overall bulk brightness of c@xmath2 and c@xmath1 varied in correlation with the bulk cn brightness , e.g. , generally brighter to the north on november 2 and generally brighter to the south on november 3 . in order to quantify the differences between the gas species ,
we show their spatial profiles in figure [ fig : profiles ] .
we plot the flux ( as measured on decontaminated , but not enhanced , images ) on november 2 as a function of distance from the nucleus in 10@xmath19 wide wedges centered along position angles ( pas ) of : 10@xmath19 ( perpendicular to the sunward direction , to the north ) , 100@xmath19 ( the sunward direction ) , 190@xmath19 ( perpendicular to the sunward direction , to the south ) , and 280@xmath19 ( the tailward direction ) .
these profiles extend to the edges of the images shown in figure [ fig : gas_morph1 ] .
the inner @xmath0800 km are not plotted as there are very few pixels in each wedge at these distances , causing small fluctuations to be exaggerated .
the northern and southern cn profiles were @xmath020% and @xmath010% brighter than the sunward profile , respectively .
the cn tailward profile was roughly the same brightness as the sunward profile .
the c@xmath2 and c@xmath1 profiles were both brightest to the north , followed by the south , then tailward , and finally sunward . the north , south , and tailward profiles were 1030% brighter than the sunward profile .
note that these brightnesses are along the line of sight and are products of projection effects , each species parentage , and excess velocities acquired .
thus , brightness enhancements in a particular direction relative to the sunward direction ( which did not have any obvious jets ) are likely only lower limits , and the actual ratio of the material originating from the nucleus in each direction may be substantially higher .
dividing the c@xmath1 and c@xmath2 images by the cn images revealed minimal differences in their morphologies , with the primary feature ( the hourglass shape ) disappearing in quotient images .
as is clearly evident in figure [ fig : profiles ] , the c@xmath2 does not extend as far as the other gas species .
this was partially due to the higher signal - to - noise in cn and c@xmath1 as compared to c@xmath2 , however , it was mostly due to the differing lifetimes of the species and their respective parentages ; c@xmath2 and its parents have much shorter lifetimes than either c@xmath1 or cn and their respective parents ( cf .
c@xmath1 appeared quite similar to cn but was more diffuse because it has multiple parents and grandparents , whereas cn is primarily a daughter species .
the multiple parentages of c@xmath1 cause its radial profile to be flatter than the profiles of either cn or c@xmath2 .
owing to the shorter observing window when the comet was brightest and the desire to acquire images in a large number of filters , we usually only acquired one or two c@xmath1 or c@xmath2 images per night . thus , we can not investigate how the morphology of these species compares as a function of rotation period .
however , on every night in which we saw cn , c@xmath2 , and c@xmath1 with sufficient signal - to - noise to discern varitions in the coma , the morphologies appeared to be completely consistent except for the parentage and lifetime effects discussed above .
this leads us to conclude that the c@xmath2 and c@xmath1 species exhibit the same rotational and seasonal morphology as cn and therefore originate from the same source region(s ) on the surface .
near the nucleus , the nh displayed some of the hourglass morphology of cn , c@xmath2 , and c@xmath1 .
the features appeared to originate from separate northern and southern sources , although the distinction between the features in the tailward direction was less obvious than for cn , c@xmath2 , or c@xmath1 .
the relative brightness of the northern and southern features varied in correlation with changes in the brightness of these features in cn .
nh had a strong asymmetry in brightness of the sunward and tailward hemispheres , with the tailward hemisphere always being substantially brighter .
the bulk brightness of the nh images displayed little variation from night to night .
the radial profiles demonstrate the extent of the tailward brightness enhancement , as the north , west , and south profiles all remained at least 20% higher than the eastern ( sunward ) profile to the edge of the images displayed in figure [ fig : gas_morph1 ] . as with c@xmath1 ,
the nh radial profiles were rather flat due to its being primarily a granddaughter species .
we used the photometry to determine the amount of reddening to apply in order to properly remove the underlying continuum from the nh .
however , it is likely that the continuum was somewhat over - removed in the innermost coma ( @xmath203000 km ) because the dust was increasingly reddened at progressively smaller distances from the nucleus , resulting in the unusually flat profile .
the morphology most different from cn was oh . the oh morphology near the nucleus showed only hints of the hourglass shape .
unlike the other gas species , the feature did not have distinct northern and southern components , but was essentially continuous from the north through the west ( tailward ) to the south .
the relative brightnesses of the north and south features of the hourglass vary much less than the other gas species . like
nh , the tailward hemisphere was much brighter than the sunward hemisphere .
furthermore , at larger distances the oh was brighter toward the north on all four nights in november , and the bulk brightness did not vary significantly from night to night .
the tailward profile was @xmath035% higher than the sunward profile , while the northern and southern profiles were 2025% higher than the sunward profile .
the striking differences between the coma morphologies of oh and cn was seen on every night in which we have usable oh data , as shown in figure [ fig : gas_morph2 ] . in each of these images
the oh signal near the nucleus showed hints of the hourglass shape seen in cn ( although the faint cn feature near the nucleus and towards the southeast on november 4 is absent in oh ) .
the tailward oh hemisphere was always much brighter than the sunward hemisphere , and the dust tail ( bc ) almost perfectly bisected the oh distribution in all images , including october 16 when the geometry was substantially different .
the hemispheric brightness asymmetry and the lack of variation in the bulk brightness from night to night suggest that there is a smearing across rotational phase in the oh signal , and that the changing viewing geometry was primarily responsible for the changes in oh morphology between october and november . as with c@xmath1 and c@xmath2
, we typically only acquired one or two sets of oh and nh on a given night so we can not determine the morphology throughout an entire rotation cycle .
however , during each night in which we observed oh and nh , their bulk coma morphologies looked similar concentrated in the tailward hemisphere and distinctly different than cn , c@xmath1 , or c@xmath2 .
the consequences of these determinations will be returned to in section [ sec : disc ] .
the dust morphology of hartley 2 was dominated by the tail throughout the apparition .
figure [ fig : r_morph ] shows enhanced r - band images monthly from 2010 august through 2011 january .
this same shape was seen , albeit with lower signal - to - noise , in the narrowband continuum filters : uv , blue , green , and red . in all cases
the dominant feature was roughly straight and in the anti - sunward direction ( the pa of the sun is labeled on each panel ) .
it does not vary with rotational phase and is consistent with the expected position of the dust tail , confirming that it is the dust tail .
overall , the brightness of the dust systematically decreased as a function of pa from the tailward direction to the sunward direction .
radial profiles of the four fundamental directions are shown in figure [ fig : profiles ] .
we detected a very faint sunward facing dust jet near the nucleus , as also reported by @xcite and @xcite . while difficult to see in figures [ fig : gas_morph1 ] and [ fig : gas_morph2 ] , it can be seen in the bc profile in figure [ fig : profiles ] , as the sunward profile is slightly higher than the northern or southern profiles out to @xmath01600 km .
the jet was nearly radial and showed minimal change in shape , position angle , or extent during the course of a night , but varied somewhat from night to night .
we show examples of the jet near the start and end of the night in enhanced continuum images on 2010 november 2 , 3 , 4 , and 7 in figure [ fig : dust_jet ] .
the jet s appearance was relatively similar on november 2 and 7 , pointing nearly due east at pas of @xmath095@xmath19 and @xmath0100@xmath19 , respectively .
it had a slightly different appearance on november 3 and 4 , when it was to the southeast at pas of @xmath0115@xmath19 and @xmath0125@xmath19 , and extended farther than on november 2 or 7 . at its maximum extent ( on november 3 ) , the dust jet was visible to 2025 arcsec from the nucleus , although signal - to - noise and our enhancement techniques may have prevented us from seeing it extend farther .
hints of the dust jet were apparent in the october 16 and 17 continuum images , but the extent was much smaller ( @xmath2010 arcsec ) than in early november despite better signal - to - noise .
the lack of a firm detection in october may have been because the jet was inactive or it may simply have been due to projection effects .
the dust jet was not detectable in our data from other months , presumably due to the significantly lower signal - to - noise and larger geocentric distance as compared to 2010 october and november . at no time
did we observe a corresponding gas feature .
surprisingly , the dust jet was easier to detect in our narrowband dust images blue continuum ( bc ) , green continuum ( gc ) , and red continuum ( rc ) than in the broadband r images .
we applied several additional enhancement techniques to the r - band images in an effort to study the dust jet with more temporal resolution than the continuum images offered .
however , we could not consistently detect the dust jet in consecutive r - band images despite the r - band images having a higher signal - to - noise than the continuum images .
all of our hartley 2 observations were focused for the hb narrowband filters , resulting in a typical r - band fwhm of @xmath06 arcsec .
given the faintness of the dust jet , this r - band defocusing appears to have been sufficient to obscure the dust jet .
since , as was the case for c@xmath2 , c@xmath1 , oh , and nh , we only obtained narrowband dust images occasionally , we do not have full rotational coverage .
thus we can not determine how the dust jet varied with rotation , but our limited data exhibit a slower rate of change than was exhibited by the gas jet(s ) .
the coma morphologies varied across the five gas species we observed . at one extreme was cn whose morphology was dominated by an hourglass shaped feature . at the opposite extreme was oh , whose morphology appeared to be an amalgamation of the cn hourglass feature and the tail - dominated dust . in between these extremes were c@xmath2 and c@xmath1 , whose morphologies were generally quite similar to the cn , and the nh , which appeared to be intermediate to the cn and the oh . the presence of the hourglass shape exhibited clearly in cn , c@xmath2 , and c@xmath1 , and to a lesser extent by nh and oh suggests that all five species originate from the same source region(s ) on the nucleus .
this is supported by the observation that the relative brightness of the northern and southern features in cn , c@xmath2 , c@xmath1 , and nh vary in concert , and also that the cn , c@xmath2 , and c@xmath1 morphologies are in agreement from night to night .
the relative brightness of the oh hourglass features did not vary as much as the other gas species , and oh also exhibited more near - nucleus coma signal in the tailward direction than the other species .
the hemispheric asymmetry of oh and nh as well as the relative uniformity of their bulk brightness ( after enhancement ) suggest that a substantial fraction of the oh and nh is derived from small icy grains which survived long enough to be subject to radiation pressure and swept tailward .
the infilling of oh and nh throughout the tailward hemisphere is likely due to the velocities in random directions acquired by the parent molecules as they leave the grains .
observed numerous individual chunks up to @xmath020 cm in radius within 30 km of the nucleus @xcite and @xcite reported a significant population of large ( @xmath21 cm ) dust grains near the nucleus .
_ suggest that these large grains break up into @xmath01 @xmath22 m solid grains , with the smaller grains providing the surface area necessary to explain hartley 2 s `` hyperactivity . ''
the bulk brightness enhancement of the oh and nh in the tailward hemisphere supports this model , with the micron - sized grains containing the water and ammonia that eventually produced the oh and nh . revealed distinctly different terrain on the nucleus , with a smooth `` waist '' connecting two rougher lobes @xcite .
ahearn _ et al_. determined that different material was coming from these two regions , with h@xmath1o vapor coming primarily from the waist and co@xmath1 , h@xmath1o ice , and organics coming primarily from an end .
thus , it is likely that the cn , c@xmath2 , c@xmath1 , oh , and nh originate near an end of a lobe , presumably from one or more of the many active regions observed by .
another very strong constraint is the observation from our photometry that the ratios of the minor species to water are normal ; therefore , the vast majority of all of our observed species must be originating at the same time and location .
the differences in the bulk morphology between the nh and oh and the cn , c@xmath2 , and c@xmath1 can be explained if the icy grains containing oh and nh are separated from the parents / grandparents of cn , c@xmath2 , and c@xmath1 ( either as smaller grains or simply as vapor ) soon after being released from the nucleus .
such a scenario suggests that there may have been an intrinsic difference in the protosolar grains out of which hartley 2 formed , with the parents of oh and nh ( primarily water and ammonia , respectively ) preferentially being contained in larger grains ( and possibly intermixed as `` dirty '' ice ) while the parents of cn , c@xmath1 , and c@xmath2 were contained in smaller grains or were deposited primarily on the surface of larger grains .
it is somewhat troubling that such a difference in grain compositions has never been detected in other comets ( especially in light of hartley 2 s `` typical '' composition ) , but given the relative uniqueness of hartley 2 s `` hyperactivity , '' it is not surprising that it may have exhibited other unusual properties ; note that no other comet showing hyperactivity has been visited by a spacecraft .
the mechanism for lifting large grains , suggested to be co@xmath1 @xcite , may be vigorous enough to lift off large chunks of the surface _ in toto _
, likely resulting in a considerably different removal of material than the canonical model of cometary activity which envisions gas leaving through pores in the surface and entraining small bits of dust in the process .
conversely , very few comets have been imaged in nh and/or oh owing to their low signal and high atmospheric extinction .
thus , the different coma morphologies of oh and nh compared with the carbon bearing species may not , in fact , be unusual . as discussed in section [ sec : dust_morphology ] , the dust jet showed little variation during a night but changed somewhat from night to night , always being in the same general direction .
these two points are consistent with the apparent non - principal axis rotation of the nucleus ( cf .
@xcite ; paper 1 ) .
the first point suggests that the source of the jet might be located near the comet s total angular momentum vector , the axis around which the nucleus was apparently `` precessing '' with a period near 18.3 hr at the time ( cf .
@xcite ) , resulting in little change in the morphology over a night .
the second point could be caused by hartley 2 s longer `` rotation '' period of @xmath055 hr at the time of the observations .
this would not have much effect on the jet s appearance over the course of one night ( only @xmath010% of a 55 hr period ) , but would cause changes from night to night ( when the rotational phase had changed by roughly 45% ) .
the lack of a gas feature corresponding to the dust jet implies that the dust is lifted from the surface by some other volatile .
the most likely candidates are co and , more likely , co@xmath1 , which @xcite concluded drives activity on hartley 2 .
presumably this source has little to no cn , c@xmath2 , c@xmath1 , oh , or nh since no corresponding gas jet was seen ; their absence implies that the source has a different composition than the source(s ) of the hourglass features and suggests some heterogeneity of the hartley 2 nucleus .
we do not see evidence of radiation pressure affecting the dust jet , which implies we are primarily seeing large dust grains .
the velocity dispersion of large grains would be expected to mask much of the rotational signature of the dust jets .
since we do not see any evidence of dust jets near the locations of the hourglass features in the gas , we infer that the gas jet(s ) do not have substantial quantities of large grains .
instead , the jet(s ) likely lifts small grains which are rapidly pushed tailward ; a similar population of small dust grains subject to radiation pressure is likely present in the dust jet and goes similarly undetected .
we next turn to the observation ( section [ sec : composition ] ) that the production rates have trended downwards steadily from 1991 to 1997/98 to 2010/11 even after accounting for the increase in perihelion distance between apparitions .
we speculate that this may be due to the relative youth of the nucleus . in the early twentieth century , hartley 2 had a perihelion distance ( @xmath23 ) near 2 au and an orbital period near 9 yr .
close approaches to jupiter in 1947 and 1971 caused a drop in the perihelion distance , putting hartley 2 in an orbit reaching @xmath01 au .
if the primary driver of activity was something more volatile than h@xmath1o such as co@xmath1 as suggested by @xcite ( co is less likely due to the extremely low abundance ; @xcite ) , then the injection into a smaller-@xmath23 orbit may be rapidly depleting the co@xmath1 . since vigorous activity
is required to lift off large chunks of particles necessary to produce the `` hyperactivity '' , a rapid decrease in the production of co@xmath1 would cause a correspondingly large drop in the production rate of h@xmath1o .
the factor of 1.53 drop in the production rates between 1997/98 and 2010/11 spanned two perihelion passages ( we are not aware of any published production rate measurements from the 2004 apparition ) , allowing additional time for production from such areas to decrease .
as there were no systematic surveys that would have been sensitive to this intrinsically faint comet on earlier , less favorable , apparitions , we can not place meaningful constraints on the production rates prior to its discovery in 1986 .
if a similar rate of decrease in production rate occurred between 1971 and 1991 , hartley 2 may have been as much as a factor of 810 more active in 1971 ( when it was first perturbed into the current orbit ) than in 2010/11 .
an alternative , but less likely , explanation for the decreased production rates since 1997/98 may simply be that the illumination of active regions was different between the apparitions .
hartley 2 exhibited a rapid spin - down coupled with non - principal axis rotation during the 2010/11 apparition ( cf .
it is possible that on previous orbits the interplay of the complex rotation and the apparently increasing and decreasing component periods ( cf .
the supporting online materials from @xcite ) could result in a different illumination of one or more active regions .
this might include the sun reaching a higher altitude , variations in topography causing less shadowing , the exposure of additional active regions , or the sun remaining above an active region s local horizon longer .
however , the nucleus complex rotation state should minimize seasonal effects ; unless the nucleus was very recently ( since 1997/98 ) excited into the present non - principal axis rotation state , large changes in the production rates due entirely to geometry are considered unlikely .
we have presented photometry and imaging of 103p / hartley 2 obtained at lowell observatory ( and one night at perth observatory ) from 1991 through 2011 .
the photometry includes three apparitions as no data were obtained on the 2004 apparition .
we find a secular decrease in brightness from the 1991 apparition to the 1997/98 apparition and then to the 2010/11 apparition and a signficant seasonal effect , with the comet reaching peak brightness @xmath010 days after perihelion and the rate of brightening steeper than the rate of fading .
we find similar results when compiling published water production rates from the literature , consistent with about a 40% decrease each orbit .
hartley 2 s composition is `` typical , '' in agreement with the results of other investigators .
we propose two scenarios for the relatively large decrease in production rates : that production rates are dropping rapidly owing to the rapid depletion of co@xmath1 ( due to the decrease in perihelion distance in the mid - twentieth century ) or , less likely , that the `` complex '' nucleus rotation resulted in progressively less illumination reaching the primary active regions from 1991 to 1997/98 to 2010/11 .
our imaging covered the 2010/11 apparition and focused on the gas coma morphology .
we previously reported on our extensive cn data set ( paper 1 ) and analyze here the morphology exhibited by other gas species ( oh , nh , c@xmath2 , and c@xmath1 ) as well as the dust .
we find that c@xmath2 and c@xmath1 exhibit coma morphology generally similar to what we previously reported for cn , with an hourglass shape in october and november that followed the rotational changes exhibited by cn , and little excess signal in the tailward direction .
we conclude that differences between the cn , c@xmath2 , and c@xmath1 coma morphologies can be explained by their different lifetimes and parentages .
the oh and nh coma morphologies differ from cn , c@xmath2 , and c@xmath1 ; while oh and nh show evidence of the hourglass shape near the nucleus , they are relatively uniform in brightness ( after enhancement ) throughout the tailward hemisphere , and the brightness in the tailward hemisphere does not vary appreciably with rotation . we conclude that oh and nh are produced from water and ammonia ices that were contained in small grains which shielded the ices long enough to be affected by radiation pressure and driven in the anti - sunward direction .
we speculate on possible explanations for why the oh and nh , but not the cn , c@xmath2 , or c@xmath1 , behave in this manner , concluding that all five species most likely originate from the same source region(s ) near the end of a lobe of the nucleus ( where saw significant active regions ) but may have come from aggregates of different sized grains or grains and vapor .
lower velocities and/or a range of velocities associated with a variety of particle sizes would also naturally cause a dilution of the rotational signature of oh and nh as compared to that seen in the carbon bearing species .
we detect the faint sunward - facing dust jet reported by @xcite and @xcite in our continuum images .
this jet is much smaller than the hourglass shaped gas feature and does not vary appreciably during a night , although it does vary from night to night .
no corresponding gas feature was seen at the location of the dust jet , possibly implying that it is driven by co@xmath1 .
the dust jet may originate from a source near the total angular momentum vector .
the flyby and supporting observations of hartley 2 have revealed that it is an unusual comet which is highly active relative to its nucleus size , has a population of large grains in the inner coma , and is in non - principal axis rotation with evolving component periods .
as one of a small handful of comets to have been visited by a spacecraft , hartley 2 represents a rare opportunity to link the macroscopic coma morphology and abundances observed remotely with the nucleus shape and active regions observed _ in situ _ via sophisticated modeling .
the strongest constraints on such models are likely the direction of the nucleus long axis as seen by during the flyby ( cf .
@xcite ) and the alignment of the rotational angular momentum at the time of the arecibo observations @xcite .
however , any comprehensive jet modeling should also incorporate the constraints discussed here and in paper 1 .
these include the morphological changes during a night ( outward motion of the material and the sense of rotation ) , differences in morphology from night to night ( due to the non - principal axis rotation ) , and evolution of the morphology from month to month ( due to the changing viewing geometry ) .
it is only by satisfying these varied constraints that a model can produce a coherent description of comet hartley 2 .
we appreciate the work of anita cochran and nalin samarasinha in carefully refereeing this manuscript and making helpful suggestions to improve the paper .
we are indebted to the following people for help in collecting data : robert millis ( 1991 ) ; tony farnham ( 1997/1998 lowell data ) ; peter birch ( 1997 perth data ) ; len bright , brian skiff , larry wasserman , and edward schwieterman ( 2010/2011 imaging data ) .
we thank allison bair for help in producing table [ t : table5 ] and for comparisons with preliminary results of the forthcoming comet database paper .
office space was generously provided for mmk by both the university of maryland department of astronomy and johns hopkins university applied physics laboratory while he conducted this work .
this work has been supported by numerous nasa grants over the years , most recently nnx08ao42 g and nnx09ab51 g . , m. f. , m. j. s. belton , w. a. delamere , l. m. feaga , d. hampton , j. kissel , k. p. klaasen , l. a. mcfadden , k. j. meech , h. j. melosh , p. h. schultz , j. m. sunshine , p. c. thomas , j. veverka , d. d. wellnitz , d. k. yeomans , s. besse , d. bodewits , t. j. bowling , b. t. carcich , s. m. collins , t. l. farnham , o. groussin , b. hermalyn , m. s. kelley , m. s. kelley , j .- y .
li , d. j. lindler , c. m. lisse , s. a. mclaughlin , f. merlin , s. protopapa , j. e. richardson , and j. l. williams 2011 . .
* 332 * , 13961400 . , k. j. , m. f. ahearn , j. a. adams , p. bacci , j. bai , l. barrera , m. battelino , j. m. bauer , e. becklin , b. bhatt , n. biver , d. bockele - morvan , d. bodewits , h. bhnhardt , j. boissier , b. p. bonev , w. borghini , j. r. brucato , e. bryssinck , m. w. buie , h. canovas , d. castellano , s. b. charnley , w. p. chen , p. chiang , y .- j .
choi , d. j. christian , y .- l .
chuang , a. l. cochran , p. colom , m. r. combi , i. m. coulson , j. crovisier , n. dello russo , k. dennerl , k. dewahl , m. a. disanti , m. facchini , t. l. farnham , y. fernndez , h. g. florn , u. frisk , t. fujiyoshi , r. furusho , t. fuse , g. galli , d. a. garca - hernndez , a. gersch , z. getu , e. l. gibb , m. gillon , e. guido , r. a. guillermo , e. hadamcik , o. hainaut , h. b. hammel , d. e. harker , j. k. harmon , w. m. harris , p. hartogh , m. hashimoto , b. husler , t. herter , a. hjalmarson , s. t. holland , m. honda , s. hosseini , e. s. howell , n. howes , h. h. hsieh , h .- y .
hsiao , d. hutsemkers , s. m. immler , w. m. jackson , s. v. jeffers , e. jehin , t. j. jones , m. d. j. ovelar , h. m. kaluna , t. karlsson , h. kawakita , j. v. keane , l. d. keller , m. s. kelley , d. kinoshita , n. n. kiselev , j. kleyna , m. m. knight , h. kobayashi , h. a. kobulnicky , l. kolokolova , m. kreiny , y .- j . kuan , m. kppers , j. m. lacruz , w. b. landsman , l. m. lara , a. lecacheux , a. c. levasseur - regourd , b. li , j. licandro , r. ligustri , z .- y .
lin , m. lippi , d. c. lis , c. m. lisse , a. j. lovell , s. c. lowry , h. lu , s. lundin , k. magee - sauer , p. magain , j. manfroid , e. mazzotta epifani , a. mckay , m. d. melita , h. mikuz , s. n. milam , g. milani , m. min , r. moreno , b. e. a. mueller , m. j. mumma , m. nicolini , m. c. nolan , h. l. nordh , p. b. nowajewski , odin team , t. ootsubo , l. paganini , c. perrella , j. pittichov , e. prosperi , y. l. radeva , w. t. reach , a. j. remijan , m. rengel , t. e. riesen , m. rodenhuis , d. p. rodrguez , r. w. russell , d. k. sahu , n. h. samarasinha , a. snchez caso , a. sandqvist , g. sarid , m. sato , d. g. schleicher , e. w. schwieterman , a. k. sen , d. shenoy , j .- c .
shi , y. shinnaka , j. skvarc , c. snodgrass , m. l. sitko , s. sonnett , s. sosseini , g. sostero , s. sugita , b. m. swinyard , s. szutowicz , n. takato , p. tanga , p. a. taylor , g .-
tozzi , r. trabatti , j. m. trigo - rodrguez , c. tubiana , m. de val - borro , w. vacca , b. vandenbussche , j. vaubaillion , f. p. velichko , s. f. velichko , r. j. vervack , jr .
, m. j. vidal - nunez , g. l. villanueva , c. vinante , j .- b .
vincent , m. wang , l. h. wasserman , j. watanabe , h. a. weaver , p. r. weissman , s. wolk , d. h. wooden , c. e. woodward , m. yamaguchi , t. yamashita , p. a. yanamandra - fischer , b. yang , j .- s .
yao , d. k. yeomans , t. zenn , h. zhao , and j. e. ziffer 2011 . .
* 734 * , l1 . , l. m. , m. f. ahearn , d. g. schleicher , t. l. farnham , j. p. mcmullin , m. c. h. wright , j. m. veal , l. e. snyder , i. de pater , j. r. forster , p. palmer , y. kuan , w. r. williams , c. c. cheung , and b. r. smith 2002 . . *
157 * , 193204 .
+ 2010 aug 13&03:2811:48&1.1&@xmath1275.9&1.46&0.57&30&22&r , cn , bc , c3,c2,gc&photometric + 2010 aug 14&03:0311:48&1.1&@xmath1274.9&1.45&0.56&30&21&r , cn , bc , c3,c2,gc&photometric + 2010 aug 15&03:1512:10&1.1&@xmath1273.9&1.45&0.55&30&20&r , cn , bc&clouds + 2010 aug 16&07:4011:54&1.1&@xmath1272.8&1.44&0.54&31&19&r , cn , bc&clouds + 2010 aug 17&07:4811:01&1.1&@xmath1271.9&1.43&0.53&31&18&r , cn , bc&clouds + 2010 sep 9&02:5312:16&1.1&@xmath1248.9&1.25&0.33&37&357&r , cn , bc , c3,c2,gc , oh&photometric + 2010 sep 10&02:3612:09&1.1&@xmath1248.0&1.24&0.32&37&357&r , cn , bc , c3,c2,gc , oh&photometric + 2010 sep 11&02:3312:08&1.1&@xmath1247.0&1.24&0.31&37&356&r , cn , bc , c3,c2,gc , oh&photometric + 2010 sep 12&02:3312:11&1.1&@xmath1246.0&1.23&0.31&38&356&r , cn , bc&clouds + 2010 sep 13&02:3512:11&1.1&@xmath1244.9&1.22&0.30&38&355&r , cn , bc , c3&clouds + 2010 oct 12&03:1212:34&0.8&@xmath1215.9&1.08&0.13&49&43&r , cn&photometric + 2010 oct 13&03:1312:43&0.8&@xmath1214.9&1.08&0.13&49&48&r , cn&photometric + 2010 oct 14&03:1312:40&0.8&@xmath1213.9&1.08&0.13&50&52&r , cn&photometric + 2010 oct 15&03:0605:33&0.8&@xmath1213.1&1.07&0.13&50&56&r , cn&clouds + 2010 oct 16&05:0112:21&1.1&@xmath1211.9&1.07&0.12&51&61&r , cn , bc , c3,c2,gc , oh , uc , nh&thin cirrus + 2010 oct 17&05:0012:38&1.1&@xmath1210.9&1.07&0.12&51&65&r , cn , bc&clouds + 2010 oct 19&10:5612:24&1.1&@xmath128.8&1.07&0.12&52&73&r , cn , bc&clouds + 2010 oct 31&07:1012:36&0.8&@xmath243.2&1.06&0.14&58&99&r , cn&thin cirrus + 2010 nov 1&07:1512:45&0.8&@xmath244.2&1.06&0.14&59&100&r , cn&thin cirrus + 2010 nov 2&06:4512:54&1.1&@xmath245.2&1.06&0.15&59&102&r , cn , bc , c3,c2,gc , oh , uc , nh , rc&photometric + 2010 nov 2&07:2710:32&0.8&@xmath245.1&1.06&0.15&59&102&r , cn&photometric + 2010 nov 3&06:4113:01&1.1&@xmath246.2&1.06&0.15&59&103&r , cn , bc , c3,c2,gc , oh , uc , nh , rc&photometric + 2010 nov 4&06:3913:07&1.1&@xmath247.2&1.06&0.16&59&104&r , cn , bc , c3,c2,gc , oh , uc , h20+,rc&thin cirrus + 2010 nov 5&07:4410:45&0.8&@xmath248.1&1.06&0.16&59&105&r , cn&photometric + 2010 nov 6&07:4110:58&0.8&@xmath249.1&1.07&0.16&59&106&r , cn&thin cirrus + 2010 nov 7&06:4813:09&1.1&@xmath2410.2&1.07&0.17&59&107&r , cn , bc , c3,c2,gc , oh , uc , nh , rc&photometric + 2010 nov 10&07:5913:11&0.8&@xmath2413.2&1.07&0.18&58&109&r , cn&clouds + 2010 nov 12&07:5612:30&0.8&@xmath2415.2&1.08&0.19&58&111&r , cn&clouds + 2010 nov 13&08:0913:10&0.8&@xmath2416.2&1.08&0.20&57&112&r , cn&clouds + 2010 nov 16&08:0813:08&0.8&@xmath2419.2&1.09&0.21&56&114&r , cn&photometric + 2010 nov 26&07:4512:43&0.8&@xmath2429.2&1.13&0.26&51&123&r , cn&photometric + 2010 nov 27&07:4812:46&0.8&@xmath2430.2&1.14&0.27&50&124&r , cn&photometric + 2010 dec 9&07:0213:19&1.1&@xmath2442.2&1.21&0.33&42&136&r , cn , bc&thin cirrus + 2010 dec 10&06:4808:50&1.1&@xmath2443.1&1.21&0.34&41&137&r , cn , bc&thin cirrus + 2010 dec 15&08:5709:13&1.1&@xmath2448.1&1.25&0.36&38&143&cn&clouds + 2011 jan 7&04:0210:51&1.1&@xmath2471.1&1.42&0.51&25&179&r , cn , bc&clouds + 2011 jan 8&07:1207:57&1.1&@xmath2472.1&1.43&0.52&25&181&r , cn&clouds + 2011
jan 9&07:0908:47&1.1&@xmath2473.1&1.44&0.53&25&183&r ,
cn&clouds + 2011 jan 11&04:0110:50&1.1&@xmath2475.1&1.45&0.54&24&187&r , cn , bc&thin cirrus + [ t : imaging_circ ] + 1991&oct&11.44&+30.2&1.050&1.063&56.4&+0.47&+10.5&@xmath1214.507&@xmath1213.121&@xmath1212.370 + 1997&nov&2.13&@xmath1249.1&1.241&1.019&50.9&+0.47&@xmath1212.7&@xmath1214.735&@xmath1213.328&@xmath1212.648 + 1997&dec&1.53&@xmath1219.7&1.071&0.918&58.8&+0.46&@xmath126.5&@xmath1214.839&@xmath1213.177&@xmath1212.489 + 1997&dec&4.16&@xmath1217.1&1.062&0.908&59.4&+0.46&@xmath125.8&@xmath1214.818&@xmath1213.172&@xmath1212.495 + 1998&feb&26.23&+66.9&1.370&1.091&45.7&+0.46&+14.4&@xmath1214.504&@xmath1213.397&@xmath1212.642 + 2010&jul&12.35&@xmath12107.9&1.742&0.958&29.0&+0.38&@xmath1215.4&@xmath1215.215&@xmath1213.607&@xmath1212.943 + 2010&aug&11.30&@xmath1278.0&1.480&0.588&29.9&+0.38&@xmath1214.7&@xmath1215.011&@xmath1213.471&@xmath1212.799 + 2010&aug&12.24&@xmath1277.0&1.472&0.578&30.0&+0.38&@xmath1214.7&@xmath1215.006&@xmath1213.466&@xmath1212.793 + 2010&sep&7.21&@xmath1251.1&1.266&0.344&35.8&+0.42&@xmath1212.4&@xmath1214.745&@xmath1213.346&@xmath1212.670 + 2010&sep&30.16&@xmath1228.1&1.128&0.189&44.4&+0.46&@xmath128.1&@xmath1214.870&@xmath1213.220&@xmath1212.519 + 2010&oct&1.17&@xmath1227.1&1.123&0.183&44.7&+0.46&@xmath127.9&@xmath1214.873&@xmath1213.215&@xmath1212.516 + 2010&oct&31.30&+3.0&1.060&0.140&58.2&+0.47&+1.0&@xmath1214.857&@xmath1213.223&@xmath1212.597 + 2010&nov&16.38&+19.1&1.091&0.211&56.0&+0.47&+5.8&@xmath1214.650&@xmath1213.135&@xmath1212.400 + 2010&dec&13.34&+46.1&1.232&0.352&39.3&+0.44&+11.6&@xmath1214.561&@xmath1213.272&@xmath1212.527 + 2011&jan&5.34&+69.1&1.405&0.497&26.0&+0.35&+14.2&@xmath1214.532&@xmath1213.417&@xmath1212.668 + 2011&feb&1.17&+95.9&1.635&0.742&21.8&+0.31&+15.3&@xmath1214.656&@xmath1213.558&@xmath1212.793 + 2011&feb&2.25&+97.0&1.645&0.754&21.8&+0.31&+15.3&@xmath1214.662&@xmath1213.564&@xmath1212.799 + 2011&feb&23.14&+117.9&1.831&1.025&24.2&+0.33&+15.4&@xmath1214.757&@xmath1213.658&@xmath1212.896 +
[ t : phot_circ ] + 1991&oct&8.50&35.3&4.13&@xmath129.70&@xmath1210.55&@xmath129.85&@xmath1210.12&@xmath129.88&@xmath1213.42& ... &@xmath1213.22&32.34&30.03&29.99&29.40&29.98 + 1991&oct&11.44&35.3&4.13&@xmath129.78&@xmath1210.72&@xmath129.99&@xmath1210.29&@xmath1210.00&@xmath1213.45& ... &@xmath1213.27&32.23&29.90&29.88&29.25&29.89 + 1997&nov&2.13&81.4&4.48&@xmath1210.37&@xmath1211.33&@xmath1210.71&@xmath1210.81&@xmath1210.67&@xmath1214.33&@xmath1213.87&@xmath1213.88&31.83&29.47&29.40&28.84&29.33 + 1997&dec&1.53&109.9&4.56& ... & ...
&@xmath129.89&@xmath1210.31&@xmath129.90& ... &@xmath1213.29& ... & ... & ... &29.97&29.12&29.88 + 1997&dec&4.16&146.7&4.68&@xmath129.54&@xmath1210.26&@xmath129.74&@xmath1210.19&@xmath129.75&@xmath1213.00&@xmath1213.06&@xmath1213.17&32.64&30.28&30.12&29.23&30.01 + 1998&feb&26.23&114.7&4.66&@xmath129.92&@xmath1211.24&@xmath1210.48&@xmath1210.91&@xmath1210.51&@xmath1213.62&@xmath1213.57&@xmath1213.81&32.11&29.68&29.69&28.88&29.63 + 2010&jul&12.32&97.2&4.53&@xmath1211.65&@xmath1212.55&@xmath1211.65&@xmath1211.69&@xmath1211.79&@[email protected]&@xmath1214.39&@xmath1214.28&30.98&28.47&28.71&28.21&28.45 + 2010&jul&12.38&97.2&4.53&@xmath1211.60&@xmath1213.74&@xmath1211.68&@xmath1211.72&@xmath1211.76&@xmath1214.65&@xmath1214.55&@xmath1214.44&31.03&27.28&28.68&28.17&28.48 + 2010&aug&11.30&97.2&4.32&@xmath1210.99&@xmath1211.85&@xmath1211.15&@xmath1211.15&@xmath1211.16&@xmath1214.36&@xmath1214.23&@xmath1214.22&31.01&28.60&28.64&28.18&28.52 + 2010&aug&11.31&97.2&4.32&@xmath1211.01&@xmath1211.91&@xmath1211.14&@xmath1211.15&@xmath1211.17&@xmath1214.24&@xmath1214.09&@xmath1214.24&30.99&28.55&28.64&28.17&28.51 + 2010&aug&12.20&155.9&4.51&@xmath1210.66&@xmath1211.34&@xmath1210.86&@xmath1211.00&@xmath1210.85&@xmath1215.13&@xmath1214.21&@xmath1215.02&31.32&29.09&28.90&28.31&28.80 + 2010&aug&12.23&97.2&4.31&@xmath1210.98&@xmath1211.89&@xmath1211.16&@xmath1211.20&@xmath1211.18&_und_&@xmath1214.40&@xmath1214.28&31.00&28.55&28.61&28.11&28.48 + 2010&aug&12.26&97.2&4.31&@xmath1210.99&@xmath1211.90&@xmath1211.20&@xmath1211.15&@xmath1211.15&@xmath1214.43&@xmath1214.15&@xmath1214.34&30.99&28.54&28.57&28.16&28.51 + 2010&aug&12.28&48.6&4.01&@xmath1211.49&@xmath1212.36&@xmath1211.67&@xmath1211.59&@xmath1211.67&@xmath1214.62&@xmath1214.58&@xmath1214.44&30.49&28.08&28.09&27.72&27.98 + 2010&sep&7.13&62.4&3.89&@xmath1210.60&@xmath1211.57&@xmath1210.92&@xmath1210.76&@xmath1210.92&@xmath1214.06&@xmath1213.71&@xmath1213.68&30.67&28.29&28.27&27.97&28.15 + 2010&sep&7.18&62.4&3.89&@xmath1210.60&@xmath1211.60&@xmath1210.92&@xmath1210.72&@xmath1210.93&@xmath1214.22&@xmath1213.69&@xmath1213.73&30.66&28.27&28.27&28.01&28.14 + 2010&sep&7.22&62.4&3.89&@xmath1210.61&@xmath1211.57&@xmath1210.93&@xmath1210.74&@xmath1210.92&@xmath1214.14&@xmath1213.71&@xmath1213.74&30.66&28.30&28.26&27.98&28.15 + 2010&sep&7.24&97.2&4.08&@xmath1210.29&@xmath1211.25&@xmath1210.63&@xmath1210.50&@xmath1210.61&@xmath1214.13&@xmath1213.53&@xmath1213.58&30.98&28.62&28.56&28.23&28.46 + 2010&sep&7.26&62.4&3.89&@xmath1210.62&@xmath1211.63&@xmath1210.95&@xmath1210.78&@xmath1210.92&@xmath1214.09&@xmath1213.71&@xmath1213.73&30.65&28.24&28.24&27.95&28.15 + 2010&sep&7.28&155.9&4.29&@xmath129.96&@xmath1210.96&@xmath1210.32&@xmath1210.45&@xmath1210.28&_und_&@xmath1213.43&@xmath1213.82&31.30&28.91&28.88&28.28&28.80 + 2010&sep&30.11&62.4&3.63&@xmath1210.24&@xmath1211.02&@xmath1210.46&@xmath1210.22&@xmath1210.42&@xmath1213.54&@xmath1213.08&@xmath1213.12&30.63&28.21&28.06&27.89&28.03 + 2010&sep&30.17&48.6&3.52&@xmath1210.43&@xmath1211.25&@xmath1210.61&@xmath1210.37&@xmath1210.60&@xmath1213.56&@xmath1213.17&@xmath1213.20&30.45&27.97&27.91&27.73&27.85 + 2010&sep&30.19&155.9&4.03&@xmath129.57&@xmath1210.36&@xmath129.76&@xmath129.73&@xmath129.76&@xmath1213.05&@xmath1212.74&@xmath1212.81&31.31&28.86&28.76&28.38&28.69 + 2010&sep&30.20&97.2&3.82&@xmath129.91&@xmath1210.71&@xmath1210.10&@xmath129.97&@xmath1210.10&@xmath1213.32&@xmath1212.94&@xmath1212.96&30.96&28.52&28.42&28.13&28.35 + 2010&oct&1.17&97.2&3.81&@xmath129.86&@xmath1210.63&@xmath129.98&@xmath129.84&@xmath1210.04&@xmath1213.33&@xmath1212.88&@xmath1212.94&30.99&28.56&28.51&28.24&28.38 + 2010&oct&31.27&62.4&3.50&@xmath129.90&@xmath1210.70&@xmath1210.06&@xmath129.74&@xmath129.94&@xmath1213.14&@xmath1212.66&@xmath1212.72&30.70&28.26&28.28&28.05&28.20 + 2010&oct&31.28&126.7&3.81&@xmath129.30&@xmath1210.12&@xmath129.53&@xmath129.32&@xmath129.41&@xmath1212.80&@xmath1212.40&@xmath1212.46&31.29&28.84&28.80&28.47&28.73 + 2010&oct&31.30&48.6&3.39&@xmath1210.04&@xmath1210.87&@xmath1210.25&@xmath129.90&@xmath1210.13&@xmath1213.23&@xmath1212.80&@xmath1212.83&30.56&28.10&28.09&27.89&28.01 + 2010&oct&31.30&38.5&3.29&@xmath1210.20&@xmath1211.06&@xmath1210.42&@xmath1210.06&@xmath1210.30&@xmath1213.35&@xmath1212.89&@xmath1212.91&30.40&27.91&27.92&27.74&27.84 + 2010&oct&31.32&97.2&3.69&@xmath129.54&@xmath1210.34&@xmath129.75&@xmath129.49&@xmath129.62&@xmath1212.95&@xmath1212.52&@xmath1212.55&31.06&28.62&28.59&28.30&28.52 + 2010&nov&16.31&97.2&3.87&@xmath129.47&@xmath1210.36&@xmath129.63&@xmath129.69&@xmath129.74&@xmath1212.94&@xmath1212.66&@xmath1212.71&31.27&28.87&28.87&28.48&28.78 + 2010&nov&16.32&62.4&3.68&@xmath129.80&@xmath1210.72&@xmath129.94&@xmath129.91&@xmath1210.06&@xmath1213.17&@xmath1212.87&@xmath1212.91&30.94&28.51&28.55&28.26&28.46 + 2010&nov&16.35&38.5&3.47&@xmath1210.15&@xmath1211.06&@xmath1210.28&@xmath1210.22&@xmath1210.42&@xmath1213.41&@xmath1213.08&@xmath1213.12&30.60&28.17&28.22&27.95&28.10 + 2010&nov&16.39&77.8&3.77&@xmath129.63&@xmath1210.49&@xmath129.77&@xmath129.79&@xmath129.88&@xmath1213.01&@xmath1212.77&@xmath1212.82&31.12&28.74&28.73&28.39&28.64 + 2010&nov&16.40&97.2&3.87&@xmath129.48&@xmath1210.34&@xmath129.63&@xmath129.65&@xmath129.74&@xmath1213.08&@xmath1212.69&@xmath1212.75&31.27&28.89&28.86&28.52&28.78 + 2010&nov&16.41&48.6&3.57&@xmath129.98&@xmath1210.85&@xmath1210.11&@xmath1210.03&@xmath1210.22&@xmath1213.37&@xmath1212.96&@xmath1213.01&30.76&28.38&28.39&28.15&28.30 + 2010&nov&16.42&62.4&3.68&@xmath129.80&@xmath1210.68&@xmath129.94&@xmath129.90&@xmath1210.06&@xmath1213.31&@xmath1212.87&@xmath1212.91&30.95&28.55&28.56&28.27&28.46 + 2010&nov&16.43&126.7&3.99&@xmath129.29&@xmath1210.15&@xmath129.45&@xmath129.53&@xmath129.55&@xmath1212.96&@xmath1212.59&@xmath1212.66&31.45&29.08&29.05&28.64&28.97 + 2010&nov&16.45&77.8&3.77&@xmath129.63&@xmath1210.50&@xmath129.76&@xmath129.77&@xmath129.87&@xmath1213.19&@xmath1212.77&@xmath1212.84&31.12&28.74&28.73&28.40&28.65 + 2010&dec&13.27&38.5&3.69&@xmath1210.38&@xmath1211.42&@xmath1210.65&@xmath1210.61&@xmath1210.81&@xmath1213.87&@xmath1213.41&@xmath1213.42&30.72&28.39&28.42&28.12&28.26 + 2010&dec&13.28&48.6&3.79&@xmath1210.19&@xmath1211.28&@xmath1210.47&@xmath1210.48&@xmath1210.63&@xmath1213.66&@xmath1213.32&@xmath1213.29&30.91&28.53&28.60&28.25&28.44 + 2010&dec&13.34&77.8&4.00&@xmath129.86&@xmath1210.91&@xmath1210.16&@xmath1210.21&@xmath1210.26&@xmath1213.48&@xmath1213.09&@xmath1213.11&31.24&28.91&28.91&28.51&28.81 + 2010&dec&13.35&24.5&3.50&@xmath1210.70&@xmath1211.77&@xmath1210.96&@xmath1210.88&@xmath1211.14&@xmath1214.12&@xmath1213.59&@xmath1213.61&30.40&28.04&28.11&27.84&27.93 + 2010&dec&13.36&48.6&3.79&@xmath1210.19&@xmath1211.28&@xmath1210.47&@xmath1210.43&@xmath1210.61&@xmath1213.74&@xmath1213.29&@xmath1213.34&30.92&28.53&28.60&28.29&28.46 + 2010&dec&13.37&97.2&4.09&@xmath129.71&@xmath1210.81&@xmath1210.02&@xmath1210.09&@xmath1210.12&@xmath1213.34&@xmath1213.03&@xmath1213.10&31.40&29.01&29.05&28.64&28.95 + 2010&dec&13.38&126.7&4.21&@xmath129.54&@xmath1210.60&@xmath129.86&@xmath129.98&@xmath129.94&@xmath1213.20&@xmath1212.92&@xmath1213.00&31.57&29.22&29.21&28.74&29.13 + 2010&dec&13.40&62.4&3.90&@xmath1210.02&@xmath1211.10&@xmath1210.31&@xmath1210.32&@xmath1210.43&@xmath1213.56&@xmath1213.24&@xmath1213.28&31.08&28.71&28.76&28.40&28.64 + 2010&dec&13.41&77.8&4.00&@xmath129.86&@xmath1210.90&@xmath1210.15&@xmath1210.18&@xmath1210.25&@xmath1213.52&@xmath1213.14&@xmath1213.20&31.24&28.91&28.92&28.54&28.82 + 2011&jan&5.30&62.4&4.05&@xmath1210.39&@xmath1211.69&@xmath1210.88&@xmath1210.93&@xmath1210.94&@xmath1214.12&@xmath1213.64&@xmath1213.65&30.99&28.57&28.63&28.21&28.54 + 2011&jan&5.31&38.5&3.84&@xmath1210.71&@xmath1212.08&@xmath1211.24&@xmath1211.19&@xmath1211.33&@xmath1214.14&@xmath1213.84&@xmath1213.73&30.66&28.17&28.26&27.95&28.15 + 2011&jan&5.36&77.8&4.15&@xmath1210.20&@xmath1211.55&@xmath1210.74&@xmath1210.84&@xmath1210.78&@xmath1213.78&@xmath1213.47&@xmath1213.49&31.17&28.71&28.77&28.30&28.71 + 2011&jan&5.37&97.2&4.24&@xmath1210.06&@xmath1211.39&@xmath1210.59&@xmath1210.69&@xmath1210.64&@xmath1213.81&@xmath1213.43&@xmath1213.45&31.31&28.87&28.92&28.44&28.84 + 2011&feb&1.13&62.4&4.23&@xmath1210.91&@xmath1212.46&@xmath1211.29&@xmath1211.43&@xmath1211.40&@xmath1214.35&@xmath1214.02&@xmath1214.15&30.94&28.29&28.69&28.19&28.56 + 2011&feb&1.14&97.2&4.42&@xmath1210.57&@xmath1211.82&@xmath1210.99&@xmath1211.09&@xmath1211.13&@xmath1214.08&@xmath1213.68&@xmath1213.62&31.28&28.92&29.00&28.53&28.83 + 2011&feb&1.21&97.2&4.42&@xmath1210.59&@xmath1211.94&@xmath1211.01&@xmath1211.11&@xmath1211.07&@xmath1214.13&@xmath1213.83&@xmath1213.96&31.25&28.81&28.97&28.51&28.90 + 2011&feb&2.23&126.7&4.54&@xmath1210.42&@xmath1211.95&@xmath1210.88&@xmath1211.01&@xmath1210.96&@xmath1214.30&@xmath1213.79&@xmath1213.81&31.45&28.81&29.13&28.63&29.02 + 2011&feb&2.26&48.6&4.12&@xmath1211.11&@xmath1212.96&@xmath1211.58&@xmath1211.64&@xmath1211.71&@xmath1214.53&@xmath1214.04&@xmath1214.16&30.75&27.81&28.42&28.00&28.27 + 2011&feb&23.11&62.4&4.37&@xmath1211.34&@xmath1212.58&@xmath1211.74&@xmath1211.63&@xmath1211.83&@xmath1214.72&@xmath1214.22&@xmath1214.35&30.89&28.55&28.63&28.37&28.51 + 2011&feb&23.13&77.8&4.46&@xmath1211.19&@xmath1213.05&@xmath1211.55&@xmath1211.58&@xmath1211.63&@xmath1214.42&@xmath1214.22&@xmath1214.26&31.03&28.08&28.82&28.42&28.71 + 2011&feb&23.15&97.2&4.56&@xmath1211.00&@xmath1212.34&@xmath1211.39&@xmath1211.48&@xmath1211.51&@xmath1214.43&@xmath1213.82&@xmath1213.95&31.23&28.79&28.98&28.52&28.84 + 2011&feb&23.18&62.4&4.37&@xmath1211.32&@xmath1212.78&@xmath1211.67&@xmath1211.66&@xmath1211.89&@xmath1214.75&@xmath1214.21&@xmath1214.06&30.91&28.34&28.70&28.33&28.45 + [ t : phot_flux ] + 1991&oct&8.50&+27.2&0.014&4.13&28.50.01&26.42.01&25.99.00&25.33.00&26.18.00&2.43.01& ... &2.39.01&28.63 + 1991&oct&11.44&+30.2&0.021&4.13&28.39.01&26.28.01&25.88.00&25.17.01&26.08.00&2.42.02& ... &2.36.01&28.51 + 1997&nov&2.13&@xmath1249.1&0.094&4.48&27.55.01&25.39.02&24.98.01&24.46.02&25.10.01&1.39.13&1.52.04&1.53.03&27.64 + 1997&dec&1.53&@xmath1219.7&0.030&4.56& ... & ... &25.42.00&24.75.01&25.52.00& ... &1.80.02 & ... & ... + 1997&dec&4.16&@xmath1217.1&0.026&4.68&28.06.01&25.85.00&25.44.00&24.82.01&25.52.00&2.28.02&1.89.01&1.80.01&28.18 + 1998&feb&26.23&+66.9&0.137&4.66&27.62.02&25.38.04&25.06.01&24.36.04&25.20.01&2.07.09&1.79.05&1.57.05&27.69 + 2010&jul&12.32&@xmath12107.9&0.241&4.53&26.77.09&24.48.21&24.33.04&23.73.12&24.27.06&@xmath201.2&1.19.14&1.32.11&26.78 + 2010&jul&12.38&@xmath12107.9&0.241&4.53&26.82.07&23.30.89&24.30.03&23.69.11&24.30.06&1.26.28&1.04.18&1.16.14&26.83 + 2010&aug&11.30&@xmath1278.0&0.170&4.32&27.05.02&24.87.05&24.50.01&23.90.04&24.58.02&1.19.16&1.00.10&1.03.09&27.10 + 2010&aug&11.31&@xmath1278.0&0.170&4.32&27.03.02&24.82.05&24.51.01&23.90.04&24.57.02&1.32.12&1.13.07&1.01.09&27.08 + 2010&aug&12.20&@xmath1277.1&0.168&4.51&27.06.02&25.04.03&24.49.08&23.86.08&24.58.01&0.20.79&0.80.15&0.01.44&27.11 + 2010&aug&12.23&@xmath1277.0&0.168&4.31&27.05.02&24.83.06&24.48.02&23.84.05&24.55.02&_und_&0.81.15&0.96.11&27.10 + 2010&aug&12.26&@xmath1277.0&0.168&4.31&27.04.02&24.82.05&24.45.02&23.89.04&24.58.02&1.11.17&1.06.08&0.90.11&27.09 + 2010&aug&12.28&@xmath1277.0&0.168&4.01&27.03.03&24.86.07&24.44.02&23.79.05&24.53.03&1.23.14&0.94.11&1.10.08&27.08 + 2010&sep&7.13&@xmath1251.1&0.102&3.89&27.31.01&25.19.03&24.72.01&24.14.02&24.80.01&1.32.09&1.34.03&1.39.02&27.40 + 2010&sep&7.18&@xmath1251.1&0.102&3.89&27.31.01&25.16.03&24.72.01&24.19.01&24.80.01&1.16.10&1.36.02&1.34.03&27.39 + 2010&sep&7.22&@xmath1251.0&0.102&3.89&27.30.01&25.19.02&24.71.01&24.16.01&24.81.01&1.24.09&1.35.03&1.34.03&27.39 + 2010&sep&7.24&@xmath1251.0&0.102&4.08&27.31.01&25.18.02&24.71.01&24.19.01&24.81.01&1.06.12&1.33.02&1.30.03&27.39 + 2010&sep&7.26&@xmath1251.0&0.102&3.89&27.29.01&25.13.03&24.69.01&24.13.01&24.80.01&1.29.07&1.34.02&1.34.02&27.37 + 2010&sep&7.28&@xmath1251.0&0.102&4.29&27.31.01&25.14.04&24.72.02&24.04.04&24.83.00&_und_&1.22.03&0.85.07&27.39 + 2010&sep&30.11&@xmath1228.2&0.052&3.63&27.64.01&25.47.01&24.87.01&24.37.01&25.05.00&1.48.03&1.61.01&1.59.01&27.75 + 2010&sep&30.17&@xmath1228.1&0.052&3.52&27.64.01&25.43.01&24.90.00&24.37.01&25.05.01&1.57.03&1.63.01&1.62.01&27.75 + 2010&sep&30.19&@xmath1228.1&0.052&4.03&27.66.00&25.45.01&24.94.00&24.39.00&25.07.00&1.57.02&1.55.01&1.50.01&27.77 + 2010&sep&30.20&@xmath1228.1&0.052&3.82&27.65.00&25.45.01&24.92.00&24.37.01&25.05.00&1.51.03&1.56.01&1.56.01&27.76 + 2010&oct&1.17&@xmath1227.1&0.050&3.81&27.70.00&25.52.01&25.03.00&24.50.00&25.10.00&1.48.03&1.60.01&1.57.01&27.81 + 2010&oct&31.27&+3.0&0.025&3.50&27.89.02&25.71.01&25.27.00&24.69.00&25.40.00&1.70.03&1.85.01&1.81.01&28.02 + 2010&oct&31.28&+3.0&0.025&3.81&27.97.01&25.76.01&25.30.00&24.72.00&25.42.00&1.73.03&1.80.01&1.76.01&28.10 + 2010&oct&31.30&+3.0&0.025&3.39&27.94.01&25.74.01&25.27.00&24.69.01&25.39.00&1.71.03&1.82.01&1.81.01&28.07 + 2010&oct&31.30&+3.0&0.025&3.29&27.95.01&25.73.01&25.27.01&24.68.01&25.39.00&1.70.03&1.83.01&1.83.01&28.07 + 2010&oct&31.32&+3.1&0.025&3.69&27.93.00&25.74.01&25.26.00&24.69.00&25.40.00&1.69.02&1.80.01&1.78.01&28.05 + 2010&nov&16.31&+19.0&0.038&3.87&27.87.01&25.70.01&25.27.00&24.66.01&25.38.00&1.90.06&1.85.02&1.83.02&27.98 + 2010&nov&16.32&+19.1&0.038&3.68&27.85.01&25.67.02&25.27.00&24.67.01&25.38.00&1.87.06&1.84.02&1.83.01&27.97 + 2010&nov&16.35&+19.1&0.038&3.47&27.87.01&25.70.01&25.28.00&24.65.01&25.37.01&1.84.04&1.85.01&1.82.01&27.98 + 2010&nov&16.39&+19.1&0.038&3.77&27.87.00&25.74.00&25.29.00&24.68.00&25.40.00&1.93.01&1.85.01&1.82.01&27.99 + 2010&nov&16.40&+19.1&0.038&3.87&27.86.00&25.73.00&25.27.00&24.70.00&25.39.00&1.76.02&1.83.01&1.79.01&27.98 + 2010&nov&16.41&+19.1&0.038&3.57&27.86.00&25.73.01&25.28.00&24.70.00&25.40.00&1.78.02&1.86.01&1.83.01&27.97 + 2010&nov&16.42&+19.2&0.038&3.68&27.86.00&25.71.01&25.28.00&24.68.00&25.38.00&1.73.02&1.84.01&1.83.01&27.97 + 2010&nov&16.43&+19.2&0.038&3.99&27.86.00&25.72.00&25.28.00&24.70.00&25.40.00&1.77.02&1.82.01&1.77.01&27.98 + 2010&nov&16.45&+19.2&0.038&3.77&27.87.00&25.73.00&25.29.00&24.70.00&25.41.00&1.75.02&1.84.01&1.80.01&27.99 + 2010&dec&13.27&+46.0&0.091&3.69&27.68.03&25.61.04&25.18.01&24.55.02&25.22.01&1.71.11&1.84.03&1.84.02&27.77 + 2010&dec&13.28&+46.0&0.091&3.79&27.71.02&25.58.02&25.19.01&24.54.01&25.24.01&1.82.06&1.83.02&1.88.01&27.80 + 2010&dec&13.34&+46.1&0.091&4.00&27.69.01&25.60.01&25.18.00&24.57.01&25.28.00&1.79.05&1.86.01&1.85.01&27.78 + 2010&dec&13.35&+46.1&0.091&3.50&27.70.01&25.60.03&25.20.01&24.55.01&25.23.01&1.65.06&1.86.01&1.85.02&27.79 + 2010&dec&13.36&+46.1&0.091&3.79&27.71.01&25.58.02&25.20.00&24.59.01&25.26.01&1.74.04&1.86.01&1.83.01&27.80 + 2010&dec&13.37&+46.1&0.091&4.09&27.70.00&25.54.01&25.17.00&24.58.01&25.27.00&1.83.03&1.81.01&1.77.01&27.79 + 2010&dec&13.38&+46.1&0.091&4.21&27.69.00&25.56.01&25.16.00&24.58.01&25.28.00&1.86.03&1.81.01&1.76.01&27.77 + 2010&dec&13.40&+46.1&0.091&3.90&27.69.01&25.57.01&25.18.00&24.56.01&25.26.00&1.81.03&1.80.01&1.78.01&27.78 + 2010&dec&13.41&+46.1&0.091&4.00&27.70.01&25.61.01&25.19.00&24.60.01&25.29.00&1.76.05&1.81.02&1.77.02&27.79 + 2011&jan&5.30&+69.0&0.148&4.05&27.42.01&25.25.04&24.88.01&24.21.03&24.99.01&1.51.11&1.67.03&1.67.03&27.48 + 2011&jan&5.31&+69.0&0.148&3.84&27.45.02&25.22.06&24.85.01&24.22.03&24.95.02&1.70.08&1.67.03&1.81.02&27.51 + 2011&jan&5.36&+69.1&0.148&4.15&27.45.01&25.23.03&24.87.01&24.19.02&25.01.01&1.75.06&1.74.03&1.73.03&27.51 + 2011&jan&5.37&+69.1&0.148&4.24&27.44.01&25.23.03&24.88.01&24.24.02&24.99.01&1.63.08&1.68.03&1.68.03&27.50 + 2011&feb&1.13&+95.9&0.214&4.23&27.17.04&24.77.17&24.74.02&24.02.07&24.81.03&1.59.17&1.59.07&1.48.09&27.20 + 2011&feb&1.14&+95.9&0.214&4.42&27.21.02&25.08.07&24.76.01&24.15.05&24.79.02&1.67.14&1.74.05&1.82.04&27.24 + 2011&feb&1.21&+95.9&0.214&4.42&27.19.01&24.97.07&24.73.01&24.14.04&24.85.02&1.61.12&1.58.05&1.48.07&27.21 + 2011&feb&2.23&+97.0&0.216&4.54&27.19.01&24.78.12&24.71.02&24.14.06&24.80.02&1.34.27&1.52.07&1.53.07&27.22 + 2011&feb&2.26&+97.0&0.216&4.12&27.16.03&24.46.29&24.64.03&23.95.09&24.69.04&1.53.19&1.69.06&1.59.07&27.19 + 2011&feb&23.11&+117.8&0.263&4.37&26.96.05&24.86.19&24.52.04&24.05.10&24.60.06&1.46.28&1.62.09&1.51.11&26.96 + 2011&feb&23.13&+117.9&0.263&4.46&26.95.04&24.23.40&24.56.03&24.00.09&24.65.05&1.66.20&1.53.11&1.51.11&26.95 + 2011&feb&23.15&+117.9&0.263&4.56&26.99.03&24.78.16&24.58.03&24.01.11&24.63.04&1.55.23&1.84.05&1.73.07&27.00 + 2011&feb&23.18&+117.9&0.263&4.37&26.98.04&24.66.24&24.59.04&24.02.10&24.54.06&1.43.28&1.63.08&1.81.05&26.98 + [ t : phot_rates ] + nh&@[email protected]&@[email protected]&@xmath122.18&+0.01&+0.09 + cn&@[email protected]&@[email protected]&@xmath122.55&+0.01&+0.07 + c@xmath2&@[email protected]&@[email protected]&@xmath123.15&+0.01&+0.07 + c@xmath1&@[email protected]&@[email protected]&@xmath122.46&+0.01&+0.07 + uc&@[email protected]&@[email protected]&@xmath1225.88&+0.03&+0.23 + bc&@[email protected]&@[email protected]&@xmath1225.81&+0.04&+0.25 + gc&@[email protected]&@[email protected]&@xmath1225.84&+0.04&+0.26 + uc&@[email protected]&@[email protected]&@xmath1225.43&+0.03&+0.21 + bc&@[email protected]&@[email protected]&@xmath1225.44&+0.04&+0.22 + gc&@[email protected]&@[email protected]&@xmath1225.48&+0.03&+0.21 + [ t : table5 ] .
other oh - based results include _ hst_/fos data by @xcite , data from the _ swift _ satellite ( bodewits private communication ; @xcite ) , and radio oh measurements from nancey @xcite .
forbidden oxygen data were obtained by @xcite at the 1997 apparition , while the granddaughter hydrogen was measured using the lyman alpha line with _
soho_/swan ( @xcite ; @xcite ) . in the ir ,
water measurements have become more common , with space - based data from _
iso_/isophot @xcite and _ herschel _ @xcite , and ground - based data with keck / nirspec @xcite .
see the key to associate the symbols with these references ; only the first author is listed in the key due to space constraints . as discussed in the text ,
the ensemble of data indicate a larger drop in production rates between 1997/98 and 2010/11 and a larger amount of variability than we inferred from our own data alone .
, width=604 ] from the nucleus within a 10@xmath19 wide wedge centered on a given position angle ( pa ) .
the wedges are centered at pas 10@xmath19 ( green ) , 100@xmath19 ( black ) , 190@xmath19 ( red ) , and 280@xmath19 ( blue ) .
the sun is at a pa of 102@xmath19 .
the species is given in the top right corner of each plot .
the flux has units of erg cm@xmath13 s@xmath26 for the gas species and erg cm@xmath13 s@xmath26 @xmath26 for bc .
a 1/@xmath10 profile is shown as a dashed line in the bc plot .
note that the slopes of the gas species can be compared directly , as they are plotted with the same @xmath17log(flux ) , but with different ranges.,height=642 ] 8,000 km across , centered on the nucleus , enhanced by division of an azimuthal median profile , and smoothed with a boxcar smooth .
north is up , east is left , and the position angle ( pa ) of the sun is near 105@xmath19 ( exact pas for each night are given in table [ t : imaging_circ ] ) .
all images have the same stretch
. the dust jet can be seen at pas of @xmath095@xmath19 on november 2 , @xmath0115@xmath19 on november 3 , @xmath0125@xmath19 on november 4 , and @xmath0100@xmath19 on november 7 .
diagonal streaks are trailed stars and the bright white area in the west half of each image is the dust tail.,width=226 ] | we report on photometry and imaging of comet 103p / hartley 2 obtained at lowell observatory from 1991 through 2011 .
we acquired photoelectric photometry on two nights in 1991 , four nights in 1997/98 , and 13 nights in 2010/11 .
we observed a strong secular decrease in water and all other observed species production in 2010/11 from the 1991 and 1997/98 levels .
we see evidence for a strong asymmetry with respect to perihelion in the production rates of our usual bandpasses , with peak production occurring @xmath010 days post - perihelion and production rates considerably higher post - perihelion .
the composition was `` typical , '' in agreement with the findings of other investigators .
we obtained imaging on 39 nights from 2010 july until 2011 january .
we find that , after accounting for their varying parentage and lifetimes , the c@xmath1 and c@xmath2 coma morphology resemble the cn morphology we reported previously .
these species exhibited an hourglass shape in october and november , and the morphology changed with rotation and evolved over time .
the oh and nh coma morphology showed hints of an hourglass shape near the nucleus , but was also enhanced in the anti - sunward hemisphere . this tailward brightness enhancement did not vary significantly with rotation and evolved with the viewing geometry .
we conclude that all five gas species likely originate from the same source regions on the nucleus , but that oh and nh were derived from small grains of water and ammonia ice that survived long enough to be affected by radiation pressure and driven in the anti - sunward direction .
we detected the faint , sunward facing dust jet reported by other authors , and did not detect a corresponding gas feature .
this jet varied little during a night but exhibited some variations from night to night , suggesting it is located near the total angular momentum vector .
overall , our imaging results support the conclusions of other authors that hartley 2 s `` hyperactivity '' is caused by icy particles of various sizes that are lifted off the surface and break up in the coma to greatly increase the effective active surface area . * keywords * : comets ; comets , coma ; comets , composition |
the ultraviolet ( uv ) divergences of a quantum filed theory are related to the short distance behaviour of the vacuum expectation values of product of the field operators . in case of the free scalar field in minkowski background ,
the correlator is given by @xmath4 in the limit @xmath5 the correlator diverges leading to the bad uv behaviour .
the one possible way to remove uv divergences is to introduce a cut - off on some sufficiently high energy scales . in this approach momentum integrations
have to be limited to the scale of cut - off @xmath6 .
then field theoretical description of the field is effective and valid up to the scale of cut - off @xmath6 , where a new physical description has to be introduced .
another way to introduce cut - off is to perform averaging of the field on the sufficiently small distances .
then we introduce a new field @xmath7 which is averaged version of the field @xmath8 with some averaging function @xmath9 which fulfils the normalisation condition @xmath10 .
the @xmath11 is determinant of the spatial part of the metric . in our considerations
we take the frw metric , then @xmath12 where @xmath13 is the scale factor .
it turns out that vacuum expectation values of product of the new fields @xmath7 becomes finite for @xmath5 . therefore the uv divergence is removed . in the model considered in this paper
we assume smoothing gaussian function in the form @xmath14 the parameter of this function is dispersion @xmath15 , which is the scale of smoothing .
all the structures of the field below the scale @xmath15 are smoothed out .
it is worth to stress that the smoothing is over the fixed physical distance .
therefore we have introduced the factor @xmath16 in formula ( [ gauss ] ) .
the scale of smoothing can be related either with the properties of the field or with the properties of the space .
namely , it can be for example a scale where the structure of a field becomes polymerised , as in the case studied in @xcite
. then effective field theoretical description is valid only at the scales larger than @xmath15 .
another source of smoothing can be the discrete nature of space . in this case
averaging has to be performed on the scales where the quantum gravitational effects becomes significant , namely on the planck scale .
subsequent possible source of smoothing is decoherence of the quantum fluctuations @xcite . in this case
the parameter @xmath15 encodes the details of interaction between the quantum fluctuations and the environment .
process of diffusion is also the natural smoothing factor . in this paper
we study how the effect of smoothing modify the power spectrum of the scalar field . in our consideration
we apply the smoothing to the quantum fluctuations generated during the phase of inflation . in particular we show how power spectrum from de sitter inflation is regularised by the smoothing .
then we investigate possible observational consequences of this effect .
now we will derive the spectrum of the scalar field in the flat frw background in presence of smoothing .
we will consider classical scalar field now , but the results obtained are the same like in the quantum case . the quantum case is discussed in more details in appendix where also an example of the smoothed free scalar field in the minkowski background is shown . the scalar field @xmath17 can be smoothed by performing the integration @xmath18 where @xmath19 is the shooting function .
taking the fourier transform of equation ( [ conv ] ) we obtain @xmath20 where the fourier transform of the smoothing function ( [ gauss ] ) is given by @xmath21 and where @xmath22 . squaring the equation ( [ phif ] ) and taking the fourier transform of the smoothing function ,
we find @xmath23 the power spectrum of the field @xmath24 is defined as follows @xmath25 the spectrum of the smoothed field @xmath26 can be expressed taking the ( [ phif2 ] ) .
based on the definition ( [ power ] ) we obtain @xmath27 the effect of the smoothing is exponential suppression of the power spectrum on the scales @xmath28 comparable with @xmath15 . in the next section
we show how this effect modify the inflationary spectrum of quantum fluctuations .
in this section we will consider quantum fluctuations of the scalar field in de sitter inflation . during this phase of evolution the scale factor
is given by @xmath29 where @xmath30 is hubble factor which is constant .
parameter @xmath31 is conformal time defined as @xmath32 .
the vacuum is given by the bunch - davies mode @xmath33 applying this to the definition ( [ powerf ] ) and introducing the smoothing as in equation ( [ psigma ] ) we find @xmath34 e^{-p^2\sigma^2}. \label{pfds1}\ ] ] here we have introduced physical momentum @xmath35 in order to simplify the notation . in fig .
[ pf ] we plot function ( [ pfds1 ] ) for two values of @xmath30 .
we also assumed that @xmath36 . in the super - horizontal limit @xmath37
this spectrum takes the classical form @xmath38 however for the sub - horizontal modes the spectrum is significantly modified by the smoothing effect . in the case with no smoothing , the spectrum is approximated , in this range , by the minkowski vacuum and therefore @xmath39 .
such a behaviour leads to the quadratic uv divergence of the correlation function .
however due to the smoothing , the spectrum is suppressed and uv divergence is removed .
the transition between the minkowski vacuum range and the smoothing domain is in the form of the bump in the power spectrum .
maximum of this bump is located in @xmath40 however while @xmath41 then the maximum of the spectrum is placed in @xmath42 . in this case
bump is not present in the power spectrum .
the spectrum is not scale - invariant in this bump range .
it is namely `` blue '' on the left side with the spectral index @xmath43 .
moreover amplitude of perturbations here can be much bigger that in the super - horizontal domain .
this is interesting feature because e.g. models of primordial black holes formation require such a `` blue '' tilted spectrum in order to resolve problem of the supermassive black holes @xcite .
the spectrum obtained here is free from the uv divergences .
however the ir ( logarithmic ) divergence remains and can be removed only by taking superposition of the positive and negative bunch - davies modes ( see e.g.@xcite ) .
this is naturally obtained e.g. in the models with the contracting ( bounce ) phase before the inflation ( see @xcite ) . in this case
the inflationary spectrum is suppressed also on the large scales ( small @xmath44 ) and ir divergence is removed .
if the lack of divergences is condition for the physical relevance then inflationary model with the bounce and smoothing is favoured .
these both effects can be a result of quantum nature of gravity .
it is interesting to see whether the studied effect of smoothing can lead to any observational imprints . in order to investigate it
we will calculate observed primordial power spectrum .
for the particular @xmath45 mode , the amplitude of perturbations can be calculated during the horizon crossing , namely for @xmath46 .
this gives us @xmath47 however this is very idealised case . in real ,
the modes become _ frozen _ for @xmath48 . in this case spectrum
is given by ( [ pds ] ) and any effects of smoothing are seen .
namely the smoothing factor is now @xmath49 where @xmath50 . if @xmath51 as expected for the standard inflation and @xmath52 then @xmath53 .
therefore the exponential factor is with very high precision equal to one .
therefore we conclude that the effects of smoothing from the inflation are unable to detect .
this can be seen as no - go result , this way of searching for the quantum gravity effects is rather inaccessible .
in the previous section we have shown that in the case of de sitter inflation , the effect of smoothing can be suppression of the total power spectrum . however
when @xmath54 then this effect is negligible and can not be studied observationally . in this section
we will check whether effects of suppression in any other form can be probed observationally .
in particular the _ instantaneous smoothing _ can lead to more significant modifications . in this case smoothing is effective only in some short period of time , therefore the smoothing factor can be written as @xmath55 where @xmath56 is the value of the scale factor during the smoothing take place .
based on the cmb observations we can only constraint the joined factor @xmath57 .
the example we will examine in details in this section is direct smoothing of the primordial perturbations .
we will study here rather phenomenological form of modifications of the power spectrum and results obtained can be later related with some specific model . in this case modification will be given by the factor @xmath0 .
for the perturbations generated during the phase of inflation the spectrum can be parametrised in the following form @xmath58 introducing directly effect of smoothing we obtain the spectrum @xmath59 we have used here equation ( [ psigma ] ) with @xmath60 . based on this spectrum
we compute spectrum of the cmb anisotropy and polarisation .
we also find confidence intervals for the parameters of the model , namely on @xmath61 , @xmath62 and @xmath15 . in the numerical calculations we use the publicly available camb @xcite code and markov chain monte carlo ( mcmc )
package cosmo mc @xcite .
the codes were suitably modified to investigate the spectrum ( [ pphen ] ) .
to test our model we use the current astronomical observations from wmap5 @xcite .
we take the standard cosmological parameters as follows @xmath63 and the pivot scale @xmath64 . in fig .
[ cmb ] we show the obtained tt , te and ee spectra of cmb . [ cols="^ " , ] we show how these spectra are modified by the presence of smoothing .
the evident signature is suppression of the spectra on the high multipoles .
this effect is opposite to the effect of low multipoles suppression , usually considered in context of searching for the departures from the standard cosmological model .
in particular the effect of the bounce can lead to suppression of low multipoles @xcite .
however for the low multipoles , the so - called _ cosmic variance _ becomes significant .
this is serious obstacle in testing cosmological models beyond the standard inflation . on the high multipoles , the effect of _ cosmic variance _ becomes negligible .
however instrumental effects becomes significant here .
resolution of observations can be nonetheless considerably improved by the new missions like planck @xcite . in fig .
[ constraints ] we show the obtained constraints on the parameters @xmath61 , @xmath62 and @xmath15 . , @xmath62 and @xmath15 . 2d plots : solid lines show the 68 and 95 @xmath2 confidence intervals .
1d plots : dotted lines are mean likelihoods of samples , solid lines are marginalised probabilities.,width=453 ] we see that values of parameters @xmath62 and @xmath61 are constrained from the both sides . in case of the parameter @xmath15
only upper bound can be found .
based on the five years observations of wmap satellite we find @xmath65 this constraint can be significantly ( by a few orders of magnitude ) improved with the future observations e.g. from planck .
the scale @xmath66 corresponds to the present intergalactic distances .
one can apply the obtained limit to the case of _ instantaneous smoothing _ , then @xmath67 .
let us study a case when the smoothing takes place at the end of inflation , it can be e.g. due to decoherence . taking the @xmath68 gev to be the energy scale at the end of inflation we find @xmath69 . based on this
we find that at the end of inflation , the scale of smoothing is constrained by @xmath70 this is far from the planck scale , however relatively close to the scale of the micro - world where the quantum effects becomes significant .
therefore we speculate that the method can be used to constraint some models of decoherence of the quantum fluctuations .
also the scale of diffusion can be constrained with this result .
in this paper we have considered effects of the scalar field smoothing on the power spectrum .
we considered model where averaging function is given by gaussian shape . based on this
we have showed that smoothing modifies power spectrum by the factor @xmath71 , where @xmath15 is dispersion of the gaussian distribution .
we have applied the smoothing to the quantum fluctuations generated during the phase of cosmic inflation .
we have shown that the spectrum is regularised by the effect of smoothing and uv divergence is removed .
subsequently we have investigated whether smoothing on the planck scale can be probed observationally
. this would be possible way to study quantum gravitational effects .
unfortunately , our investigations suggest rather no - go perspective for this method .
later we have investigated whether smoothing in any other form can be studied observationally .
we have considered phenomenological modification of the inflationary power spectrum due to the smoothing .
we have shown that smoothing can lead to suppression of the high multipoles in the spectrum of the cmb anisotropy and polarisation .
based on five years data from wmap satellite we have derived constraint for the parameter @xmath15 .
we showed that @xmath72 mpc ( 95@xmath2 cl ) .
this size is comparable with distances between galaxies in the galactic cluster .
we have applied this constraint to the model of _ instantaneous smoothing _ at the end of inflation .
we showed that the constraint @xmath3 m is satisfied then .
this is far from the planck scale and can not be used to probe quantum gravity effects .
however models of other physical processes like diffusion or decoherence can be , in principle , constrained .
the quantum scalar field in frw background can be decomposed for the fourier modes as follows @xmath73 here @xmath74 and @xmath75 are the so - called mode functions .
moreover creation and annihilation operator fulfils relations @xmath76=\delta^{(3)}({\bf k - p})$ ] and @xmath77= [ \hat{a}^{\dagger}_{\bf k},\hat{a}^{\dagger}_{\bf p } ] = 0 $ ] .
the averaging of the quantum field @xmath78 is performed in the same way like in the classical case , namely @xmath79 now it is straightforward to calculate the correlation function .
we obtain @xmath80 where the power spectrum is defined as follows @xmath81 absorbing the factor @xmath71 in the integral ( [ cor1 ] ) we can define the smoothed spectrum as in equation ( [ psigma ] ) .
therefore the results obtained here as in the classical case are the same . as an example we consider free scalar field in the minkowski space . in this case
@xmath60 , @xmath82 and @xmath83 .
the correlation function for the field without the smoothing is equal to @xmath84 this is exactly the function ( [ cm ] ) .
the expression was calculated introducing regularizator @xmath85 and taking @xmath86 after the integration .
this correlation function is uv divergent in the limit @xmath87 .
the correlator of the smoothed scalar field does not exhibit such a divergence .
namely it can expressed as follows @xmath88 where @xmath89 and @xmath90 is imaginary error function defined as @xmath91 .
we can check the uv limit of this function . using the limit for the imaginary error function @xmath92
we find @xmath93 this value is finite and therefore uv divergence does not occur .
g. m. hossain , v. husain and s. s. seahra , arxiv:0906.2798 [ astro-ph.co ] . g. m. hossain , v. husain and s. s. seahra , arxiv:0906.4046 [ hep - th ] . c. kiefer and d. polarski , arxiv:0810.0087 [ astro - ph ] .
d. sudarsky and a. de unanue , j. phys .
* 174 * ( 2009 ) 012059 [ arxiv:0901.2884 [ gr - qc ] ] .
n. duechting , phys .
d * 70 * ( 2004 ) 064015 [ arxiv : astro - ph/0406260 ] .
l. h. ford , arxiv : gr - qc/9707062 .
j. mielczarek , arxiv:0908.4329 [ gr - qc ] .
a. lewis , a. challinor and a. lasenby , astrophys . j. * 538 * ( 2000 ) 473 [ arxiv : astro - ph/9911177 ] .
a. lewis and s. bridle , phys . rev .
d * 66 * ( 2002 ) 103511 [ arxiv : astro - ph/0205436 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , astrophys . j. suppl .
* 180 * ( 2009 ) 330 [ arxiv:0803.0547 [ astro - ph ] ] .
[ planck collaboration ] , arxiv : astro - ph/0604069 . | in this paper we investigate power spectrum of a smoothed scalar field . the smoothing leads to the regularisation of the uv divergences and can be related with the internal structure of the considered field or the space itself .
we apply procedure of smoothing to the quantum fluctuations generated during the phase of cosmic inflation .
we study whether this effect can be probed observationally and conclude that the modifications of the power spectrum due to the smoothing on the planck scale are negligible and far beyond the observational abilities .
subsequently we investigate whether smoothing in any other form can be probed observationally .
we introduce phenomenological smoothing factor @xmath0 to the inflationary spectrum and investigate its effects on the spectrum of cmb anisotropies and polarisation .
we show that smoothing can lead to suppression of high multipoles in the spectrum of the cmb .
based on five years observations of wmap satellite we indicate that the present scale of high multipoles suppression is constrained by @xmath1 mpc ( 95@xmath2 cl ) .
this corresponds to the constraint @xmath3 m at the end of inflation . despite this value
is far above the planck scale , the other processes of smoothing can be possibly studied with this constraint , e.g. diffusion or decoherence of primordial perturbations . |
accurate shape measurements of faint , small galaxies are crucial for certain applications , most notably the measurement of weak gravitational lensing . quantifying the slight distortion of background galaxies by foreground matter relies on detecting small but coherent changes in the shapes of many galaxies ( see refregier 2003 for a recent review ) . to extract the lensing signal ,
it is crucial to remove instrumental effects from galaxies measured shapes . on the _ hubble space telescope _ ( hst ) , these include : * convolution of an image with the telescope s point spread function ( psf ) .
* geometric distortion of an image .
this is particularly large in the _ advanced camera for surveys _ ( acs ) because of its location off hst s optical axis . *
trailing of faint objects in the ccd readout direction due to degraded charge transfer efficiency ( cte ) . in this proceeding , we describe a method to model and correct for the telescope s temporally and spatially varying psf .
the geometric distortion has already been shown to be successfully removed during image processing by ( koekemoer et al .
2002 ) . removing
the distortion does change the psf , and we present recommendations to minimize stochastic changes introduced during the repixellization stage of image processing .
the effect of continuing cte degradation on galaxy shapes is only becoming apparent as the acs spends longer in orbit , and is not yet completely understood .
that is therefore beyond the scope of this proceeding .
a separate method to remove cte effects will be presented in rhodes et al .
( 2006 ) , and the application of all these corrections in a weak lensing analysis will be presented in massey et al .
other branches of astronomy , including stellar photometry in crowded fields , the study of agn , and proper motions also require detailed knowledge of the psf and will benefit from the models we describe here . in weak lensing , to deconvolve galaxy shapes from the psf , we must accurately know the shape of the psf at the position of the object and at the time of the observation for example , see rhodes , refregier & groth ( 2000 ) for a description of the method we use on the cosmic evolution survey ( cosmos ; scoville et al .
2006 ) images we use to test the psf models we describe in this paper .
if the hst psf were stable over time , it would be straightforward to build a catalog of stellar images across the entire field of view .
however , thermal fluctuations in hst that change its effective focus ( the distance between the primary and secondary mirrors ) lead to temporal psf variations . as an example
, figure [ fig : cosmospsfs ] shows the psf pattern in two sets of cosmos images .
the left hand panel shows stars from images taken when the telescope was near optimal focus , and the right hand panel shows stars observed when the telescope was several microns below optimal focus .
each tick mark in the figure represents the ellipticity of one star , measured using the standard weak lensing definition , @xmath1 where star s weighted second order moments @xmath2 involve summations over all pixels .
@xmath3 is the intensity of a pixel , @xmath4 is some weighting function ( in our case a gaussian with a width of about the fwhm of the psf ) , and @xmath5 is the distance of a pixel from the centroid of the object .
it is apparent from figure [ fig : cosmospsfs ] that changes in the psf over time are sufficient for a temporally stable acs psf model to be inadequate in demanding applications , when using data collected over a period of more than a few days .
other effects , including cte , introduce additional variation on longer time scales .
this proceeding is organized as follows . in [ tinytim ]
we introduce the software package that we used for psf modeling , and discuss modifications that we have made . in [ multidrizzle ] we discuss and how to minimize the aliasing of point sources that occurs during distortion removal . in [ models ]
we show how we have used our models to quantify the temporal variation of the acs psf , and describe how the same models can be used to correct for it . in [ conclusions ] we draw conclusions and outline a plan to release our psf models .
we have adapted version 6.3 of the software package ( krist & hook 2004 ) to create simulated images of stars .
creates fits images containing one or more stars that include the effects of diffraction , geometric distortion , and charge diffusion within the ccds . by default
the images appear as they would in raw acs data ; they are highly distorted and have a pixel scale of 0.05 arcseconds per pixel .
we have written an idl wrapper to undo the distortion , resample the images , and combine adjacent psfs to mimic the effects of dithering .
the wrapper can also run multiple times and create a grid of psf models across the whole acs field of view .
we insert our artificial stars into blank images with the same dimensions and fits structure as real acs data , thereby manufacturing arbitrarily dense starfields .
this basic pipeline calculates a diffraction pattern ( spot diagram ) , distorts it , and adds charge diffusion ; all three effects usually depend upon the position of the star in the acs field of view . we have made two versions of artificial starfields with important changes to this basic pipeline .
the deviations from this default are : * in order to examine the effects of the distortion removal process , we have created a version of our starfields where each star has an identical diffraction pattern and charge diffusion , but a geometric distortion determined by the location of the psf within the acs field of view . once the geometric distortion is removed ( for example by running the field through )
, these stars should all appear identical . * in order to correct data , we have created a second version of our starfields that do not contain the effects of geometric distortion at all , instead modeling stars as they would appear after a perfect removal of geometric distortion .
conversion between distorted and non - distorted frames , which is necessary to simulate charge diffusion in the raw ccd , was performed using very highly oversampled images .
this avoids stochastic aliasing of the psf ( see [ multidrizzle ] ) , and minimizes noise in the psf models .
we describe the application of these simulations in the following two sections .
is used to combine dithered exposures , remove cosmic rays and bad pixels , and eliminate the large geometric distortion in acs wfc images ( koekemoer et al .
. however , the transformation of pixels from the distorted input image to the undistorted output plane can introduce significant `` aliasing '' of pixels if the output pixel scale is comparable to the input scale .
when transforming a single input image to the output plane , point sources can be enlarged , and their ellipticities changed by several percent , depending upon their sub - pixel position .
this is one of the fundamental reasons why dithering is recommended for observations , since the source is at a different sub - pixel position in different exposures , hence the effects are mitigated to some extent when all the exposures are combined .
however , the remaining effect in combined images is still quite sufficient to prevent the measurement of small , faint galaxies at the precision required for weak lensing analysis .
of course , such pixellization effects are unavoidable during the initial exposure , when the detector discretely samples an image .
however , it is clearly desirable to minimize related effects during data reduction .
the effect on each individual object depends on how the input and output pixel grids line up .
indeed , this can be mitigated by using a finer grid of output pixels ( _ e.g. _ lombardi et al .
the reduction in pixel scale ( which will cause a corresponding increase in computer overheads ) can be performed in conjunction with simultaneously `` shrinking '' the area of the input pixels that contains the signal , by making use of the ` pixfrac ` parameter and convolution kernel .
we have run a series of tests on the simulated psf grids described in [ tinytim ] to determine the optimal values of the parameters specifically for weak lensing science . to this end
, we first produced a grid of stars that ought to look identical after the removal of geometric distortion .
figure [ fig : aliasing ] shows the `` aliasing '' produced when the distortion is removed from a single image .
we have also created a series of four dithered input images ( with the linear dither pattern used for the cosmos survey ) ; the scatter in the ellipticities of the output stars is then approximately half as big .
this confirms the idea that the repixellization adds stochastic noise to the ellipticity when the four sub - pixel positions are uncorrelated . for weak lensing purposes , this noise is still substantial . with enough dithered input images ,
the scatter of ellipticities could be reduced further , but this is not feasible for most observations .
we then ran a series of tests using on the same input image but with a range of output pixel scales , convolution kernels , and values of ` pixfrac ` .
we then measured the scatter in the ellipticity values in the output images .
the smaller that scatter , the more accurately the psf is represented .
we found the results were not strongly dependent on the choice of`pixfrac ` and settled on a value of 0.8 for that parameter .
we show in figure [ fig : multidrizzle ] that psf stability is improved dramatically by reducing the output pixel scale from 0.05 arcseconds ( the default ) to 0.03 arcseconds .
there is only a very slight gain in going to smaller output pixel sizes and the storage requirements rapidly become problematic .
the gain in going to smaller pixel scales is more stable with a gaussian kernel for than with the default square kernel .
therefore , for weak lensing work , we recommend an output pixel scale of 0.03 arcseconds , ` pixfrac`=0.8 , and a gaussian kernel in order to best preserve the psf during this stage of image reduction .
we note that , despite its clear advantages for weak lensing studies , the gaussian kernel does have some general drawbacks , such as the introduction of more correlated noise which may not be desirable for other types of science where minimization of correlated noise is important .
we can measure hst s effective focus at the time of an observation using @xmath0 fairly bright stars in the field .
we first created dense grids of artificial stars across the acs field of view . by changing the separation of the primary and secondary mirrors in s raytracing model
, these models were made at successive displacements of the focus from nominal , from @xmath6 m to @xmath7 m in @xmath8 m increments .
these are reasonable bounds on the maximum extent of physical variations in hst .
the stars are created without geometric distortion , to avoid the noise that would have been introduced had it been necessary to carry out geometric transformations on the stellar fields .
we then compare the @xmath0 bright stars in each cosmos field to the psf grids at each focus value . minimizing the difference in ellipticity between the models and the data finds the best fit value of telescope focus for that particular field .
tests on observations of dense stellar fields that contain many suitable stars show that this procedure is repeatable with an rms error @xmath9 m , when using ten different stars repeatedly selected at random .
+ figure [ fig : focus ] shows our estimation of hst s focus in microns away from nominal during several months in cycles 12 and 13 , using a uniform set of cosmos images .
hst was not manually refocussed during this time , but the apparent focus still oscillates . at times , the oscillations seem periodic , but there are also sharp jumps and more erratic behavior .
the random component probably depends in a complex fashion upon the orientation of hst with respect to the sun and the earth during preceding exposures , and we do not believe that it can be easily predicted in advance .
note that the uncertainty on the focus value during any single pointing is quite large ; more so than the tests on dense stellar fields would suggest .
a major component of this error is undoubtedly the @xmath10 m thermal fluctuations in focus that hst experiences during each orbit due to `` breathing '' .
the cosmos images are all taken with a total exposure time of one orbit , and the apparent focus therefore represents the integral of a gradually changing psf .
we have not been able to investigate focus changes on short time periods and , given this behavior , it is even more curious that long term patterns are so clearly present in figure [ fig : focus ] .
figure [ fig : focuspattern ] shows the focus values determined for all of the cosmos fields .
the obvious clustering of focus values in that plot is due to the observing strategy used for cosmos , in which data was typically taken in chunks of a few fields at a time .
adjacent fields are likely to have been taken at similar times , and therefore tend to have a similar focus values .
figure [ fig : comparison ] shows the models at focus @xmath11 m and all the stars from the cosmos fields with a best - fit focus value of @xmath11 m .
the cosmos stars have been averaged in a spatial grid of approximately 600 @xmath12 600 0.03 arcsecond pixels .
there is good agreement between the models and the stars over most of the field .
the agreement is not as good in the boxed area near the center of the field .
we believe this is due to a degradation of the cte of the acs ccds .
this degradation causes trailing of low flux objects in the readout direction ( the @xmath13 direction ) .
the effect is most pronounced the further away the object is from the readout registers at the bottom and top of the field ( mutchler & sirianni 2005 ; riess & mack 2004 ) .
this causes the objects to be elongated vertically .
thus , fainter cosmos stars appear more elongated in the @xmath13 direction at the center of the field than the models , which do not include the effects of cte .
note that this does not affect the estimation of focus positions , because the bright stars matched to our psf models are less affected by cte than the faint sources .
we are currently exploring ways to correct for cte in all objects , and will publish the results in rhodes et al.(2006 ) .
we have shown that can produce model psfs that are very close to those observed in real data ( for example the cosmos 2-square degree survey ) .
this required some modifications to the code , most importantly adding the ability to mimic the distortion correction and dithering normally implemented via , and to produce grids of psfs across the entire acs wfc field of view .
we used model stars to find the best values for and found that using a gaussian kernel , ` pixfrac ` @xmath14 , and an output pixel scale of 0.03 arcseconds greatly reduced the `` aliasing '' of point sources introduced during repixellization . discrepancies between our models and the cosmos data can be attributed largely to a degradation in the acs cte since launch .
we are currently studying this problem and will present a complete psf solution including how to correct for cte in rhodes et al .
we plan to correct science images for cte on a pixel - by - pixel basis , like the bristow code developed for stis ( bristow et al./ 2002 ) , moving charge back to where it belongs , rather than including the effects of cte in our model psf ( like rhodes et al .
thus , the model psfs we present here are the ones we plan to use in our weak lensing analysis . at the time of press
, we have thoroughly tested psf models in only the f814w filter . however , our idl routines preserve s ability to create psfs in other filters , and for sources of any colors . our routines are therefore easily adaptable to other data sets .
the whole method is intentionally designed to be as adaptable as possible for many methods .
the desire to know the psf at any arbitrary position on the sky is far from unique to weak lensing .
but even in lensing , advanced methods like shapelets will , in the near future , be able to take advantage of more detailed information about the psf shape than it is reasonable to expect from interpolation between a few stars ( massey & refregier 2005 ; refregier & bacon 2004 ) .
this is even more exaggerated when considering higher order shape parameters , with an intrinsically lower signal to noise .
the creation of noise - free , oversampled stars at any position on an image allows such analysis in any acs data . in the near future
, we plan to release our psf models to the community along with the wrapper we have written for which will allow users to create psf models in different filters and at user - defined positions in the acs field of view .
interested readers are advised to contact the authors for these resources .
we would like to thank catherine heymans for sharing her knowledge of the acs psf .
we are grateful to john krist for guiding our poor lost souls through the underbelly of .
great thanks go to andy fruchter and marco lombardi for useful discussions about .
adam riess and marco sirianni provided expert knowledge about cte effects .
richard ellis and alexandre refregier have engaged us in many useful and interesting discussions about the cosmos field .
we are also pleased to acknowledge the continuing support of nick scoville , patrick shopbell and the whole cosmos team in obtaining and analyzing the cosmos images that were used to test our psf models .
bristow et al . 2002 , _ modelling charge transfer on the stis ccd _ , in the 2002 _ hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system _ , eds arribas , koekemoer , & whitmore , stsci koekemoer , a. m. , fruchter , a. , hook , r. & hack , w. , 2002 , ` multidrizzle : an integrated pyraf script for registering , cleaning and combining images , ' in the 2002 _ hst calibration workshop : hubble after the installation of the acs and the nicmos cooling system _ , eds arribas , koekemoer , & whitmore , stsci , p. 337 krist , j. & hook , r. , 2004 , ` the tinytim user s guide ' , stsci , 339 lombardi , m. et al . , 2005 ,
apj , 623 , 42 massey r. & refregier , a. 2005 , mnras 363 , 197 massey r. et al .
2006 , in preparation mutchler , m. & sirianni , m. , 2005 , _ instrument science report _
acs 2005 - 003 , + ( baltimore : stsci ) , available through http://www.stsci.edu/hst/acs refregier , a. , 2003 , ara&a , 41 , 645 refregier , a. & bacon , d. , 2002 , mnras , 338 , 48 rhodes , j. et al .
2006 , in preparation rhodes et al .
2004 , apj , 605 , 29 rhodes , j. , refregier , a. & groth , e. , 2000 , apj , 536 , 79 riess , a. & mack , j. , 2004 , isr acs 2004 - 006,stsci scoville , n. et al .
2006 , in preparation | the ability to accurately measure the shapes of faint objects in images taken with the _ advanced camera for surveys _ ( acs ) on the _ hubble space telescope _ ( hst ) depends upon detailed knowledge of the point spread function ( psf ) .
we show that thermal fluctuations cause the psf of the acs wide field camera ( wfc ) to vary over time .
we describe a modified version of the psf modeling software to create artificial grids of stars across the acs field of view at a range of telescope focus values .
these models closely resemble the stars in real acs images . using @xmath0 bright stars in a real image ,
we have been able to measure hst s apparent focus at the time of the exposure .
can then be used to model the psf at any position on the acs field of view .
this obviates the need for images of dense stellar fields at different focus values , or interpolation between the few observed stars .
we show that residual differences between our models and real data are likely due to the effects of charge transfer efficiency ( cte ) degradation .
furthermore , we discuss stochastic noise that is added to the shape of point sources when distortion is removed , and we present parameters that are optimal for weak lensing science . specifically , we find that reducing the output pixel scale and choosing a gaussian kernel significantly stabilizes the resulting psf after image combination , while still eliminating cosmic rays / bad pixels , and correcting the large geometric distortion in the acs .
we discuss future plans , which include more detailed study of the effects of cte degradation on object shapes and releasing our models to the astronomical community . |
quantum chromodynamics ( qcd ) calculation predicted an exotic quark - gluon matter in qcd phase diagram @xcite may be created in the early stage of heavy ion collisions at ultra - relativistic energy @xcite .
for mapping the qcd phase diagram and locating qcd phase boundary and critical point @xcite , one needs to find a way to vary temperature ( @xmath2 ) as well as chemical potential ( @xmath3 ) . the na61 collaboration and na49-future collaboration @xcite suggested that it can be achieved via a systematic energy ( e ) and system - size ( a ) ( @xmath4 ) scan .
jet quenching phenomenon has been theoretically predicted @xcite and experimentally observed @xcite .
so far , dihadron azimuthal correlations have been demonstrated as a good method to reconstruct particle and energy distribution induced by the quenched jet . in experiment ,
a double - peak structure was found on the away side of dihadron azimuthal correlation functions @xcite and the indication of conical emission of charged hadrons was reported by the star collaboration @xcite .
the centrality and transverse momentum dependences of double - peak structure of away - side correlation functions were experimentally investigated by rhic - bnl @xcite and theoretically simulated in ref .
@xcite .
these interesting phenomena have attracted some theorists to explain the physical mechanisms for the origin of the double - peak structure .
these mechanisms include a cherenkov - like gluon radiation model @xcite , medium - induced gluon bremsstrahlung radiation @xcite , shock wave model in hydrodynamic equations @xcite , waking the colored plasma and sonic mach cones @xcite , sonic booms and diffusion wakes in thermal gauge - string duality @xcite , jet deflection @xcite and strong parton cascade mechanism and so on @xcite .
recently , gyulassy and his collaborators suggest that the conical emission can stem from universal flow - driven mechanism @xcite . in ref .
@xcite , the shock wave phenomena are discussed in viscous fluid dynamics and kinetic theory .
renk and neufeld @xcite presented their systematic study of dependence of the mach cone signal on the energy deposition into the medium in linearized hydrodynamics .
the path - length effect on the correlations relative to the reaction plane are studied , respectively , by jia et al . in a simple model @xcite and by ma et al . in ampt model @xcite . in the shock wave model
@xcite , the emission angle relative to jet is calculated to be about 1.23 rad for qgp , 1.11 rad for hadronic gas and zero for mixed phase .
the gluon radiation mechanism for the double - peak structure @xcite suggests that the more energetic the jet , the smaller the emission angle . in ref .
@xcite , it is suggested that the suppression or even disappearance of mach cone at the qcd critical point due to the attenuation of the sound mode .
while it remains unclear what the main mechanism for the emergence of the double peak structure is , all the experimental and theoretical works suggest that it should depend on the nature of the hot and dense matter created in the collisions @xcite . in this paper , we study the properties of hot - dense matter produced by different system size by investigating the system - size dependence of dihadron azimuthal correlations . in this paper , we present participant path - length , defined as @xmath5 = 2@xmath6/@xmath7 @xcite ( @xmath6 and @xmath7 are the number of binary collision and participants , respectively ) , dependence of the double - peak structure of away - side correlation function in the most central collisions ( 0 - 10% ) .
the structure of away - side correlation function changes near @xmath8ca + @xmath8ca collisions at @xmath9 = 200 gev in central collisions ( 0 - 10% ) .
the results show obvious degree of freedom dependence in the system with a partonic phase or with a pure hadron gas @xcite , which implies information of the onset of deconfinement .
in this work , a multi - phase transport model ( ampt ) @xcite , which is a hybrid dynamic model , is employed to study dihadron azimuthal correlations .
it includes four main components to describe the physics in relativistic heavy ion collisions : 1 ) the initial conditions from hijing model @xcite , 2 ) partonic interactions modeled by a parton cascade model ( zpc ) @xcite , 3 ) hadronization ( discussed later ) , 4 ) hadronic rescattering simulated by a relativistic transport ( art ) model @xcite .
excited strings from hijing are melted into partons in the ampt version with string melting mechanism @xcite ( abbr . _ the melt ampt version _ ) and a simple quark coalescence model is used to combine the partons into hadrons . in the default version of ampt model @xcite ( abbr .
_ the default ampt version _ ) , minijet partons are recombined with their parent strings when they stop interactions and the resulting strings are converted to hadrons via the lund string fragmentation model @xcite . the melt ampt version undergoes a partonic phase , while a pure hadron gas is in the default ampt version .
details of the ampt model can be found in a review paper @xcite and previous works @xcite .
the analysis method for dihadron azimuthal correlations is similar to that used in previous experiments @xcite , which describes the azimuthal correlation between a high @xmath10 particle ( trigger particle ) and low @xmath10 particles ( associated particles ) .
the raw signal can be obtained by accumulating pairs of trigger and associated particles into @xmath11 distributions in the same event .
the background which is expected mainly from elliptic flow is simulated by mixing event method @xcite . to reconstruct the background , we accumulate pairs of one fake trigger particle ( high @xmath12 ) in one event and another fake associated particle ( low @xmath12 ) in another event to obtain the @xmath13 distribution as the corresponding background , the centralities of the above two events are requested very closed .
then the background is subtracted from raw signal by using a zero yield at minimum " ( zyam ) assumption as that used in experimental analysis @xcite ( see our detailed analysis in ref .
@xcite ) . recently ,
wang et al .
@xcite discussed the background in the correlations and presented an analytical form for flow background to jet - like azimuthal correlations in a cluster approach . and it is suggested that the collision geometry fluctuations and triangular flow should be taken into account in the correlation analysis @xcite . but those go beyond our discussion in this paper .
the participant path - length , defined as @xmath5 = 2@xmath6/ @xmath7 @xcite , can describe degree of multiple collisions between participants in the early stage of heavy ion collisions and characterize the size of the reaction zone .
the @xmath14 represents average collision number of partons in the melt ampt version .
the values of @xmath5 and @xmath14 significantly increase with varying collision system ( @xmath15 ) from `` small '' size to `` large '' size at @xmath9 = 200 gev in the most central collisions ( 0 - 10% ) as shown in table [ talble - samecentral ] . from this table
, we can see the multiple collisions are more frequent in `` large '' size collision system than in `` small '' size one .
the values of @xmath7 , @xmath6 and @xmath5 are comparable to those from the glauber model @xcite for both @xmath16cu + @xmath16cu and @xmath1au + @xmath1au collisions .
when the mixed background which mainly stems from elliptic flow is subtracted from the raw dihadron correlation signal taken in the same events , we can get the correlation function .
[ corr - fun ] shows dihadron azimuthal correlation functions of different collision systems in the most central ( 0 - 10% ) collisions at @xmath9 = 200 gev .
the correlation functions are calculated in the kinetic windows , 1 @xmath17 3 gev/@xmath18 as well as 2.5 @xmath19 6 gev/@xmath18 and @xmath20 1 .
it shows that the structure of away - side correlation function changes from the gaussian - like distribution to double - peak structure near @xmath8ca + @xmath8ca collisions with varying collision system from @xmath0n + @xmath0n to @xmath1au + @xmath1au collisions in the melt ampt version . in this figure ,
the amplitude of the correlation function becomes higher with the increasing of collision system size .
the associated particles in the melt ampt version are more abundant than those in the default ampt version . .
@xmath21 , @xmath22 , @xmath23 = @xmath24 , @xmath14 in different collision system at @xmath9 = 200 gev for centrality 0 - 10 % , the value in blanket is taken from glauber model @xcite . [
cols="^,^,^,^,^,^,^,^",options="header " , ] [ talble - samecentral ] the default ampt version is used to compare the properties of the double - peak structure in partonic phase and in hadron gas . for investigating the properties of collision system - size dependences of away - side dihadron azimuthal correlations , we extract the associated particle yield @xmath25 , splitting parameter ( @xmath26 ) ( half distance between double peaks on the away side ) and root mean square width ( @xmath27 ) of away - side associated particles , which will be discussed in the following sections , respectively . the @xmath23 dependence of @xmath25 is shown in fig .
[ yield ] from the melt/ default ampt version , respectively .
it presents a significant increasing trend of @xmath25 with varying the collision system from @xmath0n + @xmath0n to @xmath1au + @xmath1au collisions in the most central collisions ( 0 - 10% ) at @xmath9 = 200 gev in the melt ampt version .
the default ampt version , with a hadronic gas , does not result in a rapid increasing dependence trend . in the melt ampt version ,
the dependence trend indicates the jet correlation information can be inherited by and transmitted to more particles in a partonic phase than in a hadronic gas , especially in `` large '' size collision system .
furthermore it implies that the interaction strength in `` large '' size collision system is more significant than that in `` small '' size collision system , and while strong parton cascade plays a dominant role to push more away - side associated particles in the melt ampt version .
it is interesting that the increasing slope of @xmath25 vs @xmath23 from the linear fitting in the melt ampt version is quicker above @xmath8ca + @xmath8ca collision system , where clear double - peak structure emerges , than that in small systems .
this property indicates the double - peak ( mach - like ) structure can enhance associated particles yields of jet correlations partially .
root mean square width ( @xmath27 ) of away - side correlation function is defined as @xmath28 where @xmath29 is the mean @xmath13 of away - side correlation function and it approximates to @xmath30 .
@xmath31 can describe the diffusion degree of the associated particles relative to the direction of back jet .
the @xmath23 dependences of @xmath27 in the melt/ default ampt version are shown in fig .
[ rms ] , respectively .
@xmath31 from the melt ampt version are consistent with phenix data @xcite for cu + cu and au + au collisions .
@xmath32 increases from @xmath0n + @xmath0n collisions to @xmath1au + @xmath1au collisions in the melt ampt version as well as the default ampt version , but the increasing trend is not so quick in the later , especially for systems larger than @xmath8ca + @xmath8ca .
the increasing trend of @xmath31 shows broadening of away - side correlation functions with increasing size of collision system .
it indicates that the jet correlation information can reach faraway relative to direction of jet with changing the collision system from `` small '' size one to `` large '' size in a partonic phase .
it is remarkable that the increasing trend of @xmath27 from the linear fitting in the melt ampt version shows two different slope after and before @xmath8ca + @xmath8ca collision system , where clear double - peak structure emerges .
the double peak ( mach - like ) phenomenon from quenched jet can enhance the diffusion degree of the associated particles relative to back jet in the `` large '' size collision system .
this suggests that the back jet modification in the medium created in heavy ion collisions with a partonic phase is more distinct in the `` large '' size collision system than in `` small '' size one .
the splitting parameter ( @xmath26 ) is another useful observable to characterize the structure of the double - peak of away - side correlation function , and further discloses essential of jet modification .
the @xmath23 dependence of splitting parameters ( @xmath26 ) in the melt / default ampt version are shown in fig .
[ splitd ] for the most central collisions at @xmath9 = 200 gev . for lighter systems from @xmath0n + @xmath0n to @xmath33al + @xmath33al collisions , the splitting parameter is not extracted since there is no observable double - peak structure of away - side correlation functions . in both simulation cases , the splitting parameter ( @xmath26 ) increases from `` small '' size collision system to `` large '' size one , which indicates that there exists stronger jet - medium interaction in `` large '' system .
it is also remarkable that the splitting parameter ( @xmath26 ) is larger in the melt ampt version than that in the default ampt version .
the melt ampt results are comparable to phenix data @xcite for cu + cu and au + au collisions due to effect of parton cascade in the melt ampt version @xcite .
the parton interaction cross section is taken to be 10 mb in this work , which is also reasonable for reproducing elliptic flow as well as dihadron azimuthal correlations in the melt ampt version @xcite .
this calculation reflects that it is necessary to pass a strong partonic stage to reproduce large enough double - peak structure as the experimental data demonstrate . from these results
, it can be concluded that a considerable `` large '' collision system is necessary and the strong parton cascade is essential for the formation of the double - peak structure of away - side correlation function .
an onset of the observable double - peak structure occurs in the mass range of @xmath8ca + @xmath8ca collision .
this phenomenon indicates that the correlation is sensitive to @xmath23 and @xmath14 , i.e. the correlation depends on the collision system size and the violent degree of the partonic interaction in a partonic phase .
different results obtained in a partonic phase and a pure hadron gas imply the double - peak structure and jet modification are sensitive to the effective degree of freedom of the dense medium created in relativistic heavy ion collisions , which can give us some hints of the onset of deconfinement in the system - size viewpoint .
in summary , the present work discusses the collision system - size dependence of dihadron azimuthal correlations at @xmath9 = 200 gev by a multi - phase transport model .
the yields of associated particles , width of away - side correlation functions and splitting parameter show significant system - size dependence .
the away - side correlation function becomes more broadening with the increasing of collision system size and displays the onset of double - peak structure near @xmath8ca + @xmath8ca collisions after the system undergoes a strong partonic transport stage .
these results also present the degree of freedom dependence , which might be related to onset of deconfinement .
we would remark that these observations do not assume any dynamical mechanism for the formation of the double hump structure .
the ampt model would include collective mach - like effects associated to the particles , but it would also include eccentricity fluctuating effects and triangle flow components .
the result is , in this sense , robust .
this work was supported in part by the national natural science foundation of china under grant no .
11035009 , 11047116 , 10905085 , and 10875159 , 10705043 and 10705044 , and the shanghai development foundation for science and technology under contract no .
09jc1416800 , and the knowledge innovation project of the chinese academy of sciences under grant no .
kjcx2-ew - n01 , y155017011 and o95501p0 - 11 and the project - sponsored by srf for rocs , sem .
o819011012 .
i. arsene et al .
( brahms collaboration ) , nucl .
phys . a * 757 * , 1 ( 1990 ) ; b. b. back et al .
( phobos collaboration ) , nucl .
phys . a * 757 * , 28 ( 2005 ) ; j. adams et al . ( star collaboration ) , nucl .
phys . a * 757 * , 102 ( 2005 ) ; s. s. adcox et al . (
phenix collaboration ) , nucl .
a * 757 * , 184 ( 2005 ) . et al .
( star collaboration ) , phys .
lett . * 91 * , 072304 ( 2003 ) ; j. adams et al .
( star collaboration ) , phys .
c * 73 * , 064907(2006 ) ; j. adams et al .
( star collaboration ) , j. phys .
g * 34 * , 799 ( 2007 ) ; s. s. adler et al .
( phenix collaboration ) , phys .
lett . * 91 * , 072301(2003 ) ; s. s. adler et al .
( phenix collaboration ) , phys .
c * 73 * , 054903(2006 ) ; s. s. adler et al .
( phenix collaboration ) , phys .
c * 71 * , 051902(2005 ) . | system - size dependence of dihadron azimuthal correlations in ultra - relativistic heavy ion collision is simulated by a multi - phase transport model .
the structure of correlation functions and yields of associated particles show clear participant path - length dependences in collision systems with a partonic phase .
the splitting parameter and root - mean - square width of away - side correlation functions increase with collision system size from @xmath0n+@xmath0n to @xmath1au+@xmath1au collisions .
the double - peak structure of away - side correlation functions can only be formed in sufficient `` large '' collision systems under partonic phase .
the contrast between the results with partonic phase and with hadron gas could suggest some hints to study onset of deconfinment .
di - hadron azimuthal correlation , splitting parameter , partonic transport |
when a ferromagnetic material is subject to a time dependent oscillating magnetic field , the system may not respond to the external magnetic field instantaneously , and this situation leads to some interesting phenomena .
namely , dynamic phase transitions ( dpts ) and dynamic hysteresis behavior originate due to a competition between time scales of the relaxation time of the system and oscillation period of the external magnetic field . on the other hand , at high temperatures and for the high amplitudes of the periodic magnetic field ,
the system is able to follow the external field with some delay while this is not the case for low temperatures and small magnetic field amplitudes .
this spontaneous symmetry breaking indicates the presence of a dpt @xcite which shows itself in the dynamic order parameter ( dop ) which is defined as the time average of the magnetization over a full period of the oscillating field . according to early experimental studies , dating back to last century
, there exists an empirical law for the hysteresis loop areal scaling @xcite . up to now ,
a fairly well numerous theoretical studies have been dedicated to dpt and hysteresis properties of kinetic ising models by using various methods such as monte carlo ( mc ) simulations @xcite , effective - field theory ( eft ) @xcite and mean - field theory ( mft ) @xcite . apart from these , there exist a detailed geometrical description @xcite , and also an hysteresis criterion based on rate competition between the critical curvature and the potential - barrier height @xcite . by employing standard metropolis mc algorithm with periodic boundary conditions , the pure kinetic ising model in a two dimensional square lattice
has been simulated by lo and pelcovits @xcite , and they found some evidences of a dpt .
they also investigated the behavior of dynamic hysteresis for varying hamiltonian parameters , however they did not intend to make any classification for dynamic hysteresis behavior .
afterwards , by making use of mc simulations and mft , acharyya and chakrabarti @xcite have presented a comprehensive investigation of the pure kinetic ising model and they showed that the hysteresis loops are asymmetric in the dynamically ordered phase while in a dynamic disordered phase , they are symmetric around the origin in a magnetization versus field amplitude plane . in ref .
@xcite , the attention has been focused on the behavior of dynamic hysteresis and dynamic loop area ( dla ) of a pure kinetic ising model by using both mft and mc calculations .
in addition , the frequency variations of the coercive field have been analyzed below the zero field critical temperatures .
it has been concluded that the numerical results for the dynamic coercive field show a power law frequency variation for both cases .
however , the coercive field calculated by mft becomes frequency independent in the low frequency regime .
differences between the two methods ( mft and mc ) have been explained by using a kind of landau - type double well free energy mechanism .
moreover , there exist many previously published studies such as the domain wall motion and nucleation theory @xcite regarding the hysteresis loop mechanism @xcite . on the experimental picture
, it can be said that a great many of experimental studies have been devoted to dpts and hysteresis behaviors of co films on a cu ( 001 ) surface @xcite , [ co(4@xmath0)/pt(7@xmath0 ) ] multi - layer system with strong perpendicular anisotropy @xcite , thin polycrystalline ni@xmath1fe@xmath2 films @xcite , ultrathin ferromagnetic fe / au(001 ) films @xcite , epitaxial fe / gaas(001 ) thin films @xcite , epitaxial single ferromagnetic fcc nife(001 ) , fcc co(001 ) , and fcc nife / cu / co(001 ) layers @xcite , fe@xmath3zn@xmath4f@xmath5 @xcite , finemet thin films @xcite , and epitaxial fe / gaas(001 ) and fe / inas(001 ) ultrathin films @xcite .
after some detailed experimental investigations , it has been observed that experimental non - equilibrium dynamics of considered real magnetic systems strongly resemble the dynamic behavior predicted from theoretical calculations of a kinetic ising model .
it is clear from these works that there exists a strong evidence of qualitative consistency between theoretical and experimental studies . besides
, the lattice models including impurities and defects have attracted considerable attention , because they are very useful to investigate the behavior of disordered systems in nature . for this purpose ,
more effective models have been introduced to examine the influences of disorder on the thermal and magnetic properties of real magnetic materials .
for example , the effects of impurities on driving - rate - dependent energy loss in a ferromagnet under the time dependent magnetic field have been analyzed by zheng and li @xcite by using several well defined models within the frameworks of mft and mc , and they found using mft that , the hysteresis loop area is a power law function of the linear driving rate as @xmath6 , where , @xmath7 and @xmath8 are the static hysteresis loop area , the linear driving rate and scaling exponent of the system , respectively .
very recently , the quenched site and bond diluted kinetic ising models under the influence of a time dependent oscillating magnetic field have been analyzed by making use of eft @xcite on a two dimensional honeycomb lattice . in ref .
@xcite , the global phase diagrams including the reentrant phase transitions are presented by the authors for site diluted kinetic ising model , and they showed that the coexistence regions disappear for sufficiently weak dilution of lattice sites .
following the same methodology , the authors have concentrated on the influences of quenched bond dilution process on the dynamic behavior of the system .
after some detailed analysis , it has been found that the impurities in a bond diluted kinetic ising model give rise to a number of interesting and unusual phenomena such as reentrant phenomena and the impurities to have a tendency to destruct the first - order transitions and the dynamic tricritical point @xcite .
furthermore , it has also been shown that dynamically ordered phase regions get expanded with decreasing amplitude which is more evident at low frequencies .
very recently , frequency dependencies of the dynamic phase boundary ( dpb ) and magnetic hysteresis of ising model in an oscillating magnetic field have been examined by punya _
_ @xcite . based on mft
, they have represented the phase diagram in a magnetic field versus temperature plane for varying frequencies , in order to show the influence of the oscillation frequency of the external field on the hysteretic response of the system . according to their calculations
, it has been found that the frequency dispersion of the dla , the remanence and coercivity can be categorized into three distinct types for a fixed temperature .
although the detailed investigations and classifications of the hysteretic behaviors of pure kinetic ising model have been done within the mft @xcite , as far as we know , the dynamic hysteretic behavior within the eft has not been studied for disordered kinetic ising models .
actually , the main motivation for our work comes from the recent analysis of ref .
@xcite for the investigation of hysteresis behavior and dynamic phase diagrams of kinetic ising model .
therefore in this work , we intend to probe the effects of the quenched bond dilution process on the dynamic hysteresis behavior of kinetic ising model in the presence of a time - dependent oscillating external magnetic field by using the eft with correlations based on the exact van der waerden identity for a spin-1/2 system .
present method is quite superior to mft , since thermal fluctuations are partly considered in eft .
the outline of the paper can be summarized as follows : the dynamic equation of motion , dop and dla of the quenched bond diluted kinetic ising model have been introduced in the next section .
the numerical results and related discussions are given in section [ results ] , and finally section [ conclude ] contains our conclusions .
the kinetic ising model is given by the time dependent hamiltonian @xmath9 where the spin variables @xmath10 are defined on a honeycomb lattice @xmath11 and the first sum in eq .
( [ eq1 ] ) is over the nearest neighbor pairs of spins .
we assume that the nearest - neighbor interactions are randomly diluted on the lattice according to the probability distribution function @xmath12 where @xmath13 denotes the concentration of active bonds .
the term @xmath14 in eq .
( [ eq1 ] ) is a time dependent external magnetic field which is defined as @xmath15 where @xmath16 and @xmath17 represent the amplitude and the angular frequency of the oscillating field , respectively .
if the system evolves according to a glauber - type stochastic process @xcite at a rate of @xmath18 which represents the transitions per unit time then the dynamic equation of motion can be obtained as follows @xmath19\right\rangle,\ ] ] where @xmath20 is the local field acting on the lattice site @xmath21 , and @xmath22 and @xmath23 denote the boltzmann constant and temperature , respectively .
if we apply the differential operator technique @xcite in eq .
( [ eq4 ] ) by taking into account the random configurational averages we get @xmath24 where @xmath25 , @xmath26 , and @xmath27 represents the average magnetization .
@xmath28 is a differential operator , and the inner @xmath29 and the outer @xmath30 brackets represent the thermal and configurational averages , respectively . when the right - hand side of eq .
( [ eq5 ] ) is expanded , the multispin correlation functions appear . the simplest approximation , and one of the most frequently adopted is to decouple these correlations according to @xmath31 for @xmath32 @xcite . if we expand the right - hand side of eq .
( [ eq5 ] ) with the help of eq .
( [ eq6 ] ) then we obtain the following dynamic effective - field equation of motion for the magnetization of the quenched bond diluted kinetic ising model @xmath33 where the coefficients @xmath34 can be easily calculated by employing the mathematical relation @xmath35 .
( [ eq7 ] ) describes the non - equilibrium behavior of the system in the eft formalism .
moreover , the time dependence of magnetization can be one of two types according to whether they obey the following property or not @xmath36 a solution satisfying eq .
( [ eq8 ] ) is called symmetric solution and it corresponds to a paramagnetic or disordered phase . in this type of solution , the time dependent magnetization oscillates around zero value . the second type of solutions which does not satisfy eq .
( [ eq8 ] ) is called non - symmetric solution which corresponds to ferromagnetic ( i.e. ordered ) phase where the time dependent magnetization oscillates around a non - zero value .
the time averaged magnetization over a full cycle of the oscillating magnetic field acts as dop which is defined as follows @xcite @xmath37 where @xmath38 is a stable and periodic function .
on the other hand , the dla corresponding to energy loss due to the hysteresis is defined as @xcite @xmath39 the behavior of dop and dla as a function of the temperature for selected hamiltonian parameters can be obtained by solving eqs .
( [ eq7 ] ) , ( [ eq9 ] ) and ( [ eq10 ] ) numerically .
the remanent magnetization and coercivity can be calculated by benefiting from hysteresis loop ( @xmath40 curve ) in order to understand and clarify the behavior of the quenched bond diluted system .
in this section , in order to understand the dynamic evolution of the magnetic system in detail , we will focus our attention on the hysteretic response and dpt properties of the quenched - bond diluted kinetic ising model under a time - dependent applied magnetic field .
it is well known that depending on the value of the applied field frequency , the magnetic system can exist in a dynamically ordered or disordered phase . from this point of view
, we give the dpb in a @xmath41 plane with four selected amplitudes of the applied field @xmath42 , and for some selected values of the active bond concentration in figs .
[ fig1](a - d ) , in order to clarify the influence of applied field frequency on the dpt properties . at first sight
, one can easily observe from figs .
[ fig1](a - d ) that the dynamically ordered phases get expanded with increasing applied field frequency for all considered active bond concentrations .
actually , this is an expected result since as the frequency increases then the time - dependent magnetization can not follow applied magnetic field immediately , hence the dynamically ordered phase region moves upward with the increasing frequency . we can also mention that aforementioned situations are also reported in previously published works @xcite .
+ + on the other hand , as the value of @xmath13 decreases , starting from pure case ( i.e. @xmath43 ) , the magnetic system exhibits unusual and interesting behaviors .
for instance , dynamically disordered phases occur in the low temperature regions and also reentrant behavior emerges .
another important consequence of decreasing @xmath13 is that the dynamically ordered phases get narrower for all values of the considered applied field amplitudes .
these observations can also be explained by the following mechanism : the energy contribution which comes from the spin - spin interaction gets smaller when the value of @xmath13 is decreased .
so , the system can undergo a dpt at a lower critical temperature , since the energy contribution originating from the temperature and ( or ) magnetic field overcomes the ferromagnetic spin - spin interactions . as a result of this mechanism ,
the dynamically ordered phase regions in @xmath41 plane get narrower . as discussed above ,
the existence or absence of the reentrant behavior strongly depends on the applied field amplitude , as well as oscillation frequency . as an example , let us consider the curves with @xmath44 to show the influences of the applied field amplitude on the dpb , for selected values of @xmath16 . as seen in figs .
[ fig1](a - d ) , for @xmath45 and @xmath46 , the magnetic system exhibits reentrant behavior whereas this behavior disappears with further increase in @xmath16 , such as for @xmath47 and @xmath48 .
it is known that applied field frequencies naturally affect the dynamic evolution of a magnetic system . in order to show the influences of the applied field frequencies on the system , the temperature dependencies of the dop * ( @xmath49 ) * and dla * ( @xmath50 ) * are depicted in figs .
[ fig2](a - d ) for values of @xmath44 and @xmath45 corresponding to fig .
[ fig1](a ) .
in accordance with this purpose , we separated the frequency space into four regions : the order of the magnetic system is dynamically paramagnetic in first region for @xmath51 .
one can see from fig .
[ fig2](a ) that the dla gradually increases and exhibits a local maxima with a schottky - like rounded hump and then tends to diminish with increasing temperature . on the other side ,
it is clear from fig .
[ fig2](b ) that there exists two successive second order dpts for a selected frequency value of @xmath52 , and this type of treatment corresponds to second region .
the first dpt is from paramagnetic to ferromagnetic phase while the second dpt is from ferromagnetic to paramagnetic phase .
for the same hamiltonian parameters , the dla exhibits two local maxima and two whiskers and the positions of the whiskers may correspond to dpt points . in fig .
[ fig2](c ) , we represent the thermal variation of the dop and dla curves for @xmath53 which indicates the third region .
it can be easily seen from fig .
[ fig2](c ) that only one dpt emerges in magnetic system for @xmath53 .
dop curve for @xmath53 increases form its ground state value with increasing temperature which originates from thermal agitations and it decreases with further increasing temperatures and then falls to zero at a second order dpt temperature . similarly , with the increasing temperature , starting from zero , the dla decreases and shows a local minima at the dynamically ordered phase region and then it exhibits a sharp peak near the dpt point . in the following analysis ,
let us consider the higher value of the applied field frequency such as @xmath54 which refers to fourth region . as we can see from fig .
[ fig2](d ) that the magnetic system exhibits a second order dpt .
however , we should also mention that the value of phase transition temperature observed for @xmath54 is slightly higher than that of @xmath53 .
actually , this is an expected result because an increase in the applied field frequency makes the occurrence of phase transition difficult , hence the dynamically ordered phase gets expanded with the increasing applied field frequency . for given hamiltonian parameters
, the dla shows a peak above the second order dpt point .
we should note that the existence of a peak on the dla near dpt point has also been indicated in previous investigations for various models @xcite .
+ + we give the frequency dispersion curves of the dla in figs .
[ fig3](a - d ) for four selected @xmath13 values with a variety of @xmath16 values .
we should note that the reduced temperature in this study is fixed as @xmath55 which is lower than the transition temperature @xmath56 of a pure system with eft for @xmath57 in the absence of external field .
this selection of the temperature allows the system to undergo an amplitude , frequency or disorder induced dynamic phase transition .
fig [ fig3](a ) is plotted for @xmath43 where all ferromagnetic spin - spin exchange interactions @xmath58 are active . at first sight
, one can clearly observe in fig .
[ fig3](a ) that dla curves can exhibit two types of characteristic behavior .
the first one corresponds to a dynamically ordered phase where the dla increases as the applied field frequency increases and then exhibits a characteristic rounded peak ( or local maxima ) for the relatively low field amplitude values such as @xmath59 and @xmath46 .
these curves are identical to type - i curves of ref . @xcite .
as is expected , the dla is relatively close to zero at the low and high applied field frequency regions in the dynamically ordered phases .
on the other hand , for @xmath60 and @xmath48 values , dla curves start from a nonzero value and increases with increasing frequency values then exhibit a finite cusp at a second order phase transition temperature at which the shape of hysteresis loop changes from symmetric to asymmetric and also a rounded hump in high frequency regime which corresponds to second type of characteristic behavior .
these curves correspond to type - ii curves in ref .
finite cusps are found to be discontinuous for low @xmath16 and weak disorder .
the rounded humps observed in dla curves in high frequency regime disappears for increasing @xmath16 and ( or ) decreasing @xmath13 values .
for instance , for @xmath61 with @xmath43 , the dla exhibits a discontinuous cusp and a rounded distinct hump while it displays only a continuous cusp for @xmath62 and @xmath43 .
we can mention that the characteristic cusp of @xmath63 curve arises from the frequency evolution of symmetric hysteresis loop , whereas the rounded hump of this curve originates from the frequency evolution of asymmetric hysteresis loop . on the other side , it is known that the phase transition temperature decreases with increasing amount of disorder .
the effects of the aforementioned situation on the frequency dispersion can be seen in figs .
[ fig3](b - d ) which have been plotted for @xmath64 and @xmath65 .
one can observe in figs .
[ fig3](b - d ) that as the disorder effects become dominant in the system then the representative type - i curves tend to evolve into type - ii , and type - ii curves have a tendency to turn into type - iii curves defined in ref .
@xcite which represent the paramagnetic dla dispersion curves .
in contrast to type - i curves , type - iii curves have a finite residual value in the limit of @xmath66 .
in addition , the type - iii curves are characterized by a rounded hump , as type - i curves .
based on these observations , we see that dla dispersion curves clearly reveal the signs of the oscillation frequency and disorder induced dynamic phase transitions in the present system . + + in figs . [ fig4](a - d ) , we present the frequency dependence of remanent magnetization curves corresponding to dla curves in fig .
[ fig3 ] . here
we select four typical values of disorder parameter @xmath13 with a variety of field amplitude @xmath16 values .
remanent magnetization is defined as the residual magnetization of a ferromagnetic material after an external magnetic field is removed .
hence the remanence data can be easily extracted from calculated hysteresis loops . at first glance , it can be easily seen that depending on amplitude and frequency of the external magnetic field , as well as the amount of disorder , remanence curves can exhibit three distinct characteristics corresponding to type - i , type - ii and type - iii profiles of dla curves .
corresponding to type - i dla curves , either branches of remanent magnetization curves are positively valued which means that the system is ordered ferromagnetically in the whole range of oscillation frequency @xmath67 . increasing the amount of disorder or amplitude of the external field , representative type - i remanence curves may evolve to type - ii and type - iii curves in order .
type - ii remanence curves exhibit equally valued branches with opposite signs in the low frequency region indicating that the system is in a dynamically paramagnetic state . as the frequency increases then the negative branch of a representative type - ii remanence curve changes its sign at a critical value of frequency ,
hence we understand that the corresponding hysteresis loop loses its symmetry and becomes dynamically ordered .
hence this behavior can be considered as a frequency induced phase transition . for sufficiently high amplitudes and ( or ) for a sufficiently large amount of disorder , the observed type - ii curves evolve to type - iii characteristics . in this case , either branches of remanence curves are symmetric with opposite signs which indicates that the system is dynamically disordered for the whole range of oscillation frequency spectrum .
remanent magnetization is a vital property for materials used in magnetic information storage devices .
namely , in case of some fixed values of external field amplitude and oscillation frequency , as well as an amount of existing disorder , for a device made of a material with higher remanence property there will be less chance that the information stored in the media will be erased due to some external sources @xcite . from this point of view
, it is important to note that according to fig .
[ fig4](d ) , there exists an optimum field frequency for a disordered material under an oscillating magnetic field at which the type - iii remanence ( i.e. the distance between positive and negative remanent magnetizations ) exhibit a maximum value , and as the field amplitude increases then the optimum frequency value shifts towards higher frequency values .
+ + as another characteristic feature of hysteresis loops , we have examined the coercivity of the system which is defined as the minimum required reversal magnetic field , in order to reduce the remanent magnetization to zero . for storage media , a material with a high coercivity
is desired to preserve the stored data from unwanted external factors .
we represent the variation of the coercive field curves as a function of oscillation frequency in figs .
[ fig5](a - d ) for some selected values of disorder parameter @xmath13 and external field amplitude @xmath16 . for the dla curves with type - i characteristics , there is no coercivity for any field amplitude , oscillation frequency , or disorder , because all hysteresis curves corresponding to type - i behavior have asymmetric shape .
this means that the instantaneous magnetization can not follow the external magnetic field . for type - ii characteristics
however , we see that at low frequencies , absolute values of either branches in a coercivity curve of the system increase for a while then tend to decrease with further increasing frequencies .
after a critical frequency value , these branches lose their symmetric shape then coercive fields disappear and a dynamic phase transition from a dynamic paramagnetic phase to a ferromagnetic one takes place .
we note that nonsymmetric coercivity data have not been shown in fig .
we again see that an oscillation frequency induced dynamic phase transition takes place in representative type - ii curves . if the magnetic field amplitude is large enough or if there exists a sufficiently strong disorder ( see figs .
[ fig5]c and d ) , coercivity curves may evolve from type - ii to type - iii characteristics . since the type - iii hysteresis curves are symmetric , coercive field
can be observed for the whole range of oscillation frequency values .
however , for type - iii behavior , coercivity is rather small at low frequencies . as the field frequency increases then coercivity of the system increases and in the limit of @xmath68 it saturates to its maximum value .
the maximum value of coercivity is noting but just the applied amplitude of the external magnetic field for sufficiently high frequencies .
hence , we can conclude that the greater amplitude of the external magnetic field , we get more enhanced coercivity . + + the two fundamental tools , namely coercivity and remanence properties allow us to make predictions on the shape of the hysteresis loops .
effect of the field frequency on the shape of typical hysteresis loops corresponding to figs .
[ fig3 ] , [ fig4 ] , and [ fig5 ] are depicted in fig .
[ fig6 ] for a wide variety of oscillation frequencies within the range @xmath69 with given values of @xmath13 and @xmath16 . in fig .
[ fig6](a ) , we present the hysteresis loop curves corresponding to type - i characteristics of dla curves . as seen in fig .
[ fig6](a ) , hysteresis curves exhibit positive valued narrow loops at low frequencies then evolve into a horizontal line with increasing frequencies .
type - ii curves are shown in fig .
[ fig6](b ) , where the low frequency paramagnetic loops turn into high frequency ferromagnetic loops due to a frequency induced dynamic symmetry loss in the system .
moreover , according to fig .
[ fig6](c ) , symmetric type - iii hysteresis curves keep their dynamic paramagnetic state and exhibit remanence and coercivity with branches having both positive and negative values .
we can conclude from fig .
[ fig6](c ) that low frequency thin and @xmath70-shaped hysteresis loops evolves into broad square - like shapes then turn into elliptical loops for sufficiently high frequencies .
if we increase the frequency further , these elliptical loops become horizontal lines .
related to the frequency evolutions of the hysteresis curves , we should also note that a qualitative consistency is found between our numerical findings and experimental observations @xcite .
+ in the following analysis , let us focus on the influence of disorder on the frequency dispersion of hysteretic properties of the system . in figs .
[ fig7](a ) and [ fig7](b ) , we represent variation of coercivity and remanence curves as a function of oscillation frequency for some selected values of disorder parameter @xmath13 with a given field amplitude @xmath71 . from fig .
[ fig7](a ) , we see that as the concentration of diluted bonds increases then the zero frequency branches of coercivity curves vanish as @xmath72 . in type - ii curves ,
critical frequency at which the system exhibit a dynamic phase transition increases with increasing disorder .
this frequency value can be regarded as a cut - off frequency at which the symmetric loops turn into asymmetric loops .
hence , the non - symmetric coercivity data has not been considered in fig .
[ fig7](a ) .
moreover , coercivity curves exhibiting type - ii characteristics may evolve into type - iii behavior , if the concentration of diluted bonds exceeds a critical value ( i.e. percolation threshold ) .
hence , we understand that the system exhibits a disorder induced dynamic phase transition . in fig .
[ fig7](b ) , we illustrate the dependence of frequency dispersion of remanence curves on the concentration of disordered bonds with a fixed external field amplitude @xmath71 corresponding to fig . [ fig7](a ) .
type - i curves in fig .
[ fig7](b ) exhibit a positive remanence which indicates that the system stays in a dynamic ferromagnetic order for the whole range of frequency spectrum . by increasing disorder
, type - i curves turn into type - ii characteristics where we can observe that either branches of remanence curves tend to lose their symmetric shape with increasing frequency which signals the appearance of a frequency induced dynamic phase transition . in a similar manner
, we observe in fig .
[ fig7](b ) that for sufficiently low concentration of magnetic bonds , the remanent magnetization curves may exhibit type - iii characteristics where the system always exhibit a dynamically disordered phase for the whole range of oscillation frequency spectrum . with some selected values of disorder parameter @xmath13.,title="fig:",width=226 ] with some selected values of disorder parameter @xmath13.,title="fig:",width=226 ] + as a final investigation
, we discuss the evolution of some typical hysteresis loops corresponding to remanence and coercivity properties depicted in fig .
[ fig7 ] . in fig .
[ fig8 ] we fix the amplitude of the external magnetic field as @xmath71 .
[ fig8](a ) corresponds to low frequency situation @xmath73 . from fig .
[ fig8]a , we see that a disorder induced dynamic phase transition takes place in the system with increasing disorder .
hysteresis loops exhibit a positive valued horizontal line in the pure case whereas as the amount of disorder increases then the loop area gets narrower and we observe s - shaped thin loops . on the other hand ,
the situation is depicted for relatively high frequency values such as @xmath74 in fig . [ fig8](b ) . in this case
, asymmetric loops corresponding to ferromagnetic order become elliptical symmetric loops in the dynamic paramagnetic phase with increasing disorder .
hence , we can conclude that influence of disorder on the shape of hysteresis loops strictly depends on the magnitude of the applied field frequency .
aforementioned results are also consistent with the observations illustrated in fig .
[ fig7 ] .
in conclusion , by using eft with correlations , we have investigated frequency dispersion features of dynamic loop area , coercivity and remanent magnetization of a kinetic ising model in the presence of quenched - diluted bonds .
first of all , in order to gain some general insight regarding the influence of bond disorder , external field amplitude , and oscillation frequency on the dynamic phase transition properties of the system we have demonstrated the phase diagrams of our model in a @xmath41 plane for weak @xmath75 , intermediate @xmath76 and large @xmath77 values of the external field amplitude with some selected values of disorder parameter @xmath13 . we have found that critical frontiers between symmetric and asymmetric phases in @xmath41 plane are strongly dependent on amplitude and frequency of oscillating field , as well as amount of disorder .
eft is superior to conventional mft predictions owing to the fact that the single spin correlations , in other words thermal fluctuations are partially considered .
hence , we take the recently published work performed by ref.@xcite one step forward by considering single - site correlations between different lattice sites , and by considering the presence of disordered bonds .
however , it is worth to note that the approximation method used in the present work , does not take into account the local structure of the lattice . in other words
, eft can not distinguish between a two dimensional triangular lattice and a three dimensional simple cubic lattice , since both have a coordination number @xmath78 .
in this context , the results obtained in the present paper depend on the coordination number of the system .
the results obtained in the present paper qualitatively support the recent findings based on mft @xcite for the pure case .
namely , we have observed that frequency dispersion of dynamic loop area ( dla ) can be classified into three distinct types . in type - i behavior ,
the system always exhibits asymmetric hysteresis loops at any oscillation frequency for small field amplitudes and weak amount of disorder whereas type - ii curves are characterized by a frequency induced phase transition where the low frequency symmetric loops become asymmetric loops with increasing oscillation frequency . on the other hand , type - iii behavior
is essentially represented by symmetric hysteresis loops for the whole range of frequency spectrum when the external field amplitude is sufficiently large and ( or ) if there exists sufficiently strong dilution of ferromagnetic nearest - neighbor bonds .
therefore we can conclude that type - ii curves exhibit frequency induced dynamic phase transitions whereas evolution of dla curves from type - i to type - iii characteristics is due to a disorder induced effect or as a consequence of large amplitudes of the external magnetic field .
aforementioned observations have also been supported by detailed investigation of frequency dispersions of coercivity and remanent magnetizations . from the technological point of view , in magnetic information storage devices a material with high coercivity and large remanence is desired to preserve the stored data from unwanted external factors . in this regard we have found that there exists an optimum field frequency for a disordered material under an oscillating magnetic field at which the type - iii remanence ( i.e. the distance between positive and negative remanent magnetizations corresponding to type - iii ) exhibit a maximum value , and as the field amplitude increases then the optimum frequency value shifts towards higher frequency values .
finally , we have observed that influence of disorder on the shape of hysteresis loops strictly depends on the magnitude of the applied field frequency . as a conclusion , we hope that the results obtained in this work would shed light on the further investigations of the dynamic nature of the critical phenomena in disordered systems and would be beneficial from both theoretical and experimental points of view .
the authors ( e.v . and y.y . ) would like to thank the scientific and technological research council of turkey ( tbitak ) for partial financial support .
the numerical calculations reported in this paper were performed at tbitak ulakbim ( turkish agency ) , high performance and grid computing center ( truba resources ) .
chakrabarti , and m. acharyya , rev .
phys . * 71 * , 847 ( 1999 ) , and the references therein .
steinmetz , trans .
electr . eng .
* 9 * , 3 ( 1892 ) .
w.s . lo , and r.a .
pelcovits , phys .
a * 42 * , 7471 ( 1990 ) .
m. acharyya , and b.k .
chakrabarti , phys . rev .
b * 52 * , 6550 ( 1995 ) .
m. acharyya , physica a * 253 * , 199 ( 1998 )
. m. acharyya , phys .
e * 58 * , 179 ( 1998 ) .
m. rao , h.r .
krishnamurthy , and r. pandit , phys .
b * 42 * , 856 ( 1990 ) .
sides , p.a .
rikvold , and m.a .
novotny , phys .
e * 59 * , 2710 ( 1999 ) .
s.w . sides , p.a .
rikvold , and m.a .
novotny , phys .
81 * , 834 ( 1998 ) .
s.w . sides , p.a .
rikvold , and m.a .
novotny , phys .
e * 57 * , 6512 ( 1998 ) .
g. korniss , p.a .
rikvold , and m.a .
novotny , phys .
e * 66 * , 056127 ( 2002 ) .
h. zhu , s. dong , and j .-
liu , phys .
b * 70 * , 132403 ( 2004 ) .
f. zhong , phys .
b * 66 * , 060401(r ) ( 2002 ) .
m. acharyya , phys .
e * 59 * , 218 ( 1999 ) .
x. shi , g. wei , and l. li , phys .
a * 372 * , l5922 ( 2008 ) .
b. deviren , o. canko , and m. keskin , chin .
b * 19 * , 050518 ( 2010 ) .
t. tom , and m.j .
de oliveira , phys .
a * 41 * , 4251 ( 1990 ) .
m. acharyya , j. phys .
a : math . gen . * 27 * , 1533 ( 1994 ) .
a. punya , r. yimnirun , p. laoratanakul , and y. laosiritaworn , physica b * 405 * , 3488 ( 2010 ) .
lyuksyutov , t. nattermann , and v. pokrovsky , phys .
b * 59 * , 4260 ( 1999 ) .
t. nattermann , v. pokrovsky , and v.m .
vinokur , phys .
. lett . * 87 * , 197005 ( 2001 ) .
t. nattermann , and v. pokrovsky , physica a * 340 * , 625 ( 2004 ) .
a. misra , and k. chakrabarti , europhys . lett . * 52 * , 311 ( 2000 ) .
f. schtze , and t. nattermann , phys .
b , * 83 * 024412 ( 2011 ) .
l. wang , b.h .
teng , y.h .
rong , y. l , and z.c .
wang , sol . state comm .
( 2012 ) http://dx.doi.org/10.1016/j.ssc.2012.04.076 m.f .
zimmer , phys .
e * 47 * , 3950 ( 1993 ) .
luse , a. zangwill , phys .
e * 50 * , 224 ( 1994 ) . j .-
liu , h.l.w .
chan , c.l .
choy , and c.k .
ong , phys .
b * 65 * , 014416 ( 2001 ) .
sariyer , a. kabakcioglu , and a.n .
berker , arxiv.1206.0230v1 i.d .
mayergoyz , mathematical model of hysteresis , springer - verlag , berlin , ( 1991 ) .
r. gilmore , phys .
a * 20 * , 2510 ( 1979 ) . q. jiang , h .-
yang , and g .- c .
wang , phys .
b * 52 * , 14911 ( 1995 ) .
robb , y.h .
xu , o. hellwig , j. mccord , a. berger , m.a .
novotny , and p.a .
rikvold , phys .
b * 78 * , 134422 ( 2008 )
choi , w.y .
lee , a. samad , and j.a.c .
bland phys .
b * 60 * , 11906 ( 1999 ) .
he , and g .- c .
wang , phys .
. lett . * 70 * , 2336 ( 1993 ) .
w.y . lee , b .- ch .
choi , j. lee , c.c .
yao , y.b .
hasko , and j.a.c .
bland , appl .
* 74 * , 1609 ( 1999 ) .
w.y . lee , b .- ch .
choi , y. b. xu , and j.a.c .
bland , phys .
b * 60 * , 10216 ( 1999 ) .
rivera , j.m .
ferreira , and f.c .
montenegro , j. magn .
magn . mat . * 226 - 230 * , 1309 ( 2001 ) .
l. santi , r.l .
sommer , a. magni , g. durin , f. colaiori , and s. zapperi , ieee trans .
* 39 * , 2666 ( 2003 ) .
moore , j. rothman , y.b .
xu , and j.a.c .
blanda , j. appl . phys . * 89 * , 7018 ( 2001 ) . g .-
zheng , and m. li , phys .
b * 66 * , 054406 ( 2002 ) .
u. akinci , y. yksel , e. vatansever , and h. polat , physica a ( 2012 ) , doi : 10.1016/j.physa.2012.06.060 .
e. vatansever , b.o .
aktas , y. yksel , u. akinci , and h. polat , j. stat .
phys . * 147 * , 1068 ( 2012 ) .
glauber , j. math .
phys . * 4 * , 294 ( 1963 ) .
r. honmura , and t. kaneyoshi , j. phys .
c * 12 * , 3979 ( 1979 ) .
t. kaneyoshi , acta phys . pol .
a * 83 * , 703 ( 1993 ) .
i. tamura , and t. kaneyoshi , prog .
. phys . * 66 * , 1892 ( 1981 ) .
z. huang , z. chen , f. zhang , and y. du , phys .
a * 338 * , 485 ( 2005 ) . c. altavilla , e. ciliberto ,
inorganic nanoparticles : synthesis , applicaions , and prespectives , taylor&francis , crc press , p33 ( 2010 ) . j .-
liu , q. xiao , z.g .
liu , h.l.w .
chan , and n.b .
ming , mat .
phys . * 82 * , 733 ( 2003 ) . | frequency evolutions of hysteresis loop area and hysteresis tools such as remanence and coercivity of a kinetic ising model in the presence of quenched bond dilution are investigated in detail .
the kinetic equation describing the time dependence of the magnetization is derived by means of effective - field theory with single - site correlations .
it is found that the frequency dispersions of hysteresis loop area , remanence and coercivity strongly depend on the quenched bond randomness , as well as applied field amplitude and oscillation frequency .
in addition , the shape of the hysteresis curves for a wide variety of hamiltonian parameters is studied and some interesting behaviors are found .
finally , a comparison of our observations with those of recently published studies is represented and it is shown that there exists a qualitatively good agreement . |
team decision theory has its roots in control theory and economics .
marschak @xcite was perhaps the first to introduce the basic elements of teams and to provide the first steps toward the development of a _
team theory_. radner @xcite provided foundational results for static teams , establishing connections between person - by - person optimality , stationarity , and team - optimality . the work of witsenhausen @xcite , @xcite , @xcite , @xcite , @xcite on dynamic teams and characterization of information structures has been crucial in the progress of our understanding of dynamic teams .
further discussion on the design of information structures in the context of team theory and economics applications are given in @xcite and @xcite , among a rich collection of other contributions not listed here .
establishing the existence and structure of optimal policies is a challenging problem .
existence of optimal policies for static and a class of sequential dynamic teams has been studied recently in @xcite .
more specific setups and non - existence results have been studied in @xcite , @xcite . for a class of teams which are _ convex _
, one can reduce the search space to a smaller parametric class of policies ( see @xcite , and for a comprehensive review , see @xcite ) . in this paper ,
our aim is to study the approximation of static and dynamic team problems using finite models which are obtained through the uniform discretization , on a finite grid , of the observation and action spaces of agents . in particular , we are interested in the asymptotic optimality of quantized policies . in the literature
relatively few results are available on approximating static or dynamic team problems .
we can only refer the reader to @xcite,@xcite and a few references therein . with the exception of @xcite , these works in general study a specific setup ( the witsenhausen counterexample ) and are mostly experimental ; that is , they do not rigorously prove the convergence of approximate solutions . in @xcite ,
a class of static team problems are considered and the existence of smooth optimal strategies is studied . under fairly strong assumptions ,
the existence of an optimal strategy with lipschitz continuous partial derivatives up to some order is proved . by using this result , an error bound on the accuracy of near optimal solutions
is established , where near optimal strategies are expressed as linear combinations of basis functions with adjustable parameters . in @xcite ,
the same authors investigated the approximation problem for witsenhausen s counterexample , which does not satisfy the conditions in @xcite ; the authors derived an analogous error bound on the accuracy of the near optimal solutions . for the result in @xcite both the error bound and the near optimal solutions depend on the knowledge of the optimal strategy for witsenhausen s counterexample .
moreover , the method devised in @xcite implicitly corresponds to the discretization of only the action spaces of the agents .
therefore , it involves only the approximation with regard to the action space , and does not correspond to a tractable approximation for the set of policies / strategies .
particular attention has been paid in the literature to witsenhausen s counterexample .
this problem has puzzled the control community for more than 40 years with its philosophical impact demonstrating the challenges that arise due to a non - classical information structure , and its formidable difficulty in obtaining an optimal or suboptimal solution .
in fact , optimal policies and their value are still unknown , even though the existence of an optimal policy has been established using various methods @xcite . some relevant results on obtaining approximate solutions can be found in @xcite .
certain lower bounds , that are not tight , building on information theoretic approaches are available in @xcite , see also @xcite . in this paper , we show that finite models obtained through the uniform quantization of the observation and action spaces lead to a sequence of policies whose cost values converge to the optimal cost .
thus , with high enough computation power , one could guarantee that for any @xmath1 , an @xmath0-optimal policy can be constructed .
we note that the operation of quantization has typically been the method to show that a non - linear policy can perform better than an optimal linear policy , both for witsenhausen s counterexample @xcite and another interesting model known as the gaussian relay channel problem @xcite .
our findings show that for a large class of problems , quantized policies not only may perform better than linear policies , but that they are actually almost optimal .
we finally note that although finding optimal solutions for finite models for witsenhausen s counterexample as the ones constructed in this paper was shown to be np - complete in @xcite , the task is still computationally less demanding than the method used in @xcite . loosely speaking , to obtain a near optimal solution using the method in @xcite , one has to compute the optimal partitions of the observation spaces and the optimal representation points in the action spaces .
in contrast , the partitions of the observation spaces and the available representation points in the action spaces used by our method are known _
a priori_. we also note that if one can establish smoothness properties of optimal policies such as differentiability or lipschitz continuity ( e.g. , as in @xcite ) , the methods developed in our paper can be used to provide rates of convergence for the sequence of finite solutions as the finite models are successively refined . * contributions of the paper . *
( i ) we establish that finite models asymptotically represent the true models in the sense that the solutions obtained by solving such finite models lead to cost values that converge to the optimal cost of the original model .
thus , our approach can be viewed to be constructive ; even though the computational complexity is typically at least exponential in the cardinality of the finite model .
( ii ) the approximation approach here provides , to our knowledge , the first rigorously established result showing that one can construct an @xmath0-optimal strategy for any @xmath1 through an explicit solution of a simpler problem for a large class of static and dynamic team problems , in particular for the witsenhausen s celebrated counterexample .
the rest of the paper is organized as follows . in section [ sub1sec1 ]
we review the definition of witsenhausen s _ intrinsic model _ for sequential team problems . in section [ boundedcase ]
we consider finite _ observation _ approximations of static team problems with compact observation spaces and bounded cost , and prove the asymptotic optimality of strategies obtained from finite models . in section [ unboundedcase ]
an analogous approximation result is obtained for static team problems with non - compact observation spaces and unbounded cost functions . in section [ sec3 ]
we consider finite observation approximations of dynamic team problems via the static reduction method . in sections [ sec4 ] and [ gaussrelay ]
we apply the results derived in section [ sec3 ] to study finite observation space approximations of witsenhausen s celebrated counterexample and the gaussian relay channel .
discretization of the action spaces is considered in section [ discact ] .
section [ sec5 ] concludes the paper .
in this section , we introduce the model as laid out by witsenhausen , called _ the intrinsic model _ @xcite ; see @xcite for a more comprehensive overview and further characterizations and classifications of information structures . in this model
, any action applied at any given time is regarded as applied by an individual agent , who acts only once .
one advantage of this model , in addition to its generality , is that the definitions regarding information structures can be compactly described .
suppose that in the decentralized system , there is a pre - defined order in which the agents act .
such systems are called _ sequential systems _ ( for non - sequential teams , we refer the reader to @xcite , @xcite and @xcite , in addition to @xcite ) . in the following , all spaces are assumed to be borel spaces ( i.e. , borel subsets of complete and separable metric spaces ) endowed with borel @xmath2-algebras . in the context of a sequential system , the _ intrinsic model _ has the following components : * a collection of _ measurable spaces _
@xmath3 , specifying the system s distinguishable events , and the action and measurement spaces . here
@xmath4 is the number of actions taken , and each of these actions is supposed to be taken by an individual agent ( hence , an agent with perfect recall can also be regarded as a separate decision maker every time it acts ) .
the pair @xmath5 is a measurable space on which an underlying probability can be defined .
the pair @xmath6 denotes the measurable space from which the action @xmath7 of agent @xmath8 is selected .
the pair @xmath9 denotes the measurement ( or observation ) space of agent @xmath8 . *
a _ measurement constraint _ which establishes the connection between the observation variables and the system s distinguishable events .
the @xmath10-valued observation variables are given by @xmath11 , where @xmath12 , @xmath13 is a stochastic kernel on @xmath10 given @xmath14 , and @xmath15 denotes the action of agent @xmath16 . * a _ design constraint _ , which restricts the set of admissible @xmath17-tuple control strategies @xmath18 , also called _ policies _ , to the set of all measurable functions , so that @xmath19 , where @xmath20 is a measurable function .
let @xmath21 denote the set of all admissible policies for agent @xmath8 and let @xmath22 . * a _ probability measure _ @xmath23 defined on @xmath5 which describes the measures on the random events in the model .
we note that the intrinsic model of witsenhausen gives a set - theoretic characterization of information fields ; however , for borel spaces , the model above is equivalent to the intrinsic model for sequential team problems . under this intrinsic model , a sequential team problem is _ dynamic _ if the information available to at least one agent is affected by the action of at least one other agent . a decentralized problem is
, if the information available at every decision maker is only affected by state of the nature ; that is , no other decision maker can affect the information at any given decision maker . information structures ( iss ) can also be classified as classical , quasi - classical , and nonclassical .
an is is _ classical _ if @xmath24 contains all of the information available to agent @xmath16 for @xmath25 .
an is is _ quasi - classical _ or _ partially nested _ , if whenever @xmath15 ( for some @xmath25 ) affects @xmath24 , then agent @xmath8 has access to @xmath26 .
an is which is not partially nested is _
nonclassical_. for any @xmath27 , we let the cost of the team problem be defined by @xmath28,\ ] ] for some measurable cost function @xmath29 , where @xmath30 and @xmath31 .
[ def : tb1 ] for a given stochastic team problem , a policy ( strategy ) @xmath32 is an _ optimal team decision rule _
if @xmath33 the cost level @xmath34 achieved by this strategy is the _ optimal team cost_. [ def : tb2 ] for a given stochastic team problem , a policy @xmath35 constitutes a _
nash equilibrium _
( synonymously , a _ person - by - person optimal _
solution ) if , for all @xmath36 and all @xmath37 , the following inequalities hold : @xmath38 where we have adopted the notation @xmath39 unless otherwise specified , the term ` measurable ' will refer to borel measurability in the rest of the paper . in
what follows , the terms _ policy _ , _ measurement _ , and _ agent _ are used synonymously with _
, _ observation _ , and _ decision maker _ , respectively . in this section , for the ease of reference we state some well - known results in measure theory and functional analysis that will be frequently used in the paper .
the first theorem is lusin s theorem which roughly states that any measurable function is _ almost _ continuous .
( lusin s theorem ( * ? ? ?
* theorem 7.5.2))[lusin ] let @xmath40 and @xmath41 be two borel spaces and let @xmath42 be a probability measure on @xmath40 .
let @xmath43 be a measurable function from @xmath40 into @xmath41 .
then , for any @xmath44 there is a closed set @xmath45 such that @xmath46 and the restriction of @xmath43 to @xmath47 is continuous .
the second theorem is the dugundji extension theorem which is a generalization of the tietze extension theorem @xcite .
( dugundji extension theorem ( * ? ? ?
* theorem 7.4))[dugundji ]
let @xmath40 be a borel space and let @xmath47 be a closed subset of @xmath40 .
let @xmath41 be a convex subset of some locally convex vector space
. then any continuous @xmath48 has a continuous extension on @xmath40 .
the next theorem originally states that the closed convex hull of a compact subset in a locally convex vector space is compact if the vector space is completely metrizable .
since the closure of a convex set is convex and a closed subset of a compact set is compact , we can state the theorem in the following form .
* theorem 5.35)[clconv ] in a completely metrizable locally convex vector space @xmath49 , the closed convex hull of a compact set is convex and compact .
the same statement also holds when @xmath49 is replaced with any of its closed and convex subsets .
in this section , we consider the finite observation approximation of static team problems . in
what follows , the static team problem is formulated in a state - space form which can be reduced to the intrinsic model introduced in section [ sub1sec1 ] .
let @xmath50 be a probability space representing the state space , where @xmath51 is a borel space and @xmath52 is its borel @xmath2-algebra .
we consider an @xmath17-agent static team problem in which agent @xmath8 , @xmath53 , observes a random variable @xmath24 and takes an action @xmath7 , where @xmath24 takes values in a borel space @xmath10 and @xmath7 takes values in a borel space @xmath54 . given any state realization @xmath55 , the random variable @xmath24 has a distribution @xmath56 ; that is , @xmath56 is a stochastic kernel on @xmath10 given @xmath51 @xcite .
the team cost function @xmath57 is a non - negative function of the state , observations , and actions ; that is , @xmath58 , where @xmath59 and @xmath60 .
for agent @xmath8 , the set of strategies @xmath21 is given by @xmath61 recall that @xmath62 .
then , the cost of the team @xmath63 is given by @xmath64 where @xmath65 . here , with an abuse of notation , @xmath66 denotes the joint distribution of the state and observations .
therefore , we have @xmath67 with these definitions , we first consider the case where the observation spaces are compact and the cost function is bounded . in the second part
, teams with non - compact observation spaces and unbounded cost function will be studied . in this section ,
we consider the finite observation approximation of static team problems with compact observation spaces and bounded cost function .
we impose the following assumptions on the components of the model .
[ as1 ] * * the cost function @xmath57 is bounded .
in addition , it is continuous in @xmath68 and @xmath69 for any fixed @xmath55 . * for each @xmath8
, @xmath54 is a convex subset of a locally convex vector space .
* for each @xmath8 , @xmath10 is compact .
we first prove that the minimum cost achievable by continuous strategies is equal to the optimal cost @xmath34 . to this end , for each @xmath8 , we define @xmath70 and @xmath71 .
[ prop1 ] we have @xmath72 let @xmath73 .
we prove that there exists a sequence @xmath74 such that @xmath75 as @xmath76 , which implies the proposition .
let @xmath77 denote the distribution of @xmath24 . for each @xmath78 , by lusin s theorem , there is a closed set @xmath79 such that @xmath80 and the restriction of @xmath20 to @xmath81 is continuous .
let us denote @xmath82 , and so @xmath83 is continuous . by the dugundji extension theorem
, there exists a continuous extension @xmath84 of @xmath85 .
therefore , @xmath86 and we have for @xmath87 the following @xmath88 \rp(dx , d{\bf y } ) \biggr| \nonumber \\ & \leq \int_{\sx \times ( \sy \setminus f_k ) } \bigl| c(x,{\bf y},\underline{\gamma } ) - c(x,{\bf y},\underline{\gamma}_{k } ) \bigr| \text { } \rp(dx , d{\bf y } ) \nonumber \\ & \leq 2 \|c\| \text { } \rp\bigl(\sx \times ( \sy \setminus f_k)\bigr ) , \nonumber\end{aligned}\ ] ] where @xmath89 is the maximum absolute value that @xmath57 takes . since @xmath90 , we have @xmath91 .
this completes the proof .
let @xmath92 denote the metric on @xmath10 .
since @xmath10 is compact , one can find a finite set @xmath93 such that @xmath94 is an @xmath95-net in @xmath10 ; that is , for any @xmath96 we have @xmath97 define function @xmath98 mapping @xmath10 to @xmath94 by @xmath99 where ties are broken so that @xmath100 is measurable . in the literature , @xmath98 is called the nearest neighborhood quantizer @xcite . if @xmath101 $ ] for some @xmath102 , one can choose the finite set @xmath94 such that @xmath98 becomes a uniform quantizer .
for any @xmath103 , we let @xmath104 denote the strategy @xmath105 .
define @xmath106 we let @xmath107 .
the following theorem states that an optimal ( or almost optimal ) policy can be approximated with arbitrarily small approximation error for the induced costs by policies in @xmath108 , for @xmath109 sufficiently large .
[ thm1 ] we have @xmath110 for any @xmath0 , let @xmath111 denote an @xmath0-optimal continuous strategy .
the existence of such a strategy follows from proposition [ prop1 ] .
then , we have @xmath112 where @xmath113 . note that @xmath114 as @xmath115 , for all @xmath116 since @xmath57 is continuous in @xmath68 and @xmath117
hence , by the dominated convergence theorem the second term in the last expression converges to zero as @xmath118 . since @xmath0 is arbitrary , this completes the proof . for each @xmath109 ,
define stochastic kernels @xmath119 on @xmath94 given @xmath51 as follows : @xmath120 where @xmath121 .
let @xmath122 and @xmath123 .
define @xmath124 as @xmath125 where @xmath126 , @xmath127 , @xmath128 , and @xmath129 .
[ convergence - comp ] let @xmath130 be a sequence of strategies such that @xmath131 . for each @xmath109 , define @xmath132 , where @xmath133 .
then , we have @xmath134 we have @xmath135 which converges to zero as @xmath118 by dominated convergence theorem and the fact that @xmath57 is bounded and continuous in @xmath69 .
the following theorem is the main result of this section which is a consequence of theorem [ thm1 ] .
it states that to compute a near optimal strategy for the original team problem , it is sufficient to compute an optimal ( or almost optimal if optimal does not exist ) strategy for the team problem described above .
[ thm2 ] for any @xmath44 , there exists a sufficiently large @xmath109 such that the optimal ( or almost optimal ) policy @xmath136 for the cost @xmath137 is @xmath0-optimal for the original team problem when @xmath138 is extended to @xmath139 via @xmath140 .
fix any @xmath44 . by theorem [ thm1 ] ,
there exists a sequence of strategies @xmath141 such that @xmath142 ( @xmath143 ) and @xmath144 .
define @xmath145 as the restriction of @xmath146 to the set @xmath147 .
then , we have @xmath148 for the reverse inequality , for each @xmath143 , let @xmath149 be such that @xmath150 .
define @xmath151 .
then , we have @xmath152 this completes the proof .
in this section , we consider the finite observation approximation of static team problems with noncompact observation spaces and unbounded cost function .
we impose the following assumptions on the components of the model .
[ newas1 ] * * the cost function @xmath57 is continuous in @xmath68 and @xmath69 for any fixed @xmath55 .
in addition , it is bounded on any compact subset of @xmath153 . * for each @xmath8 , @xmath54 is a closed and convex subset of a completely metrizable locally convex vector space . * for each @xmath8 , @xmath10 is locally compact .
* for any subset @xmath154 of @xmath155 , we let @xmath156 .
we assume that @xmath157 is integrable with respect to @xmath158 , for any compact subset @xmath154 of @xmath155 of the form @xmath159 . * for any @xmath73 with @xmath160 and each @xmath161 , there exists @xmath162 such that we have @xmath163 , where @xmath164 .
note that assumption [ newas1]-(d),(e ) hold if the cost function is bounded .
therefore , if the static team problem satisfies assumption [ as1 ] , then assumption [ newas1 ] ( except ( b ) ) holds as well .
hence , the results derived in this section _ almost _ includes the results proved in section [ boundedcase ] as a particular case .
however , since the analysis in this section is somewhat involved , we presented the compact and bounded case as a separate section . the following result states that there exists a near optimal strategy whose range is convex and compact . in
what follows , for any compact subset @xmath154 of @xmath155 , we let @xmath165 [ newprop2 ] suppose assumption [ newas1]-(a),(b),(c),(e ) hold . then , for any @xmath44 , there exists a compact subset @xmath154 of @xmath155 of the form @xmath159 , where each @xmath166 is convex and compact , such that @xmath167 fix any @xmath44 .
let @xmath73 with @xmath168 .
we construct the desired @xmath154 iteratively . by assumption
[ newas1]-(e ) there exists @xmath169 such that @xmath170 .
let @xmath171 be a compact set containing @xmath172 .
we define @xmath173 define also @xmath174 , @xmath175 , and @xmath176 . then , we have @xmath177 + e\bigl [ c(x,\by,\bu ) 1_{\{y^1 \notin m_1\ } } \bigr ] \nonumber \\ & \phantom{xxxxxxxxxxx}- e\bigl [ c(x,\by,\bu^{-1},\tilde{u}^1 ) 1_{\{y^1 \in m_1\ } } \bigr ] \nonumber \\ & \phantom{xxxxxxxxxxxxxxxxxx}- e\bigl [ c(x,\by,\bu^{-1},\tilde{u}^1 ) 1_{\{y^1 \notin m_1\ } } \bigr ] \biggr| \nonumber \\ & \leq e\bigl [ c(x,\by,\bu ) 1_{\{y^1 \notin m_1\ } } \bigr ] + e\bigl [ c(x,\by,\bu^{-1},\tilde{u}^1 ) 1_{\{y^1 \notin m_1\ } } \bigr ] \nonumber \\ & = \int_{\sx\times\sy\times\su^{-1}\times ( g^1)^c } c(x,\by,\bu ) \text { } \delta_{\underline{\gamma}}(d\bu ) \text { } \rp(dx , d\by ) \nonumber \\ & \phantom{xxx}+ \int_{\sx\times\sy\times\su^{-1}\times ( g^1)^c } c(x,\by,\bu ) \text { } \delta_{(\underline{\gamma}^{-1},\gamma^1_{u^{1,*}})}(d\bu ) \text { } \rp(dx , d\by ) \nonumber,\end{aligned}\ ] ] where @xmath178 denotes the complement of the set @xmath179
, @xmath180 denotes the point mass at @xmath181 , and @xmath182 .
recall that @xmath183 by assumption [ newas1]-(e ) .
hence , the last expression can be made smaller than @xmath184 by properly choosing @xmath185 since @xmath186 is a borel space ( * ? ? ?
* theorem 3.2 ) .
since the closed convex hull of the set @xmath185 is compact by theorem [ clconv ] , we can indeed take @xmath185 convex without loss of generality . by replacing @xmath187 with @xmath188 and applying the same method as above ,
we can obtain @xmath189 , and a convex and compact @xmath190 such that @xmath191 and @xmath192 . continuing this way ,
we obtain @xmath159 and @xmath193 such that @xmath194 , where @xmath166 is convex and compact for all @xmath53 .
hence , we have @xmath195 , completing the proof .
recall that @xmath196 denotes the set of continuous strategies .
for any @xmath197 , we define @xmath198 ; that is , @xmath199 is the set of continuous strategies having range inside @xmath154 .
[ newprop1 ] suppose assumption [ newas1 ] holds .
then , for any @xmath44 , there exists a compact subset @xmath154 of @xmath155 of the form @xmath159 , where each @xmath166 is convex and compact , such that @xmath200 fix any @xmath44 . by lemma [ newprop2 ] , there exists a compact subset @xmath159 of @xmath155 , where @xmath166 is convex and compact , and @xmath201 such that @xmath202 recall that @xmath77 denotes the distribution of @xmath24
. let @xmath203 which will be specified later .
analogous to the proof of proposition [ prop1 ] , we construct a continuous strategy @xmath204 which is _ almost _ equal to @xmath187 . for each @xmath8
, lusin s theorem implies the existence of a closed set @xmath205 such that @xmath206 and the restriction , denoted by @xmath207 , of @xmath20 on @xmath208 is continuous . by the dugundji extension theorem there exists a continuous extension @xmath209 of @xmath207 .
therefore , @xmath210 . let @xmath211 .
then , we have @xmath212 hence , we obtain @xmath213 \rp(dx , d{\bf y } ) \biggr| \nonumber \\ & \leq \int_{\sx \times ( \sy \setminus f ) } \bigl [ c(x,{\bf y},\underline{\gamma } ) + c(x,{\bf y},\underline{\tgamma } ) \bigr ] \text { } \rp(dx , d{\bf y } ) \nonumber \\ & \leq 2 \int_{\sx \times ( \sy \setminus f ) } w_g(x,{\bf y } ) \text { } \rp(dx , d{\bf y } ) . \nonumber\end{aligned}\ ] ] by assumption [ newas1]-(d ) @xmath157 is @xmath23-integrable so that the measure @xmath214 is absolutely continuous with respect to @xmath23 .
since @xmath215 as @xmath216 , we obtain @xmath217 since @xmath218 , there exists a sufficiently small @xmath219 such that @xmath220 .
this completes the proof . since for each @xmath8
, @xmath10 is a locally compact separable metric space , there exists a nested sequence of compact sets @xmath221 such that @xmath222 and @xmath223 ( * ? ? ?
* lemma 2.76 ) , where @xmath224 denotes the interior of the set @xmath179 .
recall that @xmath92 denotes the metric on @xmath10 . for each @xmath225 ,
let @xmath226 be an @xmath95-net in @xmath227 . recall that if @xmath228 is an @xmath95-net in @xmath227 , then for any @xmath229 we have @xmath230 for each @xmath231 and @xmath109 , let @xmath232 be the nearest neighborhood quantizer
; that is , for any @xmath229 @xmath233 where ties are broken so that @xmath234 is measurable .
if @xmath235 \subset \sy^i = \r$ ] for some @xmath102 , the finite set @xmath228 can be chosen such that @xmath234 becomes an uniform quantizer .
we let @xmath236 denote the extension of @xmath234 to @xmath10 given by @xmath237 where @xmath238 is some auxiliary element .
define @xmath239 ; that is , @xmath240 is defined to be the set of all strategies @xmath241 of the form @xmath242 , where @xmath103 .
define also @xmath243 .
the following theorem states that an optimal ( or almost optimal ) policy can be approximated with arbitrarily small approximation error for the induced costs by policies in @xmath244 for sufficiently large @xmath231 and @xmath109 .
[ newthm1 ] for any @xmath44 , there exist @xmath245 and @xmath246 such that @xmath247 by proposition [ newprop1 ] , there exists @xmath248 such that @xmath249 , where @xmath159 and each @xmath166 is convex and compact . for each @xmath231 and @xmath109 , we define @xmath250 and @xmath251 .
define also @xmath252 .
let @xmath253 denote the collection of all subsets of @xmath254 except the empty set .
for any @xmath255 , we define @xmath256 recall that @xmath257 is the strategy which maps any @xmath258 to @xmath259 .
let @xmath260 and observe that @xmath261 note that since the range of the strategy @xmath262 is contained in @xmath154 , we have @xmath263 for all @xmath255 by assumption [ newas1]-(d ) .
hence , there exists a sufficiently large @xmath231 such that @xmath264 let @xmath265 .
then , we have @xmath266 note that in the last expression , the integrands in the third and fourth terms are less than @xmath157 .
since @xmath157 is @xmath23-integrable by assumption [ newas1]-(d ) , @xmath267 on @xmath227 as @xmath118 ( recall that @xmath20 is continuous ) , and @xmath57 is continuous in @xmath268 , the third and fourth terms in the last expression converge to zero as @xmath118 by dominated convergence theorem .
hence , there exists a sufficiently large @xmath269 such that the last expression becomes less than @xmath270 .
therefore , @xmath271 , completing the proof .
the above result implies that to compute a near optimal policy for the team problem it is sufficient to choose a strategy based on the quantized observations @xmath272 for sufficiently large @xmath231 and @xmath109 .
furthermore , this nearly optimal strategy can have a compact range of the form @xmath159 , where @xmath166 is convex and compact for each @xmath53 .
however , to obtain a result analogous to the theorem [ thm2 ] , we need to impose a further assumption . to this end
, we first introduce a finite observation model . for each @xmath273 ,
let @xmath274 ( i.e. , the output levels of @xmath275 ) and define the stochastic kernels @xmath276 on @xmath277 given @xmath51 as follows : @xmath278 where @xmath279 .
let @xmath280 and @xmath281 .
define @xmath282 as @xmath283 where @xmath126 , @xmath127 , @xmath284 , and @xmath285 .
note that the probability measure @xmath286 can also be treated as a measure on @xmath287 . in this case
, it is not difficult to prove that @xmath286 converges to @xmath23 weakly as @xmath288 . for any compact subset @xmath154 of @xmath155
, we also define @xmath289 .
[ nnewas1 ] for any compact subset @xmath154 of @xmath155 of the form @xmath159 , we assume that the function @xmath157 is uniformly integrable with respect to the measures @xmath290 ; that is , @xmath291 [ convergence - unbounded ] let @xmath292 be a sequence of strategies such that @xmath293 , where @xmath294 and each @xmath166 is convex and compact . for each @xmath231 and @xmath109 , define @xmath295 , where @xmath296 .
then , we have @xmath297 let us introduce the following finite measures on @xmath287 : @xmath298 since @xmath286 converges to @xmath23 weakly , by ( * ? ? ?
* theorem 3.5 ) and assumption [ nnewas1 ] we have @xmath299 weakly as @xmath300 .
hence , the sequence @xmath301 is tight .
therefore , there exists a compact subset @xmath302 of @xmath287 such that @xmath303 and @xmath304 for all @xmath305 .
then , we have @xmath306 the first term in the last expression goes to zero as @xmath307 by dominated convergence theorem and the fact that @xmath57 is bounded and continuous in @xmath69 .
the second term is less than @xmath0 .
since @xmath0 is arbitrary , this completes the proof .
the following theorem is the main result of this section which is a consequence of theorem [ newthm1 ] .
it states that to compute a near optimal strategy for the original team problem , it is sufficient to compute an optimal ( or an almost optimal policy if an optimal one does not exist ) policy for the team problem described above .
[ newthm2 ] suppose assumptions [ newas1 ] and [ nnewas1 ] hold .
then , for any @xmath44 , there exists a pair @xmath245 and a compact subset @xmath159 of @xmath155 such that an optimal ( or almost optimal ) policy @xmath308 in the set @xmath309 for the cost @xmath310 is @xmath0-optimal for the original team problem when @xmath138 is extended to @xmath139 via @xmath311 .
fix any @xmath44 . by lemma [ newprop2 ] and
theorem [ newthm1 ] , there exists compact subset @xmath154 of @xmath155 of the form @xmath312 such that @xmath313 for each @xmath314 , let @xmath315 be such that @xmath316 . define @xmath317 as the restriction of @xmath318 to the set @xmath319 .
then , we have @xmath320 for the reverse inequality , for each @xmath314 , let @xmath293 be such that @xmath321 .
define @xmath295 .
then , we have @xmath322 this completes the proof .
the results for the static case apply also to the dynamic case , through a static reduction .
first we review the equivalence between sequential dynamic teams and their static reduction ( this is called _ the equivalent model _
@xcite ) . consider a dynamic team setting according to the intrinsic model where there are @xmath17 decision epochs , and agent @xmath8 observes @xmath323 , and the decisions are generated as @xmath324 .
the resulting cost under a given team policy @xmath187 is @xmath325.\ ] ] this dynamic team can be converted to a static team provided that the following absolute continuity condition holds .
[ as2 ] for every @xmath53 , there exists a function @xmath326 , where @xmath327 , and a probability measure @xmath328 on @xmath10 such that for all @xmath329 we have @xmath330 therefore , for a fixed choice of @xmath187 , the joint distribution of @xmath331 is given by @xmath332 where @xmath333 .
the cost function @xmath334 can then be written as @xmath335 where @xmath336 and @xmath337 .
the observations now can be regarded as independent , and by incorporating the @xmath338 terms into @xmath57 , we can obtain an equivalent _ static team _ problem .
hence , the essential step is to appropriately adjust the probability space and the cost function .
note that in the static reduction method , some nice properties ( such as continuity and boundedness ) of the cost function @xmath57 of the original dynamic team problem can be lost , if the @xmath338 functions in assumption [ as2 ] are not well - behaved .
however , the observation channels between @xmath339 and the @xmath24 are quite well - behaved for most of the practical models ( i.e , additive gaussian channel ) which admits static reduction .
therefore , much of the nice properties of the cost function are preserved for most of the practical models .
the next theorem is the main result of this section .
it states that for a class of dynamic team problems , finite models can approximate an optimal policy with arbitrary precision . in what follows
, @xmath340 denotes the distribution of state and observations in the finite model approximation of the static reduction .
[ thm4 ] suppose assumptions [ newas1]-(a),(b),(c),(e ) and [ as2 ] hold .
in addition , for each @xmath53 , @xmath341 is continuous in @xmath342 and @xmath24 , and @xmath343 for all compact @xmath197 of the form @xmath159
. then , the static reduction of the dynamic team model satisfies assumptions [ newas1 ] and [ nnewas1 ] .
therefore , theorems [ newthm1 ] and [ newthm2 ] hold for the dynamic team problem .
in particular , theorems [ newthm1 ] and [ newthm2 ] hold for the dynamic team problems satisfying assumptions [ newas1 ] , [ nnewas1 ] and [ as2 ] , if @xmath338 is bounded and continuous in @xmath342 and @xmath24 for each @xmath53 .
an important dynamic information structure is the _ partially nested _ information structure .
an is is partially nested if whenever @xmath15 affects @xmath24 for some @xmath344 , agent i has access to @xmath26 ; that is , there exists a measurable function @xmath345 such that @xmath346 for all @xmath73 and all realizations of @xmath347 .
for such team problems , one talks about _ precedence relationships _ among agents : agent @xmath16 is _ precedent _ to agent @xmath8 ( or agent @xmath16 _ communicates _ to agent @xmath8 ) , if the former agent s actions affect the information of the latter , in which case ( to be partially nested ) agent @xmath8 has to have the information based on which the action - generating policy of agent @xmath16 was constructed .
dynamic teams with such an information structure always admit a static reduction through an informational equivalence . for such partially nested ( or quasi - classical ) information structures
, a static reduction was studied by ho and chu in the context of lqg systems @xcite and for a class of non - linear systems satisfying restrictive invertibility properties @xcite . for such dynamic teams ,
the cost function does not change as a result of the static reduction , unlike in the static reduction in section [ staticred ] .
therefore , if the partially nested dynamic team satisfies assumptions [ newas1 ] and [ nnewas1 ] , then its static reduction also satisfies it .
hence , theorems [ newthm1 ] and [ newthm2 ] hold for such problems . before proceeding to the next section , we prove an auxiliary result which will be used in the next two sections .
[ uniform ] let @xmath43 and @xmath348 be non - negative real functions defined on metric spaces @xmath40 and @xmath41 , respectively .
suppose @xmath349 for some sequence of probability measures @xmath350 and @xmath351 .
then , we have @xmath352 let @xmath353 \coloneqq \int f d\mu_n$ ] and @xmath354 \coloneqq \int g d\nu_n$ ] . it is easy to prove that @xmath355 \eqqcolon a < \infty$ ] and @xmath356 \eqqcolon b < \infty$ ] . note that @xmath357 . hence , @xmath358 \int_{\{f > \sqrt{r}\ } } f \text { } d\mu_n + e_n[f ] \int_{\{g >
\sqrt{r}\ } } g \text { } d\nu_n \nonumber \\ &
\phantom{xxxx}\leq b \int_{\{f > \sqrt{r}\ } } f \text { } d\mu_n + a \int_{\{g > \sqrt{r}\ } } g \text { } d\nu_n . \nonumber\end{aligned}\ ] ] since the last term converges to zero as @xmath359 by assumption , this completes the proof .
in witsenhausen s celebrated counterexample @xcite ( see fig . [ fig1 ] ) , thare are two decision makers : agent @xmath360 observes a zero mean and unit variance gaussian random variable @xmath361 and decides its strategy @xmath362 .
agent @xmath363 observes @xmath364 , where @xmath365 is a standard ( zero mean and unit variance ) gaussian noise independent of @xmath361 , and decides its strategy @xmath366 .
= [ draw , fill = white!20 , minimum size=3em ] = [ pin edge = to-,thin , black ] = [ draw , circle ] \(a ) @xmath367 ; ( b ) [ left of = a , node distance=2 cm , coordinate ] a ; ( d ) [ right of = a ] @xmath368 ; ( c ) [ right of = d ] @xmath369 ; ( end ) [ right of = c , node distance=2 cm ] ; ( e ) [ below of = d , node distance=1 cm ] @xmath365 ; ( e ) edge node ( d ) ; ( b ) edge node @xmath361 ( a ) ; ( a ) edge node @xmath362 ( d ) ; ( d ) edge node @xmath370 ( c ) ; ( c ) edge node @xmath366 ( end ) ; the cost function of the team is given by @xmath371 where @xmath372 . here
, the state of the nature @xmath347 can be regarded as a degenerate random variable .
we let @xmath373 .
then we have @xmath374 where @xmath375 denotes the lebesgue measure .
let @xmath376 so that @xmath377 .
the static reduction proceeds as follows : for any policy @xmath187 , we have @xmath378 where @xmath379 denotes the standard gaussian distribution .
hence , by defining @xmath380 and @xmath381 , we can write @xmath334 as @xmath382 hence , in the static reduction of witsenhausen s counterexample , agents observe independent zero mean and unit variance gaussian random variables . in this section
we study the approximation problem for witsenhausen s counterexample by using the static reduction formulation .
we show that the conditions in theorem [ thm4 ] hold for witsenhausen s counterexample , and therefore , theorems [ newthm1 ] and [ newthm2 ] can be applied .
the cost function of the static reduction is given by @xmath383 where @xmath384 is given in ( [ eq20 ] ) .
note that the strategy spaces of the original problem and its static reduction are identical , and same strategies induce the same team costs . recall that @xmath379 denotes the standard gaussian distribution .
[ newlemma1 ] for any @xmath385 with @xmath386 , we have @xmath387 < \infty$ ] .
to prove the lemma , we use the original problem setup instead of its static reduction . fix any strategy @xmath388 with finite cost .
since @xmath389<\infty$ ] and @xmath390 , we have @xmath391<\infty$ ] via @xmath391^{1/2 } \leq e\bigl [ ( u^1-y^1)^2 \big]^{1/2}+e\bigl [ ( y^1)^2 \big]^{1/2}$ ] .
then , we also have @xmath392<\infty$ ] since @xmath393<\infty$ ] and @xmath392^{1/2 } \leq e\bigl [ ( u^1)^2 \big]^{1/2 } + e\bigl [ ( u^1-u^2)^2 \big]^{1/2}$ ] , completing the proof . for any @xmath394
, we define @xmath395 $ ] .
let @xmath396 denote the uniform quantizer on @xmath397 having @xmath109 output levels ; that is , @xmath398 where @xmath399 .
let us extend @xmath396 to @xmath400 by mapping @xmath401 to @xmath402 .
for each @xmath273 , let @xmath403 ( i.e. , output levels of the extended @xmath396 ) and define the probability measure @xmath286 on @xmath319 as @xmath404 moreover , let @xmath405 and define @xmath406 with the help of lemma [ newlemma1 ] , we now prove the following result .
[ newprop3 ] witsenhausen s counterexample satisfies conditions in theorem [ thm4 ] .
assumption [ newas1]-(a),(b),(c ) and assumption [ as2 ] clearly hold . to prove assumption [ newas1]-(e ) , we introduce the following notation . for any strategy @xmath388 , we let @xmath407 denote the corresponding expectation operation .
pick @xmath388 with @xmath386 . since @xmath408 \biggr ]
= e_{\gamma^1,\gamma^2 } \biggl [ \bigl ( u^1 - u^2 \bigr)^2 \biggr ] \nonumber\end{aligned}\ ] ] by the law of total expectation , there exists @xmath169 such that @xmath409 < \infty$ ] .
let @xmath410 .
then we have @xmath411 \nonumber \\ & \phantom{xxxxx}=e_{\gamma^1,\gamma^2}\biggl [ l \bigl(u^1 - y^1\bigr)^2 + \bigl(u^1\bigr)^2 \biggr ] < \infty \text { ( by lemma~\ref{newlemma1 } ) } \nonumber \\ \intertext{and } & j(\gamma_{u^{1,*}}^1,\gamma^2 ) = e_{\gamma_{u^{1,*}}^1,\gamma^2}\biggl [ l \bigl(u^{1 } - y^1\bigr)^2 + \bigl(u^{1 } - u^2\bigr)^2 \biggr ] \nonumber \\ & \phantom{xx}=e_{\gamma^1,\gamma^2}\biggl [ l \bigl(u^{1 , * } - y^1\bigr)^2 \biggr ] + e_{\gamma^1,\gamma^2}\biggl[\bigl(u^{1 } - u^2\bigr)^2 \biggl| u^1 = u^{1 , * } \biggr ] \nonumber \\ & \phantom{xx}<\infty .
\nonumber\end{aligned}\ ] ] therefore , assumption [ newas1]-(e ) holds .
note that for the @xmath412-uniform integrability condition , it is sufficient to consider compact sets of the form @xmath413 ^ 2 $ ] for some @xmath102 , since any compact set in @xmath414 is contained in @xmath413 ^ 2 $ ] for sufficiently large @xmath102 .
let @xmath102 .
we have @xmath415 ^ 2 } l(u^1-y^1)^2 + ( u^2-u^1)^2 \nonumber \\ & = l \bigl ( m + |y^1|)^2 + 4 m^2 \nonumber\end{aligned}\ ] ] and @xmath416 ^ 2 } f(u^1,y^2 ) & \leq \sup_{(u^1,u^2 ) \in [ -m , m]^2 } \exp{\{y^2 u^1\ } } \nonumber \\ & = \exp{\{m |y^2|\ } } \eqqcolon w_2(y^2 ) .
\nonumber\end{aligned}\ ] ] for functions @xmath417 and @xmath418 , we have @xmath419 \text { } d\rp_{\sg}=0 \label{nneq5 } \\
\intertext{and } & \lim_{r \rightarrow
\infty } \sup_{n , l } \int_{\{w_2(y^2 ) > r\ } } w_2(y^2 ) \text { } d\rp_{l , n } \nonumber \\ & \phantom{xxxx}= \lim_{r \rightarrow \infty } \sup_{n , l } \int_{\{w_2(y^2 ) > r\ } } w_2(q_{l , n}(y^2 ) ) \text { } d\rp_{\sg } \nonumber \\ & \phantom{xxxx}\leq \lim_{r \rightarrow \infty } \int_{\{w_2(y^2 ) > r\ } } \exp{m(|y^2|+1 ) } \text { } d\rp_{\sg}=0 , \label{nneq6}\end{aligned}\ ] ] where ( [ nneq5 ] ) and ( [ nneq6 ] ) follow from the facts that @xmath420 and the integrability of @xmath417 and @xmath418 with respect to the @xmath379 . by lemma [ uniform ] ,
the product @xmath421 is @xmath412-uniformly integrable .
therefore , @xmath422 ^ 2 } \tilde{c}(y^1,y^2,u^1,u^2)$ ] is also @xmath412-uniformly integrable .
since @xmath423 is arbitrary , this completes the proof .
proposition [ newprop3 ] and theorem [ thm4 ] imply that theorems [ newthm1 ] and [ newthm2 ] is applicable to witsenhausen s counterexample .
therefore , an optimal strategy for witsenhausen s counterexample can be approximated by strategies obtained from finite models .
the theorem below is the main result of this section .
it states that to compute a near optimal strategy for witsenhausen s counterexample , it is sufficient to compute an optimal strategy for the problem with finite observations obtained through uniform quantization of the observation spaces .
[ thm3 ] for any @xmath44 , there exists @xmath245 and @xmath424 such that an optimal policy @xmath425 in the set @xmath426 for the cost @xmath310 is @xmath0-optimal for witsenhausen s counterexample when @xmath425 is extended to @xmath427 via @xmath428 , @xmath429 , where @xmath430 $ ] and @xmath431 .
an important dynamic team problem which has attracted interest is the gaussian relay channel problem ( see fig .
[ fig2 ] ) @xcite . here
, agent @xmath360 observes a noisy version of the state @xmath55 which has gaussian distribution with zero mean and variance @xmath432 ; that is , @xmath433 where @xmath434 is a zero mean and variance @xmath435 gaussian noise independent of @xmath55 .
agent @xmath360 decides its strategy @xmath362 based on @xmath361 . for @xmath436 , agent @xmath8 receives @xmath437 ( a noisy version of the decision @xmath438 of agent @xmath439 ) , where @xmath440 is a zero mean and variance @xmath441 gaussian noise independent of @xmath442 , and decides its strategy @xmath443 .
= [ draw , fill = white!20 , minimum size=3em ] = [ pin edge = to-,thin , black ] = [ draw , circle , scale=0.8 ] the cost function of the team is given by @xmath444 where @xmath445 for all @xmath446 . to ease the notation in the sequel , we simply take @xmath447 .
recall that @xmath448 .
then we have @xmath449 recall also that @xmath450 , where @xmath451 is defined in ( [ eq20 ] ) .
then , for any policy @xmath187 , we have @xmath452 \text { } \rp_{\sg}^{n+1}(dx , d\by ) , \nonumber\end{aligned}\ ] ] where @xmath453 denotes the product of @xmath454 zero mean and unit variance gaussian distributions .
therefore , in the static reduction of gaussian relay channel , we have the components @xmath455 $ ] and @xmath456 . analogous to witsenhausen s counterexample , the agents observe independent zero mean and unit variance gaussian random variables . in this subsection , the approximation problem for the gaussian relay channel is considered using the static reduction formulation .
analogous to section [ sec4 ] , we prove that the conditions of theorem [ thm4 ] hold for gaussian relay channel , and so theorems [ newthm1 ] and [ newthm2 ] can be applied .
the cost function of the static reduction is given by @xmath457 \nonumber\end{aligned}\ ] ] where @xmath451 is given in ( [ eq20 ] ) .
recall the uniform quantizer @xmath396 on @xmath458 $ ] ( @xmath394 ) having @xmath109 output levels from section [ sec4 ] .
we extend @xmath396 to @xmath400 by mapping @xmath401 to @xmath402 .
recall also the set @xmath403 and the probability measure @xmath286 on @xmath319 given by @xmath459 define @xmath405 and @xmath460 where @xmath461 .
define @xmath462 .
[ newprop4 ] the gaussian relay channel problem satisfies the conditions in theorem [ thm4 ] .
it is clear that assumption [ newas1]-(a),(b),(c ) and assumption [ as2 ] hold . for any strategy @xmath187 ,
let @xmath463 denote the corresponding expectation operation . to prove assumption [ newas1]-(e ) ,
pick @xmath187 with @xmath160 .
analogous to lemma [ newlemma1 ] , one can prove that @xmath464 < \infty$ ] .
for any @xmath53 , by the law of total expectation we can write @xmath465 \biggr ] \nonumber \\ \intertext{and } e_{\underline{\gamma } } \biggl [ \bigl ( u^n \bigr)^2 \biggr ] & = e_{\underline{\gamma } } \biggl [ e_{\underline{\gamma } } \bigl [ \bigl ( u^n \bigr)^2 \bigl| u^i \bigr ] \biggr ] .
\nonumber\end{aligned}\ ] ] therefore , for each @xmath53 , there exists @xmath162 such that @xmath466 < \infty$ ] and @xmath467 < \infty$ ] .
then we have @xmath468 \nonumber \\ & \phantom{xxx}\leq e_{\underline{\gamma}^{-i},\gamma_{u^{i,*}}^i}\biggl [ 2 x^2 + 2 \bigl ( u^n \bigr)^2 + \sum_{j=1}^{n-1 } l_j \bigl(u^j\bigr)^2\biggr ] \nonumber \\ & \phantom{xxx}=e_{\underline{\gamma}}\biggl [ \sum_{j=1}^{i-1 } l_j \bigl(u^j\bigr)^2 + 2 x^2 \biggr ] \nonumber \\ & \phantom{xxxxxxxx}+ e_{\underline{\gamma}}\biggl [ 2 \bigl ( u^n \bigr)^2 + \sum_{j = i}^{n-1 } l_j \bigl(u^j\bigr)^2 \biggr| u^i = u^{i , * } \biggr ] < \infty . \nonumber\end{aligned}\ ] ] therefore , assumption [ newas1]-(e ) holds .
analogous to the proof of proposition [ newprop3 ] , for @xmath340-uniform integrability , it is sufficient to consider compact sets of the form @xmath413^n$ ] for some @xmath102 .
choose any @xmath102 .
then we have @xmath469^n } \bigl ( u^n - x \bigr)^2 + \sum_{i=1}^{n-1 } l_i \bigl(u^i\bigr)^2 \nonumber \\ & = \bigl ( m + x)^2 + \sum_{i=1}^{n-1 } l_i m^2 \nonumber\end{aligned}\ ] ] and @xmath470^n } \prod_{i=2}^n f(u^{i-1},y^i ) & \leq \sup_{\bu \in [ -m , m]^n } \prod_{i=2}^n \exp{\{y^i u^{i-1}\ } } \nonumber \\ & = \prod_{i=2}^n \exp{\{m |y^i|\ } } \eqqcolon \prod_{i=2}^n w_{i,2}(y^i ) .
\nonumber\end{aligned}\ ] ] it can be proved as in the proof of proposition [ newprop3 ] that @xmath471 is @xmath286-uniformly integrable for each @xmath436 .
let @xmath472 and so @xmath473 then , we have @xmath474 hence , by lemma [ uniform ] , the product @xmath475 is @xmath340-uniformly integrable .
therefore , @xmath476^n } \tilde{c}(x,{\bf y},{\bf u})$ ] is also @xmath340-uniformly integrable .
since @xmath423 is arbitrary , this completes the proof . the preceding proposition and theorem [ thm4 ]
imply , via theorems [ newthm1 ] and [ newthm2 ] , that an optimal strategy for gaussian relay channel can be approximated by strategies obtained from finite models .
the following theorem is the main result of this section .
[ thm5 ] for any @xmath44 , there exists @xmath245 and @xmath424 such that an optimal policy @xmath477 in the set @xmath478 for the cost @xmath310 is @xmath0-optimal for gaussian relay channel when @xmath477 is extended to @xmath139 via @xmath428 , @xmath53 , where @xmath430 $ ] and @xmath431 .
for computing near optimal strategies for static team problems using numerical algorithms , the action spaces @xmath54 must be finite . in this section , we show that the action spaces can be taken to be finite in finite observation models , if a sufficiently large number of points are used for accurate approximation . in this section
, we consider the most general case studied in section [ unboundedcase ] .
we note that the results that will be derived in this section can be applied to the dynamic teams which admit static reduction and satisfy conditions in theorem [ thm4 ] .
recall the finite observation models constructed in section [ unboundedcase ] . for each @xmath273 , the finite model have the following components : @xmath479 , where @xmath51 is the state space , @xmath277 is the observation space for agent @xmath8 , @xmath54 is the action space for agent @xmath8 , @xmath276 observation channel from state to the observation of agent @xmath8 , @xmath57 is the cost function , and @xmath23 is the distribution of the state .
furthermore , strategy spaces are defined as @xmath280 and @xmath281 . then the cost function @xmath282 is given by @xmath283 where @xmath126 , @xmath127 , @xmath284 , and @xmath285 .
the theorem below is the main result of this section which states that one can approximate optimal strategy in @xmath480 by strategies taking values in a finite set .
[ newthm3 ] suppose that original static team problem satisfies assumptions [ newas1 ] and [ nnewas1 ] .
then , for each @xmath273 and for any @xmath44 , there exist finite sets @xmath481 for @xmath53 such that @xmath482 where @xmath483 .
fix any @xmath273 and @xmath0 .
let us choose @xmath484 such that @xmath485 note that for any @xmath53 , the range of @xmath486 is a finite subset of @xmath54 and so , is contained in some compact and convex subset @xmath166 of @xmath54 .
define @xmath159 .
let @xmath487 denote the metric on @xmath54 .
since @xmath166 is compact , one can find a finite set @xmath488 which is a @xmath489-net in @xmath166 .
define @xmath490 and @xmath491 . for @xmath486 ,
we let @xmath492 hence @xmath493 we define @xmath494 .
then we have @xmath495 the last integral converges to zero as @xmath76 by the dominated convergence theorem since : ( i ) @xmath496 as @xmath76 by ( [ eq5 ] ) and assumption [ newas1]-(a ) , ( ii ) @xmath497 for all @xmath498 , and ( iii ) @xmath157 is @xmath286-integrable .
therefore , there exists sufficiently large @xmath499 such that the last expression is less than @xmath0 . by choosing @xmath500 , for @xmath53
, the proof is complete .
consider the finite _ observation _ models introduced in section [ sec4 ] that approximate the witsenhausen s counterexample .
for any @xmath424 and @xmath501 , let @xmath502 denote the uniform quantizer with @xmath16 output levels ( recall that @xmath430 $ ] ) .
note that @xmath503 is a quantizer applied to subsets of action spaces @xmath504 , @xmath429 ( not to be confused with @xmath396 in section [ sec4 ] ) .
the preceding theorem implies that for each @xmath273 and @xmath44 , there exists a @xmath424 and @xmath501 such that @xmath505 where @xmath506 and @xmath507 is the set of output levels of @xmath503 .
therefore , to compute a near optimal strategy for witsenhausen s counterexample , it is sufficient to compute an optimal strategy for the finite model that is obtained through uniform quantization of observation and action spaces ( i.e. , @xmath400 ) on finite grids when the number of grid points is sufficiently large .
in particular , through constructing the uniform quantization so that both the _ granular region _ and the _ granularity _ of the quantizers are successively refined ( that is the partitions generated by the quantizers are successively nested ) , we have the following proposition which lends itself to a numerical algorithm . there exists a sequence of finite models obtained through a successive refinement of the measurement and action set partitions generated by uniform quantizers whose optimal costs will converge to the cost of the witsenhausen s counterexample .
approximation of both static and dynamic team problems by finite models was considered . under mild technical conditions
, we showed that the finite model obtained by quantizing uniformly the observation and action spaces on finite grids provides a near optimal strategy if the number of grid points is sufficiently large . using this result ,
an analogous approximation results were also established for the well - known counterexample of witsenhausen and gaussian relay channel .
our approximation approach to the witsenhausen s counterexample thus provides , to our knowledge , the first rigorously established result that for any @xmath1 , one can construct an @xmath0 optimal strategy through an explicit solution of a conceptually simpler problem .
the authors are grateful to professor tamer baar for his technical comments and encouragement .
a. gupta , s. yksel , t. basar , and c. langbort , `` on the existence of optimal policies for a class of static and sequential dynamic teams , '' _
siam j. control optim .
_ , vol .
53 , no . 3 , pp .
16811712 , 2015 .
y. wu and s. verd , `` witsenhausen s counterexample : a view from optimal transport theory , '' in _ proceedings of the ieee conference on decision and control , florida , usa _ , december 2011 , pp
. 57325737 . j. l. s. j.c .
krainak and s. marcus , `` static team problems
part i : sufficient conditions and the exponential cost criterion , '' _ ieee transactions automatic contr . _ , vol . 27 , pp . 839848 , april 1982 .
j. lee , e. lau , and y. ho , `` the witsenhausen counterexample : a hierarchical search approach for nonconvex optimization problems , '' _ ieee trans .
46 , no . 3 , pp . 382397 , mar .
2001 .
p. grover , s. y. park , and a. sahai , `` approximately optimal solutions to the finite - dimensional witsenhausen counterexample , '' _ ieee transactions on automatic control _
58 , no . 9 , pp . 21892204 , 2013 .
m. andersland and d. teneketzis , `` information structures , causality , and non - sequential stochastic control , i : design - independent properties , '' _ siam j. of control optimization _ , vol .
30 , p. 14471475 | in this paper , we consider finite model approximations of a large class of static and dynamic team problems where these models are constructed through uniform quantization of the observation and action spaces of the agents .
the strategies obtained from these finite models are shown to approximate the optimal cost with arbitrary precision under mild technical assumptions .
in particular , quantized team policies are asymptotically optimal . this result
is then applied to witsenhausen s celebrated counterexample and the gaussian relay channel problem . for the witsenhausen s counterexample , our approximation approach provides , to our knowledge , the first rigorously established result that one can construct an @xmath0-optimal strategy for any @xmath1 through a solution of a simpler problem . |
ubiquitous , secure , and high data rate communication is a basic requirement for the next generation wireless communication systems .
the rapid growth of wireless data traffic in the past decades has heightened the energy consumption in both transmitters and receivers . as a result ,
multiuser multiple - input multiple - output ( mimo ) has been proposed in the literature for facilitating energy efficient wireless communication . although the energy dissipation of the transmitters may be significantly reduced by multiuser mimo technology , mobile communication devices and sensor devices are still often powered by batteries with limited energy storage capacity .
frequently replacing the device batteries can be costly and inconvenient in difficult - to - access environments , or even infeasible for medical sensors embedded inside the human body .
hence , the limited lifetime of communication networks constitutes a major bottleneck in providing quality of service ( qos ) to the end - users .
recently , energy harvesting based mobile communication system design has drawn significant interest from both academia and industry since it enables self - sustainability of energy constrained wireless devices .
traditionally , wind , solar , and biomass , etc .
are the major sources for energy harvesting .
although these renewable energy sources are perpetual , their availability usually depends on location and climate which may not be suitable for mobile devices . on the other hand , wireless energy transfer technology , which allows receivers to scavenge energy from the ambient radio frequency ( rf ) signals ,
has attracted much attention lately although the concept can be traced back to nikola tesla s work in the early 20th century @xcite .
there have been some preliminary applications of wireless energy transfer such as wireless body area networks ( wban ) for biomedical implants , passive radio - frequency identification ( rfid ) systems , and wireless sensor networks .
indeed , the combination of rf energy harvesting and communication provides the possibility of simultaneous wireless information and power transfer ( swipt ) which imposes many new and interesting challenges for wireless communication engineers @xcite@xcite . in @xcite , the trade - off between channel capacity and harvested energy
was studied for near - field communication over a frequency selective channel . in @xcite ,
the authors investigated the performance limits of a three - node wireless mimo broadcast channel for swipt .
in particular , the tradeoffs between maximal information rate versus energy transfer were characterized by the boundary of a rate - energy ( r - e ) region . in @xcite , a power splitting receiver and a separated receiver were proposed to realize concurrent information decoding and energy harvesting for narrow - band single - antenna communication systems . in @xcite@xcite , different resource allocation algorithms were proposed for improving the utilization of limited system resources .
optimal beamforming and power allocation design was studied for multiuser narrow - band systems with multiple transmit antennas in @xcite , while the resource allocation algorithm design for wide - band swipt systems was studied in @xcite@xcite . in @xcite and @xcite , the energy efficiency of multi - carrier modulation with swipt
was investigated for single user and multiuser systems , respectively .
in particular , it was shown in @xcite that the energy efficiency of a communication system can be improved by integrating an energy harvester into a conventional information receiver which further motivates the deployment of swipt in practice .
besides , a power allocation algorithm was designed for the maximization of spectral efficiency of swipt systems employing power splitting receivers in @xcite . the results in @xcite@xcite reveal that the amount of harvested energy at the receivers can be increased by increasing the transmit power of the information signals .
however , a high signal power may also lead to substantial information leakage due to the broadcast nature of the wireless communication channel and facilitate eavesdropping .
the notion of physical ( phy ) layer security in swipt systems has recently been pursued in @xcite@xcite . by exploiting multiple antennas ,
transmit beamforming and artificial noise generation can be utilized for providing communication security while guaranteeing qos in wireless energy transfer to energy harvesting receivers .
however , the resource allocation algorithms in @xcite@xcite were designed for a single information receiver and single - antenna eavesdroppers .
the results in @xcite@xcite may not be applicable to the case of multiple - antenna eavesdroppers . besides
, the works in @xcite@xcite did not take into account fairness issues in transferring energy to energy harvesting receivers .
nevertheless , fairness is an essential qos figure of merit for wireless energy transfer . in this paper , we focus on the resource allocation algorithm design for providing fairness to energy receivers in the wireless energy transfer process while guaranteing communication secrecy to information receivers . the resource allocation algorithm design is formulated as a non - convex optimization problem . in particular , we promote the dual use of artificial noise for secrecy communication and efficient wireless energy transfer provisioning .
the considered non - convex problem is solved optimally by semidefinite programming ( sdp ) relaxation and our simulation results unveil the potential performance gain achieved by the proposed optimization framework .
we use boldface capital and lower case letters to denote matrices and vectors , respectively .
@xmath0 , @xmath1 , @xmath2 , and @xmath3 represent the hermitian transpose , trace , rank , and determinant of matrix @xmath4 , respectively ; @xmath5 denotes the maximum eigenvalue of matrix @xmath4 ; @xmath6 and @xmath7 indicate that @xmath4 is a positive definite and a positive semidefinite matrix , respectively ; @xmath8 is the @xmath9 identity matrix ; @xmath10 denotes the set of all @xmath11 matrices with complex entries ; @xmath12 denotes the set of all @xmath9 hermitian matrices .
the circularly symmetric complex gaussian ( cscg ) distribution is denoted by @xmath13 with mean vector @xmath14 and covariance matrix @xmath15 ; @xmath16 indicates distributed as " ; @xmath17 denotes statistical expectation ; @xmath18 represents the absolute value of a complex scalar .
@xmath19^+$ ] stands for @xmath20 .
single - antenna information receivers and @xmath21 multiple - antenna energy harvesting receivers.,width=336 ] we consider a multiuser downlink communication system with swipt .
in particular , a base station ( /transmitter ) equipped with @xmath22 antennas serving @xmath23 information receivers and @xmath24 energy harvesting receivers is considered .
the @xmath23 information receivers are low complexity single - antenna devices while the @xmath24 energy harvesting receivers are equipped with @xmath25 antennas .
we assume that @xmath26 .
a possible application of the considered system model is cognitive radio . in particular
, the energy harvesting receivers may be the primary users which are temporally connecting to a secondary transmitter with the intend to harvest energy from the received signals for extending the lifetime of the primary network . in turn , the primary network may grant the secondary network spectrum usage rights for the energy supply services . in each time slot , @xmath23
independent signal streams are transmitted simultaneously to @xmath23 information receivers . to prevent the @xmath24 energy harvesting receivers ( potential eavesdroppers ) from eavesdropping the information for
the @xmath27 intended information receivers , artificial noise is transmitted concurrently with the information signals for interfering the reception of the energy harvesting receivers . as a result , the transmitted signal vector , @xmath28 , consists of the @xmath23 desired information signals and artificial noise , and can be expressed as @xmath29 where @xmath30 and @xmath31 are the symbol and the beamforming vector intended for information receiver @xmath32 , respectively .
variable @xmath33 is the artificial noise vector generated by the transmitter and is modelled as @xmath34 with zero mean and covariance matrix @xmath35 , @xmath36 . without loss of generality
we assume that @xmath37 . in this paper
, we focus on a frequency flat slow fading channel . the received signals at the information receivers and the energy harvesting receivers are given by @xmath38 respectively .
the channel vector between the transmitter and desired receiver @xmath32 is denoted by @xmath39 .
the channel matrix between the transmitter and energy harvesting receiver @xmath40 is denoted by @xmath41 .
the channel vectors and matrices capture the joint effects of multipath fading and path loss . @xmath42 and @xmath43 are additive white gaussian noises ( awgn ) caused by the thermal noises in the antennas of the information receivers and the energy harvesting receivers , respectively .
@xmath44 denotes the noise power at the receiver .
in the considered system , the energy harvesting receivers scavenge power from the rf .
the total amount of power harvested by energy harvesting receiver @xmath40 is given by @xmath45 where @xmath46 is a constant which denotes the energy conversion efficiency of energy harvesting receiver @xmath40 .
given perfect channel state information ( csi ) at the receiver , the channel capacity ( bit / s / hz ) between the transmitter and information receiver @xmath32 is given by @xmath47 is the receive signal - to - interference - plus - noise ratio ( sinr ) at information receiver @xmath32 and @xmath48 .
on the other hand , we focus on an unfavourable scenario for the decoding capability of the energy harvesting receivers for providing communication security to the information receivers .
specifically , we assume that energy harvesting receiver @xmath40 performs interference cancellation to remove all multiuser interference and eavesdrops the message intended for information receiver @xmath32 .
therefore , the channel capacity between the transmitter and energy harvesting receiver @xmath40 for decoding the signal of information receiver @xmath32 can be represented as @xmath49 where @xmath50 is the interference - plus - noise covariance matrix for energy harvesting receiver @xmath40 assuming the worst case for communication secrecy .
thus , the achievable secrecy capacity of information receiver @xmath32 is given by @xmath51^+.\end{aligned}\ ] ] the system objective is to maximize the minimum harvested power among all energy harvesting receivers while providing qos for communication security .
the resource allocation algorithm design is formulated as an optimization problem which is given by @xmath52 constraint c1 indicates that the receive sinr at information receiver @xmath32 is required to be larger than a given threshold , @xmath53 , for guaranteeing reliable communication .
the upper limit @xmath54 in c2 is imposed to restrict the channel capacity of energy harvesting receiver @xmath40 if it attempts to decode the message of information receiver @xmath55 .
constant @xmath56 in constraint c3 limits the radiated power from the transmitter accounting for the power budget of the transmitter .
constraint c4 and @xmath57 ensure that the covariance matrix @xmath35 is a positive semidefinite hermitian matrix . problem ( [ eqn : cross - layer ] ) is a non - convex optimization problem . in particular ,
the non - convexity arises from constraints c1 and c2 . to overcome the non - convexity , we first propose the following proposition and then recast the considered problem into a convex optimization problem using sdp relaxation . [ prop : relaxed_c2 ] for @xmath58 , the following implication on constraint c2 holds : @xmath59 where @xmath60 and @xmath61 is an auxiliary constant .
@xmath62 is a linear matrix inequality ( lmi ) constraint .
we note that constraints @xmath62 and @xmath63 are equivalent if @xmath64 . _
proof : _ please refer to the appendix for the proof.@xmath65 now , we apply proposition [ prop : relaxed_c2 ] to ( [ eqn : cross - layer ] ) and replace constraint @xmath66 with constraint @xmath62 . by setting @xmath67 , @xmath60 , and @xmath68 ,
we can rewrite the optimization problem in its hypograph form : @xmath69 where @xmath70 is an auxiliary optimization variable . then
, we adopt sdp relaxation by removing constraint @xmath71 from the problem formulation which results in a convex sdp problem .
the relaxed sdp problem formulation of ( [ eqn : rank_one ] ) is given by @xmath72 we note that the relaxed problem in ( [ eqn : sdp_relaxation ] ) can be solved efficiently by numerical solvers such as cvx @xcite . if the obtained solution for ( [ eqn : sdp_relaxation ] ) , @xmath73 , admits a rank - one matrix , then the problems in ( [ eqn : cross - layer ] ) , ( [ eqn : rank_one ] ) , and ( [ eqn : sdp_relaxation ] ) share the same optimal solution and the same optimal objective value . in the following ,
we investigate if @xmath74 holds for the solution of ( [ eqn : sdp_relaxation ] ) .
it can be shown that the problem in ( [ eqn : rank_one ] ) satisfies slater s constraint qualification .
thus , strong duality holds and solving the dual problem is equivalent to solving the primal problem .
the lagrangian of ( [ eqn : sdp_relaxation ] ) is expressed as : @xmath75\notag\\ -&&\hspace*{-5mm}\sum_{k=1}^k\delta_k\big[-\frac{\operatorname{\mathrm{tr}}(\mathbf{h}_k\mathbf{w}_k)}{\gamma_{\mathrm{req}_k } } + \hspace*{-0.5 mm } \operatorname{\mathrm{tr}}\big(\mathbf{h}_k\big(\sum\limits _ { \substack{m\neq k}}^k\mathbf{w}_m\hspace*{-0.5mm}+\hspace*{-0.5mm}\mathbf{v}\big)\big)\hspace*{-0.5mm}+\hspace*{-0.5mm}\sigma_{\mathrm{s}}^2\big]\notag\\ + & & \hspace*{-5mm}\operatorname{\mathrm{tr}}(\mathbf{y}\mathbf{v})\hspace*{-0.5mm}+\hspace*{-0.5 mm } \sum_{k=1}^k\operatorname{\mathrm{tr}}(\mathbf{z}_k\mathbf{w}_k ) \hspace*{-0.5mm}-\hspace*{-0.5mm}\lambda\big[\hspace*{-0.5mm}\operatorname{\mathrm{tr}}\big(\mathbf{v}\hspace*{-0.5mm}+\hspace*{-0.5mm}\sum_{k=1}^k\mathbf{w}_k\big)-
p_{\max}\hspace*{-0.5mm}\big]\notag\\ -&&\hspace*{-5mm}\sum_{j=1}^j\sum_{k=1}^k\operatorname{\mathrm{tr}}\big\{\mathbf{x}_{j , k}\big[\mathbf{g}_j^h\mathbf{w}_k\mathbf{g}_j- \alpha_{\mathrm{er}_{j , k}}\mathbf{q}_{j}\big ] \big\}\notag,\end{aligned}\ ] ] where @xmath76 , @xmath77 , and @xmath78 are the dual variable matrices for constraints @xmath62 , c4 , and c5 , respectively .
@xmath79 , @xmath80 , and @xmath81 are the scalar dual variables for constraints c1 , c3 , and c7 respectively .
let @xmath82 , and @xmath83 .
besides , we denote the orthonormal basis of the null space of @xmath84 as @xmath85 , and @xmath86 , where @xmath87 , denotes the @xmath88-th column of @xmath89 . hence , @xmath90 and @xmath91 .
also , it can be shown that @xmath92 holds for the optimal solution .
now , we introduce the following theorem for revealing the tightness of the sdp relaxation adopted in ( [ eqn : sdp_relaxation ] ) .
[ thm : rankone_condition ] suppose the optimal solution of ( [ eqn : sdp_relaxation ] ) is denoted by @xmath93 , @xmath94 , and @xmath54 .
the optimal solution , @xmath95 , can be expressed as @xmath96 where @xmath97 and @xmath98 are positive scalars and @xmath99 , @xmath100 , satisfies @xmath101 .
if @xmath102 , i.e. , @xmath103 , then we can construct another solution of ( [ eqn : sdp_relaxation ] ) , denoted by @xmath104 , which not only achieves the same objective value as @xmath93 , but also admits a rank - one matrix , i.e. , @xmath105 .
the new optimal solution is given as @xmath106 with @xmath107 , where @xmath108 and @xmath109 can be easily found by substituting the variables in ( [ eqn : rank - one - structure ] ) into the relaxed version of ( [ eqn : sdp_relaxation ] ) and solving the resulting convex optimization problem for @xmath108 and @xmath109 . _ proof : _ the proof of theorem 1 is similar to the proof of ( * ? ? ?
* proposition 4.1 ) and omitted here for brevity.@xmath65 by combining the results of proposition 1 and theorem 1 , the global optimal solution of ( [ eqn : cross - layer ] ) can be obtained by solving ( [ eqn : sdp_relaxation ] ) even though sdp relaxation is applied .
carrier center frequency & 915 mhz + small - scale fading distribution & rician fading with rician factor @xmath110 db + total noise variance , @xmath44 & + transmit power budget , @xmath56 & @xmath111 + number of receive antennas at each er , @xmath112 & @xmath113 + receive antenna gain & @xmath114 db + max .
tolerable channel capacity at ers , @xmath115 & @xmath116 bit / s / hz + rf energy to electrical energy conversion efficiency for er @xmath40 , @xmath117 & @xmath118 +
in this section , we study the system performance of the proposed resource allocation scheme via simulation .
there are @xmath119 information receivers and @xmath21 energy harvesting receivers , which are uniformly distributed in the range between a reference distance of @xmath113 meters and a maximum distance of @xmath120 meters .
the detailed simulation parameters can be found in table [ tab : feedback ] .
we assume that all information receivers require the same minimum sinr , i.e. , @xmath121 , for illustration .
besides , we solve the optimization problem in ( [ eqn : cross - layer ] ) and obtain the average system performance by averaging over different channel realizations . in figure [
fig : hp_sinr ] , we study the average minimum harvested power per energy harvesting receiver of the optimal scheme versus the minimum required sinr , @xmath122 , for different numbers of transmit antennas and different resource allocation schemes . it can be observed that the minimum harvested power decreases with @xmath122 .
indeed , for satisfying a more stringent sinr requirement , the transmitter is forced to allocate a higher power to the information signal and to steer the direction of transmission towards the information receivers . besides
, some degrees of freedom have to be sacrificed for artificial noise transmission for reducing the interference to the information receivers .
this leads to a smaller amount of rf energy for energy harvesting . on the other hand , when the number of antennas increase from @xmath123 to @xmath124 , a significant energy harvesting gain can be achieved by the proposed optimal scheme .
in fact , the degrees of freedom for resource allocation increase with the number of transmit antennas , which enables a more power efficient energy transfer to the energy harvesting receivers .
.,width=336 ] for comparison , we also show the performance of two simple suboptimal baseline schemes for @xmath125 . for baseline scheme 1 ,
the artificial noise is transmitted into the null space spanned by the channel of the @xmath23 information receivers . in other words ,
the artificial noise does not interfere with the desired information receivers .
then , we maximize the minimum harvested power at the energy receivers by optimizing @xmath73 and the power of artificial noise subject to the same constraints as in ( [ eqn : sdp_relaxation ] ) .
we note that the obtained @xmath73 for baseline scheme 1 may not be a rank - one matrix . as for baseline scheme 2 , it shares the same resource allocation policy as baseline scheme 1 except that the direction of beamforming matrix , @xmath73 , is fixed .
in particular , we calculate the null space of @xmath126 for desired information receiver @xmath32 where @xmath127 $ ] .
then , we project the vector @xmath128 onto the null space of @xmath126 and use the resulting vector as the direction of beamforming vector @xmath129 .
we note that @xmath73 in baseline scheme 2 is a rank - one matrix by construction
. it can be observed in figure [ fig : hp_sinr ] that for the proposed optimal scheme , the energy harvesting receivers are able to harvest more energy compared to the two baseline schemes , due to the joint optimization of @xmath73 and @xmath35 . besides
, the performance gain of the optimal scheme over the two baseline schemes is further enlarged for an increasing number of transmit antennas @xmath130 .
this can be explained by the fact that the optimal scheme can fully utilize the degrees of freedom offered by the system for resource allocation .
in contrast , although multiuser and artificial noise interference are eliminated at the information receivers for baseline scheme 1 and baseline scheme 2 , respectively , the degrees of freedom for resource allocation in the baseline schemes are reduced which results in a lower harvested energy at the energy harvesting receivers .
figure [ fig : cap_sinr ] illustrates the average secrecy capacity per information receiver versus the minimum required sinr @xmath122 of the information receivers for different numbers of transmit antennas and different resource allocation schemes .
it can be observed that the average system secrecy capacity , i.e. , @xmath131 , is a non - decreasing function with respect to @xmath122 .
in fact , the channel capacity of energy harvesting receiver @xmath40 for decoding information receiver @xmath32 is constrained to be less than @xmath132 bit / s / hz , cf .
table i. besides , we note that baseline scheme 1 is unable to meet the minimum required of secrecy capacity as specified by constraints c1 and c2 . in other words , there are time instants for baseline 1 such that @xmath133 . thereby , although baseline scheme 1 satisfies constraint @xmath62 , unlike the optimal scheme , it does not necessarily satisfy constraint c2 . on the other hand , both the proposed algorithm and baseline scheme 2 are able to meet the minimum required secrecy capacity due to the rank - one solution for beamforming matrix @xmath73 .
however , the exceedingly large secrecy capacity achieved by baseline scheme 2 comes at the expense of a smaller harvested power compared to the proposed optimal scheme , cf .
figure [ fig : hp_sinr ] .
in this paper , we studied the resource allocation algorithm design for swipt .
the algorithm design was formulated as a non - convex optimization problem to ensure the max - min fairness in energy transfer to the energy harvesting receivers .
the proposed problem formulation enabled the dual use of artificial noise for efficient energy transfer and secure communication .
sdp relaxation was adopted to obtain the optimal solution of the considered non - convex optimization problem .
simulation results unveiled the potential gain in harvested energy of our proposed optimal resource allocation scheme compared to baseline schemes .
we start the proof by rewriting constraint c2 as @xmath134 then , we propose a lower bound on the left hand side of ( [ eqn : det_ineq2 ] ) by introducing the following lemma .
d. w. k. ng , e. s. lo , and r. schober , `` energy - efficient resource allocation in multiuser ofdm systems with wireless information and power transfer , '' in _ proc .
ieee wireless commun . and networking conf . _ , 2013 .
d. w. k. ng , e. s. lo , and r. schober , `` multi - objective resource allocation for secure communication in cognitive radio networks with wireless information and power transfer , '' _ submitted for possible journal publication _ , mar .
[ online ] .
available : http://arxiv.org/abs/1403.0054 .
q. li and w. k. ma , `` spatially selective artificial - noise aided transmit optimization for miso multi - eves secrecy rate maximization , '' _ ieee trans . signal process .
_ , vol .
27042717 , may 2013 . | this paper considers max - min fairness for wireless energy transfer in a downlink multiuser communication system .
our resource allocation design maximizes the minimum harvested energy among multiple multiple - antenna energy harvesting receivers ( potential eavesdroppers ) while providing quality of service ( qos ) for secure communication to multiple single - antenna information receivers .
in particular , the algorithm design is formulated as a non - convex optimization problem which takes into account a minimum required signal - to - interference - plus - noise ratio ( sinr ) constraint at the information receivers and a constraint on the maximum tolerable channel capacity achieved by the energy harvesting receivers for a given transmit power budget .
the proposed problem formulation exploits the dual use of artificial noise generation for facilitating efficient wireless energy transfer and secure communication . a semidefinite programming ( sdp ) relaxation approach
is exploited to obtain a global optimal solution of the considered problem .
simulation results demonstrate the significant performance gain in harvested energy that is achieved by the proposed optimal scheme compared to two simple baseline schemes . |
finite temperature investigations of spontaneously broken gauge theories are of importance to the physics of the very early universe .
two prime examples are the inflationary universe and the generation of the baryon asymmetry . although symmetry restoring phase transitions in spontaneously broken gauge theories are crucial for these areas , our knowledge about them comes chiefly from perturbation theory .
motivated by the desire to learn more about their non - perturbative aspects , lattice investigations@xcite of these theories at finite temperature have been made .
our investigation of the @xmath0 spin model on anisotropic lattices is one more step in this direction .
recall that this model is obtained from the fundamental @xmath9 higgs - gauge model in its weak gauge coupling limit .
both the models are expected to be trivial , giving rise to an upper bound on the higgs mass in their respective scaling regions .
the finite temperature investigation is aimed at studying the model in this scaling region to find out about the symmetry restoring phase transition and to obtain a lower bound on the symmetry restoration temperature @xmath2 .
employing anisotropic lattices has the advantage of being able to distinguish the finite temperature effects , which could be called as a special type of finite size effects , from arbitrary finite size effects since the former have to be independent of the anisotropy in the scaling region .
in addition , anisotropic lattices allow one to study the finite temperature effects at a correlation length of order unity .
on anisotropic lattices @xmath10 , the @xmath11 spin model is defined by the action @xmath12 where @xmath13 , @xmath14 , and @xmath15(or @xmath16 for @xmath0 ) is the coupling on the isotropic lattices for which the anisotropy coupling @xmath17 is unity .
the physical volume and temperature are respectively given by @xmath18 and @xmath19 . since @xmath20 , varying @xmath21 on the lattices above amounts to holding the temperature constant in units of @xmath22 , apart from possible quantum renormalization effects .
in both the large @xmath1 approximation and the numerical simulations our procedure to investigate finite temperature effects was the following .
for a given value of the anisotropy coupling @xmath17 , we obtained the critical coupling on an @xmath10 lattice by setting @xmath17 to its classical value @xmath21 .
our results justify this choice _ a posteriori_. @xmath23 in the large @xmath1 limit is obtained by solving numerically the saddle point equation @xmath24 for @xmath25 , where @xmath26 is given by @xmath27 with the momenta @xmath28 given by @xmath29 , @xmath30 , where @xmath31 and @xmath32 .
the prime on the sum in eq .
( 2 ) indicates that the zero mode , @xmath33 , is being left out . in monte carlo(mc ) simulations the unique crossing point of the cumulant @xmath34 for various volumes @xmath35 yields @xmath36 . here
@xmath37 is the order parameter , defined by @xmath38 .
alternatively , one may use the peak position of the susceptibility , @xmath39 , to define @xmath40 . using the critical exponents of the @xmath0 model in three dimensions , @xmath36
can then be obtained using the finite size scaling theory .
the higgs mass at zero temperature was then determined at the @xmath36 by studying the zero momentum connected correlation functions of the spin variables on @xmath41 lattices in both the spatial and the temporal directions . from the exponential fall - off of the correlation functions , the higgs mass , @xmath42 ,
can be obtained using standard methods , such as fits or local distance dependent masses .
demanding euclidean invariance and by appropriately scaling the temporal direction to match these correlation functions , we determined corrections to the relation @xmath43 .
they were found to be 5 - 10% for all @xmath17-values we studied , indicating that the quantum corrections to the anisotropy are small .
the same conclusion was also obtained in the large @xmath1 limit in the symmetric phase .
1 exhibits our results for both @xmath44 and @xmath45 on @xmath46 and @xmath47 lattices for @xmath48 .
we used the spectral density method to obtain the smooth curves shown from our data , shown by crosses .
similar results have also been obtained for other values of @xmath21 and @xmath49 .
we have used @xmath50 , 1.5 and 2 , @xmath51 , 4 , 6 and @xmath52 and 24 . in each case
we obtained @xmath36 by using both the crossing point of @xmath44 and the finite size scaling of the peak position of the susceptibility .
both estimates were always found to be consistent , although we preferred to use the former for determining @xmath3 . at each coupling
, the higgs mass @xmath3 was obtained from the plateau in the local distance - dependent masses , defined as @xmath53 , where @xmath54 is the zero momentum correlation function .
again we checked that a fit to the data of an exponential form yielded consistent results with these estimates . using these results ,
the ratio @xmath55 shown in table 1 is obtained for various @xmath21 and @xmath49 .
the @xmath21-independence of the ratio is obvious .
recall that @xmath56 , as one approaches @xmath57 , i.e. , as @xmath49 grows .
thus , depending on the choice of value of the correlation length up to which an effective theory can be defined , one obtains a lower bound on the ratio @xmath5 . from table 1 , one sees this bound to be @xmath4 for a correlation length of @xmath58 2 , which decreases by 10% for @xmath8 considering the fluctuations around the saddle point of the large @xmath1 limit , one can obtain the higgs mass @xmath3 at @xmath59 , while the corresponding renormalized vacuum expectation value of the field is given by @xmath60 .
2 shows these large @xmath1 results for @xmath5 and @xmath61 .
both are seen to be clearly independent of @xmath21 .
further , the latter seems to be independent of @xmath49 for @xmath62 .
qualitatively , the large @xmath1 limit seems to reproduce all the features of the monte carlo(mc ) data well .
however , quantitatively , the large @xmath1 results seem to lie systematically lower than the mc results by @xmath5815 %
. it would be interesting to check whether the early scaling evident in the mc data for @xmath51 is real by simulating the theory at more @xmath21 values and also by studying the @xmath63 ratio .
9 h. g. evertz , j. jersk and k. kanaya , nucl . phys .
* b285[fs19 ] * ( 1987 ) 229 ; p. h. damgaard and u. m. heller , nucl . phys .
* b304 * ( 1988 ) 63 .
u. m. heller , phys . lett .
* b191 * ( 1987 ) 109 ; k. jansen and p. seuferling , nucl . phys .
* b343 * ( 1990 ) 507 . | results of investigations of the @xmath0 spin model at finite temperature using anisotropic lattices are presented . in both the large @xmath1 approximation and the numerical simulations using the wolff cluster algorithm
we find that the ratio of the symmetry restoration temperature @xmath2 to the higgs mass @xmath3 is independent of the anisotropy .
we obtain a lower bound of @xmath4 for the ratio , @xmath5 , at @xmath6 , which is lowered further by about @xmath7 at @xmath8 |
anisotropies of the cosmic microwave background radiation ( cmb ) are directly related to the origin of structure in the universe .
galaxies and clusters of galaxies eventually formed by gravitational instability from primordial density fluctuations , and these same fluctuations left their imprint on the cmb .
recent balloon @xcite and ground - based interferometer @xcite experiments have produced reliable estimates of the power spectrum of the cmb temperature anisotropies . while they helped eliminate certain candidate theories for the primary source of cosmic perturbations ,
the power spectrum data are still compatible with theoretical estimates of a relatively large variety of models , such as @xmath0cdm , quintessence models or some hybrid models including cosmic defects .
these models , however , differ in their predictions for the statistical distribution of the anisotropies beyond the power spectrum .
the map ( currently in space ) and planck ( scheduled for launch in 2007 ) satellite missions will provide high - precision data allowing definite estimates of non - gaussian signals in the cmb .
it is therefore important to know precisely what are the predictions of all candidate models for the statistical quantities that will be extracted from the new data , and to identify specific signatures of the various models .
there are two main classes of models of structure formation_passive _ and _ active _ models . in passive models , density inhomogeneities
are set as initial conditions at some early time , and while they subsequently evolve as described by einstein - boltzmann equations , no additional perturbations are seeded . on the other hand , in active models
a time - dependent source generates new density perturbations through all time .
most realizations of passive models are based on the idea of inflation . in simplest inflationary models
it is assumed that there exists a weakly coupled scalar field @xmath1 , called the inflaton , which `` drives '' the ( approximately ) exponential expansion of the universe .
quantum fluctuations of @xmath1 are stretched by the expansion to scales beyond the horizon , thus `` freezing '' their amplitude .
inflation is followed by a period of thermalization , or reheating , during which energy is transferred to usual particles . because of the spatial variations of @xmath1 introduced by quantum fluctuations , thermalization occurs at slightly different times in different parts of the universe .
such fluctuations in the thermalization time give rise to density fluctuations .
because of their quantum nature and because of the fact that initial perturbations are assumed to be in the vacuum state and hence well described by a gaussian distribution , perturbations produced during inflation are expected to follow gaussian statistics to a high degree , or either be products of gaussian random variables .
this is a fairly general prediction that will be tested shortly with map and more thoroughly in the future with planck data .
active models of structure formation are motivated by cosmic topological defects with the most promising candidates being cosmic strings .
it is widely believed that the universe underwent a series of phase transitions as it cooled down due to the expansion .
if our ideas about grand unification are correct , then some cosmic defects , such as domain walls , strings , monopoles or textures , should have formed during phase transitions in the early universe @xcite .
once formed , cosmic strings could survive long enough to seed density perturbations .
generically , perturbations produced by active models are not expected to be gaussian .
defect models possess the attractive feature that they have no parameter freedom , as all the necessary information is in principle contained in the underlying particle physics model .
this is a feature that fascinated dennis sciama : during his days at trieste , he would always be willing to discuss these `` marvelous cosmic defects '' with everyone .
he even encouraged a few students _ sans liaisons contraignantes _ to pursue studies on this promising subject and facilitated all necessary means for it .
the relevant , ubiquitous dimensionless quantity for every astrophysical and cosmological signature left by strings is @xmath2 , where @xmath3 is newton s constant and @xmath4 stands for the mass per unit length of the string , with a value of roughly @xmath5 for gut strings .
this happens to be just the correct order of magnitude of the level of cmb anisotropies .
back in 1996 , the power spectrum plot of @xmath6 vs. @xmath7 consisted of some scattered points and strings were considered a likely candidate for the origin of structure formation .
moreover , the error bars were so big that dennis would never miss the opportunity to tease inflation fans who wanted to see a peak in that plot : `` come on '' , he would say , `` you do not claim you actually _ see _ a peak there , do you .. ! '' . to close this section ,
let us mention that , in addition to purely active or passive scenarios , perturbations could be seeded by some combination of the two mechanisms . for example , cosmic strings could have formed just before the end of inflation and partially contributed to seeding density fluctuations .
it has been shown @xcite that such hybrid models are very successful in fitting the cmb power spectrum data .
cosmic strings are generally expected to produce distinguishing non - gaussian features in the cmb and it will soon become possible to look for them in the data from map and planck .
calculating cmb anisotropies sourced by topological defects is a rather difficult task . in inflationary scenario
the entire information about the seeds is contained in the initial conditions for the perturbations in the metric . in the case of cosmic defects ,
perturbations are continuously seeded over the period of time from the phase transition that had produced them until today .
the exact determination of the resulting anisotropy requires , in principle , the knowledge of the energy - momentum tensor [ or , if only two point functions are being calculated , the unequal time correlators @xcite ] of the defect network and the products of its decay at all times .
this information is simply not available !
instead , a number of clever simplifications , based on the expected properties of the defect networks ( _ e.g. _ scaling ) , are used to calculate the source . the latest data from boomerang @xcite , maxima @xcite and dasi @xcite experiments clearly disagree with the predictions of these simple models of defects @xcite .
the shape of the cmb angular power spectrum is determined by three main factors : the geometry of the universe , coherence and causality .
the curvature of the universe directly affects the paths of light rays coming to us from the surface of last scattering @xcite . in a closed universe , because of the lensing effect induced by the positive curvature , the same physical distances between points on the sky would correspond to larger angular scales . as a result ,
the peak structure in the cmb angular power spectrum would shift to the larger angular scales or the smaller values of @xmath7 .
the prediction of the cosmic string model of @xcite for @xmath8 is shown in fig .
[ omega ] .
= 0.6 0.1 truecm as can be seen , the main peak in the angular power spectrum can be matched by choosing a reasonable value for @xmath9 .
however , even with the main peak in the right place the agreement with the data is far from satisfactory .
the peak is significantly wider than that in the data and there is no sign of a rise in power at @xmath10 suggested by the data .
the situation is even worse for global strings which fail to produce a distinct peak of any kind . for local strings , the sharpness and the height of the main peak in the angular spectrum
can be enhanced by including the effects of the gravitational radiation @xcite and wiggles @xcite .
more precise high - resolution numerical simulations of string networks in realistic cosmologies with a large contribution from @xmath11 are needed to determine the exact amount of small - scale structure on the strings and the nature of the products of their decay .
it is , however , unlikely that including these effects alone would result in a sufficiently narrow main peak or any presence of secondary peaks .
it seems that now dennis would not have had strong arguments to tease his friends .
let us next consider issues of causality and coherence and how the random nature of the string networks comes into the calculation of the anisotropy spectrum .
both experimental and theoretical results for the cmb power spectra involve calculations of averages . to estimate the correlations of the observed temperature anisotropies , experimentalists compute the average over all available patches on the sky .
when calculating the predictions of their models , theorists find the average over the _ ensemble _ of possible outcomes provided by the model . in inflationary models , as in all passive models ,
only the initial conditions for the perturbations are random .
the subsequent evolution is the same for all members of the ensemble . for wavelengths higher than the hubble radius ,
the linear evolution equations for the fourier components of such perturbations have a growing and a decaying solution .
the modes corresponding to smaller wavelengths have only oscillating solutions . as a consequence ,
prior to entering the horizon , each mode undergoes a period of phase `` squeezing '' which leaves it in a highly coherent state by the time it starts to oscillate .
coherence here means that all members of the ensemble , corresponding to the same fourier mode , have the same temporal phase .
so even though there is randomness involved , as one has to draw random amplitudes for the oscillations of a given mode , the time behavior of different members of the ensemble is highly correlated .
the total power spectrum is an ensemble averaged superposition of all fourier modes and the coherence results in an interference pattern seen as the acoustic peaks .
= 0.7 0.5 truecm in contrast , the evolution of the string network is highly non - linear .
cosmic strings are expected to move at relativistic speeds , self - intersect and reconnect in a chaotic fashion .
a consequence of this behavior is that the unequal time correlators of the string energy - momentum vanish for time differences larger than a certain coherence time .
members of the ensemble corresponding to a given mode of perturbations will have random temporal phases with the `` dice '' thrown on average once in each coherence time .
the coherence time of a realistic string network is rather short . as a result
, the interference pattern in the angular power spectrum is completely washed out .
causality manifests itself , first of all , through the initial conditions for the string sources , the perturbations in the metric and the densities of different particle species . if one assumes that defects are formed by a causal mechanism in an otherwise smooth universe , then the correct initial conditions are obtained by setting the components of the stress - energy pseudo - tensor @xmath12 to zero @xcite .
these are the same as the isocurvature initial conditions @xcite .
a generic prediction of isocurvature models ( assuming perfect coherence ) is that the first acoustic peak is almost completely hidden .
the main peak is then the second acoustic peak and in flat geometries it appears at @xmath13 .
this is due to the fact that after entering the horizon a given fourier mode of the source perturbation requires time to induce perturbations in the photon density .
causality also implies that no superhorizon correlations in the string energy density are allowed .
the correlation length of a `` realistic '' string network is normally between 0.1 and 0.4 of the horizon size .
an interesting study was performed by magueijo _
_ where they have constructed a toy model of defects with two parameters : the coherence length and the coherence time .
the coherence length was taken to be the scale at which the energy density power spectrum of the strings turns from a power law decay for large values of @xmath14 into a white noise at low @xmath14 .
this is essentially the scale corresponding to the correlation length of the string network .
the coherence time was defined in the sense described in the beginning of this section , in particular , as the time difference needed for the unequal time - correlators to vanish .
their study showed ( see fig .
[ magetal ] ) that by accepting any value for one of the parameters and varying the other ( within the constraints imposed by causality ) one could reproduce the oscillations in the cmb power spectrum .
unfortunately for cosmic strings , at least as we know them today , they fall into the parameter range corresponding to the upper right corner in fig .
[ magetal ] . in order to fit the observations , cosmic strings
must either be more coherent or they have to be stretched over larger distances , which is another way of making them more coherent . to understand this , imagine that there were only one long straight string stretching across the universe and moving with some velocity .
the evolution of this string would be linear and the induced perturbations in the photon density would be coherent . by increasing the correlation length of the string network
we would move closer to this limiting case of just one long straight string and so the coherence would be enhanced .
= 0.7 0.1 truecm the question of whether or not defects can produce a pattern of the cmb power spectrum similar to , and including the peaks of , that produced by the adiabatic inflationary models was repeatedly addressed in the literature . in particular
, it was shown that one can construct a causal model of active seeds which for certain values of parameters can reproduce the oscillations in the cmb spectrum .
the main problem today is that current realistic models of cosmic strings fall out of the parameter range that is needed to fit the observations .
in addition to purely active or passive models , perturbations could be seeded by some combination of the two mechanisms @xcite .
figure [ mixed ] ( borrowed from bouchet et al . ) shows two uncorrelated spectra produced by an inflationary @xmath0cdm model ( dot - dashed line ) and a global string model ( dashed line ) , both normalized on the cobe data , as well as a weighted sum of the two ( solid line ) .
the best fit to the latest boomerang , maxima and dasi data is obtained with a non - negligible string contribution of @xmath15% . at the moment ,
models combining strings with inflation @xcite or models of varying speed of light @xcite are the only ones involving topological defects that to some extent can match the observations .
one possible way to distinguish their predictions from those of inflationary models would be to compute a non - gaussian statistic , such as the cmb bispectrum .
let us now turn to a computation of the bispectrum of cmb temperature anisotropies as generated by simulated cosmic strings . to calculate the sources of perturbations we use an updated version of the wiggly string model of ref.@xcite ( see also @xcite ) .
the network is represented by a collection of uncorrelated , straight string segments moving with random , uncorrelated velocities , and are produced at some early epoch . at every subsequent epoch , a certain fraction of the number of segments decays in a way that maintains network scaling .
the length of each segment at any time is taken to be equal to the correlation length of the network which , together with the root mean square velocity of segments , are computed from the velocity - dependent one - scale model of martins and shellard @xcite .
the positions of segments are drawn from a uniform distribution in space and their orientations are chosen from a uniform distribution on a two - sphere . according to the one - scale model @xcite , the string network at any time
can be characterized by a single length scale , the correlation length @xmath16 , defined by @xmath17 , where @xmath18 is the energy density in the string network .
expansion of space stretches the strings without altering their mass per unit length .
if simply stretched in that manner , strings would soon come to dominate the energy density of the universe .
however , at the same time with the expansion , long strings reconnect and chop off loops which later decay
. these two competing effects result in an intermediate steady - state ( the so - called `` scaling solution '' ) in which the characteristic scale of the string network remains constant relative to the horizon size .
the total energy of the string network in a volume @xmath19 at any time @xmath20 is @xmath21 where @xmath22 is the total number of string segments at that time , @xmath23 , @xmath24 is the expansion factor and @xmath25 is a constant simulation volume . from eq.([vrho ] ) it follows that @xmath26 , where we have defined @xmath27 .
the comoving length @xmath28 is approximately proportional to the conformal time @xmath20 and implies that the number of strings @xmath29 within the simulation volume @xmath25 falls as @xmath30 . to calculate the cmb anisotropy we need to evolve the string network over at least four orders of magnitude in cosmic expansion .
hence we would have to start with @xmath31 string segments in order to have one segment left at the present time .
keeping track of such a large number of segments is numerically impossible .
fortunately , a way around this difficulty was suggested in ref .
@xcite .
the suggestion is to consolidate all string segments that decay at the same epoch .
the number of segments that decay at the ( discretized ) conformal time @xmath32 is @xmath33 \label{eq : nd}\ ] ] where @xmath34^{-3}$ ] is the number density of strings at time @xmath35 .
then , with the formalism we employ , instead of summing over @xmath36 segments we now sum over only @xmath37 segments , making @xmath37 a parameter .
such a simplification only works for computation of two- and three - point correlation functions under the assumption that string segments are statistically independent .
we refer the interested reader to our recent article @xcite , where we explain the procedure in more detail . as a final comment ,
let us note that the simulation model in this form does not allow computation of cmb sky maps .
this is because the method of finding the two- and three - point functions we describe involves `` consolidated '' quantities which do not correspond to the energy - momentum tensor of a real string network .
these quantities are auxiliary and specially prepared to give the correct two- or three - point functions after ensemble averaging .
having clarified this , the stress - energy from the simulated string model can be found and used to compute the appropriate density perturbations which lead to the expected cmb temperature anisotropies .
it is conventional to expand cmb temperature fluctuations into spherical harmonics : @xmath38 the coefficients @xmath39 can be decomposed into fourier modes , @xmath40 given the sources @xmath41 produced by the string model , @xmath42 are found by solving linearized einstein - boltzmann equations and integrating along the line of sight @xcite .
this procedure can be written as the action of a linear operator @xmath43 on the source energy - momentum tensor , @xmath44 .
the third moment of @xmath39 s can be expressed as @xmath45 because of homogeneity we have @xmath46 given scalar values @xmath47 , @xmath48 , @xmath49 , there is a unique ( up to an overall rotation ) triplet of directions @xmath50 for which the rhs of eq .
( [ p3lvector ] ) does not vanish .
the quantity @xmath51 is invariant under an overall rotation of all three vectors @xmath52 and therefore may be equivalently represented by a function of scalar values @xmath47 , @xmath48 , @xmath49 .
hence , we get @xmath53 the triangle constraint @xmath54 must be satisfied . then , in terms of the simulation volume @xmath25 , we have @xmath55 given an arbitrary direction @xmath56 and the magnitudes @xmath57 , @xmath58 and @xmath59 , the directions @xmath60 and @xmath61 are specified up to rotations by the triangle constraint .
both sides of eq .
( [ p3l0 ] ) are function of scalar @xmath62 only .
the expression on the rhs will be computed numerically by averaging over different realizations of the string network _ and _ over all directions @xmath56 .
because of isotropy and since the allowed sets of directions @xmath63 are planar , it is enough to restrict the numerical calculation to directions @xmath63 in a fixed plane .
this significantly reduces the computational expense . substituting eqs .
( [ p3lvector ] ) to ( [ p3l0 ] ) into ( [ eq:3alm-1 ] ) , fourier transforming the dirac delta and using the rayleigh identity , we can perform all angular integrations analytically and obtain a compact form for the third moment @xmath64 where , noting @xmath65 a wigner 3@xmath66-symbol , we have @xmath67 and where we have defined , in terms of spherical bessel functions @xmath68 , @xmath69 our numerical procedure consists of computing the rhs of eq .
( [ eq:3alm - res ] ) by evaluating the necessary integrals and averaging over many realizations of the strings sources .
readers wishing more details can consult ref .
@xcite . from eq .
( [ eq:3alm - res ] ) we easily derive the cmb bispectrum @xmath70 , defined as @xcite @xmath71 we now restrict our calculations to the angular bispectrum @xmath72 in the ` diagonal ' case , _
i.e. _ @xmath73 .
this is a representative case and , in fact , the one most frequently considered in the literature . the power spectrum is usually plotted in terms of @xmath74 which , apart from constant factors , is the contribution to the mean squared anisotropy of temperature fluctuations per unit logarithmic interval of @xmath7 . in full analogy with this ,
the relevant quantity to work with in the case of the bispectrum is @xmath75 for large values of the multipole index @xmath7 , @xmath76 .
note also what happens with the 3@xmath77-symbols appearing in the definition of the coefficients @xmath78 : the symbol @xmath79 is absent from the definition of @xmath72 , while in eq .
( [ eq : qtp ] ) the symbol @xmath80 is squared .
hence , there are no remnant oscillations due to the alternating sign of @xmath81 .
however , even more important than the value of @xmath82 itself is the relation between the bispectrum and the cosmic variance associated with it . in fact , it is their comparison that tells us about the observability ` in principle ' of the non - gaussian signal .
the cosmic variance constitutes a theoretical uncertainty for all observable quantities and is due to having just one realization of the stochastic process , in our case , the cmb sky .
the way to proceed is to employ an estimator @xmath83 for the bispectrum and compute the variance from it . by choosing an unbiased estimator
we ensure it satisfies @xmath84 .
however , this condition does not isolate a unique estimator . the proper way to select the _
best unbiased _ estimator is to compute the variances of all candidates and choose the one with the smallest value .
the estimator with this property was computed in ref .
@xcite and is @xmath85 the variance of this estimator , assuming a mildly non - gaussian distribution , can be expressed in terms of the angular power spectrum @xmath86 as follows @xcite @xmath87 the theoretical signal - to - noise ratio for the bispectrum ( in the diagonal case @xmath73 ) is then given by latexmath:[\[(s / n)_{l_{1}l_{2}l_{3 } } = incorporating all the specifics of the particular experiment , such as sky coverage , angular resolution , etc .
, will allow us to give an estimate of the particular non - gaussian signature associated with a given active source and , if observable , indicate the appropriate range of multipole @xmath7 s where it is best to look for it . in fig .
[ fig:1 ] we show the results for @xmath89 [ cf .
( [ eq : qtp ] ) ] .
it was calculated using the string model with @xmath90 consolidated segments in a flat universe with cold dark matter and a cosmological constant .
only the scalar contribution to the anisotropy has been included .
vector and tensor contributions are known to be relatively insignificant for local cosmic strings and can safely be ignored in this model @xcite .
the plots are produced using a single realization of the string network by averaging over @xmath91 directions of @xmath52 .
the comparison of @xmath89 ( or equivalently @xmath92 ) with its cosmic variance [ cf .
( [ eq : sigma ] ) ] clearly shows that the bispectrum ( as computed from our cosmic string model ) lies hidden in the theoretical noise and is therefore undetectable for any given value of @xmath7 .
let us note , however , that in its present stage our string code describes brownian , wiggly long strings in spite of the fact that long strings are very likely not brownian on the smallest scales .
in addition , the presence of small string loops @xcite and gravitational radiation into which they decay were not yet included in our model .
these are important effects that could , in principle , change our predictions for the string - generated cmb bispectrum on very small angular scales .
the imprint of cosmic strings on the cmb is a combination of different effects . prior to the time of recombination strings
induce density and velocity fluctuations on the surrounding matter . during the period of last scattering these fluctuations
are imprinted on the cmb through the sachs - wolfe effect : namely , temperature fluctuations arise because relic photons encounter a gravitational potential with spatially dependent depth .
in addition to the sachs - wolfe effect , moving long strings drag the surrounding plasma and produce velocity fields that cause temperature anisotropies due to doppler shifts . while a string segment by itself is a highly non - gaussian object , fluctuations induced by string segments before recombination are a superposition of effects of many random strings stirring the primordial plasma .
these fluctuations are thus expected to be gaussian as a result of the central limit theorem . as the universe becomes transparent
, strings continue to leave their imprint on the cmb mainly due to the kaiser - stebbins effect .
this effect creates line discontinuities in the temperature field of photons passing on opposite sides of a moving long string .
however , this effect can produce non - gaussian perturbations only on sufficiently small scales .
this is because on scales larger than the characteristic inter - string separation at the time of the radiation - matter equality , the cmb temperature perturbations result from superposition of effects of many strings and are likely to be gaussian . currently , strongest constraints on cosmic string models come from the cmb power spectrum data . predictions of strings - only models are inconsistent with data from boomerang , maxima and dasi
. however , the data can still be fitted reasonably well by `` hybrid '' models that combine strings and adiabatic inflationary perturbations as sources for primordial fluctuations .
presence of cosmic strings could , in principle , be distinguished by non - gaussian statistics .
the new high - resolution cmb measurements from map and planck will allow a precise estimate of non - gaussian signals in the cmb .
the most straightforward non - gaussian discriminator is the three - point function of temperature , or equivalently ( in fourier space ) the angular bispectrum . in ref .
@xcite , we have developed a numerical method for computing the cmb angular bispectrum from first principles in any active model of structure formation , such as cosmic defects , where the energy - momentum tensor is known or can be simulated .
the method does not use cmb sky maps and requires a moderate amount of computations .
we applied this method to the computation of some relevant components of the bispectrum produced from a model of cosmic strings and did not find a non - gaussian signal that would be observable in the forthcoming satellite - based cmb missions .
this indicates that either the model we used to simulate cosmic strings is inadequate for extracting the non - gaussian signature , or that the angular bispectrum is not an appropriate statistic for detecting cosmic strings .
avelino _ et al . _
@xcite applied several non - gaussian tests to the perturbations seeded by cosmic strings .
they found the density field distribution to be close to gaussian on scales larger than @xmath93 mpc , where @xmath94 is the fraction of cosmological matter density in baryons and cdm combined .
scales this small correspond to the multipole index of order @xmath95 .
we have not attempted a calculation the cmb bispectrum on these scales because the linear approximation is almost guaranteed to fail at such small scales , and because of increased computational cost for higher @xmath7 multipoles .
the interest in cosmic strings has considerably declined since they have been ruled out as the primary seed for structure formation in the universe .
this , however , is not at all sufficient to rule out strings as an interesting topic of research . studying formation , evolution , properties and cosmological implications of cosmics strings and other topological defects remains to be both exciting and important .
certain cosmological puzzles , such as baryogenesis and ultra - high energy cosmic rays , could be resolved with the help of topological defects @xcite .
recent experimental and theoretical progress in understanding properties of topological defects in low - temperature condensed matter systems , such as superfluid helium and superconductors , has led to an intriguing possibility of studying cosmology in the lab @xcite .
any new experimental constraint on the population and properties of cosmic defects is potentially relevant to building viable cosmological and particle physics models .
since current and future cmb experiments will provide the most precise cosmological data , it is absolutely essential to use them maximally in learning as much as we can about cosmic defects .
s. hanany , _ et al .
_ , astrophys .
j. * 545 * ( 2000 ) l5 ; preprint astro - ph/0005123 ; + a.h .
jaffe , _ et al .
_ , phys .
* 86 * ( 2001 ) 3475 - 3479 ; preprint astro - ph/0007333 ; a. t. lee , _ et al .
_ , preprint astro - ph/0104459 .
a. vilenkin , e.p.s .
shellard , _ `` cosmic strings and other topological defects '' _ , paperback ( 2nd ) edition , cambridge u. press , 2000 ; + m.b .
hindmarsh and t.w.b .
kibble , _ `` cosmic strings '' _ , rept .
58 * , 477 - 562 ( 1995 ) ; preprint hep - ph/9411342 ; + a. gangui , _
`` topological defects in cosmology '' _ , lecture notes for the first bolivian school on cosmology , to appear in the proceedings ( 2002 ) ; preprint astro - ph/0110285 . c. contaldi , m. hindmarsh , j. magueijo , phys
* 82 * ( 1999 ) 2034 ; r. battye and j. weller , phys . rev .
* d61 * , 043501 ( 2000 ) ; f.r .
bouchet , p.p .
peter , a. riazuelo , m. sakellariadou , astro - ph/0005022 ; m. sakellariadou , astro - ph/0111501 . | we briefly review certain aspects of cosmic microwave background anisotropies as generated in passive and active models of structure formation .
we then focus on cosmic strings based models and discuss their status in the light of current high - resolution observations from the boomerang , maxima and dasi collaborations .
upcoming megapixel experiments will have the potential to look for non - gaussian features in the cmb temperature maps with unprecedented accuracy .
we therefore devote the last part of this review to treat the non - gaussianity of the microwave background and present a method for computation of the bispectrum from simulated string realizations .
cosmic strings ; cosmic microwave background radiation ; bispectrum |
proton scattering has been widely used as a means to study both collective and microscopic aspects of nuclear structure @xcite .
the study is consistent only if a well - defined effective nucleon - nucleon ( nn ) interaction is applied in the analysis .
also , with the advent of radioactive nuclear beams there is constant enhancement of our knowledge frontiers on the structure and reaction dynamics of the known stable nuclei as well as their less known unstable counterparts @xcite .
this rapidly developing field provides a testing ground for different nuclear reaction theories and effective interactions . scattering involving @xmath0ne and @xmath0o are interesting because not only are they mirror nuclei , but also @xmath0o is a stable nucleus while @xmath0ne is its radioactive counterpart . in this work ,
proton scattering on @xmath0ne and @xmath0o has been studied at low energies ( @xmath1100 mev / a ) @xcite in a folding model approach .
the folding model is well known as a powerful tool for analyzing nucleus - nucleus scattering data at relatively low incident energies @xcite .
it directly links the density profile of the nucleus with the scattering cross sections and is thus very appropriate for studying nuclei , especially those with exotic matter distributions .
a semi - microscopic analysis in the optical model ( om ) framework is carried out . in the dwba calculations of nuclear excitation , with transferred angular momentum @xmath2 ,
the form factors are obtained by taking the derivatives of the potentials used .
the nucleon - nucleus potential can be obtained by single folding the density distribution of the nucleus with the nucleon - nucleon effective interaction @xcite as , @xmath3 where @xmath4 is density of the nucleus at @xmath5 and @xmath6 is the effective interaction between two nucleons at the sites @xmath7 and @xmath5 .
two different forms of effective interactions have been employed in this work .
we perform a comparative study between the modified seyler - blanchard ( sbm ) and density dependent m3y ( ddm3y ) effective nn interactions .
the finite range , density , momentum and isospin dependent effective interaction sbm has different strengths for pp ( or nn ) and pn interactions and its form is @xcite , @xmath8\ ] ] where , the subscripts ` @xmath2 ' and ` @xmath9 ' refer to like - pair ( nn or pp ) and unlike - pair ( np ) interactions , respectively . here ` @xmath10 ' is the range of the two - body interaction , ` @xmath11 ' is a measure of the strength of repulsion with relative momentum ` @xmath12 ' , while ` @xmath13 ' and ` @xmath14 ' are two parameters determining the strength of density dependence .
@xmath15 and @xmath4 are densities at the sites of the two interacting nucleons .
the values of parameters @xmath16 are given in table 1 .
these constants are found to reproduce the bulk properties of nuclear matter and of finite nuclei @xcite and are known also to explain the p + @xmath17he,@xmath18li scattering data successfully @xcite .
the parameters are determined without exchange effects and thus they contain the effect indirectly though in a very approximate way .
the finite range m3y effective interaction @xmath19 appearing in eqn
. 1 is given by @xcite @xmath20 this interaction is based upon a realistic g - matrix , which was constructed in an oscillator representation .
effectively it is an average over a range of nuclear densities as well as energies and therefore the m3y has no explicit density or energy dependence .
the only energy dependent effect that arises from its use is a rather weak one contained in an approximate treatment of single - nucleon knock - on exchange . the density and energy averages are adequate for the real part of the optical potential for heavy ions at lower energies . although , it is important to consider the density and energy dependence explicitly for scattering at higher energies , where the effects of a nuclear rainbow are seen and hence the scattering becomes sensitive to the potential at small radii .
such cases were studied introducing suitable and semirealistic explicit density dependence @xcite into the m3y interaction which was then called the ddm3y and was very successful for interpreting consistently the high energy elastic @xmath21 and heavy ion scattering data .
the present calculations use the density dependent m3y effective interaction ( ddm3y ) supplemented by a zero - range pseudo potential . in ddm3y ,
the effective nucleon - nucleon interaction @xmath19 is assumed to be density and energy dependent and therefore becomes functions of density and energy and is given by @xmath22 where @xmath23 is the same m3y interaction given by eqn . 3 but supplemented by a zero range pseudo - potential @xcite @xmath24 where the zero - range pseudo - potential representing the single - nucleon exchange term is given by @xmath25 and the density dependent
part has been taken to be @xcite @xmath26 which takes care of the higher order exchange effects and the pauli blocking effects .
@xmath27 is energy per nucleon .
the constants of this interaction @xmath28 and @xmath29 when used in single folding model description , can be determined by nuclear matter calculations @xcite as 2.07 and 1.624 @xmath30 respectively .
the density dependence parameter @xmath29 has the dimension of cross - section .
the term @xmath31 reduces the strength of the interaction and changes sign at high densities making it repulsive .
this is a direct consequence of the pauli blocking effect .
thus @xmath31 represents the probability of non - interaction arising due to collision probability @xmath32 of a nucleon in nuclear medium of density @xmath33 .
the parameter @xmath29 can be identified as the ` in medium ' effective nucleon - nucleon interaction cross section @xmath34 .
this value of @xmath29 along with nucleonic density of infinite nuclear matter @xmath35 can also provide the nuclear mean free path @xmath36 .
the nuclear ground state densities have been calculated in the framework of spherical hartree fock plus bcs calculations in co - ordinate space using two different parameter sets of the skyrme interactions given by @xcite @xmath37 + t_2(1 + x_2{\bf p}_{\sigma } ) { \bf k'}\delta({\bf r } ) { \bf k } + t_3(1 + x_3{\bf p}_{\sigma})\rho^{\alpha } ( { \bf r})\delta({\bf r})\ ] ] there are negligible differences in the ground state densities of the nuclei with the two different parameter sets ( for siii @xcite and skm * @xcite ) provided in table 2 .
we show skm * parameterization for the @xmath0o ground state density ( fig . 1 ) and used it for calculations of the single folding potential and form factor .
we carried out the same procedure for @xmath0ne and for skm * the binding energies per nucleon are in good agreement ( within 0.1 mev ) with the experimental values for both the nuclei .
the potentials used while calculating phenomenological best fits @xcite have the following form , @xmath38 where , @xmath39^{-1}$ ] , @xmath40 and @xmath41 .
the subscripts @xmath42 denote real , volume imaginary , surface imaginary and spin - orbit respectively and @xmath43 , @xmath44 ( @xmath45 ) and @xmath46 are the strengths of the real , volume ( surface ) imaginary and spin - orbit potentials respectively .
@xmath47 is the coulomb potential of a uniformly charged sphere of radius 1.20 @xmath48 . in semi - microscopic analysis both the volume
real ( @xmath49 ) and volume imaginary ( @xmath50 ) parts of the potentials ( generated microscopically by folding model ) are assumed to have the same shape , as in ref .
@xcite , i.e. @xmath51 = ( @xmath52 + @xmath53)@xmath54(@xmath55 ) where , @xmath52 and @xmath56 are the renormalization factors for real and imaginary parts respectively @xcite . in fig .
2 the p + @xmath0o renormalized real folded potentials ( employing the sbm and ddm3y interactions ) are shown at @xmath57 = 24.5@xmath58 mev , along with the best fit phenomenological real part . thus the potentials for elastic scattering analysis include real and volume imaginary terms ( folded potentials ) and also surface imaginary and spin - orbit terms ( best fit phenomenological potentials ) . for each angular distribution ,
best fits are obtained by minimizing @xmath59/n , where @xmath59 = @xmath60 ^ 2 $ ] , where @xmath61 and @xmath62 are the theoretical and experimental cross sections respectively , at angle @xmath63 , @xmath64 is the experimental error and n is the number of data points .
the potentials for elastic scattering analysis are subsequently used in the dwba calculations of inelastic scattering with transferred angular momentum @xmath2 .
the calculations are performed using the code dwuck4 @xcite .
the derivative of the potentials ( @xmath65 ) are used as the form factors .
the microscopic real and imaginary form factors have the same shape with strengths @xmath66 and @xmath67 respectively , where @xmath68 = @xmath69 , where the radius parameter @xmath70 is the rms radius of the folded potential .
in addition , form factors derived from phenomenological surface imaginary and spin - orbit potentials are included .
the deformation parameters @xmath71 are determined by fitting the inelastic scattering angular distribution .
the renormalizations required for the potentials are reminiscent of those for deuteron and @xmath72li scattering and it may be surmised that it is for the same reasons ; weak binding and ease of breakup and other reaction channels .
table 1 gives the parameters of the interactions used here .
both interactions provide incompressibility of @xmath73 300 mev for spin and isospin symmetric cold infinite nuclear matter .
moreover , in case of ddm3y , @xmath29 value of 1.624 @xmath30 obtained from nuclear matter calculations , along with nucleonic density of 0.16 @xmath74 provides a mean free path of about 4 fm which is in excellent agreement with other theoretical estimates @xcite .
table 3 gives the phenomenological best fit optical model parameters while table 4 gives the renormalization factors , @xmath71 values , @xmath59/n for the folding model analysis .
the relationship between the reduced electric quadrupole transition rate @xmath75 for the ground state to the @xmath76 state in units of @xmath77 and the quadrupole deformation parameter @xmath71 is given by @xcite @xmath78 where @xmath79 fm and @xmath80 is the atomic number .
the quadrupole deformations listed in reference @xcite were obtained by using eqn .
10 but keeping only the terms up to first order in @xmath71 .
we have recalculated these values by keeping terms up to third order in @xmath71 .
the recalculated quadrupole deformations thus obtained from the experimental @xmath75 values listed in reference @xcite are 0.33078 and 0.59284 for @xmath0o and @xmath0ne respectively .
as can be seen from table 4 , the quadrupole deformation obtained from the present analysis for @xmath0o is in excellent agreement while that for @xmath0ne is significantly underestimated due to lack of experimental data at forward angles .
the inelastic scattering is more sensitive at forward angles due to its relative purity compared to data corresponding to backward angular range where other non - elastic processes also contribute .
the quality of elastic as well as the inelastic fits deteriorate at backward angles and similar deterioration of fits are also seen for proton scattering from other nuclei @xcite .
the reason for this is probably that the full cross section is ascribed to potential scattering while quasi - compound - nucleus formation feeds back into the elastic channel and whose energy - dependence is controlled by barrier - top resonances .
the backward angular range classically corresponds to smaller impact parameters .
this fact suggests higher compound nuclear formation probabilities at backward angles while those at forward angles are expected to be negligibly small . since relative contributions of compound elastic and direct elastic are not disentangled , @xmath59 was calculated only upto about 90@xmath81 in center of mass during fitting the @xmath0o data .
it may be noted that since we could not acquire the experimental data for @xmath0o , they have been read quite accurately from the plots in the original paper @xcite and a 5@xmath82 uniform error has been assumed .
the technique used for extracting the data from plots along with relevant co - ordinate transformations are described in the appendix .
in the present study we find that the parametrized sbm effective interaction , and the realistic ddm3y effective interaction obtained from sophisticated g - matrix calculations provide equivalent descriptions for the elastic and inelastic scattering of protons from the mirror nuclei @xmath0ne and @xmath0o ( fig . 3 , 4 ) .
the values of the deformation parameters have been extracted from the calculations for these nuclei . even though the analysis reported here is quite detailed , measurements at higher energies may still be useful in distinguishing various effective interactions .
the lack of sensitivity is due to the fact that at such energies , nuclear densities primarily probed are near the surface , which makes the density dependent effects less realizable .
the form of density dependence used here is more physical as compared to other forms .
the parameter @xmath29 can be interpreted as ` in medium ' nucleon - nucleon interaction cross section while @xmath31 as the non - interaction probability arising due to higher order exchange and pauli blocking effects . in summary , a consistent folding model analysis of proton scattering on a = 18 nuclei
is carried out using two different effective nn interactions .
the conventional way of generating the form factors is followed , that is , by taking the derivatives of the potentials ( microscopic real and imaginary as well as phenomenological surface imaginary and spin - orbit potentials ) .
deformation parameters ( @xmath71 ) are extracted from the analyses .
the results obtained for the deformation parameter are in good agreement with the available results .
the density dependence parameter obtained from nuclear matter calculations , which has been used in the single folding model description for the analysis of elastic and inelastic scattering of protons , also provides excellent estimate for nuclear mean free path .
we know that the ddm3y effective interaction , has profound theoretical basis .
it provides unified description of cluster radioactivity , scatterings of @xmath21 and heavy ion when used in a double folding model , and nuclear matter when used in a single folding model .
we find that it also provides reasonable description for elastic and inelastic scattering of protons .
the authors gratefully acknowledge l. a. riley for sending the experimental data in a tabular form .
* table 1 : * + parameters of the sbm and ddm3y interactions + [ cols="^,^,^,^,^,^,^ " , ] k. amos , p. j. dortmans , h. v. von geramb , s. karataglidis , adv . in nucl .
25 ( 2000 ) 275 .
d. gupta , c. samanta , jour .
g : nucl . part .
28 , 85 ( 2002 ) ; see references therein i. tanihata , h. hamagaki , o. hashimoto , y. shida , n. yoshikawa , k. sugimoto , o. yamakawa , t. kobayashi , n. takahashi , phys .
55 ( 1985 ) , 2676 .
d. gupta , c. samanta , r. kanungo , nucl .
a 674 ( 2000 ) , 77 ; see references therein l. a. riley , j. k. jewell , p. d. cottle , t. glasmacher , k. w. kemper , n. alamanos , y. blumenfeld , j. a. carr , m. j. chromik , r. w. ibbotson , f. marechal , w. e. ormand , f. petrovich , h. scheit , and t. suomijarvi , phys .
82 , 4196 ( 1999 ) .
j. l. escudie , r. lombard , m. pignanelli , f. resmini , and a. tarrats , phys .
c 10 , 1645 ( 1974 ) .
f. petrovich , s. k. yoon , m. j. threapleton , r. j. philpott , j. a. carr , nucl . phys . a 563 ( 1993 ) , 387 .
h. rebel , g. hauser , g. w. schweimer , g. nowicki , w. wiesner and d. hartmann , nucl .
phys . a 218 ( 1974 ) , 13 . c. samanta , y. sakuragi , m. ito , and m. fujiwara , j. phys .
g : nucl . part .
( 1997 ) , 1697 ; see references therein .
g. r. satchler and w. g. love , phys .
55 ( 1979 ) , 183 .
d. bandyopadhyay , c. samanta , and s. k. samaddar and j. n. de , nucl .
phys . a 511 ( 1990 ) , 1 .
c. samanta , d. bandyopadhyay and j. n. de , phys .
b 217 ( 1989 ) , 381 .
r. kanungo , c. samanta , subinit roy , s. k. samaddar , nucl .
phys.a 581 ( 1995 ) , 294 .
r. kanungo , and c. samanta , nucl .
a 617 ( 1997 ) , 265 ; see references therein .
r. kanungo , c. samanta , j. phys .
g : nucl . part .
( 1998 ) , 1611 .
r. kanungo , i. tanihata , c. samanta , prog . of theo .
( 1999 ) , 1133 .
g. bertsch , j. borysowicz , h. mcmanus and w.g .
love , nucl .
a 284 , 399 ( 1977 ) .
a.m. kobos , b.a .
brown , r. lindsay and g.r .
satchler , nucl .
phys . a 425 , 205 ( 1984 ) .
chaudhuri , nucl .
phys . a 449 , 243 ( 1986 ) .
d. n. basu , jour .
g : nucl . part .
30 , b7 ( 2004 ) j. bartel , p. quentin , m. brack , c. guet , h. b. hakansson , nucl .
phys . a 386 , 79 ( 1982 ) m. beiner , h. flocard , n. v. giai , p. quentin , nucl .
phys . a 238 , 29 ( 1975 ) n. alamanos , a. pakou , a. lagoyannis , a. musumarra , nucl .
phys . a 660 , 406 ( 1999 ) p. d. kunz , computer code dwuck4 , unpublished b. sinha , phys .
50 , 91 ( 1983 ) s. raman et.al .
, atomic data and nuclear data tables 36 , 1 ( 1987 ) d. t. khoa , e. khan , g. colo , n. van giai , nucl .
a 706 , 61 ( 2002 )
this appendix is aimed at providing a simple co - ordinate transformation formula for general purpose use in extracting data from plots using computer graphics . in many instances
the values of the experimental data are not available in literatures while the plots are shown .
particularly , for old plots , tabulated data are often very difficult to acquire . in such cases ,
data values can be read in quite accurately by using computer programs . in some cases
both linear and logarithmic scales are involved . a general formula is thus required which will convert the co - ordinates of the data points as shown by the software to their actual values .
it should also be taken into account that the co - ordinate system of the printed graph may be rotated , albeit small , with respect to the software co - ordinate system .
thus a handy formula would be extremely useful for such purposes .
we assume ( @xmath83 ) to be the co - ordinates of the printed graph while ( @xmath84 ) are those of the computer program such as gsview . in fig . 5 , ( @xmath85 ) is the origin of the printed graph while ( @xmath86 ) and ( @xmath87 ) are two points on the @xmath88 and @xmath89 axes , respectively . since the ratio of two linear lengths measured in one co
- ordinate system should be equal to that in the other system , @xmath90sin\theta}{(x_1 - x_0)sec\theta}$ ] where , @xmath91 = @xmath92 .
a little manipulation gives , @xmath93\ ] ] we consider x scale to be linear and y scale to be logarithmic .
therefore , + @xmath94cos\theta}{(y_2 - y_0)sec\theta}$ ] which gives , @xmath95 if both axes are linear , then eqn .
11 holds while @xmath96\ ] ] if both axes are logarithmic then eqn .
12 holds while , @xmath97 the expressions provided above , though simple , will be immensely useful for experimentalists as well as theoreticians for extracting data where their explicit values are not available . | the elastic and inelastic scattering of protons from mirror nuclei @xmath0ne and @xmath0o are studied in a folding model approach . for comparison ,
two different effective interactions are folded with hartree - fock densities to obtain the nuclear interaction potentials .
both of them provide equivalent descriptions to the data and the deformation parameters extracted from inelastic scattering are reasonable .
the density dependence parameters obtained from nuclear matter calculations and used for present analysis also provide a good estimate for the nuclear mean free path .
the present formalism unifies radioactivity , nuclear matter and nuclear scattering .
.2 true cm keywords : elastic and inelastic proton scattering ; effective interaction ; folding model ; sbm ; ddm3y ; mirror nuclei pacs numbers : 25.40.cm , 25.40.ep , 21.30.fe , 25.60.-t |
the submillimeter array ( sma ) is a joint venture of the smithsonian astrophysical observatory ( sao ) and the academia sinica institute of astronomy and astrophysics ( asiaa ) .
the interferometer has been under construction since the early 1990s .
first fringes at 230 ghz were initially obtained with a single baseline at haystack observatory in october 1998 .
this milestone was repeated on mauna kea in late september 1999 .
all eight 6-meter antennas will be commissioned by the fall of 2003 .
the surface of the primary reflector is composed of aluminum panels with a carbon fiber backup structure .
there are 24 antenna pads arranged into four tangential rings approximating reuleaux triangles with a maximum baseline of 508 m. the cryostat can accomodate up to eight sis receiver inserts , which provides the capability to eventually cover the entire range from 175 to 950 ghz that is accessible from the ground .
further description of the sma antennas , receivers , and if / lo systems can now be found in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
getting the array operating successfully in the 650 ghz band was a high initial priority for the sma in order to enable observations at frequencies substantially higher than any of the existing millimeter interferometers .
a theoretical plot of the opacity in this band in good conditions is shown in figure [ fig1 ] .
for some actual measurements with a fourier transform spectrometer , see @xcite . on july 20 , 2002 , we saw first fringes on venus with a single 16 m ne / sw baseline at 672/682 ghz . although the data were acquired during the late afternoon and early evening ( usually the worst part of the diurnal weather pattern on mauna kea ) the rms phase noise was only @xmath2 after removing a fit to the residual error in the baseline length ( figs .
[ fig2 ] and [ fig3 ] ) .
several other strong dust continuum sources were detected during the night of july 20th , including iras 1629a , [email protected] , and g34.26 + 0.14 ( fig [ fig4 ] ) .
by late summer , we had three antennas ( numbers 4 , 5 , and 6 ) working in the 650 ghz band , in a configuration with with baseline lengths up to 25.2 m. a 682.5 ghz beacon for holography had been installed on the exterior of the nearby subaru telescope @xcite , and the first phase closure tests were performed on september 20th using this signal ( fig [ fig5 ] ) .
the first convincing three baseline detections on a celestial source were obtained two days later , with the array re - tuned to place the @xmath0co(6 - 5 ) line in the usb .
the zenith opacity in this frequency band was approximately 1.2 which causes a large elevation dependence in the system temperature ( fig [ fig6 ] ) .
the phase closure on uranus was noisy , but the value averaged to the expected result of zero .
saturn was easily detected on all baselines , but since it was highly resolved , the closure phase was not constant .
strong @xmath0co(6 - 5 ) emission was detected strongly in the orion bn / kl hot core ( fig .
[ fig7 ] ) .
a convincing closure phase was also seen in the continuum on callisto ( fig .
[ fig8 ] ) .
one of the biggest problems for interferometric imaging in the 650 ghz band is the lack of strong point sources usable for the derivation of the complex gain as a function of time .
nearly all of the quasars used at millimeter wavelengths will be too weak for use as calibrators .
compact thermal sources such as small planets , planetary moons and asteroids will have to be used instead . as a star towards doing this , we took advantage of the close proximity of jupiter to the evolved star irc+10@xmath4216 last year . while jupiter itself was far too large ,
the galilean moons were only slightly resolved . on december 8th
, we were able to detect ( fig [ fig9 ] ) make a calibrated image of cs(14 - 13 ) at 685 ghz in irc+10@xmath4216 using mars for passband calibration and callisto as a complex gain calibrator ( fig .
[ fig10 ] ) .
cs(14 - 13 ) has a high critical density ( @xmath5 @xmath6 ) and was expected to be unresolved on our current short baselines . on december 12th
, we tuned down to 658 ghz , and made the first interferometric observations of the strong vibrationally - excited water line there @xcite .
the line was detected in vy cma , u ori and w hya ( fig . [ fig11 ] ) . | phase closure at 682 ghz and 691 ghz was first achieved using three antennas of the submillimeter array ( sma ) interferometer located on mauna kea , hawaii .
initially , phase closure was demonstrated at 682.5 ghz on sept .
19 , 2002 using an artificial ground - based `` beacon '' signal .
subsequently , astronomical detections of both saturn and uranus were made at the frequency of the @xmath0co(6 - 5 ) transition ( 691.473 ghz ) on all three baselines on sept .
22 , 2002 .
while the larger planets such as saturn are heavily resolved even on these short baselines ( 25.2 m , 25.2 m and 16.4 m ) , phase closure was achieved on uranus and callisto .
this was the first successful experiment to obtain phase closure in this frequency band .
the @xmath0co(6 - 5 ) line was also detected towards orion bn / kl and other galactic sources , as was the vibrationally - excited 658 ghz h@xmath1o maser line toward evolved stars .
we present these historic detections , as well as the first arcsecond - scale images obtained in this frequency band . |
one of the most challenging problems in mathematical fluid dynamics is to understand whether a solution of the 3d incompressible euler equations can develop a finite time singularity from smooth initial data with finite energy .
a main difficulty is due to the presence of the vortex stretching term , which has a formal quadratic nonlinearity in vorticity .
this problem has attracted a lot of attention in the mathematics community and many people have contributed to its understanding , see the recent book by majda and bertozzi @xcite for a review of this subject .
an important development in recent years is the work by constantin , fefferman , and majda who showed that the local geometric regularity of vortex lines can lead to depletion of nonlinear vortex stretching @xcite .
inspired by the work of @xcite , deng , hou , and yu @xcite obtained more localized non - blowup criteria by exploiting the geometric regularity of a vortex line segment whose arclength may shrink to zero at the potential singularity time . to obtain these results ,
deng - hou - yu @xcite used a lagrangian approach and explored the connection between the local geometric regularity of vortex lines and the growth of vorticity .
guided by this local geometric non - blowup analysis , hou and li @xcite performed large scale computations with resolution up to @xmath0 to re - examine some of the most well - known blow - up scenarios , including the two slightly perturbed anti - parallel vortex tubes that was originally investigated by kerr @xcite .
the computations of hou and li @xcite provide strong numerical evidence that the geometric regularity of vortex lines , even in an extremely localized region near the support of maximum vorticity , can lead to depletion of vortex stretching .
we refer to a recent survey paper @xcite for more discussions on this topic . in this paper
, we derive new growth rate estimates of maximum vorticity for the 3d incompressible euler equations .
we use a framework similar to that adopted by deng - hou - yu @xcite .
the main innovation of this work is to introduce a method of analysis to study the dynamic evolution of the integral of the absolute value of vorticity along a local vortex line segment . specifically , we derive a dynamic estimate for the quantity : @xmath1 where @xmath2 is a parameterization of a vortex line segment , @xmath3 , and @xmath4 is the arclength of @xmath3 .
the assumption on @xmath2 is less restrictive than that in @xcite . as in @xcite , we assume that the vorticity along @xmath3 is comparable to the maximum vorticity , i.e. @xmath5 .
let @xmath6 , and @xmath7 . here
@xmath8 be the unit vorticity vector of @xmath3 , and @xmath9 the unit normal vector . under the assumption that @xmath10 and @xmath11
, we derive a relatively sharp growth estimate for @xmath12 , which can be used to obtain an upper bound on the growth rate of the maximum vorticity : @xmath13 where @xmath14 and @xmath15 depend on @xmath16 .
if we further assume that @xmath4 has a positive lower bound , the above estimate implies no blow - up up to @xmath17 , if @xmath18 .
this generalizes the result of cordoba and fefferman @xcite .
the above estimate extends the result of deng - hou - yu in @xcite .
in fact , it is easy to check that under the assumption that @xmath19 and @xmath20 with @xmath21 , the right hand side of ( [ vorticity_growth ] ) remains bounded up to the time @xmath17 , implying no blow - up up to @xmath17 .
our result can be also applied to the critical case when @xmath22 , which was considered in @xcite . in this case
, we have @xmath23 if we further assume that there exists @xmath24 such that @xmath25 where @xmath26 depends on @xmath16 , and the scaling constants in @xmath27 , @xmath28 and @xmath4 , then our growth estimate implies that @xmath29 application of the beale - kato - majda non - blow - up criterion @xcite would exclude blow - up at @xmath17 since @xmath30 implies @xmath31 . of particular interest
is the case when the vorticity has a local clebsch representation . in this case
, the vorticity can be represented by the two clebsch variables @xmath32 and @xmath33 near the support of maximum vorticity as follows : @xmath34 where @xmath32 and @xmath33 are carried by the flow , that is @xmath35 where @xmath36 is the velocity field .
in addition to the geometric regularity assumption on @xmath3 , if we further assume that one of the clebsch variables has a bounded gradient and @xmath37 , then we prove that the maximum vorticity can not grow faster than double exponential in time , i.e. @xmath38 . as an application of this result , we re -
examine the computations of the 3d incompressible euler equations with the two slightly perturbed anti - parallel vortex tubes initial data by hou and li @xcite . by examining the vorticity field carefully near the support of maximum vorticity ( see fig .
1 ) , the vorticity field seems to have a local clebsch representation .
one of the clebsch variables may be chosen along the vortex tube direction , which appears to be regular . moreover
, the vortex lines within the support of maximum vorticity seem to be quite smooth and has length of order one , implying that @xmath4 has a positive lower bound .
thus the result that we described above may apply .
one of the important findings of the hou - li computations is that the maximum vorticity does not grow faster than double exponential in time .
our new estimate on the vorticity growth may offer a theoretical explanation to the mechanism that leads to this dynamic depletion of vortex stretching . .
computation from hou and li @xcite for the 3d incompressible euler equations with two slightly perturbed anti - parallel vortex tubes initial data.,width=302 ] we also apply our method of analysis to the surface quasi - geostrophic model ( sqg ) @xcite . as pointed out in @xcite , a formal analogy between the sqg model and the 3d euler equations can be established by considering @xmath39 as the corresponding vorticity in the 3d euler equations . here
@xmath40 is a scalar quantity that is transported by the flow : @xmath41 let @xmath3 be a level set segment of @xmath40 along which @xmath42 is comparable to @xmath43 and denote by @xmath8 the unit tangent vector of @xmath3 . under the assumption that @xmath10 and @xmath11 , we obtain a much better growth estimate for @xmath44 : @xmath45 in particular , if @xmath37 , the above estimate implies that @xmath46 .
this seems to be consistent with the numerical results obtained in @xcite , see also @xcite .
the rest of the paper is organized as follows . in section 2 ,
we derive our estimate on the integral of vorticity over a vortex line segment for the 3d euler equations , and apply this estimate to obtain an upper bound for the dynamic growth rate of maximum vorticity . in section 3 ,
we generalize our analysis to the sqg model . in the appendix , we prove a technical result for the 3d euler equations which states that the maximum velocity is bounded by @xmath47 when the vorticity field has a local clebsch representation and one of the clebsch variables has a bounded gradient .
in this section , we derive a new dynamic growth estimate of the maximum vorticity for the 3d incompressible euler equations .
we adopt a framework similar to that used in @xcite .
let @xmath48 .
we consider , at time @xmath49 , a vortex line segmant @xmath3 along which the maximum of @xmath50 ( denoted by @xmath51 in the following ) is comparable to @xmath52 .
we use @xmath53 to parameterize @xmath3 with @xmath54 being the arclength variable . in our paper
, we do not assume that @xmath3 is a subset of @xmath55 , the flow image of @xmath56 at time @xmath49 , for @xmath57 .
this assumption was required in the analysis of @xcite .
further , we denote by @xmath4 the arclength of @xmath3 . the unit tangential and normal vectors are defined as follows:@xmath58 , the unsigned curvature is defined as @xmath59 , and @xmath60 .
finally , we denote @xmath6 , and @xmath7 .
[ lemma euler ] let @xmath61 be a family of vortex line segments which come from the same vortex line .
define @xmath12 as the mean of @xmath62 over @xmath3 , @xmath63 then , we have @xmath64|\omega|({\mathbf{x}}(l , t),t)\nonumber\\ & & + \frac{1}{l}\left[|\omega({\mathbf{x}}(l , t),t)|-|\omega({\mathbf{x}}(0,t),t)|\right]\left(\frac{d { \mathbf{x}}}{dt}\cdot { \mbox{\boldmath$\xi$}}+{\mathbf{u}}\cdot { \mbox{\boldmath$\xi$}}\right)({\mathbf{x}}(0,t),t)\nonumber\\ & & + \frac{l_t}{l}\left(|\omega({\mathbf{x}}(l , t),t)|-q\right ) . \quad\quad\end{aligned}\ ] ] differentiating @xmath12 with respect to @xmath49 yields @xmath65 let @xmath66 be the arclength parameter of this vortex line at time @xmath67
. then we can write , for this specific vortex line , @xmath68 .
let @xmath69 be the corresponding coordinates of the end points of @xmath3 , i.e. @xmath70 first , we can change the integral variable from @xmath54 to @xmath66 in ( [ dqw ] ) , @xmath71 in @xcite , deng - hou - yu proved the following equality , @xmath72 substituting the above relation to ( [ dqw-1 ] ) yields @xmath73 where we have used @xmath74 with @xmath75 @xcite .
note that the arclength @xmath4 can be expressed as follows : @xmath76 differentiating the both sides with respect to @xmath49 , we get @xmath77 substituting the above relation to ( [ dqw-1 ] ) , we obtain @xmath78 observe that @xmath79 substituting the above equality to ( [ dqw-1-update ] ) , we get @xmath80 now , we have @xmath81 using @xmath82 and integrating by parts , we obtain @xmath83 ) gives @xmath64|\omega|({\mathbf{x}}(l , t),t)\nonumber\\ & & + \frac{1}{l}\left[|\omega({\mathbf{x}}(l , t),t)|-|\omega({\mathbf{x}}(0,t),t)|\right]\left(\frac{d { \mathbf{x}}}{dt}\cdot { \mbox{\boldmath$\xi$}}+{\mathbf{u}}\cdot { \mbox{\boldmath$\xi$}}\right)({\mathbf{x}}(0,t),t)\nonumber\\ & & + \frac{l_t}{l}\left(|\omega({\mathbf{x}}(l , t),t)|-q\right ) . \quad\quad\end{aligned}\ ] ] this completes the proof of lemma [ lemma euler ] .
now we are ready to state the main result of this paper .
[ theorem 3d euler ] assume there is a family of vortex line segments @xmath61 which come from the same vortex line and @xmath84 , such that @xmath85 for some @xmath86 for all @xmath87 and @xmath88 .
further we assume that there exist constants @xmath89 , such that the following condition is satisfied : @xmath90 then , the maximum vorticity @xmath52 satisfies the following growth estimate : @xmath91 where @xmath92 , @xmath93 and @xmath94 . without the loss of generality
, we may assume that @xmath4 is monotonically decreasing , i.e. @xmath95 and @xmath96 is sufficiently small .
in lemma 1 of @xcite , deng - hou - yu proved the following equality : @xmath97 it follows from the assumption @xmath98 that @xmath99 where @xmath94 . by lemma [ lemma euler ] , we have @xmath100|\omega|({\mathbf{x}}(l , t),t)\nonumber\\ & & + \frac{1}{l}\left[|\omega({\mathbf{x}}(l , t),t)|-|\omega({\mathbf{x}}(0,t),t)|\right]\left(\frac{d { \mathbf{x}}}{dt}\cdot { \mbox{\boldmath$\xi$}}+{\mathbf{u}}\cdot { \mbox{\boldmath$\xi$}}\right)({\mathbf{x}}(0,t),t)\nonumber\\ & & + \frac{l_t}{l}\left(|\omega({\mathbf{x}}(l , t),t)|-q\right)\nonumber\\ & \equiv & i_1+i_2+i_3+i_4.\end{aligned}\ ] ] recall that we choose the endpoint @xmath101 of @xmath3 such that @xmath102 which implies that @xmath103
. we also have @xmath95 by our assumption .
thus , we conclude that @xmath104 to estimate @xmath105 , we use the assumption @xmath106 , which implies that @xmath107 it remains to estimate @xmath108 and @xmath109 on the right hand side of ( [ dqw - update ] ) .
first of all , @xmath108 can be estimated as follows : @xmath110 as for @xmath109 , we proceed as follows : : @xmath111|\omega|({\mathbf{x}}(l , t),t ) \le2c_1vq / l .\end{aligned}\ ] ] now , combining ( [ dqw-2 ] ) , ( [ dqw-1 - 3 ] ) , ( [ dqw-1 - 1 ] ) and ( [ dqw-1 - 2 ] ) , we obtain the following estimate for @xmath112 , @xmath113 where @xmath92 , @xmath93 .
it follows from the above inequality that @xmath114 therefore , we have proved that @xmath115 this completes the proof of theorem [ theorem 3d euler ] .
if we further assume @xmath4 has a positive lower bound , then the above growth estimate for the maximum vorticity implies no blowup up to @xmath17 , if @xmath116 is integrable from 0 to @xmath117 .
this extends the result of cordoba and fefferman @xcite . in the critical case
when @xmath118 , if we further assume that there exists a positive constant @xmath119 such that @xmath120 then the solution remains regular up to time @xmath117 .
[ euler - corollary ] using theorem [ theorem 3d euler ] and the assumption ( [ assumption - critical ] ) , we have @xmath121 since @xmath122 .
then , the beale - kato - majda non - blowup criterion @xcite implies that there is no blowup up to time @xmath117 .
we remark that corollary [ euler - corollary ] generalizes the result of deng - hou - yu in @xcite with less restrictive requirement on the scaling constants . more specifically
, if there is @xmath123 $ ] and positive constants @xmath124 , such that @xmath125 then corollary [ euler - corollary ] implies that there is no blowup up to time @xmath117 , as long as the following condition is satisfied : @xmath126 [ theorem - euler - l0 ] suppose that all the assumptions in theorem [ theorem 3d euler ] hold .
if we further assume @xmath127 then the maximum vorticity is bounded by the following growth estimate : @xmath128 where @xmath129 .
the assumption @xmath130 implies that @xmath131 substituting the above inequality to ( [ dq - final ] ) in the proof of theorem [ theorem 3d euler ] , we obtain @xmath132 where @xmath129 .
solving the above differential inequality gives @xmath133 which immediately yields the desired growth estimate for @xmath134 : @xmath135 this completes the proof of theorem [ theorem - euler - l0 ] .
the assumption @xmath130 may appear strong .
we remark that under certain assumption on the local vorticity structure around the vortex line segments @xmath3 , this property can be justified . specifically , suppose that the vorticity field admits a clebsch representation in a region @xmath136 with diameter @xmath137 containing @xmath3 .
this implies that there exist two level set functions @xmath138 such that the vorticity can be represented as follows : @xmath139 where @xmath32 and @xmath33 are carried by the flow , that is @xmath140 with smooth initial data that decay rapidly at infinity .
if we further assume that one of the level set functions has a bounded gradient and there exists a small constant @xmath141 such that @xmath142 , where @xmath143 is a ball whose center is @xmath144 and radius is @xmath145 , then we can show that the maximum velocity over @xmath3 satisfies @xmath146 the proof of this results will be given in the appendix .
one immediate consequence of theorem [ theorem - euler - l0 ] is the following corollary . if in the statement of theorem [ theorem - euler - l0 ] we further assume that @xmath147 then @xmath52 can not grow faster than double exponential in time .
in this section , we will apply the method of analysis presented in the previous section to study the dynamic growth of @xmath44 for the sqg model .
first , we state an estimate for the maximum velocity obtained by d. cordoba in @xcite .
[ vel - qg ] for the sqg model , there exists a generic constant @xmath148 such that for @xmath149 , @xmath150 let @xmath151 .
we consider , at time @xmath49 , a level set segmant @xmath3 along which the maximum of @xmath152 ( denoted by @xmath51 in the following ) is comparable to @xmath52 .
we use the same notations as in the previous section .
first , we prove the corresponding estimate for @xmath12 for the sqg model .
[ lemma sqg ] let @xmath61 be a family of level set segments which come from the same level set , and @xmath12 be the average of @xmath42 over @xmath3 , @xmath153 then , we have @xmath154|{\nabla ^\bot \theta}|({\mathbf{x}}(l , t),t)\nonumber\\ & & + \frac{1}{l}\left[|{\nabla ^\bot \theta}({\mathbf{x}}(l , t),t)|-|{\nabla
^\bot \theta}({\mathbf{x}}(0,t),t)|\right]\left(\frac{d { \mathbf{x}}}{dt}\cdot { \mbox{\boldmath$\xi$}}+{\mathbf{u}}\cdot { \mbox{\boldmath$\xi$}}\right)({\mathbf{x}}(0,t),t)\nonumber\\ & & + \frac{l_t}{l}\left(|{\nabla ^\bot \theta}({\mathbf{x}}(l , t),t)|-q\right ) . \quad\quad\end{aligned}\ ] ] the proof follows exactly the same procedure as in the proof of lemma [ lemma euler ] in the previous section by using the equality @xmath155 which holds for the sqg model , see @xcite .
we will not reproduce the proof here . by following the same procedure as in the proof of theorem [ theorem - euler - l0 ] ,
we can obtain the following growth estimate for the sqg model : [ theorem - sqg ] assume there is a family of level set segments @xmath61 and @xmath156 , such that @xmath85 for some @xmath86 and @xmath88 for @xmath157 .
further , we assume that there exist constants @xmath158 such that @xmath159 for @xmath160 .
then , the maximum of @xmath161 is bounded by the following estimate : @xmath128 where @xmath129 , @xmath162 is the constant given in lemma [ vel - qg ] , @xmath94 , @xmath163 are same as those defined in theorem [ theorem 3d euler ] .
in addition to the assumptions stated in theorem [ theorem - sqg ] , if we further assume that @xmath4 has a positive lower bound , i.e. @xmath164 , then @xmath52 does not grow faster than double exponential in time .
more precisely , we have @xmath165
assume that @xmath166 has a local clebsch representation in a region @xmath136 containing @xmath3 , i.e. there exist two level set functions , @xmath167 such that the vorticity can be expressed as follows : @xmath168 where @xmath32 and @xmath33 are carried by the flow , that is @xmath140 with smooth initial data that decay rapidly at infinity .
if one of the level set functions has a bounded gradient , and there exists a small constant @xmath141 such that @xmath142 , where @xmath143 is a ball whose center is @xmath144 and radius is @xmath145 , then the maximum velocity over @xmath3 satisfies the following estimate : @xmath169 without the loss of generality , we may assume that @xmath170 . by the well - known biot - savart law @xcite ,
we have @xmath171 define a smooth cut - off function @xmath172 $ ] , such that @xmath173 for @xmath174 and @xmath175 for @xmath176 .
let @xmath177 be a small positive parameter to be determined later .
then we have @xmath178 by a direct calculation , we get @xmath179 to estimate @xmath109 , we split it into two terms as follows : @xmath180 we first estimate @xmath181 .
integration by parts gives @xmath182d\mathbf{y}\right|\nonumber\\ & \le & \frac{1}{4\pi}\left| \int_{\mathbb{r}^3}\left[{\mathbf{u}}({\mathbf{x}}+\mathbf{y},t)\times \nabla\left(1-\chi\left(\frac{|\mathbf{y}|}{\rho}\right)\right)\right ] \frac{\mathbf{y}}{|\mathbf{y}|^3}d\mathbf{y}\right|\nonumber\\ & & + \frac{1}{4\pi}\left| \int_{\mathbb{r}^3}\left(1-\chi\left(\frac{|\mathbf{y}|}{\rho}\right)\right ) \left[\left({\mathbf{u}}({\mathbf{x}}+\mathbf{y},t)\times\nabla\right)\times \frac{\mathbf{y}}{|\mathbf{y}|^3}\right]d\mathbf{y}\right|\nonumber\\ & \equiv & a+b.\end{aligned}\ ] ] to estimate @xmath105 , using the assumptions @xmath184 and @xmath142 , we can to split @xmath105 into three terms for any @xmath185 : @xmath186 d\mathbf{y}\right|\nonumber\\ & \le & \frac{1}{4\pi}\left| \int_{\mathbb{r}^3}\left(1-\chi\left(\frac{|\mathbf{y}|}{\delta}\right)\right ) \left[\left(\left(\phi\nabla \psi\right)({\mathbf{x}}+\mathbf{y},t)\times \nabla\chi\left(\frac{|\mathbf{y}|}{\rho}\right)\right)\times \frac{\mathbf{y}}{|\mathbf{y}|^3}\right ] d\mathbf{y}\right|\nonumber\\ & & + \frac{1}{4\pi}\left| \int_{\mathbb{r}^3}\chi\left(\frac{|\mathbf{y}|}{\rho}\right ) \left[\left(\left(\phi\nabla \psi\right)({\mathbf{x}}+\mathbf{y},t)\times \nabla\left(1-\chi\left(\frac{|\mathbf{y}|}{\delta}\right)\right)\right)\times \frac{\mathbf{y}}{|\mathbf{y}|^3}\right ] d\mathbf{y}\right|\nonumber\\ & & + \frac{1}{4\pi}\left| \int_{\mathbb{r}^3}\chi\left(\frac{|\mathbf{y}|}{\rho}\right ) \left(1-\chi\left(\frac{|\mathbf{y}|}{\delta}\right)\right ) \left[\left(\left(\phi\nabla \psi\right)({\mathbf{x}}+\mathbf{y},t)\times \nabla\right)\times \frac{\mathbf{y}}{|\mathbf{y}|^3}\right ] d\mathbf{y}\right|\nonumber\\ & \equiv & c+d+e,\hspace{1 cm } \forall { \mathbf{x}}\in l^t.\end{aligned}\ ] ] by a direct calculation , we get @xmath187 by taking @xmath188 and using the assumption @xmath189 and the fact that @xmath190 , @xmath191 are bounded , we prove that @xmath192 for some constant @xmath148 as long as @xmath193 . | by performing estimates on the integral of the absolute value of vorticity along a local vortex line segment , we establish a relatively sharp dynamic growth estimate of maximum vorticity under some assumptions on the local geometric regularity of the vorticity vector .
our analysis applies to both the 3d incompressible euler equations and the surface quasi - geostrophic model ( sqg ) . as an application of our vorticity growth estimate
, we apply our result to the 3d euler equation with the two anti - parallel vortex tubes initial data considered by hou - li @xcite . under some additional assumption on the vorticity field , which seems to be consistent with the computational results of @xcite
, we show that the maximum vorticity can not grow faster than double exponential in time .
our analysis extends the earlier results by cordoba - fefferman @xcite and deng - hou - yu @xcite . + + * keywords * 3d euler equations ; sqg equation ; finite time blow - up ; growth rate of maximum vorticity ; geometric properties .
+ + * mathematics subject classification * primary 76b03 ; secondary 35l60 , 35m10 |
the description of inhomogeneous charge and spin phases in the hubbard model it a topic of current interest , mainly because it is now generally accepted that the high - t@xmath1 compounds , at least in the underdoped regime , are intrinsic electronically inhomogeneous systems .
a powerful tool for the investigation of such inhomogeneous electronic states is the unrestricted hartree - fock ( hf ) scheme which allows for the diagonalization of reasonable cluster sizes .
recently an extension of this approach based on the gutzwiller wave function @xcite has shown to significantly improve the hf solutions which strongly underestimate the attraction between charge carriers . in this paper
we extend the approach of ref .
@xcite to include transversal spin degrees of freedoms which permits the description of coplanar and three - dimensional inhomogeneous spin textures . among coplanar
spin structures homogeneous spiral solutions have been studied by a large variety of methods ( see e.g. @xcite ) which show that a small amount of holes doped into the half - filled system leads to a @xmath2 spiral phase which changes its direction to @xmath3 above some critical concentration . relaxing the constraint of a homogeneous charge distribution the formation of coplanar
, vortexlike phases has been investigated in ref .
@xcite using an unrestricted hf approach .
these are configurations where the antiferromagnetic ( af ) spin order rotates by multiples of 2@xmath4 around the localized hole . due to this twist in the magnetization their energy increases @xmath5 which implies their instability in large clusters .
a three - dimensional spin texture which is known to be topologically stable is the skyrmion as a solution of the o(3 ) non - linear @xmath6-model @xcite .
it has been studied by several authors also for an af background @xcite mainly concentrating on small clusters .
however , it is still controversal wether skyrmion solutions exist also on discrete lattices . whereas unrestricted hf theory for the 2-d hubbard model predicts the decay of skyrmions into conventional spin - polarons @xcite , exact diagonalization studies of a small cluster within the tj - model ref . @xcite
seem to support their existence even when one takes the contribution of the skyrmion far field into acount . within our slave - boson approach
we will show below that skyrmions textures are stable solutions of the discrete two - dimensional hubbard model , when the system is doped with two holes away from half - filling .
connected with the discovery of static stripe order in la@xmath7nio@xmath8 and la@xmath9nd@xmath10sr@xmath11cuo@xmath12 @xcite a lot of work has been done in order to understand the different domain wall structures in these compounds @xcite . whereas in the ni - doped compounds one finds the stripes along the diagonal with one hole per ni - site the charge and spin order in the nd - doped cuprates is along the copper - oxygen bond direction with one hole per every second copper site only . since half - filled horizontal walls are at odds with hf calculations the inclusion of correlations turns out to be a necessary ingredient for the study of domain walls in these systems @xcite .
in addition long - range coulomb interactions can play an important role in stabilizing half - filled vertical stripes @xcite .
based on a landau free - energy analysis of coupled charge and spin - density - wave order parameters it has been shown in ref .
@xcite that within some region of parameter domain walls may have a spiral component .
whereas this type of ordering is not observed in the ni - oxides , some spiral contribution can not be rigorously excluded to be present in the nd - doped compounds @xcite . according to our analysis presented below , elliptical stripes are not stable for concentrations around @xmath13 but may be formed in the very low doping limit .
the rest of the paper is organized as follows : in sec .
ii we give a detailed description of the formalism , in sec .
iii we present the results for vortex , skyrmion and elliptical domain wall solutions respectively , and in sec .
iv we summarize our conclusions .
we consider the two - dimensional hubbard model on a square lattice , with hopping restricted to nearest neighbors ( indicated by the bracket @xmath14 ) @xmath15 where @xmath16 destroys ( creates ) an electron with spin @xmath6 at site i , and @xmath17 .
u is the on - site hubbard repulsion and t the transfer parameter . for the calculations in sec .
iii we take t=1 . in the following we use a spin - rotation - invariant form @xcite of the slave - boson representation introduced by kotliar and ruckenstein in ref .
@xcite . the subsidiary boson fields @xmath18
, @xmath19 stand for the annihilation ( creation ) of empty and doubly occupied sites , respectively , whereas the matrix @xmath20 represents the case of a singly occupied site .
since we consider the mean - field limit all boson operators will be approximated as numbers . besides the completeness condition @xmath21 the boson fields are constrained by the following relations @xmath22 where @xmath23 are the pauli spin matrices and @xmath24 . then , in the physical subspace defined by eqs .
( [ const1],[const2 ] ) the hamiltonian ( [ hm ] ) takes the form @xmath25 where @xmath26 the matrices @xmath27 and @xmath28 guarantee the correct behavior in the limit @xmath29 within the mean - field approximation and @xmath30 is the time - reversal transformed of @xmath31 .
the matrix elements of @xmath32 can be calculated by transforming to a diagonal representation for the @xmath31 ( see appendix ) .
the resulting effective one - particle hamiltonian describes the dynamics of particles which upon hopping between sites are subjected to a modulation of their spin amplitude and spin direction , respectively .
it can be diagonalized by the transformation @xmath33 where the orthogonality of the transformation requires @xmath34 given a system with @xmath35 particles we finally obtain for the total energy @xmath36 which has to be evaluated within the constraints ( [ const1],[const2 ] , [ const3 ] ) .
this is achieved by adding these constraints quadratically to eq .
( [ e1 ] ) following the procedure already applied in the gutzwiller limit @xcite .
the resulting energy functional then has to be minimized with respect to the fermionic and bosonic fields which is most conveniently done by using a standard conjugate gradient algorithm since the gradients of the energy functional can be calculated analytically . in order not to end up in pathological side minimas
we have generally started the minimization from an hf ansatz for the amplitudes @xmath37 .
since in a previous publication @xcite the unrestricted slave - boson approximation has been already applied to the description of collinear spin structures , we will restrict here to textures with two- and three dimensional spin ordering . in this section
we discuss the spin structure of vortex , skyrmion and elliptical domain wall textures which turn out to be stable energy minima within our slave - boson approach . obviously on finite lattices
one has to use open boundary conditions in order to describe higher dimensional spin structures and the cluster sizes we are considering in the following are ranging from @xmath38 up to @xmath39 .
the incorporation of transversal spin degrees of freedom allows for the definition of spin currents @xmath40 .
the flow direction of these currents is along the bonds of the lattice , however , they are additionally vectorial in spin space and within the present approach we obtain for the i - th component of the spin current flowing between sites @xmath41 : @xmath42 where @xmath43 are the pauli matrices and the hopping factors @xmath44 are defined in eq .
( [ zdef ] ) . the i - th component of @xmath45 can be thought of as measuring the spin - twist in the orthogonal directions @xmath46 so that the total current components @xmath47 , which are plotted in the results , visualize the direction of maximal twist in the spin components @xmath48 at lattice site n. the structure of vortex solutions , where the magnetization rotates in a plane by some multiples of @xmath49 around the localized holes , has already been studied in ref .
@xcite within unrestricted hf theory
. we also obtain vortex states as local minima of the energy functional ( [ e1 ] ) , where the total energy is about @xmath50 lower than in the hf approach , depending on lattice size and on - site repulsion u. however , for one hole away from half - filling vortex solutions are always higher in energy than the conventional nel ordered spin polaron . moreover , their total energy increases logarithmically with the cluster size as a consequence of the twist between neighboring magnetization vectors in agreement with ref . @xcite .
spin structure and spin currents for a vortex - antivortex pair on a @xmath0 lattice .
the hubbard on - site repulsion is @xmath51 .
this logarithmic divergency can be compensated when two holes form a vortex - antivortex pair . according to our calculations
the vortex cores are located at the center of diagonally next nearest neighbor plaquettes , thus separated by the plaquette where the two holes are localized .
1 shows the spin structure and spin current of such a vortex - antivortex pair on a @xmath0 cluster and @xmath51 .
all the spins lie in the xy - plane which means that only the z - component of the spin current has a non zero contribution .
the spin field of this texture can be described by @xmath52 where @xmath53 , ( @xmath54 ) refer to the angles between x - axis and the vectors connecting vortex ( antivortex ) core and site @xmath55 .
the af wave vector is denoted by @xmath56 . concerning the stability of the vortex - antivortex pair we obtain for the @xmath0 cluster and @xmath51 a binding energy of @xmath57 with respect to the nel type bipolaron .
taking into account the far field energy , calculated within the xy - model for the spin structure eq .
( [ vaeq ] ) and an exchange constant of @xmath58 still results in a negative binding energy of @xmath59 .
we note that the absolute value of @xmath60 slightly decreases for smaller cluster sizes since the boundary spins do not properly adjust to the solution eq .
( [ vaeq ] ) .
one can therefore safely conclude that vortex - antivortex pairs are also stable in the thermodynamic cluster limit . on a discrete 2-dimensional af lattice the spin structure of skyrmions , originally obtained as solutions of the o(3 ) non - linear @xmath6-model @xcite ,
has the form @xcite @xmath61 where @xmath62 denotes the core size of the skyrmion and its center is located at @xmath63 . in order to enhance convergence we initialized our minimization procedure with ( non self - consistent ) hf
wave functions corresponding to the spin fields eqs .
( [ skf ] ) . despite intensive search
we could not obtain skyrmion states for one hole doped in the half - filled system .
these solutions always converged towards a spin - polaron embedded into a collinear af nel state as already observed in ref .
@xcite .
charge- ( @xmath64 ) and spin - distribution for a skyrmion texture on a @xmath65 lattice for @xmath51 .
xy- , yz- and xz - projections of the spins together with the respective spin currents for the skyrmion shown in fig .
2 2 the situation changes when removing two particles from the half - filled system .
2 displays the charge- and spin structure in case of 62 particles on a @xmath65 lattice for @xmath51 .
the two holes then are localized on a @xmath66 plaquette at the skyrmion center and in their vicinity spins show a remarkable deviation from the z - direction .
this is more easily seen in fig . 3 where we have plotted the xy- , xz- and yz- spin projections , respectively , together with the corresponding spin currents .
the xy - spin components rotate by @xmath67 around the skyrmion center resulting in a circular spin current for @xmath68 .
however , all current components strongly decay for sites far away from the skyrmion center indicating that the core size parameter @xmath62 is small . upon fitting our solutions to the skyrmion field eq .
( [ skf ] ) we obtain @xmath69 for @xmath51 and cluster sizes @xmath70 , @xmath71 and @xmath0 , respectively .
this already indicates that the skyrmion state should survive the limit of large clusters . to assess the question of stability in more detail
we have also calculated the total energy using skyrmionic boundary conditions @xcite .
these are defined through an exchange field @xmath72 to which the spins at the boundary are coupled and @xmath73 has the form of eq .
( [ skf ] ) . for the exchange
constant we take the strong - coupling value of the hubbard model @xmath74 .
the total energy then is evaluated as a function of @xmath62 and in the result we substract the energy contribution of the exchange field .
this energy has to be compared with the corresponding value of a collinear bipolaron where two holes are localized on neighbored sites within the nel ordered system .
the results are plotted in fig . 4 again for @xmath51 and three different cluster sizes .
as can be seen all curves display a clear minimum at some value of @xmath62 indicating the presence of a stable skyrmion solution with a significant lower energy than the collinear bipolaron .
it should be mentioned that these minima for the corresponding one - hole doped systems are always at @xmath75 , i.e. the configuration of a conventional spin - polaron . from the results shown in fig . 4
one further sees that a lowering of energy with respect to the af bipolaron is already obtained for @xmath75 indicating that despite the imposed nel boundaries the system has a skyrmion like core .
this energy shift increases with the system size since the central spins of large lattices can more properly adjust to the skyrmion state .
from the fact that the @xmath75 results for the @xmath65 and @xmath0 coincide within the numerical error we conclude that also in the thermodynamic limit the skyrmion solution should survive .
also plotted in fig .
4 with filled symbols are the energy differences between bipolarons and skyrmions calculated with open boundary conditions .
as already mentioned one obtains the same core parameter @xmath62 for all three cluster sizes which agrees with the position of the minimum of the solid line ( @xmath0 system ) .
also this feature demonstrates that our largest cluster should already correctly describe the skyrmion structure of infinite clusters .
since the skyrmion on small lattices is very much influenced by the boundary conditions the minimum of the dotted curve ( @xmath70 cluster ) is shifted to a higher value of @xmath76 .
total energy of a two - hole - skyrmion with respect to the energy of a nel - type bipolaron as a function of the skyrmion core size parameter @xmath62 .
the boundary spins have been coupled to the skyrmion solution eq .
( [ skf ] ) via a mean - field exchange field .
solid line and circles : @xmath0 lattice ; dashed line and squares : @xmath71 lattice ; dotted line and diamonds : @xmath70 lattice .
the full symbols mark the energies for open boundary conditions .
@xmath51 . in order to compare the stability of skyrmion states with the vortex - antivortex solutions
, one has to take into account the far field contribution to the total energy .
considering again a @xmath0 system and @xmath51 the skyrmion binding energy with respect to the nel - type bipolaron ( cf .
4 ) is @xmath77 . including the far field energy ( taking again @xmath78 ) one obtains @xmath79 which is approximately half the value of the vortex - antivortex binding energy
the possibility that charge - spin coupling in correlated systems may induce the formation of elliptical domain walls was proposed by an analysis of the landau free - energy functional for coupled charge- and spin density waves @xcite .
these are coplanar spin structures where the spin components in the first harmonic are modulated as @xmath80 the eccentricity of the elliptical domain wall is determined by @xmath81 and @xmath82 correspond to the wave vectors of commensurate af and the domain wall periodicity , respectively .
the case @xmath83 describes an ideal spiral solution whereas @xmath84 reduces the spin structure to a collinear classical domain wall .
we have used ( non - self consistent ) hf states corresponding to eq .
( [ edw ] ) as starting fields for our minimization for different values of @xmath81 .
only vertical domain walls have been considered and periodic boundary conditions where applied in the x- and y - direction , respectively . in case of a completely filled domain wall ( i.e. one hole per site along the wall ) we only found the collinear solutions whereas coplanar structures become stable for half - filled walls .
5 . two possible spin structures for elliptical domain walls together with the corresponding spin currents .
the on - site repulsion is @xmath85 and periodic boundary conditions in x- and y - direction have been used .
shown are the results for a @xmath86 lattice doped with 4 holes .
this is shown in fig . 5 for a @xmath87 lattice with 68 particles where we plot two kinds of possible spin structures which are local minima of the energy functional eq . ( [ e1 ] ) .
also shown are the respective spin currents ( only z - components since the magnetization vector is completely in the xy - plane ) .
spin fields and currents display a quadrupled structure along the wall which is a necessary condition for stability within any mean - field approach @xcite . due to the stripe charge structure
the current flows are more complex than for the elliptical solution eq .
( [ edw ] ) which predicts currents flowing in a single direction orthogonal to the wall .
instead we observe also currents along the wall ( fig .
5a ) and a vortex - antivortex structure in fig . 5b .
although the elliptical stripes are local minima of the energy functional eq .
( [ e1 ] ) they are slightly higher in energy ( @xmath88 ) than collinear domain walls . these have been shown to correspond to the ground state when one takes long - range coulomb interactions into account @xcite .
however , we do not expect significant differences of a long - range contribution to collinear and elliptical stripe solutions .
the structures shown in fig . 5 correspond to systems with hole doping @xmath89 and we could not obtain elliptical solutions for higher doping
. however , one can also study the very low doping limit upon using open boundaries in the x - direction . indeed in this case
elliptical half - filled stripes become favored with repect to collinear domain walls for not too large values of the on - site repulsion u ( @xmath90 ) . fig .
the same as in fig . 5 but now for open boundaries in x - direction . in fig
6 we show the same spin structures as in fig . 5 but for open boundaries in x - direction .
the main difference between these two figures concerns the angle of spin rotation @xmath91 across the domain wall which is only @xmath92 of the expected value of @xmath4 according to eq .
( [ edw ] ) .
this is due to the fact that the system now can acquire a state which has zero total spin current , whereas for periodic boundaries the spiral component always requires a net flow in x - direction .
however , since for a regular array of stripes the charge periodicity and the spin modulation have to be related by @xmath93 @xcite the domain wall induced kink type spin rotation has to be supplemented by an additional spiral field .
assuming an exponential relaxation of the spin rotation from @xmath91 to @xmath4 by this spiral field gives an additional energy per site of @xmath94 where @xmath95 is the spin stiffness , @xmath62 defines the length scale of relaxation and l is the stripe separation .
consequently this additional energy can become small enough in the low doping regime so that elliptical stripe configurations are more stable than collinear solutions for small hole concentrations .
summarizing , we have presented the structure of spin textures which we obtained by applying an unrestricted slave - boson mean - field approximation in its spin - rotation invariant form to the 2d hubbard model .
this approach is suited for calculating the charge and spin distribution of electrons ( holes ) in inhomogeneous and strongly correlated systems since it incorporates correlation effects beyond the standard unrestricted hf theory .
these correlation effects turn out to be especially important for the interaction among holes in the 2-d hubbard model where the hf approximation strongly underestimates their attractive potential @xcite . including the effect of transversal degrees of freedom we have shown that within our slave - boson mean - field approach two holes in the half - filled two - dimensional hubbard model are bound by forming a vortex - antivortex pair , oriented along the diagonal direction .
this texture has significantly lower energy than a conventional collinear bipolaron , also when the far field contribution is taken into account .
additionally we have found that skyrmion states can be stable even on discrete lattices .
we attribute this to the inclusion of correlation effects within our approach , since unrestricted hartree - fock theory can not account for skyrmions as self - consistent solutions ( or local minima of energy ) .
indeed it has been also observed by the authors of ref .
@xcite that within the tj - model a semiclassical description can not account for the occurence of skyrmions but that it is the quantum fluctuations which stabilize this texture .
considering the formation of elliptical domain walls it turned out that these structures only appear for half - filled walls in the low doping regime .
similar to the collinear stripes they are stabilized by a quadrupling of the period along the wall .
this can be realized either by alternating on - wall spin currents ( fig .
5a ) or by forming a vortex - antivortex structure ( fig .
5b ) .
our purpose is to calculate the matrix elements of the hopping matrices @xmath96 by transforming to a diagonal representation of the matrix @xmath31 ( eq . ( [ pi ] ) ) according to @xmath97 where the eigenvalues are given by @xmath98 .
the transformation matrix @xmath99 reads as @xmath100 applied to the hopping matrix this transformation yields @xmath101 and since @xmath102 is diagonal it is straightforward to evaluate @xmath103 and @xmath104 as @xmath105 g. seibold , e. sigmund , and v. hizhnyakov , phys .
b*57 * , 6937 ( 1998 ) .
m. fleck , a. i. liechtenstein , a. m. ole , l. hedin , and v. i. anisimov , phys .
lett . * 80 * , 2393 ( 1998 ) .
r. frsard , and p. wlfle , j. phys .
c*4 * , 3625 ( 1992 ) .
j. a. vergs , e. louis , p. s. lomdahl , f. guinea , and a. r. bishop , phys .
rev . b*43 * , 6099 ( 1991 ) .
j. zaanen and o. gunnarsson , phys .
rev . b*40 * , 7391 ( 1989 ) .
a. a. belavin , and a. m. polyakov , jetp lett .
* 22 * ( 10 ) 245 ( 1975 ) .
b. i. shraiman , e. d. siggia , phys .
lett . * 61 * , 467 ( 1988 ) .
r. j. gooding , phys .
lett . * 66 * , 2266 ( 1991 ) .
s. haas , f - c .
zhang , f. mila , t. m. rice , phys .
lett . * 77 * , 3021 ( 1996 ) .
j. m. tranquada , d. j. buttrey , v. sachan , and j. e. lorenzo , phys .
lett . * 73 * , 1003 ( 1994 ) ; j. m. tranquada , b. j. sternlieb , j. d. axe , y. nakamura , and s. uchida , nature * 375 * , 561 ( 1995 ) ; j. m. tranquada , j. d. axe , n. ichikawa , a. r. moodenbaugh , y. nakamura , and s. uchida , phys . rev
. lett . * 78 * , 338 ( 1997 )
. j. m. tranquada , j. d. axe , n. ichikawa , y. nakamura , s. uchida , and b. nachumi , phys .
rev . b*54 * , 7489 ( 1996 ) .
j. zaanen and p. b. littlewood , phys .
b*50 * , 7222 ( 1994 ) ; z. g. yu , j. zang , j. t. gammel , and a. r. bishop , phys .
b*57 * , 3241 ( 1998 ) .
j. zaanen and m. ole , ann .
physik * 5 * , 224 ( 1996 ) .
g. seibold , c. castellani , c. di castro , and m. grilli , cond - mat/9803184 .
s. r. white and d. j. scalapino , cond - mat/9705128 ; s. r. white and d. j. scalapino , phys .
80 * , 1272 ( 1998 ) .
o. zachar , s. a. kivelson , and v. j. emery , phys .
rev . b*57 * , 1422 ( 1998 ) . t. li , p. wlfle , and p. j. hirschfeld , phys .
rev . b*40 * , 6817 ( 1989 ) ; r. frsard , and p. wlfle , int .
b*6 * , 237 ( 1992 ) .
g. kotliar and a. e. ruckenstein , phys .
lett . * 57 * , 1362 ( 1986 ) . | the spin and charge texture around doped holes in the two - dimensional hubbard model is calculated within an unrestricted spin rotational invariant slave - boson approach . in the first part
we examine in detail the spin structure around two holes doped in the half - filled system where we have studied cluster sizes up to @xmath0 .
it turns out that the most stable configuration corresponds to a vortex - antivortex pair which has lower energy than the nel - type bipolaron even when one takes the far field contribution into account .
we also obtain skyrmions as local minima of the energy functional but with higher total energy than the vortex solutions .
additionally we have investigated the stability of elliptical domain walls for commensurate hole concentrations .
we find that ( i ) these phases correspond to local minima of the energy functional only in case of partially filled walls , ( ii ) elliptical domain walls are only stable in the low doping regime .
pacs numbers : 71.27.+a,71.10.fd,75.10.-b,75.60.ch 2 |
the extraction of the ground - state parameters from the operator product expansion ( ope ) series for a relevant correlator is a cumbersome procedure : even if several terms of the ope for the correlator are known precisely , the numerical procedures of the method of dispersive sum rules @xcite can not determine the true exact value of the bound - state parameter . instead , the method should provide the band of values such the true hadron parameter has a flat probability distribution within this band @xcite .
this band is a systematic , or intrinsic , sum - rule uncertainty .
the method of sum rules in qcd contains a set of prescriptions ( see e.g. @xcite ) which are believed to provide such a systematic error . in qcd this , however , always remains a conjecture it is impossible to prove that the range provided by the standard sum - rule procedures indeed contains the actual value of the bound - state parameter .
the only possibility to acquire an unbiased judgement of the reliability of the error estimates in sum rules is to apply the method to a problem where the parameters of the theory may be fixed and the corresponding parameters of the ground state may be calculated independently and exactly .
presently , only quantum - mechanical potential models provide such a possibility . a simple harmonic - oscillator ( ho ) potential model @xcite possesses the essential features of qcd confinement and asymptotic freedom @xcite and has the following advantages : ( i ) the bound - state parameters ( masses , wave functions , form factors ) are known precisely ; ( ii ) direct analogues of the qcd correlators may be calculated exactly . ( for a discussion of many aspects of sum rules in quantum mechanics we refer to @xcite ) .
making use of this model , we have already studied the extraction of ground - state parameters from different types of correlators : namely , the ground - state decay constant from two - point vacuum - to - vacuum correlator @xcite , the form factor from three - point vacuum - to - vacuum correlator @xcite , and form factor from vacuum - to - hadron correlator @xcite .
we have demonstrated that the standard adopted procedures for obtaining the systematic errors do not work properly : for all types of correlators the true known value of the bound - state parameter was shown to lie outside the band obtained according to the standard criteria .
these results give a solid ground to claim that also in qcd the actual accuracy of the method turns out to be much worse than the accuracy expected on the basis of the standard criteria applied .
thus , our results reported in @xcite mainly contained cautious messages concerning the application of sum rules to hadron properties .
however , understanding the problem is the necessary first step in solving the problem .
we have realized that the main origin of this failure of the method lies in an over - simplified model for hadron contnuum which is described as a perturbative contribution above a constant borel - parameter independent effective continuum threshold .
we have introduced the notion of the _ exact _ effective continuum threshold , which corresponds to the true bound - state parameters : in a ho model the true hadron parameters decay constant and form factor are known and the exact effective continuum thresholds for different correlators may be calculated .
we have demonstrated that the exact effective continuum threhsold ( i ) is not a universal quantity and depends on the correlator considered ( i.e. it is in general different for two - point and three - point vacuum - to - vacuum correlators ) , and ( ii ) depends on the borel parameter and , for the form - factor case , also on the momentum - transfer .
moreover , we have shown that the `` borel stability criterium '' combined with the assumption of a borel - parameter independent effective continuum threshold leads to the extraction of the wrong bound - state parameters . in view of these results
a natural question arises : it is possible to formulate modified procedures which would improve the actual accuracy of hadron parameters extracted from sum rules , or the method is only suitable for obtaining numerical estimates with uncontrollable errors which may amount to 20 - 30% ?
we believe the answer to the first part of this question is positive . in a recent paper @xcite we proposed a new algorithm for extracting the parameters of the ground state .
the simple idea of @xcite is to relax the standard assumption of a borel - parameter independent effective continuum threshold , and allow for a borel - parameter dependent quantity . in @xcite we have illustrated the way the procedure works for the extraction of the form factor from three - point vacuum - to - vacuum correlator . in the present paper
we develop this idea and provide details of its application to the extraction of hadron parameters from dispersive sum rules .
we present for the first time the application of the new algorithm to the two - point correlator and present a detailed analysis of the extraction of the form factor from three - point correlator .
the new algorithm is shown to lead to a significant improvement of the actual accuracy of the extracted bound - state parameters . even though the full control over the systematic uncertainties , in the strict mathematical sense , is still not feasible ( and presumably can not be achieved in principle ) , our findings suggest that it may be possible to shed light on the actual accuracy of the method of sum - rules in qcd .
the paper is organized as follows : section [ sect : ho ] recalls the results for bound states in the ho model ; section [ sect : pi ] gives the ope for the polarization operator .
section [ sect : gamma ] presents detailed results for the vertex function and the corresponding ope .
sum rules for the decay constant and the form factor are given in section [ sect : sr ] .
section [ sect : numerics ] provides our numerical results .
finally , section [ sect : conclusions ] presents our concluding remarks .
we consider a nonrelativistic model hamiltonian @xmath0 with a ho interaction potential @xmath1 , @xmath2 : @xmath3 the full green function @xmath4 and the free green function @xmath5 are related by @xmath6 the solution @xmath7 of this relation may be easily obtained as an expansion in powers of the interaction @xmath8 : @xmath9 in the ho model , all characteristics of the bound states are easily calculable . for instance , for the ground state ( g ) , @xmath10 , one finds @xmath11 where the elastic form factor of the ground state is defined in terms of the ground - state wave function @xmath12 according to @xmath13 and the current operator @xmath14 is given by the kernel @xmath15
the quantum - mechanical analogue of the polarization operator has the form @xmath16 this quantity is used in the sum - rule approach for the extraction of the wave function at the origin ( i.e. , of the decay constant ) of the ground state @xcite .
a detailed analysis of the corresponding procedure for the ho model can be found in @xcite .
for the ho potential , the analytic expression for @xmath17 is well - known @xcite : @xmath18^{3/2}}.\end{aligned}\ ] ] apart from the overall factor , @xmath17 is a function of one parameter @xmath19 .
the average energy of the polarization function is defined as @xmath20 at @xmath21 both @xmath17 and @xmath22 diverge . for large values of @xmath23
the contributions of the excited states to the correlator vanish and therefore @xmath24 for @xmath25 .
the deviation of the energy from @xmath26 at finite values of @xmath23 measures the `` contamination '' of the correlator by the excited states .
the ground - state contribution to the correlator has a simple form : @xmath27 the ope series is the expansion of @xmath17 at small euclidean time @xmath23 : @xmath28 for @xmath17 this expansion is equivalent to the expansion in powers of the interaction @xmath29 .
the first term in ( [ piope ] ) does not depend on the interaction and describes the free propagation of the constituents .
@xmath30 may be written as the spectral integral @xcite @xmath31 with @xmath32 the known spectral density of the one - loop two - point feynman diagram of the nonrelativistic field theory @xcite .
the rest of the series represents power corrections , which may be obtained just as the difference between the exact correlator and the free - propagation term : @xmath33
the basic quantity for the extraction of the form factor in the method of dispersive sum rules is the correlator of three currents @xcite .
the analogue of this quantity in quantum mechanics reads @xcite @xmath34 [ with the operator @xmath14 defined in ( [ j ] ) ] and its double borel ( laplace ) transform under @xmath35 and @xmath36 @xmath37 for large @xmath38 and @xmath39 the correlator is dominated by the ground state : @xmath40 let us notice the ward identity which relates the vertex function at zero momentum to the polarization operator : @xmath41 this expression follows directly from the current conservation relation @xmath42 we obtain the exact @xmath44 by using the known expression for the green function in configuration space @xmath45 all necessary integrals are gaussian and we easily derive an explicit expression for @xmath44 . for our further investigation
, we consider the vertex function for equal times @xmath46 , which reads @xmath47 the correlator @xmath48 is a function of two dimensionless variables @xmath49 and @xmath50 .
setting , without loss of generality , @xmath51 we still have 2 dimensionless variables @xmath49 and @xmath52 .
this will be done later for the numerical analysis .
the corresponding average energy is defined as follows @xmath53 [ cols="^,^ " , ] the ratio @xmath54 of the form factor @xmath55 extracted from the sum rule ( [ srgamma ] ) , using different approximations for @xmath56 , and the exact ground - state form factor @xmath57 , given by eq .
( [ fg ] ) , versus the momentum transfer @xmath58 .
the dots , squares and diamonds correspond , respectively , to the results obtained using a constant ( [ constant ] ) , linear ( [ linear ] ) and quadratic ( [ quadratic ] ) approximation for the @xmath23-dependence of @xmath56.,width=396 ] for a @xmath23-independent approximation ( [ constant ] ) , the minimization of @xmath59 ( [ chisq ] ) leads to @xmath60 for which the dual energy @xmath61 differs from @xmath26 by less than @xmath62 and the dual form factor @xmath63 turns out to be practically @xmath23-independent in the whole fiducial range .
such a stability , usually referred to as the borel stability , is often ( erroneously ) claimed to be the way to control the accuracy of the extracted form factor . from fig .
[ plot:5 ] it is clear , however , that the actual error of the extracted form factor turns out to be much larger than the variation of the form factor in the fiducial range of @xmath23 .
moreover , the t - independent ansatz is applicable only for not very large @xmath58 : for @xmath64 it fails to reproduce the ground - state energy in the fiducial range .
allowing for a linear ansatz ( [ linear ] ) for @xmath56 extends the range of the momentum transfers where the form factor can be extracted from the sum rule with a reasonable accuracy .
note that the _ exact _ effective continuum threshold of the ho model can be well approximated by a linear function of @xmath23 in the whole fiducial range , see fig .
[ plot:3](a ) .
nevertheless , this does not mean that the minimization of ( [ chisq ] ) finds the _ exact _ effective continuum threshold and leads to the _ exact _ form factor : the deviation of the extracted form factor from the exact one amounts to a few percent for @xmath65 and increases dramatically for @xmath66 , see fig .
[ plot:6 ] . the dramatic increase of the error at @xmath67
is directly related to the fact that the contribution of the ground state to the correlator decreases rapidly with @xmath58 in the given @xmath23-window @xmath68 ( see fig . [ plot:1 ] ) .
let us emphasize that the actual error of the extracted form factor turns out to be much larger than the variation of the form factor in the window , see fig .
[ plot:5 ] .
one may try to go further and to consider the quadratic ansatz ( [ quadratic ] ) . from fig .
[ plot:6 ] it is clear that the quadratic ansatz leads to certain instabilities in the extracted value of the form factor and does not , in general , improve the actual accuracy of the form factor extraction .
this is not strange : such a behaviour just reflects the fact that the sum - rule extraction of the ground - state parameters can be performed only with a limited accuracy , which has been known since the initial svz paper @xcite .
it should be therefore clear that there is no way to get further improvement in the accuracy of the extracted form factor by increasing the degree of the polynomial approximation of @xmath56 .
nevertheless , we have seen a clear improvement in the outcome of the extraction procedure as one goes beyond the assumption of a @xmath23-independent effective continuum threshold : first , the actual accuracy turns out to be ( much ) better ; second , the @xmath23-dependent ansatz allows one to extract the form factor in a broader range of the momentum transfer .
the crucial question for the application of sum rules to the bound - state form factors is how to understand the actual accuracy of the ground - state parameter extracted from the sum rule .
\(i ) obviously , the borel stability does not guarantee the good extraction of the ground - state parameter and the variation of the extracted ground - state parameter in the window does not provide a realistic error estimate .
\(ii ) it should be understood that even a very accurate reproduction of the ground - state energy in the window does not lead to equivalently accurate extraction of the ground - state parameters .
this is rather unpleasant since the deviation of the dual energy from the known ground - state energy is the only visible characteristic which can be controlled in the realistic cases .
what really matters for the extraction of the bound - state parameters is the deviation of the fitted effective threshold from the exact effective threshold which , however , in the realistic cases remains unknown . in the ho model ,
a realistic error estimate of the accuracy in a wide range of values of @xmath58 may be obtained by comparing to each other the form factor values extracted by assuming the linear and quadratic anstze for the effective continuum threshold ( [ linear ] ) and ( [ quadratic ] ) . whether this feature persists in qcd
is an interesting and important issue to be addressed in the future .
we studied the extraction of the ground - state form factor from the vacuum - to - vacuum correlator in the exactly solvable harmonic - oscillator model applying the standard procedures of the method of qcd sum rules .
let us summarize the main messages of our analysis : * the knowledge of the correlator in a limited range of relatively small euclidean times @xmath23 ( that is , large borel masses ) is not sufficient for the determination of the ground - state parameters .
in addition to the ope for the relevant correlator , one needs an independent criterion for fixing the effective continuum threshold . * assuming a @xmath23-independent ( i.e. , a borel - parameter independent ) effective continuum threshold , the error of the extracted form factor @xmath69 turns out to be typically much larger than ( i ) the error of the description of the exact correlator @xmath48 by the truncated ope @xmath70 and ( ii ) the variation of @xmath69 in the fiducial range of @xmath23 ( i.e. , the borel window ) . as the result
, the actual value of the hadron form factor may lie far outside the range covered by the sum - rule estimate @xmath69 within the borel window of @xmath23 .
therefore , the variation of the hadron form factor ( as well as other hadron parameters ) within the borel window can not be used as an estimate of its systematic error .
the latter point is of particular relevance for the practical applications of sum rules in qcd , since the borel stability is usually ( erroneously ) believed to control the accuracy and the reliability of the extracted ground - state parameter .
* in the cases where the ground - state mass is known ( e.g. , experimentally measured ) , the actual accuracy of the sum - rule analysis may be considerably improved by allowing for a @xmath23-dependent effective continuum threshold and finding its parameters by minimizing the deviation of the energy of the dual correlator from the known value of the ground - state energy , eq .
( [ chisq ] ) .
+ in the ho model , this procedure was shown to yield clear improvements in the extracted values of the decay constant and the form factor .
+ moreover , in the ho model the deviation between the hadron parameter values obtained by using different anstze for @xmath71 gives _ de facto _ a realistic error estimate .
this was observed for both the decay constant and the form factor .
unfortunately , it remains impossible to construct the band of values which may be _ proven _ to contain the actual form factor . in this sense , the method of sum rules is not able to provide _ rigorous _ error estimates .
nevertheless , the application of the new procedures formulated in this paper to hadron form factors in qcd seems very promising with respect to improving the _ actual _ accuracy of the method .
_ acknowledgments : _
d. m. gratefully acknowledges financial support from the austrian science fund ( fwf ) under project p20573 and from the president of russian federation under grant for leading scientific schools 1456.2008.2 .
30 m. shifman , a. vainshtein , and v. zakharov , nucl .
b * 147 * , 385 ( 1979 ) .
b. l. ioffe and a. v. smilga , phys .
b * 114 * , 353 ( 1982 ) ; v. a. nesterenko and a. v. radyushkin , phys .
b * 115 * , 410 ( 1982 ) .
w. lucha , d. melikhov , and s. simula , phys .
d * 75 * , 096002 ( 2007 ) ; phys . atom . nucl .
* 71 * , 545 ( 2008 ) .
w. lucha , d. melikhov , and s. simula , phys .
d * 76 * , 036002 ( 2007 ) ; phys .
b * 657 * , 148 ( 2007 ) ; + phys . atom .
71 * , 1461 ( 2008 ) .
w. lucha , d. melikhov , and s. simula , phys .
b * 671 * 445 , ( 2009 ) .
d. melikhov , phys .
b * 671 * 450 , ( 2009 ) .
w. lucha , d. melikhov , and s. simula , arxiv:0902.4202 [ hep - ph ] .
v. novikov , m. shifman , a. vainshtein , and v. zakharov , nucl .
b * 237 * , 525 ( 1984 ) .
a. i. vainshtein , v. i. zakharov , v. a. novikov , and m. a. shifman , sov .
j. nucl .
phys . * 32 * , 840 ( 1980 ) . v. a. novikov _ et al .
_ , phys . rep .
* 41 * , 1 ( 1978 ) ; m. b. voloshin , nucl .
b * 154 * , 365 ( 1979 ) ; j. s. bell and r. bertlmann , nucl .
b * 177 * , 218 ( 1981 ) ; nucl .
b * 187 * , 285 ( 1981 ) ; v. a. novikov , m. a. shifman , a. i. vainshtein , and v. i. zakharov , nucl .
b * 191 * , 301 ( 1981 ) .
a. le yaouanc _ et al .
_ , phys .
d * 62 * , 074007 ( 2000 ) ; phys .
b * 488 * , 153 ( 2000 ) ; phys .
b * 517 * , 135 ( 2001 ) .
d.melikhov , s.simula , phys .
d*62 * , 074012 , 2000 .
a. v. radyushkin , in _ strong interactions at low and intermediate energies _ , edited by j. l. goity ,
singapore , world scientific , pp .
91150 ( 2000 ) [ hep - ph/0101227 ] .
a. p. bakulev , acta phys .
b * 37 * , 3603 ( 2006 ) [ hep - ph/0610266 ] .
d. melikhov , eur .
j. direct c * 4 * , 2 ( 2002 ) [ hep - ph/0110087 ] ; d. melikhov and s. simula , eur .
j. c * 37 * , 437 ( 2004 ) ; w. lucha , d. melikhov , and s. simula , phys .
d * 75 * , 016001 ( 2007 ) . | we study the extraction of the ground - state parameters from vacuum - to - vacuum correlators .
we work in quantum - mechanical potential model which provides the only possibility to probe the reliability and the actual accuracy of this method : one obtains the bound - state parameters from the correlators by the standard procedures adopted in the method of sum rules and compares these results with the exact values calculated from the schrdinger equation .
we focus on the crucial ingredient of the method of sum rules the effective continuum threshold and propose a new algorithm to fix this quantity . in a quantum - mechanical model , our procedure leads to a remarkable improvement of the accuracy of the extracted ground - state parameters compared to the standard procedures adopted in the method and used in all previous applications of dispersive sum rules in qcd .
the application of the proposed procedure in qcd promises a considerable increase of the accuracy of the extracted hadron parameters . |
in the local universe , powerful double - lobed radio sources are almost always associated with elliptical galaxies ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this association is consistent with leading galaxy formation models ( e.g. * ? ? ?
* ; * ? ? ?
it has long been postulated that elliptical galaxies form through the mergers of smaller galaxies @xcite .
the merger of two gas - rich galaxies generates both star - formation and agn ( active galactic nucleus ) activity .
the merger triggers a powerful nuclear starburst ( e.g. * ? ? ?
* ; * ? ? ?
* ) , which generates strong starburst - driven winds , @xcite .
these winds will eventually blow away the enshrouding dust and lay bare the quasar core , depleting the dust and gas to form an elliptical galaxy @xcite . although there are both starburst and quasar driven winds , quasar driven winds tend to occur at a later merger stage ( e.g. * ? ? ?
* ; * ? ? ?
at the same time , the merger fuels a pre - existing quiescent black hole ( i.e. radiatively efficient quasar - mode accretion , * ? ? ? * ) .
a good example of this process in action is the ulirg f00183 - 7111 @xcite where a radio - loud agn has turned on , but the star - formation is still ongoing .
this activity ceases when the fuel supply to the central regions is exhausted , and most of the remaining gas is expelled , starving both the active galactic nucleus ( agn ) and the star - forming activity .
thereafter , the black hole is being fed more slowly ( i.e. radiatively inefficient `` radio - mode '' accretion ) .
that is to say , the merger of two gas - rich spiral forms an elliptical galaxy whilst launching radio - loud jets from the central agn ( e.g * ? ? ?
the kpc - scale double - lobed sources associated with agn are typically l@xmath9@xmath1010@xmath1whz@xmath2 .
it is not uncommon for spiral galaxies with agn to host low - power ( l@xmath1110@xmath12whz@xmath2 ) radio jets , as in seyfert galaxies .
however the majority of these radio sources are sub - kpc - scale , and confined within the optical galaxy ( e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
although accretion of matter on to a central black hole can produce an agn , a merger may be required to launch kpc - scale jets @xcite .
the process theorised to account for kpc - scale double - lobed radio sources requires the merger of two gas - rich galaxies , which eventually forms an elliptical .
this model does not result in a gas - rich spiral hosting a powerful radio - emitting agn .
powerful radio sources that are hosted by spiral and disk - dominated galaxies are known to exist at low - redshift , although to - date there are only a few tens of examples ( e.g * ? ? ?
* ) , most of which are hosted by disk - dominated or lenticular galaxies ( e.g. ngc612 , * ? ? ?
the few examples of these have been studied in detail .
table [ spirals ] gives a complete - to - date list of _ spiral _ galaxies that host large double - lobed radio sources . [ cols="<,<,<,<,<,<",options="header " , ] figure [ radec ] plots the spatial location of the 64 galaxies that are within @xmath13z = 0.003 ( @xmath13cz = 900kms@xmath2 ) of j1649 + 2635 .
there is a hint of a velocity gradient across the field , with the closest galaxies ( blue larger open circles ) predominately north of j1649 + 2635 , and the more - distant galaxies ( red smaller open circles ) predominately south - west of j1649 + 2635
. this may be further evidence for an unrelaxed system of galaxies .
perhaps the system is a group of groups , similar to the environment surrounding 0313 - 192 .
groups contain higher fractions of disk galaxies than clusters ( * ? ? ?
figure [ dct ] shows the dct image with the galaxies that are at the same redshift circled in black .
all of these galaxies show evidence for disk - like structure .
in contrast , the morphologies of cluster members are often ellipticals .
this is further evidence that this system is lower density than a classical cluster of galaxies .
however , the large optical halo around j1649 + 2635 is more reminiscent of a cluster environment . using the 1d projected velocity dispersion of the system ( figure [ hist ] ) , and following the relation from @xcite ,
we compute the dynamical mass of the system to be m@xmath14m@xmath6 , where the errors are due to galaxy membership within the aperture used .
further , @xcite note that due to the difficulty in relating the projected velocity dispersion to the true 3d velocity dispersion , the systematic error may be up to a factor of @xmath72 higher at low redshift .
consequently we estimate the dynamical mass of the system to be m@xmath8m@xmath6 .
this is consistent with the virial masses of groups of galaxies ( e.g. * ? ? ?
we have examined archival _
rosat _ all - sky survey ( rass ) data , and have found no detection towards j1649 + 2635 .
this implies an upper limit to the x - ray luminosity of @xmath15ergs@xmath2hz@xmath2 @xcite , and indicates an upper limit to the mass of @xmath16m@xmath6 @xcite , consistent with the value obtained using the velocity dispersion .
we also note that j1649 + 2635 ( m@xmath17 = 15.5 ) is one of the two brightest galaxies in this system .
the other bright galaxy , sdss j164933.52 + 265052.0 is at z = 0.0551 and is @xmath71mpc away .
it is possible that this galaxy is the cd galaxy for a group or cluster that is in the process of interacting with a group or cluster that j1649 + 2635 resides in .
there is no evidence for a larger - than - usual diffuse optical halo around the other bright galaxy . z @xmath18 0.065 .
the g - band magnitude of j1649 + 2635 is 15.5 . ]
given the disk - like morphologies of the galaxies in the vicinity of j1649 + 2635 , and the dynamical mass estimate , we find the environment of j1649 + 2635 is more like a group of galaxies , than a cluster .
this is also consistent with j1649 + 2635 being a red spiral , as red spirals are usually found in intermediate density environments .
j1649 + 2635 is clearly `` hybrid '' in nature ; it is a star - forming spiral galaxy that has large - scale radio lobes .
although this is generally unseen in the local universe , hybrid sources may be much more prevalent in the early universe ( e.g. * ? ? ?
consequently , understanding the formation of sources such as j1649 + 2635 provides us with an excellent probe for the higher - redshift universe .
here we offer three possible causes for the unusual association of a spiral galaxy with large - scale double - lobed radio emission . *
perhaps j1649 + 2635 is in the unique position of having undergone a planar minor merger that would not introduce angular momentum perpendicular to the plane , so as to not disrupt the spiral structure too much @xcite .
the merger would have to dump sufficient matter ( @xmath19m@xmath6year@xmath2 , assuming 0.1 efficiency ) onto the central black hole so as to trigger radio lobes from the agn ( perhaps the black hole spins very rapidly similar to the @xcite model ) , and it is perhaps this process that quenches the star - formation giving this source its red colour over a few gyr .
the presence of the large diffuse optical halo makes this process appear unlikely as the diffuse halo suggests multiple mergers have occurred , whereupon a spiral structure can not be sustained ( e.g * ? ? ?
* alternatively , j1649 + 2635 may be a `` normal '' elliptical galaxy that has undergone some event ( such as a merger ) that has caused it to grow spiral arms .
however , spiral arms require large reservoirs of gas to maintain .
the red colour of this galaxy suggests that the gas supply has been cut - off , thus the spiral arms may decay quite quickly .
* we offer one final possible explanation for this unique source .
seyfert galaxies are known to harbour agn , and yet are predominately disk or spiral galaxies .
seyfert galaxies are also predominately found in the field . what happens if a seyfert galaxy is in a dense environment ?
could the igm support large - scale radio lobes ?
in this paper we have presented the best example to date of a grand - design spiral hosting a kpc - scale double - lobed radio source . in many ways , j1649 +
2635 is a `` normal '' red spiral galaxy .
it has clearly visible spiral arms and infrared data from wise and nuv data from galex both suggest that there is ongoing star - formation with sfr@xmath7 0.26 2.6m@xmath6year@xmath2 .
j1649 + 2635 appears to reside in a group environment ( m@xmath8m@xmath6 ) , which is consistent with red spirals . on the other hand , j1649 + 2635 hosts a @xmath20kpc double - lobed radio source .
the association of large double - lobed radio sources with spiral galaxies is exceedingly rare in the local universe .
moreover , the galaxy hosts a @xmath796kpc diffuse optical halo whose integrated surface brightness is comparable to that of the spiral disk . the halo is similar to those found around the centrally dominant cd galaxies in massive clusters .
this is the first instance of a grand - design spiral hosting a large diffuse optical halo .
the presence of a large and diffuse optical halo and the overdensity of galaxies detected by sdss implies that j1649 + 2635 sits in a group - sized gravitational halo .
the non - gaussian redshift histogram supports the theory that the environment is unrelaxed , hinting at a merging system or unvirialised group .
the black hole mass that we estimate for this source , m@xmath3 37 @xmath4 10@xmath5m@xmath6 , is consistent with the galaxy having perhaps undergone a merger .
we considered three possible causes for the unusual association of a spiral galaxy with large - scale double - lobed radio emission .
we suggested a planar merger may be responsible for the red colour and black - hole mass .
however , if the diffuse halo is indicative of multiple mergers , this theory seems unlikely .
we considered whether j1649 + 2635 is a `` normal '' elliptical galaxy that is undergoing a merger and serendipitously appears spiral for a brief period of time .
finally , we speculate whether j1649 + 2635 may be a seyfert galaxy that is in a denser - than - normal environment .
further observations are required to test these predictions .
for example , long - slit or integrated field spectroscopic data over the source would enable us to determine the stellar population in the spiral arms .
if large velocity dispersions are observed , this may be evidence for a recent merger .
multi - object spectroscopy over the field would provide us the larger - scale clustering environment .
observations of the hi-21 cm line in the system would enable us to determine the atomic gas mass and kinematics of this source and hence deduce if the hi properties are similar to other spiral galaxies .
high - resolution radio imaging of the agn will enable us to probe the parsec - scale structure at the black hole and gain insight into the agn activity .
finally , to determine whether the large - scale diffuse radio emission from j1649 + 2635 is consistent with a re - activated radio source , similar to speca and j2345 + 0449 , observations of the diffuse radio emission are required . with the advent of citizen science projects such as galaxy zoo and radio galaxy zoo , perhaps many more of these enigmatic objects may be discovered . with this in mind , we propose calling these sources spiral dragns , the dragns acronym deriving from double - lobed radio sources associated with galactic nuclei @xcite .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
rd was a summer student at the national radio astronomy observatory .
the portion of this research for which tm is responsible was performed while he held a national research council research associateship award at the naval research laboratory .
these results made use of the discovery channel telescope at lowell observatory .
lowell is a private , non - profit institution dedicated to astrophysical research and public appreciation of astronomy and operates the dct in partnership with boston university , the university of maryland , the university of toledo and northern arizona university .
the nsf funded the construction of the lmi under grant ast-1005313 .
we are grateful to the mnras editors and the referee for their helpful feedback , and to jay blanchard , chris carilli , phil hopkins , huib intema , kristina nyland , tom pannuti , anna scaife , lornt sjouwerman , jamie stevens , gustaaf van moorsel , lisa young and many others for numerous useful and insightful discussions . | we report the discovery of a grand - design spiral galaxy associated with a double - lobed radio source . j1649 + 2635 ( z = 0.0545 ) is a red spiral galaxy with a prominent bulge that it is associated with a l@xmath010@xmath1whz@xmath2 double - lobed radio source that spans almost 100kpc .
j1649 + 2635 has a black hole mass of m@xmath3 37 @xmath4 10@xmath5m@xmath6 and sfr@xmath7 0.26 2.6m@xmath6year@xmath2 .
the galaxy hosts a @xmath796kpc diffuse optical halo , which is unprecedented for spiral galaxies .
we find that j1649 + 2635 resides in an overdense environment with a mass of m@xmath8m@xmath6 , likely a galaxy group below the detection threshold of the rosat all - sky survey .
we suggest one possible scenario for the association of double - lobed radio emission from j1649 + 2635 is that the source may be similar to a seyfert galaxy , located in a denser - than - normal environment .
the study of spiral galaxies that host large - scale radio emission is important because although rare in the local universe , these sources may be more common at high - redshifts .
[ firstpage ] galaxies : general galaxies : active galaxies : spiral radio continuum : galaxies galaxies : jets |
spintronics in molecular conductors is a field attracting more and more attention , both from fundamental physics as well as from application - oriented material science @xcite . here
the quantum - mechanical electronic spin is the central object controlling transport properties . for a conductor sandwiched between ferromagnetic leads ,
a different resistance can be observed depending on the relative orientation of the lead magnetizations . quite often , the resistance is larger in the antiparallel configuration than in the parallel one , but sometimes also the reverse situation can be observed .
it is useful to define the tunnel magnetoresistance ( tmr ) , @xmath1 , as the relative difference between the corresponding resistances .
a particularly interesting material in that context is provided by carbon nanotubes ( cnts ) , see refs .
@xcite for general reviews .
quite a number of experimental studies concerning spin transport through individual multi- ( mwnt ) or single - walled ( swnt ) nanotubes contacted by ferromagnetic leads have been reported over the past few years @xcite .
in particular , the experiments of the basel group @xcite use thin - film pdni alloys as ferromagnetic leads in order to contact either swnts or mwnts , where the shape anisotropy and the geometry of the setup allow for the study of the spin - dependence of electrical transport .
these experiments have revealed _ oscillatory _ behavior of the tmr as a function of the external gate voltage .
similar oscillations were predicted as a consequence of the gate - voltage - tunable rashba spin - orbit ( so ) interaction @xcite in a classic paper by datta and das some time ago @xcite .
since datta - das oscillations have still not been observed experimentally so far , a thorough theoretical investigation of this effect in nanotubes is called for and provided here . unfortunately , from our analysis below , we find that the weakness of so couplings in nanotubes excludes an interpretation of these data in terms of the datta - das effect they can , however , be explained in terms of quantum interference effects @xcite .
nevertheless , we show that the presence of multiple bands in cnts is not detrimental , and under certain circumstances , the effect may be sufficiently enhanced to be observable , e.g. , by a tuning of the number of bands via external gates along the lines of ref . @xcite . in the original datta - das proposal @xcite , subband mixing was ignored so that different channels just add up coherently , but subband mixing has later been argued to spoil the effect @xcite . in cnts ,
the special band structure requires a careful re - examination of the datta - das idea in this context , and we shall show that the arguments of refs .
@xcite do not necessarily apply here .
recent theoretical studies of spin - dependent transport in cnts have mainly focused on the single - channel limit , taking into account electron - electron interactions within the framework of the luttinger liquid theory @xcite ( see also @xcite for related discussions on interacting quantum wires with rashba so coupling ) . here
we confine ourselves to the noninteracting problem in order to not overly complicate the analysis , but study the many - band case and details of the band structure .
interactions can be taken into account within the luttinger liquid approach at a later stage , and may enhance the effect of so couplings @xcite .
we shall also neglect disorder effects .
mean free paths in high - quality swnts typically exceed @xmath2 m , while in mwnts this may be a more severe approximation for some samples . however , high - quality mwnts with ultra - long mean free paths have also been reported recently @xcite .
the structure of this paper is as follows . in sec .
[ sec2 ] we derive the rashba spin - orbit hamiltonian from microscopic considerations .
the resulting tight - binding so hamiltonian will be studied at low energy scales in sec .
[ sec3 ] , where we derive its continuum form . in sec .
[ sec4 ] , the consequences with regard to datta - das oscillations in the tmr are analyzed .
we shall always consider the zero - temperature limit , and ( in most of the paper ) put @xmath3 .
we start by noting that transport effectively proceeds through the outermost shell of a mwnt only , such that we can take a single - shell model even when dealing with a mwnt .
experimentally and theoretically , it is understood that such a model works very well in good - quality mwnts @xcite , essentially because only the outermost shell is electrically contacted and tunneling between different shells is largely suppressed @xcite .
naturally , a single - shell description is also appropriate for swnts , where we assume a sufficiently large radius @xmath4 such that occupation of multiple subbands can be possible .
( for a mwnt , @xmath4 denotes the radius of the outermost shell . )
depending on the electrochemical potential @xmath5 ( doping level ) , we then have to deal with @xmath0 spin - degenerate bands .
we assume full quantum coherence ( no dephasing ) , so that the usual landauer - bttiker approach applies , and exclude external magnetic fields or electric field inhomogeneities , say , due to the electrodes . we proceed to derive the rashba so interaction , @xmath6 , for this problem .
notice that this is different from the intrinsic atomic so interaction discussed in refs .
in particular , the so coupling in refs .
@xcite vanishes in the limit of large radius , which is not the case for the rashba so coupling we discuss below . though ando s so coupling @xcite could straightforwardly be included in our analysis , being gate - voltage independent it could not change our conclusions relative to the gate - voltage dependent oscillations in the magnetoresistance and is neglected in what follows .
we first define a fixed reference frame @xmath7 , with unit vector @xmath8 pointing in the axis direction and @xmath9 perpendicular to the substrate on which the cnt is supposed to be located .
next we introduce a second , local reference frame @xmath10 relative to each lattice site @xmath11 on the tube surface , where @xmath12 and @xmath13 are unit vectors along the local normal and tangential ( around the circumference ) directions at @xmath11 , respectively . using polar coordinates in the plane transverse to the tube axis ,
the relation between @xmath14 and @xmath15 is given by @xmath16 the position vector of a given carbon atom can then be written as @xmath17 . for later convenience ,
we introduce also another reference frame . for each pair of sites
@xmath18 and @xmath19 , we define @xmath20 and denote the direction perpendicular to @xmath21 and @xmath8 as @xmath22 .
then @xmath23 constitutes a new local frame @xmath24 , and one has @xmath25 \ , \hat y + \sin [ ( \varphi_i+\varphi_j)/2 ] \ , \hat z , \\ \nonumber \hat \rho_{ij } & = & -\sin [ ( \varphi_i+\varphi_j)/2 ] \ , \hat y + \cos [ ( \varphi_i+\varphi_j)/2 ] \ , \hat z .
\end{aligned}\ ] ] the @xmath26 orbital at position @xmath11 can then be represented as @xmath27 where @xmath28 , @xmath29 , @xmath30 is the bohr radius , and @xmath31 is the electron s mass .
we introduce an index @xmath32 on the orbital in order to keep track of the atom at which it is centered . the wavefunction ( [ 2pz ] ) is expected to be highly accurate for not too small @xmath4 , where hybridization with the @xmath33 orbitals is negligible . at large distances from the tube
, external gates generally produce an electric field perpendicular to the tube axis and the substrate .
as it has been shown in detail in previous works @xcite , polarization effects of the cnt itself due to a transverse field result in a reduction of the externally applied field described by @xmath34 where @xmath35 is the unscreened transverse static polarizability .
since @xmath35 is approximately proportional to @xmath36 , the factor in front of @xmath37 practically equals a constant , @xmath38 @xcite .
then , assuming homogeneity , the electric field due to the gate can be written as @xmath39 which in turn produces the ( first - quantized ) rashba spin - orbit interaction @xcite . with standard pauli matrices
@xmath40 acting in spin space , @xmath41 we proceed to derive the second - quantized spin - orbit hamiltonian within the tight - binding approximation . for that purpose , we need the matrix element of the momentum operator between two @xmath26 orbitals @xmath42 , from which we get the following form for the so hamiltonian : @xmath43 c_j , \ ] ] where the fermionic operator @xmath44 destroys an electron with spin @xmath45 in the @xmath26 orbital centered at @xmath11 , and @xmath46 . for calculational convenience , the matrix element @xmath47 can be written as @xmath48 , where the spin - orbit vectors @xmath49 and @xmath50 are defined as @xmath51 note that the modulus of @xmath49 and @xmath50 has dimension of energy , and their sum ( but not necessarily each term separately ) is antisymmetric under exchange of @xmath32 and @xmath52 .
we first observe that the spin - orbit vectors connecting a site with itself clearly vanish , since @xmath53 .
let us then discuss spin - orbit vectors connecting different sites .
since the orbitals ( [ 2pz ] ) decay exponentially , it is sufficient to consider only the case of nearest neighbors .
we start with @xmath49 .
shifting @xmath54 in eq .
( [ sovector1 ] ) and using eq .
( [ 2pz ] ) , we obtain @xmath55 using @xmath56 , we then rewrite the above integral as @xmath57 where we use @xmath58 , with the nearest - neighbor distance among carbon atoms in graphene @xmath59 .
note that @xmath60 .
the second term in the brackets is odd in @xmath61 and thus vanishes , and we obtain @xmath62 where we have used @xmath63 and the dimensionless numerical factor @xmath64 : @xmath65 for @xmath66 , we find @xmath67 notice that , up to higher orders in @xmath68 , the unit vectors @xmath69 can be replaced by @xmath70 , which makes clear that @xmath49 is normal to the tube surface . now
@xmath71|$ ] varies between zero ( when the two sites are aligned in the axis direction ) and @xmath72 ( when the two sites are aligned in the circumferential direction ) .
thus , to zeroth order in @xmath68 , @xmath49 vanishes : it is a pure curvature effect , peculiar of nanotubes , which does not exist in graphene . in practice , @xmath49 is tiny and certainly subleading to @xmath50 , which turns out to be of order @xmath73
. we shall therefore neglect it in what follows .
let us now turn to @xmath50 .
we shift @xmath74 in eq .
( [ sovector2 ] ) , and rewrite @xmath50 as the sum of two terms : @xmath75 writing again @xmath76 , the computation of the above integrals leads , to the lowest non - vanishing order in @xmath68 , to the following expressions : @xmath77 with the dimensionless numerical factors @xmath78 and @xmath79 to lowest order in @xmath68 , it does not make a difference whether we take the tangent unit vector at @xmath19 , @xmath11 , or at @xmath80 .
hence we may write @xmath81 , where @xmath82 is the unit tangent vector at @xmath83 .
we then get so couplings along the axial and along the circumferential direction , @xmath84 , \ ] ] with @xmath85 .
note that we have neglected a tiny component of @xmath86 normal to the tube surface .
the above discussion then results in the tight - binding hamiltonian @xmath87 , where @xmath88 with @xmath89 ev @xcite . here
the @xmath90 denote all sublattice - a tight - binding sites of the lattice .
furthermore , the @xmath91 are vectors connecting @xmath90 with the three nearest - neighbor sites which are all located on sublattice b @xcite .
since we consider the limit @xmath92 , the @xmath93 at each site effectively lie in the tangent plane to the tube surface at that site .
the rashba spin - orbit hamiltonian then reads @xmath94 \right ) \cdot \hat z \bigr ] c^{}_{a,\vec r } + h.c .
\nonumber\ ] ]
since we are interested in the low - energy long - wavelength properties , we now expand the electron operator around the fermi points @xmath95 in terms of bloch waves @xcite , @xmath96 where @xmath97 is the area of the unit cell , @xmath98 , and @xmath99 is the sublattice index .
the @xmath100 are slowly varying electron field operators , and we choose the fermi points at @xmath101 and @xmath102 @xcite .
we then expand @xmath103 and use the bond vectors @xmath104 these vectors are given in a fixed reference frame for a 2d graphene sheet , and we then must perform a rotation to longitudinal and circumferential directions via the chiral angle .
this rotation results in fixed phases that can be absorbed in the definition of @xmath100 and do not appear in final results .
this is of course expected from the @xmath105 symmetry emerging at low energies in the dispersion relation of graphene @xcite . after some algebra , the usual dirac hamiltonian for the kinetic term follows , @xmath106 f , \ ] ] where @xmath107 m / sec is the fermi velocity , and @xmath108 are longitudinal and circumferential coordinates , respectively , with @xmath109 .
finally , @xmath110 and @xmath111 are also pauli matrices that now act in the space of fermi ( @xmath95 ) points and sublattice space ( @xmath112 ) , respectively . for @xmath113 ,
these are defined as @xmath114 unit matrices .
the low - energy limit of the so term ( [ hso ] ) can be obtained in the following way .
first we observe that @xmath115 \times \hat z = - ( \vec \delta_a\cdot \hat x ) \hat y - \sin ( y / r ) ( \vec \delta_a \cdot \hat e_\varphi ) \hat x .\ ] ] here the only approximation is the assumption that the bond vectors lie in the plane tangent to the nanotube surface at @xmath90 .
second , by using the bond vectors ( [ delta ] ) and taking into account the chiral angle @xmath116 between the fixed direction on the graphite sheet and the circumferential direction on the nanotube , one obtains @xmath117 notice that we take into account exactly the relative orientation of the bond vectors with respect to the directions @xmath8 and @xmath82 for a generic nanotube , which is encoded in the chiral angle @xmath116 .
the constant phases @xmath118 can be absorbed by appropriately redefining the operators as @xmath119 and @xmath120 , and the final result can be written down in the form @xmath121 f , \ ] ] with @xmath122 .
for the sake of generality , we continue to use different coupling constants @xmath123 and @xmath124 .
it is worthwhile to mention that the leading term for the rashba spin - orbit coupling in a cnt , eq .
( [ hsso ] ) , does not depend on longitudinal momentum .
this is due to the peculiar band structure of graphene with its isolated fermi ( @xmath125 ) points . in the above derivation
, we also find terms that are linear in momentum , i.e. , contain spatial derivatives of the electron operators .
such terms only produce tiny renormalizations of the velocities and will be neglected here . the second term in eq .
( [ hsso ] ) allows for spin flips and mixes transverse subbands . from now on , for simplicity , we consider just a single fermi point , say , @xmath125 .
after the global @xmath126 rotation @xmath127 in spin space , we get in compact notation @xmath128 . \\
\label{soham } { \cal h}_{so } & = & u_\parallel \tau_1 \sigma_3 + u_\perp \sin(y / r ) \tau_2 \sigma_2 .\end{aligned}\ ] ] note that the exact spectrum of @xmath129 with @xmath130 can be obtained straightforwardly . in general , however , due to the smallness of the so coupling ( see below ) , it is enough to treat @xmath131 perturbatively .
the following detailed derivation is then necessary to correctly evaluate the effect of the so coupling , and moreover it is interesting and important for the generalization to the interacting case , and for the analysis of features involving the electron wavefunction ( as for instance electron - phonon interactions ) .
the eigenvalues of @xmath132 are given by @xmath133 where @xmath134 denotes the transverse momentum , @xmath135 the longitudinal one , @xmath136 labels the conduction / valence band , and @xmath137 the spin .
here @xmath138 for intrinsically metallic shells , but generally it can be taken as @xmath139 to take into account chirality gaps or orbital magnetic fields along @xmath8 .
the transverse subbands are labeled by integer values @xmath140 , where @xmath141 for ( @xmath142 ) tubes @xcite .
@xmath143 is typically much larger than the actual number @xmath144 $ ] of occupied subbands , where we define @xmath145 with the doping level @xmath5 that we assume positive here .
the velocity @xmath146 for electrons in subband @xmath147 at the fermi level ( in the absence of @xmath6 ) and the corresponding fermi momentum @xmath148 are then given by @xmath149 ^ 2},\quad q_n = k_f v_n / v.\ ] ] the eigenvalues ( [ eigzero ] ) are spin - independent and thus doubly degenerate .
the corresponding eigenstates are denoted @xmath150 , where @xmath151 and @xmath152 are respectively plane waves in circumferential and longitudinal direction . in coordinate representation they read @xmath153 with the bispinor ( in sublattice space )
@xmath154 a different , and here more convenient basis is given by the sublattice states @xmath155 .
their coordinate representation is @xmath156 where @xmath157 and @xmath158 their usefulness stems from the fact that the @xmath159 can be factorized as @xmath160 where @xmath161 is independent of @xmath147 and @xmath135 . using this basis
, we can expand the field operator @xmath162 on the tube surface as @xmath163 where the operator @xmath164 destroys an electron in the state @xmath165 , and we introduce the 1d field operators @xmath166 .
alternatively , using the basis of eigenstates of @xmath167 , @xmath162 can be expanded as @xmath168 where the operators @xmath169 destroy conduction ( @xmath170 ) or valence ( @xmath171 ) electrons with spin @xmath172 in subband @xmath147 .
notice that in what follows the spin index is left implicit .
the relation between the operators @xmath173 and @xmath174 is easily found to be @xmath175 we now proceed by treating the spin - orbit hamiltonian using perturbation theory .
first , we diagonalize @xmath176 for a fixed transverse subband @xmath147 , @xmath177 f_n \\ & = & \sum_{a=\pm } \int \frac{dq}{2\pi } [ a\epsilon_n(q ) - \mu ] c^\dagger_{na } c^{}_{na}.\end{aligned}\ ] ] next we expand around the fermi points @xmath178 defined in eq . ( [ vnqn ] ) , which introduces right- and left - movers , @xmath179 , as the relevant low - energy degrees of freedom . for small deviations
@xmath180 from @xmath178 , taylor expansion yields @xmath181 , where @xmath146 is given in eq .
( [ vnqn ] ) .
since we assumed @xmath182 , we may now restrict ourselves to the conduction band , @xmath170 .
for the hamiltonian , we then obtain @xmath183 where @xmath184 and @xmath185 .
this introduces @xmath186-moving 1d fermion operators for each subband @xmath147 ( and spin @xmath172 ) .
the relation of these 1d fermions with the original operator @xmath166 is given by @xmath187 notice that , while in general the unitary transformation from sublattice space to the conduction / valence band description depends on longitudinal momentum , in the continuum limit , one can use the transformation directly at the fermi momenta .
this is consistent with the neglect of band curvature effects implicit in the linearization of the dispersion relation , which is unproblematic away from van hove singularities associated with the onset of new subbands @xcite . at these points , the concept of @xmath186-movers breaks down , and some of our conclusions below may change .
next we express the rashba hamiltonian ( [ soham ] ) in terms of @xmath186-movers .
the first term results in @xmath188 the presence of the factor @xmath189 results from a careful treatment of the phases in eq .
( [ fn ] ) . in eq .
( [ rasb ] ) we omit an additional term mixing right- and left - movers .
this term contains a rapidly oscillating factor @xmath190 and therefore is strongly suppressed by momentum conservation . the second term in eq . ( [ soham ] )
again contains the oscillating phase factor @xmath191 , which leads to a drastic suppression of @xmath192 at low energies and long wavelengths .
of course , this argument relies in an essential way on the smallness of the coupling @xmath123 , as one expands around the hamiltonian @xmath167 .
we conclude that away from van hove singularities , the only important rashba term is given by @xmath193 in eq .
( [ rasb ] ) .
this term has the appearance of a static homogeneous but _ channel - dependent magnetic field_.
in this section we will analyze the consequences of our findings regarding spin - orbit couplings in cnts , see eqs .
( [ rasb ] ) , for the observability of spin precession effects encoded in the datta - das oscillations of the tmr . based on our expressions , it is possible to estimate the order of magnitude of this effect . for a concrete estimate ,
let us put @xmath194 , where @xmath195 is the gate - tube distance , @xmath196 the gate voltage , and @xmath197 denotes the dielectric constant of the substrate . for a given channel @xmath147 , the rashba - induced energy splitting is then easily estimated as @xmath198 where @xmath199 m is the compton length . plugging in the definition of @xmath200 , we get @xmath201 bands with small @xmath147 are only weakly split , and hence do not contribute to oscillatory tmr behavior .
this argument suggests that datta - das oscillations in principle could survive in a cnt , even when there are many channels .
the major contribution will come just from the few bands with the largest @xmath147 . to estimate the accumulated phase difference due to the different precession length of the two split eigenstates ,
let us put @xmath202 , which represents the dominant contribution , and set @xmath203 .
then eq . ( [ splitting ] ) gives as order - of - magnitude estimate @xmath204 even when assuming a very close - by gate , this gives only a tiny splitting , in retrospect justifying perturbation theory .
this splitting now translates into a momentum splitting @xmath205 , and hence into a precession phase mismatch along the cnt of length @xmath206 @xcite . for the @xmath147th band
, this phase difference is @xmath207 this phase difference should be of order @xmath208 to allow for the observation of datta - das oscillatory tmr effects @xcite .
away from a van hove singularity , eq . ( [ phase ] ) predicts that oscillations appear on a gate voltage scale of the order of @xmath209 to @xmath210 v for @xmath211 , which would make datta - das oscillations unobservable .
this argument also shows that this interpretation can be ruled out for the parameters relevant for the basel experiment @xcite . from eq .
( [ phase ] ) , we can then suggest several ways to improve the situation .
first , one should use very long cnts , while at the same time keeping the gate very close , and second , an enhancement can be expected close to van hove singularities . of course , very close to a van hove singularity , some of our arguments above break down , but the general tendency can nevertheless be read off from eq .
( [ phase ] ) .
furthermore , electron - electron interactions can also enhance spin - orbit effects @xcite . to conclude
, we have presented a detailed microscopic derivation of rashba spin - orbit coupling in carbon nanotubes .
it turns out that the rashba so coupling is small , and therefore the prospects for observing spin - precession effects like datta - das oscillations in the tunneling magnetoresistance are not too favorable .
however , for very long cnts , close - by gates , and in the vicinity of a van hove singularity , the requirements for observability of these effects could be met in practice .
we thank t. kontos and c. schnenberger for motivating this study .
support by the dfg ( gerhard - hess program ) and by the eu ( dienow network ) is acknowledged . | the rashba spin - orbit coupling in carbon nanotubes and its effect on spin - dependent transport properties are analyzed theoretically .
we focus on clean non - interacting nanotubes with tunable number of subbands @xmath0 .
the peculiar band structure is shown to allow in principle for datta - das oscillatory behavior in the tunneling magnetoresistance as a function of gate voltage , despite the presence of multiple bands .
we discuss the conditions for observing datta - das oscillations in carbon nanotubes . |
let @xmath7 be the canonical basis for the @xmath8-dimensional integer lattice @xmath0 with @xmath9 . consider a discrete - time markov chain on @xmath0 with nearest - neighbor steps , that is , with steps in @xmath10 .
for every @xmath11 and @xmath12 , denote the transition probability from @xmath13 to @xmath14 by @xmath15 and refer to the transition vector @xmath16 as the _ environment _ at @xmath13 . if the environment @xmath17 is sampled from a probability space @xmath18 , then this process is called _ random walk in a random environment _ ( rwre ) . here
, @xmath19 is the borel @xmath20-algebra corresponding to the product topology . the environment is said to be _ uniformly elliptic _ if @xmath21 for every @xmath22 , define the shift @xmath23 on @xmath24 by @xmath25 . throughout this paper
, we will assume that @xmath26 is stationary and ergodic under @xmath27 .
this condition is clearly satisfied when @xmath28 for every @xmath11 and @xmath29 , the markov chain with environment @xmath30 induces a probability measure @xmath31 on the space of paths starting at @xmath13 .
statements about @xmath31 that hold for @xmath26-a.e .
@xmath30 are referred to as _
quenched_. statements about the semidirect product @xmath32 are referred to as _ averaged _ ( or _ annealed _ ) .
expectations under @xmath33 and @xmath34 are denoted by @xmath35 and @xmath36 , respectively .
see @xcite for a survey of results on rwre .
because of the extra layer of randomness in the model , the standard questions of recurrence versus transience , the law of large numbers ( lln ) , the central limit theorem ( clt ) and the large deviation principle ( ldp)which have well - known answers for classical random walk become hard .
however , it is possible , by taking the _ point of view of the particle _ , to treat the two layers of randomness as one : if we denote the random path of the particle by @xmath37 , then @xmath38 is a markov chain ( referred to as the _ environment markov chain _ ) on @xmath24 with transition kernel @xmath39 given by @xmath40 this is a standard approach in the study of random media ; see , for example , @xcite or @xcite . instead of viewing the environment markov chain as an auxiliary construction
, one can introduce it first and then deduce the particle dynamics from it .
[ ortamkeli ] a function @xmath41 is said to be an `` environment kernel '' if @xmath42 is @xmath43-measurable for each @xmath12 and @xmath44 .
it can be viewed as a transition kernel on @xmath24 via the following identification : @xmath45 given @xmath11 , @xmath29 and any environment kernel @xmath46 , the quenched probability measure @xmath47 on the space of particle paths @xmath48 starting at @xmath13 in environment @xmath30 is defined by setting @xmath49 and @xmath50 for all @xmath51 , @xmath22 and @xmath12 .
the semidirect product @xmath52 is referred to as the averaged measure and expectations under @xmath53 and @xmath54 are denoted by @xmath55 and @xmath56 , respectively . in this paper
, we will focus on the large deviation properties of multidimensional rwre .
section [ kurdungel ] is a detailed survey of the previous results on this topic that are relevant to our purposes .
the precise statements of our results are postponed to section [ inisoyler ] because they rely heavily on the notation and theorems given in section [ kurdungel ] . in this subsection
, we will provide a short and less technical description of the key theorems in section [ kurdungel ] .
references will be omitted for the sake of brevity .
we will then highlight our main results . in the case of
quenched rwre , the ldp holds for the mean velocity @xmath57 of the particle .
rosenbluth gives a variational formula for the corresponding rate function @xmath1 .
for any @xmath58 , @xmath59 is equal to the infimum of @xmath60 , where @xmath61 is a relative entropy and @xmath62 varies over all pairs such that : ( i ) @xmath63 is an environment kernel ; ( ii ) @xmath64 is a @xmath63-invariant probability measure on @xmath24 ; ( iii ) @xmath65 on @xmath19 ; ( iv ) the asymptotic mean velocity of the walk induced by @xmath62 is equal to @xmath6 . for averaged walks in i.i.d .
environments , varadhan proves the ldp for @xmath57 and gives yet another variational formula for the corresponding rate function @xmath2 . for any @xmath66 , @xmath67 is the infimum of @xmath68 , where @xmath69 is a relative entropy [ not equal to @xmath61 ] and @xmath71 varies over all @xmath0-valued transient processes with stationary and ergodic increments in @xmath72 such that the mean drift of @xmath71 is equal to @xmath6 .
it is easily shown that ( i ) @xmath73 and ( ii ) @xmath1 , @xmath2 are not identically equal .
when @xmath3 and the walk satisfies the so - called ( * t * ) condition of sznitman , @xmath1 and @xmath2 are known to be strictly convex , analytic and equal on an open set @xmath4 . at every @xmath5 ,
varadhan s variational formula for @xmath67 has a unique minimizer .
we will assume that the environment is i.i.d .
, @xmath3 and the ( * t * ) condition of sznitman holds . for every @xmath5
, we will prove the existence of an @xmath74 that solves a certain equation involving @xmath75 and the original kernel @xmath76 of the walk ; see ( [ carbonic ] ) .
since ( [ carbonic ] ) resembles the laplace equation , we will refer to @xmath77 as _ harmonic_. we will then use @xmath78 to define a new environment kernel @xmath79 via the h - transform technique of doob ; see ( [ azdinyn ] ) . for every @xmath5 , we will prove the existence of a probability measure @xmath80 on @xmath24 that is @xmath81-invariant .
the pair @xmath82 corresponds to a stationary markov chain with values in @xmath24 .
this markov chain induces a @xmath0-valued transient process @xmath83 with stationary and ergodic increments in @xmath72 .
we will show that @xmath84 is the unique minimizer of varadhan s variational formula for @xmath67 .
the pair @xmath82 is a natural minimizer candidate for rosenbluth s variational formula for @xmath59 .
however , it is not known whether @xmath85 on @xmath19 .
we will resolve this issue by slightly modifying rosenbluth s formula so that the infimum of @xmath61 will be taken over a larger class of pairs .
finally , we will show that @xmath82 is the unique minimizer of this new formula .
recall that a sequence @xmath86 of probability measures on a topological space @xmath87 satisfies the _ large deviation principle _ ( ldp ) with rate function @xmath88 $ ] if @xmath89 is lower semicontinuous and , for any measurable set @xmath90 , @xmath91 here , @xmath92 is the interior of @xmath90 and @xmath93 its closure .
see @xcite for general background regarding large deviations . in this paper , the following theorem
will be referred to as the quenched ( level-1 ) ldp .
[ qldpgeneric ] assume ( [ ellipticity ] ) .
for @xmath26-a.e .
@xmath30 , @xmath94 satisfies the ldp with a deterministic and convex rate function @xmath1 .
( the subscript stands for `` quenched . '' ) greven and den hollander @xcite prove theorem [ qldpgeneric ] for walks on @xmath95 in i.i.d . environments .
they provide a formula for @xmath1 and show that its graph typically has flat pieces .
comets , gantert and zeitouni @xcite generalize the results in @xcite to stationary and ergodic environments . for @xmath9 ,
the first result on quenched large deviations is given by zerner @xcite .
he uses a subadditivity argument for certain passage times to prove theorem [ qldpgeneric ] in the case of _ nestling _ walks in i.i.d . environments .
[ nonnennon ] rwre is said to be _ nonnestling _ relative to a unit vector @xmath96 if @xmath97 it is said to be _ nestling _ if it is not nonnestling relative to any unit vector . in the latter case
, the convex hull of the support of the law of @xmath98 contains the origin . by a more direct use of the subadditive ergodic theorem ,
varadhan @xcite drops the nestling assumption and generalizes zerner s result to stationary and ergodic environments .
the drawback of these approaches is that they do not lead to any formula for the rate function .
kosygina , rezakhanlou and varadhan @xcite consider diffusions on @xmath99 ( with ) in stationary and ergodic environments .
they prove the analog of theorem [ qldpgeneric ] via a minimax argument and provide a variational formula for the quenched rate function .
rosenbluth @xcite adapts their work to the context of rwre .
[ see ( [ level1ratetilde ] ) below for rosenbluth s variational formula for @xmath1 . ]
the minimax argument of kosygina et al .
@xcite can be generalized to establish a quenched ldp for the so - called _ pair empirical measure _ of the environment markov chain .
below , we introduce some notation in order to give the precise statement of this theorem . for any measurable space @xmath100 , write @xmath101 [ or simply @xmath102 whenever no confusion occurs ] for the space of probability measures on @xmath100 .
consider the random walk @xmath103 on @xmath0 in a stationary and ergodic environment , let @xmath104 and focus on @xmath105 which is a random element of @xmath106 .
the map @xmath107 embeds @xmath106 into @xmath108 and we therefore refer to @xmath109 as the pair empirical measure of the environment markov chain . for any @xmath110 ,
define the probability measures @xmath111 and @xmath112 on @xmath24 by @xmath113 respectively , which are the marginals of @xmath114 when @xmath114 is seen as an element of @xmath108 . with this notation ,
let @xmath115 [ level2ldp ] assume ( [ ellipticity ] ) .
for @xmath26-a.e .
@xmath30 , @xmath116 satisfies the ldp with the rate function @xmath117 , the double convex conjugate of @xmath118 given by @xmath119 rosenbluth s quenched ldp result is a corollary of theorem [ level2ldp ] .
indeed , for any @xmath120 , set @xmath121 for any @xmath58 , define @xmath122 with this notation , @xmath123 here , ( [ level1rate ] ) follows from theorem [ level2ldp ] via the so - called contraction principle ( see @xcite ) .
note that , even though @xmath124 is convex , it may not be lower semicontinuous ( see appendix a of @xcite for an example ) . therefore , @xmath117 is not equal to @xmath125 in general .
nevertheless , ( [ level1ratetilde ] ) is valid ( see @xcite ) and it is precisely equal to the variational formula obtained by rosenbluth in @xcite .
theorem [ level2ldp ] can be generalized to establish a quenched ldp for the _ empirical process _
@xmath126 which is a random element of @xmath127 . here , @xmath128 is shorthand notation for @xmath129 .
[ level3ldp ] assume ( [ ellipticity ] ) .
for @xmath26-a.e .
@xmath30 , @xmath130 satisfies the ldp with a deterministic and convex rate function @xmath131 .
rassoul - agha and sepplinen actually obtain this result in greater generality , namely for bounded step size walks satisfying a weak ellipticity condition ( see @xcite ) . also , they show that , just as in theorem [ level2ldp ] , the rate function @xmath132 is the lower semicontinuous regularization of a relative entropy .
we choose not to state the precise formula of @xmath132 here , partly in order to keep the notation simple and partly because we will not need it in what follows . in this paper , the following theorem
will be referred to as the averaged ( level-1 ) ldp .
[ aldpgeneric ] assume ( [ ellipticity ] ) and ( [ i.i.d . ] ) .
@xmath133 satisfies the ldp with a convex rate function @xmath2 ( the subscript stands for `` averaged '' ) .
comets et al .
@xcite prove theorem [ aldpgeneric ] for @xmath134 and obtain the following variational formula for @xmath2 : @xmath135 here , the infimum is over all stationary and ergodic probability measures on @xmath24 , @xmath136 denotes the rate function for the quenched ldp when the environment measure is @xmath64 and @xmath137 is specific relative entropy . similarly to the quenched picture , the graph of @xmath2 is shown typically to have flat pieces .
varadhan @xcite proves theorem [ aldpgeneric ] for any @xmath9 .
he gives yet another variational formula for @xmath2 .
below , we introduce some notation in order to write down this formula .
an infinite path @xmath138 with nearest - neighbor steps @xmath139 is said to be in @xmath140 if @xmath141 and @xmath142 .
for any @xmath143 , let @xmath144 be the number of times @xmath145 visits the origin , excluding the last visit . by the transience assumption
, @xmath144 is finite . for any @xmath12 ,
let @xmath146 be the number of times @xmath145 jumps to @xmath147 after a visit to the origin .
clearly , @xmath148 .
if the averaged walk starts from time @xmath149 and its path @xmath150 up to the present is conditioned to be equal to @xmath145 , then the probability of the next step being equal to @xmath147 is @xmath151}{\mathbb{e } [ \prod_{z'\in u}\pi ( 0,z')^{n_{o , z ' } } ] } , \ ] ] by bayes rule .
consider the map @xmath152 that takes @xmath138 to @xmath153 .
let @xmath154 be the set of probability measures on @xmath140 that are invariant under @xmath155 and @xmath156 be the set of extremal points of @xmath154 .
each @xmath157 ( resp .
, @xmath158 corresponds to a transient process with stationary ( resp . , stationary and ergodic ) increments and induces a probability measure @xmath159 on particle paths @xmath160 .
the associated _ mean drift _ is @xmath161 .
define @xmath162 for @xmath71-a.e .
@xmath145 and @xmath12 . with this notation , @xmath163 for every @xmath66 , where @xmath164 \,{{d}}\alpha(w).\ ] ] rassoul - agha @xcite generalizes varadhan s result to a class of mixing environments and also to some other models of random walk on @xmath0 . in section [ missinglink ] , we will summarize the known qualitative properties of @xmath2 .
in particular , we will state some regularity results which are valid under a certain transience condition of sznitman .
the next subsection is devoted to introducing this condition , which involves what are called _ regeneration times_. take a unit vector @xmath165 . define a sequence @xmath166 of random times , which are referred to as _ regeneration times _ ( relative to @xmath167 ) , by @xmath168 and @xmath169\\[-8pt ] & & \hspace*{76pt}\mbox{for all } i , k\mbox { with } i < j < k \}\nonumber\end{aligned}\ ] ] for every @xmath170 .
( regeneration times first appeared in the work of kesten @xcite on one - dimensional rwre .
they were adapted to the multidimensional setting by sznitman and zerner ; see @xcite . )
if the walk is directionally transient relative to @xmath167 , that is , if @xmath171 then @xmath172 for every @xmath170 .
as shown in @xcite , the significance of @xmath173 is due to the fact that @xmath174 is an i.i.d .
sequence under @xmath175 when @xmath176 is an i.i.d . collection .
the walk is said to satisfy sznitman s transience condition ( @xmath177 ) if ( [ transience ] ) holds and @xmath178 the following lemmas list some important facts regarding regenerations .
[ madabir ] assume @xmath179 , ( [ ellipticity ] ) , ( [ i.i.d . ] ) and that ( @xmath180 ) holds for some @xmath165 .
then : a. @xmath181 and @xmath182 has finite @xmath175-moments of arbitrary order ; b. the lln holds with a limiting velocity @xmath183 such that @xmath184 ; c. ( @xmath185 ) is satisfied for every @xmath186 such that @xmath187 .
[ madaiki ] assume ( [ ellipticity ] ) and ( [ i.i.d . ] ) .
if the walk is nonnestling ( see definition [ nonnennon ] ) relative to some @xmath96 , then @xmath188<\infty\ ] ] for some @xmath189 .
in particular , ( @xmath180 ) is satisfied . on the other hand ,
if the walk is nestling , then ( [ cisembalkan ] ) fails to hold for every @xmath165 and @xmath189 . the first statement is proved in @xcite .
the second statement follows immediately from the fact that @xmath190 when the walk is nestling ( see @xcite ) .
[ madauc ] assume ( [ ellipticity ] ) and ( [ i.i.d . ] ) . if the walk is nonnestling and some @xmath191 satisfies @xmath192 , then @xmath193<\infty\ ] ] for some @xmath189 .
this is lemma 8 of @xcite .
[ mazaltov ] assume @xmath179 , ( [ ellipticity ] ) , ( [ i.i.d . ] ) and that ( @xmath177 ) holds for some @xmath165 .
since @xmath194 , there exists a @xmath12 such that @xmath195 .
then : a. @xmath196 and @xmath197 has finite @xmath175-moments of arbitrary order ; b. if the walk is nonnestling , then there exists a @xmath198 such that @xmath199<\infty;\ ] ] c. if the walk is nestling , then there exists a @xmath198 such that @xmath200<\infty.\ ] ] denote the zero - sets of @xmath1 and @xmath2 by @xmath201 and @xmath202 , respectively .
the following theorem summarizes some of the known qualitative properties of the quenched and the averaged rate functions when @xmath179 .
the rest of the known properties are given in section [ cizbord ] .
[ previousqual ] assume @xmath179 , ( [ ellipticity ] ) and ( [ i.i.d . ] ) . then : a. @xmath1 and @xmath2 are convex , @xmath203 and @xmath204 ( see @xcite ) ; b. if the walk is nonnestling , then : a. @xmath205 consists of the true velocity @xmath183 ( see @xcite ) ; b. @xmath2 is strictly convex and analytic on an open set @xmath206 containing @xmath183 ( see @xcite ) ; c. if the walk is nestling , then @xmath205 is a line segment containing the origin that can extend in one or both directions ( see @xcite ) ; it can not extend in both directions when @xmath207 ( see @xcite ) or when @xmath208 ( see @xcite ) ; d. if the walk is nestling , but ( @xmath180 ) is satisfied for some @xmath165 , then : a. the origin is an endpoint of @xmath205 ( see @xcite ) ; b. @xmath2 is strictly convex and analytic on an open set @xmath206 ( see @xcite ) ; c. there exists a @xmath209-dimensional smooth surface patch @xmath210 such that @xmath211 ( see @xcite ) ; d. the unit vector @xmath212 normal to @xmath210 ( and pointing in @xmath206 ) at @xmath183 satisfies @xmath213 ( see @xcite ) ; e. @xmath214 for every @xmath215 and @xmath216 $ ] ( see @xcite ) .
assume ( [ ellipticity ] ) and ( [ i.i.d . ] ) .
it is clear that @xmath217 for any @xmath58 , @xmath218 by jensen s inequality and fatou s lemma .
moreover , when the support of @xmath219 is not a singleton , @xmath220 at some interior points of @xmath221 ( see proposition 4 of @xcite ) .
the following theorem considers ballistic walks in dimensions four and higher , and says that the quenched and the averaged rate functions are identically equal on a set whose interior contains @xmath222 .
[ qequalsa ] assume @xmath3 , ( [ ellipticity ] ) , ( [ i.i.d . ] ) and that ( @xmath180 ) holds for some @xmath165 .
then : a. if the walk is nonnestling , @xmath223 on an open set @xmath4 containing @xmath183 ; b. if the walk is nestling : a. @xmath223 on an open set @xmath4 ; b. there exists a @xmath209-dimensional smooth surface patch @xmath224 such that @xmath225 ; c. the unit vector @xmath212 normal to @xmath224 ( and pointing in @xmath4 ) at @xmath183 satisfies @xmath226 ; d. @xmath227 for every @xmath228 and @xmath216 $ ] . assuming @xmath134 , ( [ ellipticity ] ) and ( [ i.i.d . ] ) , comets et al .
@xcite use ( [ ahubay ] ) to show that @xmath229 if and only if @xmath230 or @xmath231 . in particular , theorem [ qequalsa ] can not be generalized to @xmath9 .
it turns out that it can not be generalized to @xmath179 or @xmath232 , either . indeed , for @xmath233 , yilmaz and zeitouni @xcite provide examples of nonnestling walks in uniformly elliptic i.i.d .
environments for which the quenched and the averaged rate functions are not identically equal on any open set containing the true velocity @xmath183 . for every @xmath234 , consider the logarithmic moment generating functions @xmath235 \quad\mbox{and}\nonumber\\[-8pt]\\[-8pt ] \lambda_a(\theta ) & : = & \lim_{n\to\infty}\frac{1}{n}\log e_o [ \exp\{\langle\theta , x_n\rangle\ } ] .\nonumber\end{aligned}\ ] ] by varadhan s lemma ( see @xcite ) , @xmath236 , the convex conjugate of @xmath1 at @xmath237 .
similarly , @xmath238 . for every @xmath189 , define @xmath239 in the latter case , @xmath190 ; see theorem [ previousqual ] .
it follows from convex duality that @xmath240 in other words , @xmath241 for every @xmath242 .
the zero - level set @xmath243 of the convex function @xmath244 is convex and @xmath245 is an open ball minus this convex set . the following theorems state some of the known qualitative properties of @xmath246 and @xmath244 .
[ caveraged ] assume @xmath179 , ( [ ellipticity ] ) and ( [ i.i.d . ] ) .
recall ( [ nebiliyim ] ) .
if ( @xmath180 ) holds for some @xmath247 , then @xmath244 is analytic on @xmath248 , where @xmath249 is as in corollary [ mazaltov ] .
moreover , the hessian @xmath250 of @xmath244 is positive definite on @xmath251 .
[ cananbe ] assume @xmath3 , ( [ ellipticity ] ) and ( [ i.i.d . ] ) . recall ( [ nebiliyim ] ) .
if ( @xmath180 ) holds for some @xmath247 , then there exists a @xmath252 such that @xmath253 on @xmath254 .
in fact , the regularity properties of @xmath2 that are stated in theorem [ previousqual ] are obtained from theorem [ caveraged ] via convex duality ( see @xcite ) and @xmath255 .
similarly , note that theorem [ qequalsa ] is a corollary of theorem [ cananbe ] and @xmath256 .
in this paper , we will obtain new results concerning the large deviation properties of rwre on @xmath0 under the conditions of theorems [ qequalsa ] and [ cananbe ] .
in other words , we will assume that @xmath257 } and ( $ \mathbf{t},e_1 $ ) holds .
\end{tabular}\hspace*{-32pt}\ ] ] here , we have chosen @xmath258 for convenience . however , there is no loss of generality , that is , we could have chosen any @xmath259 ; see lemma [ madabir ] . given any @xmath234 , define @xmath260 by setting @xmath261 for every @xmath29 and @xmath12 .
our first result concerns the existence of positive harmonic functions for @xmath262 .
( here , we use the term _ harmonic _ in analogy with the continuum case where @xmath262 is replaced by a second order elliptic operator . )
[ existence ] assume ( [ kondisin ] ) .
recall theorem [ cananbe ] . for every @xmath263
, there exists an @xmath264 such that @xmath265 and @xmath266 note that @xmath262 would correspond to a markov chain on @xmath24 if @xmath267 were true for @xmath219-a.e .
however , as we will see , the latter condition is not satisfied unless @xmath268 .
nevertheless , ( [ carbonic ] ) enables us to define an environment kernel ( as in definition [ ortamkeli ] ) related to @xmath262 via the so - called _ h - transform _ technique of doob ; see @xcite .
[ doobabi ] assume ( [ kondisin ] ) .
for every @xmath269 , define a new environment kernel @xmath270 by setting @xmath271 for every @xmath29 and @xmath12 .
this technique is called _ h - transform_. recall the sets @xmath206 and @xmath4 which were introduced in theorems [ previousqual ] and [ qequalsa ] , respectively .
whenever @xmath272 , it is shown in @xcite that there is a unique minimizer of varadhan s variational formula ( [ divaneasik ] ) for @xmath67 .
our second result reveals the hidden markovian structure of this minimizer when ( [ kondisin ] ) holds and @xmath5 . before stating this theorem
, we need to introduce a family of sub-@xmath20-algebras of @xmath19 : for any @xmath273 and @xmath51 , let @xmath274 [ varminvar ] assume ( [ kondisin ] ) .
recall theorem [ qequalsa ] and definition [ doobabi ] . for every @xmath5
, there exists a unique @xmath275 such that @xmath276 .
[ by convex duality , @xmath277 .
] a. there exists a unique @xmath278 that satisfies the following : a. @xmath80 is @xmath79-invariant , that is , @xmath279 ; b. @xmath85 on @xmath280 for every @xmath51 ; see ( [ subsigsub ] ) .
+ the pair @xmath82 corresponds to a stationary markov chain ( with values in @xmath24 ) which can be identified with a @xmath281 .
the marginal on @xmath24 of @xmath282 is @xmath283 and @xmath79 is the conditional of @xmath284 given @xmath30 .
b. @xmath282 induces a @xmath0-valued transient process with stationary increments in @xmath72 via the map @xmath285 extend this process to a probability measure on doubly infinite paths @xmath286 and refer to its restriction to @xmath287 as @xmath84 . with this notation
, @xmath84 is the unique minimizer of varadhan s variational formula ( [ divaneasik ] ) . in words ,
when a particle under @xmath175 is conditioned to have asymptotic mean velocity equal to any given @xmath5 , the environment markov chain chooses to switch from its original kernel @xmath39 to the tilted kernel @xmath79 given in ( [ azdinyn ] ) , where @xmath288 . the most economical tilt in terms of averaged large deviations is realized by an h - transform .
there is an alternative characterization of @xmath282 [ see ( [ muyucananan ] ) ] which involves regeneration times . that formula ( or , rather , its analog for the marginal on @xmath289 of @xmath282 ) has already appeared in definition 9 of @xcite .
if one takes ( [ muyucananan ] ) as the definition of @xmath282 , then part ( b ) of theorem [ varminvar ] becomes essentially a restatement of theorem 10 of @xcite ( see theorem [ cizme ] of the current paper for details ) . in other words ,
the novelty of theorem [ varminvar ] lies in part ( a ) .
the quenched level-3 ldp stated in theorem [ level3ldp ] implies the quenched ( level-1 ) ldp ( i.e. , theorem [ qldpgeneric ] ) via the contraction principle .
indeed , for any @xmath58 , define @xmath290 with this notation , @xmath291 our third result is as follows .
[ rosminros ] assume ( [ kondisin ] ) .
for every @xmath5 , the measure @xmath282 ( which is obtained in theorem [ varminvar ] ) is the unique minimizer of ( [ level1ratezz ] ) .
we already know from theorem [ qequalsa ] that the quenched and the averaged rate functions @xmath1 and @xmath2 are equal on @xmath292 .
the natural interpretation of theorem [ rosminros ] is that , for @xmath26-a.e .
@xmath30 , when a particle under @xmath293 is conditioned to have asymptotic mean velocity equal to any given @xmath5 , the environment markov chain chooses to switch from its original kernel @xmath39 to the tilted kernel @xmath79 .
compare this with the last paragraph of the previous subsection . since the contraction from level-3 to level-1 may be done in two steps ( instead of one )
, the following is an immediate consequence of theorem [ rosminros ] .
[ duzbac ] assume ( [ kondisin ] ) .
for every @xmath5 , let @xmath294 be the marginal of @xmath295 . with this notation , @xmath296 is the unique minimizer of the variational formula @xmath297 given in ( [ level1rate ] ) .
recall rosenbluth s variational formula @xmath298 given in ( [ level1ratetilde ] ) .
its advantage over ( [ level1rate ] ) is that @xmath124 has a simple formula , whereas @xmath299 does not .
corollary [ duzbac ] identifies the unique minimizer of ( [ level1rate ] ) when ( [ kondisin ] ) holds and @xmath5 .
we would like to obtain an analogous result for ( [ level1ratetilde ] ) .
however , as we illustrate below , there is a problem .
we express rosenbluth s formula in the following way : @xmath300 where @xmath301 denotes relative entropy and @xmath302 in light of corollary [ duzbac ] , a natural minimizer candidate for ( [ guzelyaziss ] ) is @xmath296 .
note that @xmath296 is an element of @xmath303 if and only if its marginal @xmath80 is absolutely continuous relative to @xmath26 on @xmath19 .
however , all we know is that @xmath304 on @xmath280 for every @xmath51 ; see theorem [ varminvar ] . instead of trying to show that @xmath296 is an element of @xmath303
, we will replace @xmath303 by a larger set that contains @xmath296 .
[ omurtukendi ] a measure @xmath120 is said to be in @xmath305 if it satisfies the following conditions : a. @xmath306 ; see ( [ albaytey ] ) ; b. @xmath307 for every @xmath12 ; c. there exists a @xmath191 such that @xmath308 ; see definition [ ortamkeli ] ; d. @xmath309 on @xmath310 for every @xmath51 ; see ( [ subsigsub ] ) .
[ gobenk ] assume ( [ ellipticity ] ) .
recall ( [ halitakca ] ) and definition [ omurtukendi ] . for every , @xmath311 our last result is the following theorem .
[ tezbitti ] assume ( [ kondisin ] ) .
for every @xmath5 , @xmath312 is the unique minimizer of ( [ guzelyazisss ] ) .
note that ( [ guzelyazisss ] ) does not involve any complex conjugation and , therefore , is simpler ( i.e. , more explicit ) than ( [ level1rate ] ) .
because of this , we believe that theorem [ tezbitti ] is more useful than corollary [ duzbac ] . 1 .
when ( [ kondisin ] ) holds and @xmath5 , theorem [ varminvar ] states that @xmath313 on @xmath280 for every @xmath51 . on the other hand , it is not known if @xmath313 on @xmath19 .
is the latter statement true ?
note that , when @xmath314 , this question is of great interest ( in its own right ) because @xmath315 is the invariant measure from the point of view of the particle .
bolthausen and sznitman @xcite prove that @xmath316 on @xmath19 when @xmath314 and the disorder in the environment is low .
one expects their argument to work when @xmath317 is small .
however , their technique does not generalize to the case where the disorder is not low
the limitation of our results is that they are valid when ( [ kondisin ] ) holds and @xmath5 , and their proofs break down if any of these assumptions are weakened .
therefore , it is natural to ask the following question : in the context of multidimensional rwre , does the connection between h - transform and large deviations exist under more general conditions ?
note that such a connection has been established ( i ) for walks with bounded jumps on @xmath95 in stationary and ergodic environments ( see @xcite ) , and ( ii ) for space time walks in dimensions @xmath318 and higher ( see @xcite ) .
the rest of this paper is devoted to the proofs of our results .
most of our efforts are focused on theorems [ existence ] and [ varminvar ] , which are established in sections [ kangibir ] and [ kangiiki ] , respectively .
the remaining results ( i.e. , theorems [ rosminros ] , [ gobenk ] and [ tezbitti ] ) are obtained in section [ kangiuc ] .
assume ( [ kondisin ] ) . recall ( [ sisede ] ) and ( [ sisedelal ] ) . for every @xmath320 , @xmath234 and @xmath29 , define @xmath321\\[-8pt ] &
& \hspace*{74.8pt } \mbox{for some } k\geq1 , \beta=\infty]\nonumber\end{aligned}\ ] ] and @xmath322,\ ] ] where @xmath323 [ eskil2lemma ] assume ( [ kondisin ] ) .
there exists a @xmath252 such that @xmath324 and @xmath325 for every @xmath326 ; see ( [ nebiliyim ] ) .
this constitutes the core of the proof of theorem [ cananbe ] . for the convenience of the reader
, we will give a sketch of the argument .
see lemmas 11 and 12 of @xcite for the complete proof .
it is shown in lemma 12 of @xcite that for every @xmath327 , @xmath328=1.\ ] ] [ note that theorem [ caveraged ] follows from ( [ yalebbim ] ) by the implicit function theorem . ] for every @xmath22 , let @xmath329.\ ] ] since @xmath330 by ( [ yalebbim ] ) , @xmath331 defines a random walk @xmath332 on @xmath0 .
for every @xmath320 , @xmath333 is equal to the probability of the event @xmath334 . by renewal theory , this probability
is easily shown to converge to a nonzero limit .
in particular , ( [ lemsambir ] ) follows .
for every @xmath335 , consider two independent walks @xmath336 and @xmath337 , starting at @xmath13 and @xmath338 , respectively , in the same environment .
denote their joint quenched law and joint averaged law by @xmath339 and @xmath340 , respectively . as usual ,
@xmath341 and @xmath342 refer to expectations under @xmath343 and @xmath344 , respectively . clearly , @xmath345 . on the other hand
, the two walks do not know that they are in the same environment unless their paths intersect . in particular , for any event @xmath346 involving @xmath347 and @xmath348 , @xmath349
if @xmath350 , where @xmath351 similar to the random times @xmath352 and @xmath353 for @xmath347 , define @xmath354 and @xmath355 for @xmath348 .
the proof of ( [ lemsamiki ] ) makes use of the _ joint regeneration levels _ of @xmath347 and @xmath348 , which are elements of @xmath356 note that if the starting points @xmath13 and @xmath338 are both in @xmath357 , then @xmath358 for every @xmath320 and @xmath359 , define @xmath360 with this notation , @xmath361.\ ] ] by lemma [ nonintersectlemma ] ( stated below ) , the random paths @xmath347 and @xmath348 intersect finitely many times and the probability that they intersect far away from the origin is exponentially small .
conditioned on the first joint regeneration level after the last intersection , the right - hand side of ( [ gizyuc ] ) can be written as a product of two terms .
the first term is shown to be finite , by renewal theory , when @xmath362 with a small enough @xmath363 , and the second term is bounded from above by @xmath364 since the walks can be thought of as taking place in independent environments . as mentioned in the sketch above
, the following lemma is central to the proof of ( [ lemsamiki ] ) .
[ nonintersectlemma ] assume ( [ kondisin ] ) .
recall ( [ hatirlatmak ] ) and let @xmath365 .
then @xmath366 assume ( [ kondisin ] ) .
we saw in part ( a ) of corollary [ mazaltov ] that @xmath367 has finite moments of arbitrary order .
therefore , the second inequality follows from proposition 3.1 ( for @xmath208 ) and proposition 3.4 ( for @xmath368 ) of the recent work of berger and zeitouni @xcite .
( the proofs of these propositions are based on certain green s function estimates which fail to hold unless @xmath3 . )
since the first inequality is clear , we are done .
it is easy to see that the first infimum in ( [ eyluldebasvur ] ) is positive when @xmath233 as well .
however , we will not need this fact in what follows .
[ yenil2lemma ] assume ( [ kondisin ] ) .
recall ( [ kameriye ] ) . for every @xmath369 , @xmath370 and @xmath371 recall the notation in the sketch of the proof of lemma [ eskil2lemma ] . by definition ,
@xmath372 for every @xmath320 , @xmath269 and @xmath373 .
hence , ( [ lemsambin ] ) follows immediately from ( [ lemsambir ] ) . for every @xmath320 and @xmath269 , @xmath374\nonumber\\ & = & \sum_{k=0}^{n}\sum_{z\in\mathbb{v}_d}e_{o , o } [ f(\theta , n , x,\tilde { x}),l_1=k,\tilde{x}_{\tilde{h}_k}-x_{h_k}=z , n\in\mathcal{l } ] \nonumber\\ \label{cokheyecanli } & = & \sum_{k=0}^{n}\sum_{z\in\mathbb{v}_d}e_{o , o } [ f(\theta , k , x,\tilde { x}),l_1=k,\tilde{x}_{\tilde{h}_k}-x_{h_k}=z ] \\ & & \hspace*{34.5pt}{}\times{{e}}^{-\langle\theta , z\rangle}e_{o , z } [ f(\theta , n - k , x,\tilde{x}),n - k\in\mathcal{l } |l_1=0 ] \nonumber\\ \label{olacakbence } & \leq & e_{o , o } [ f(\theta , l_1,x,\tilde{x } ) ] \bigl(\inf_{z\in \mathbb{v}_d}p_{o , z}(l_1=0 ) \bigr)^{-1}\nonumber\\[-8pt]\\[-8pt ] & & { } \times\mathop{\sup_{0\leq k\leq n}}_{z\in \mathbb{v}_d}\mathbb{e } \{g_{n - k}(\theta,\cdot)g_{n - k}(\theta , t_z\cdot ) \}\nonumber\\ \label{madlenciko } & \leq & e_{o , o } [ f(\theta , l_1,x,\tilde{x } ) ] \bigl(\inf_{z\in \mathbb{v}_d}p_{o , z}(l_1=0 ) \bigr)^{-1}\nonumber\\[-8pt]\\[-8pt ] & & { } \times\sup_{m\geq1}\mathbb{e } \ { g_m(\theta,\cdot)^2 \}.\nonumber\end{aligned}\ ] ] indeed , we have ( [ cokheyecanli ] ) by the independence structure which is still valid for common regeneration blocks .
( [ olacakbence ] ) follows by noting that @xmath375\\ & & \qquad=\bigl(p_{o , z}(l_1=0 ) \bigr)^{-1}\mathbb{e } \{g_{n - k}(\theta,\cdot ) g_{n - k}(\theta , t_z\cdot ) \}\\ & & \qquad\leq\bigl(\inf_{z\in\mathbb{v}_d}p_{o , z}(l_1=0 ) \bigr)^{-1 } \mathop{\sup_{0\leq k\leq n}}_{z\in\mathbb{v}_d}\mathbb{e } \ { g_{n - k}(\theta , \cdot)g_{n - k}(\theta , t_z\cdot ) \}.\end{aligned}\ ] ] the third term in ( [ madlenciko ] )
is obtained using the schwarz inequality and it is finite by lemma [ eskil2lemma ] .
similarly , the second term in ( [ madlenciko ] ) is finite by lemma [ nonintersectlemma ] .
therefore , to prove ( [ lemsamikibin ] ) , it suffices to show that the first term in ( [ madlenciko ] ) is also finite . by hlder s inequality , @xmath376\nonumber\\ & & \qquad=\sum_{k=0}^\infty e_{o , o } [ \exp\{\langle\theta , x_{h_k}\rangle-\lambda_a(\theta ) h_k \}\nonumber\\ & & \qquad\quad\hspace*{36.3pt}{}\times\exp\{\langle\theta,\tilde{x}_{\tilde{h}_k}\rangle-\lambda _ a(\theta)\tilde{h}_k\ } , l_1=k ] \nonumber\\ & & \qquad \leq\sum_{k=0}^\infty e_{o , o } [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda_a(\theta)h_k \ } ] ^{1/4}\nonumber\\ & & \qquad\quad\hspace*{13.7pt}{}\times e_{o , o}[\exp\{4\langle\theta,\tilde{x}_{\tilde{h}_k}\rangle-4\lambda_a(\theta ) \tilde{h}_k\ } ] ^{1/4}p_{o , o}(l_1=k)^{1/2}\nonumber\\ & & \qquad=\sum_{k=0}^\infty e_o [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda_a(\theta)h_k \ } ] ^{1/2}p_{o , o}(l_1=k)^{1/2}.\end{aligned}\ ] ] for any @xmath377 , @xmath378\nonumber\\ & & \qquad = e_o [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda _ a(\theta)h_k \ } , h_k\leq\tau_1 ] \nonumber\\ & & \qquad\quad { } + e_o [ \exp\{4\langle \theta , x_{h_k}\rangle-4\lambda_a(\theta)h_k \},\tau_1<h_k ] \nonumber\\ & & \qquad = e_o [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda _ a(\theta)h_k \ } , h_k\leq\tau_1 ] \nonumber\\ & & \qquad\quad{}+\sum_{j=1}^{k-1}e_o [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda_a(\theta)h_k \ } , \tau_1=h_j ] \nonumber\\ & & \qquad = e_o [ \exp\{4\langle\theta , x_{h_k}\rangle-4\lambda _ a(\theta)h_k \ } , h_k\leq\tau_1 ] \nonumber\\ & & \qquad\quad { } + \sum_{j=1}^{k-1}e_o [ \exp\{4\langle\theta , x_{\tau _ 1}\rangle-4\lambda_a(\theta)\tau_1 \},\tau_1=h_j ] \nonumber\\ & & \qquad\quad\hspace*{26.4pt}{}\times e_o [ \exp\{4\langle\theta , x_{h_{k - j}}\rangle-4\lambda_a(\theta ) h_{k - j } \ } |\beta=\infty]\nonumber\\ & & \qquad \leq e_o \bigl[\sup_{1\leq n\leq\tau_1}\exp\bigl\{4|\theta & & \qquad\quad { } \times\bigl(1+\sup_{1\leq i < k}e_o [ \exp\{4\langle \theta , x_{h_i}\rangle-4\lambda_a(\theta)h_i \ } |\beta=\infty ] \bigr)\nonumber\\ \label{revirvar } & & \qquad \leq e_o \bigl[\sup_{1\leq n\leq\tau_1}\exp\bigl\{4|\theta _ { 1\leq i < k}i{{e}}^{a_1|\theta|i } \bigr)\\ \label{cubar } & & \qquad \leq k_1 \bigl(1+k{{e}}^{a_1|\theta|k } \bigr).\end{aligned}\ ] ] indeed , if the walk is nonnestling , we have @xmath379 for every @xmath380 . on the other hand , if the walk is nestling , then @xmath381 . therefore , in both cases , the first term in ( [ revirvar ] ) is finite ( provided that @xmath382 ) and it is denoted by @xmath383 in ( [ cubar ] ) . the second term in ( [ revirvar ] ) is obtained using lemma 28 of @xcite , where @xmath384 is a constant . it is shown in ( the proof of ) lemma 30 of @xcite that @xmath385<\infty$ ] for some @xmath386 . for any @xmath377 , @xmath387{{e}}^{-a_3k}=:k_2{{e}}^{-a_3k}.\ ] ] putting ( [ negobe ] ) , ( [ cubar ] ) and ( [ dondolas ] ) together , we conclude that @xmath388&\leq&\sum_{k=0}^\infty k_1^{1/2}\bigl(1+k{{e}}^{a_1|\theta|k}\bigr)^{1/2}k_2^{1/2}{{e}}^{-a_3k/2}\\ & \leq&2(k_1k_2)^{1/2}\sum_{k=0}^\infty k^{1/2}{{e}}^{(a_1|\theta & < & \infty,\end{aligned}\ ] ] provided that @xmath389 .
the constant @xmath390 is chosen in @xcite such that it satisfies @xmath391 , along with a few other conditions .
thus , ( [ lemsamikibin ] ) holds for every @xmath392 . for every @xmath393 , @xmath269 and @xmath29 , @xmath394\nonumber\\ & = & \sum_{z\in u}e_o^\omega[\exp\{\langle\theta , x_{h_n}\rangle -\lambda_a(\theta)h_n \},\nonumber\\ & & \hspace*{36pt}x_1=z , h_n=\tau_k \mbox { for some } k\geq 1 ] \nonumber\\ & = & \sum_{z\in u } \pi(0,z ) \exp\{-\lambda_a(\theta)\}\nonumber\\ & & \hspace*{14pt}{}\times e_z^\omega[\exp \{\langle\theta ,
x_{h_n}\rangle-\lambda_a(\theta)h_n \ } , h_n=\tau_k \mbox { for some } k\geq1 ] \nonumber\\ & = & \sum_{z\in u}\pi(0,z)\exp\{\langle\theta , z\rangle-\lambda_a(\theta)\ } h_{n-\langle z , e_1\rangle}(\theta , t_z\omega).\end{aligned}\ ] ] here , ( [ basel ] ) is obtained by shifting the environment by @xmath147 .
define a new function @xmath395 by @xmath396 since @xmath397 is bounded in @xmath398 by ( [ lemsamikibin ] ) , it has a subsequence @xmath399 that converges weakly to some @xmath74 .
it follows immediately from ( [ basel ] ) that @xmath400 set @xmath401 and take the weak limit of both sides of ( [ acesko ] ) as @xmath402 . since the term on the second line converges ( strongly and , hence , weakly ) to zero in @xmath403
, we conclude that @xmath404 note that @xmath405 for @xmath26-a.e .
@xmath30 . equation ( [ isitisit ] ) [ in combination with ( [ ellipticity ] ) ] implies that the set @xmath406 is invariant under @xmath407 . since ( [ i.i.d . ] ) ensures that the environment is ergodic under these shifts , @xmath408 .
however , @xmath409 by ( [ lemsambin ] ) .
therefore , we conclude that . we have thus proven theorem [ existence ] .
define a function @xmath410 by setting @xmath411}{p_o^{t_x\omega}(\beta=\infty)}\ ] ] for every @xmath269 , @xmath29 and @xmath412 .
note that @xmath413 unless @xmath414 .
the following lemma will be useful in the next section .
[ kursunlem ] for every @xmath269 , there exists a @xmath415-measurable @xmath416 such that @xmath417 recall ( [ kafes ] ) and ( [ kameriye ] ) . for every @xmath393 and @xmath269 ,
define @xmath418 analogously to ( [ analojiyap ] ) . since @xmath419 is bounded in @xmath403 by ( [ lemsamiki ] ) , it has a subsequence @xmath420 that converges weakly to some @xmath416 .
[ choose @xmath421 to be a further subsequence of the subsequence in the proof of theorem [ existence ] so that @xmath422 converges weakly to @xmath264 .
] note that @xmath423 is @xmath415-measurable for every @xmath320 since the event @xmath424 is part of the definition of @xmath423 .
hence , @xmath425 is @xmath426-measurable . for every @xmath427 , @xmath428 , @xmath269 and @xmath429 , @xmath430\nonumber\\ & = & e_o^\omega[\exp\{\langle\theta , x_{h_n}\rangle-\lambda _
\},\nonumber\\ & & \hspace*{17.8pt}|x_{\tau_1}|\geq n , h_n=\tau_k \mbox { for some } k\geq1 ] \\ & & { } + \sum_{|x|<n}e_o^\omega[\exp\{\langle\theta , x_{h_n}\rangle-\lambda_a(\theta)h_n \},\nonumber\\ & & \hspace*{56.2pt}x_{\tau_1}=x , h_n=\tau_k \mbox { for some } k\geq1 ] \nonumber.\end{aligned}\ ] ] denote the first term in ( [ dayangenc ] ) by @xmath431 .
it follows immediately from ( [ yalebbim ] ) , the renewal structure and the monotone convergence theorem that @xmath432\\[-8pt ] & & \qquad\leq \lim_{n\to\infty}e_o [ \exp\{\langle\theta , x_{\tau_1}\rangle -\lambda_a(\theta)\tau_1 \},|x_{\tau_1}|\geq n ] = 0.\nonumber\end{aligned}\ ] ] recall ( [ fifi ] ) and observe that , for every @xmath428 , @xmath433\\ & = & r_{n , n}(\theta,\omega)\\ & & { } + \sum_{|x|<n}e_o^\omega[\exp\{\langle \theta , x_{\tau_1}\rangle-\lambda_a(\theta)\tau_1 \},x_{\tau _ 1}=x ] \\ & & \hspace*{35.3pt}{}\times{{e}}^{-\langle\theta , x\rangle}e_x^\omega [ \exp\{\langle\theta , x_{h_n}\rangle-\lambda_a(\theta ) h_n \},\\ & & \hspace*{97.3pt}h_n=\tau_k \mbox { for some } k\geq1 |\beta=\infty ] \\ & = & r_{n , n}(\theta,\omega)+\sum_{|x|<n}\varphi(\theta,\omega , x)g_{n-\langle x , e_1\rangle}(\theta , t_x\omega).\end{aligned}\ ] ] therefore , whenever @xmath434 , @xmath435\\[-8pt ] & & { } + \sum_{|x|<n}\varphi(\theta,\omega , x)\frac{1}{n_k}\sum_{i = n}^{n_k}g_{i-\langle x , e_1\rangle}(\theta , t_x\omega).\nonumber\end{aligned}\ ] ] multiplying both sides of ( [ sorin ] ) by any indicator function @xmath436 , integrating against @xmath26 and letting @xmath437 tend to infinity , we arrive at the following inequality : @xmath438 finally , let @xmath439 tend to infinity .
the monotone convergence theorem and ( [ busizi ] ) imply the desired result .
we will start this section by stating two results concerning the unique minimizer of varadhan s variational formula ( [ divaneasik ] ) . we will then give a series of lemmas .
finally , we will combine everything and prove theorem [ varminvar ] .
assume ( [ kondisin ] ) .
take any @xmath272 .
since the hessian @xmath250 of @xmath244 is positive definite on @xmath440 by theorem [ caveraged ] , there exists a unique @xmath441 satisfying @xmath276 . in the next paragraph
, we define a probability measure @xmath442 by specifying the integrals of certain test functions against this measure . for every @xmath443 , take any bounded function @xmath444 such that @xmath445 is independent of @xmath446 and is measurable with respect to @xmath447 define @xmath448 by setting @xmath449 & & \hspace*{33.4pt}f(t_{x_j}\omega , z_{j+1}^\infty ) \exp\{\langle\theta , x_{\tau_j}\rangle- \lambda_a(\theta)\tau_j\ } & & { } \times \bigl(e_o [ \tau_1\exp\{\langle\theta , x_{\tau_1}\rangle- \lambda_a(\theta)\tau_1\ } |\beta=\infty]\bigr)^{-1},\nonumber\end{aligned}\ ] ] where @xmath450 and @xmath451 .
( the measure @xmath282 is well defined .
see the proof of theorem [ averagedconditioningsh ] . )
the following theorem states that the empirical process @xmath452 of the walk under @xmath175 converges to @xmath282 when the particle is conditioned to have mean velocity @xmath6 .
[ averagedconditioningsh ] assume ( [ kondisin ] ) .
for every @xmath272 , @xmath453 , @xmath443 and @xmath454 bounded such that @xmath455 is independent of @xmath446 and is @xmath456-measurable , the following holds : @xmath457\\[-8pt ] & & \hspace*{138.4pt}\hspace*{40.6pt } \biggl|\frac{x_n}{n}-\xi\biggr|\leq \delta ' \biggr)<0.\nonumber\end{aligned}\ ] ] definition 9 of @xcite introduces a probability measure @xmath458 by the formula in ( [ muyucananan ] ) , except that the test functions do not depend on @xmath459 in that case .
proposition 16 of @xcite shows that @xmath460 is well defined , and theorem 17 of @xcite establishes the analog of ( [ saglik ] ) for @xmath461 .
the proofs of these results generalize to our setting without any nontrivial change .
therefore , we omit the proof of theorem [ averagedconditioningsh ] . also , note that theorem [ averagedconditioningsh ] is proved in @xcite for the related model of space time rwre . in the first paragraph of section [ unminvar ]
, we mentioned an existence and uniqueness result for the minimizer of varadhan s variational formula ( [ divaneasik ] ) for the averaged rate function @xmath2 .
the following is the precise statement .
[ cizme ] assume ( [ kondisin ] ) .
for every @xmath272 , @xmath462 induces a @xmath0-valued transient process with stationary and ergodic increments in @xmath72 via the map @xmath463 extend this process to a probability measure on doubly infinite paths @xmath286 and refer to its restriction to @xmath287 as @xmath84 . with this notation
, @xmath84 is the unique minimizer of varadhan s variational formula ( [ divaneasik ] ) .
this is theorem 10 of @xcite , with the following difference : that result is concerned with @xmath464 ( which was mentioned in the proof of theorem [ averagedconditioningsh ] ) and it uses the map @xmath465 to induce a @xmath0-valued transient process with stationary and ergodic increments in @xmath72 .
however , since @xmath464 is the marginal of @xmath282 on @xmath466 , the @xmath467-valued process induced by @xmath464 is nothing but @xmath84 .
assume ( [ kondisin ] ) .
take any @xmath5 .
let @xmath263 denote the unique solution of @xmath468 .
recall the environment kernel @xmath79 defined in ( [ azdinyn ] ) via h - transform . for any @xmath11 and @xmath29 , abbreviate the notation introduced in definition [ ortamkeli ] by writing @xmath469 and @xmath470 instead of @xmath471 and @xmath472 , respectively .
for every @xmath320 , define @xmath473 as follows : @xmath474 for every @xmath443 and @xmath475 bounded such that @xmath455 is independent of @xmath446 and is @xmath456-measurable , @xmath476}{e_o[\exp\{\langle\theta , x_{\tau_1}\rangle-\lambda_a(\theta ) \tau_1\}]}\ ] ] for every @xmath477 . for every @xmath443 ,
take a bounded function @xmath444 such that @xmath445 is independent of @xmath446 and is @xmath478-measurable ; see ( [ kuruyorum ] ) . for every @xmath477 , @xmath479\}\nonumber\\ \label{garbir } & & \qquad = \mathbb{e } \biggl\{h(\theta,\omega)e_o^\omega\biggl[f(t_{x_n}\omega , z_{n+1}^\infty)\nonumber\\[-8pt]\\[-8pt ] & & \qquad\quad\hspace*{63pt}{}\times\exp\{\langle\theta , x_{n+k}\rangle-\lambda_a(\theta ) ( n+k)\}\frac{h(\theta , t_{x_{n+k}}\omega)}{h(\theta,\omega ) } \biggr ] \biggr\}\nonumber\\ & & \qquad = e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{n+k}\rangle-\lambda_a(\theta)(n+k)\}h(\theta , t_{x_{n+k}}\omega ) ] \nonumber\\ \label{garuc } & & \qquad = e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{h_l}\rangle-\lambda_a(\theta)h_l\}h(\theta , t_{x_{h_l}}\omega)]\\ \label{gardort } & & \qquad = \sum_{\langle x , e_1\rangle\geq1}e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{h_l}\rangle-\lambda_a(\theta ) h_l\}\nonumber\\[-8pt]\\[-8pt ] & & \qquad\quad\hspace*{90.4pt}{}\times\varphi(\theta , t_{x_{h_l}}\omega , x)g(\theta , t_{x_{h_l}+x}\omega)].\nonumber \ ] ] _ explanation _ : ( [ garbir ] ) follows from the definition of @xmath81 by noting that @xmath480 depends only on the first @xmath481 steps of the walk .
( [ garuc ] ) holds because @xmath482 is a stopping time and @xmath483 .
the representation of @xmath484 in ( [ kursunlar ] ) gives ( [ gardort ] ) .
for any @xmath11 such that @xmath485 , @xmath486\nonumber\\ \label{sonbir } & & \qquad = \sum_{\langle y , e_1\rangle = l } \mathbb{e}\bigl\{e_o^\omega [ f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , y\rangle-\lambda _ a(\theta)h_l\}\nonumber\\ & & \qquad\quad\hspace*{126.4pt}{}\times\varphi(\theta , t_y\omega , x),x_{h_l}=y]\\ & & \qquad\quad\hspace*{126.4pt}\hspace*{49.3pt}{}\times g(\theta , t_{y+x}\omega)\bigr\}\nonumber\\ \label{soniki } & & \qquad = \sum_{\langle y , e_1\rangle = l } e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , y\rangle-\lambda_a(\theta)h_l\}\nonumber\\[-8pt]\\[-8pt ] & & \qquad\quad\hspace*{112.4pt}{}\times \varphi(\theta , t_y\omega , x),x_{h_l}=y]\mathbb{e}\{g(\theta,\omega)\ } \nonumber\\ \label{sonuc } & & \qquad = e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{h_l}\rangle-\lambda_a(\theta)h_l\}\varphi(\theta , t_{x_{h_l}}\omega , x)]\nonumber\\[-8pt]\\[-8pt ] & & \qquad\quad\hspace*{0pt}{}\times\mathbb{e}\{g(\theta,\omega)\}.\nonumber\end{aligned}\ ] ] _ explanation _ : for any @xmath22 such that @xmath487 , the random quantities @xmath488 $ ] and @xmath489 appearing in ( [ sonbir ] ) are independent because the former is measurable with respect to @xmath490 , whereas the latter is @xmath491-measurable ; see lemma [ kursunlem ] . this independence ( in combination with the stationarity of @xmath26 ) gives ( [ soniki ] ) . by plugging ( [ sonuc ] ) into ( [ gardort ] ) , we see that @xmath492\\ & & \qquad\quad\hspace*{21.6pt}\hspace*{151pt}{}\times\bigl({p_o^{t_x\omega}(\beta=\infty ) } \bigr)^{-1}\bigr\}\nonumber\\ \label{leybiliki } & & \qquad = \frac{1}{p_o(\beta=\infty)}\nonumber\\ & & \qquad\quad{}\times\sum_{\langle x , e_1\rangle\geq1}\sum _ { k=1}^{l+1}e_o[f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{\tau_k}\rangle-\lambda_a(\theta)\tau_k\},\\ & & \qquad\quad\hspace*{153pt}\tau_{k-1}\leq h_l<\tau _ k , x_{\tau_k}=x]\nonumber\\ \label{leybiluc } & & \qquad = \frac{1}{p_o(\beta=\infty)}e_o[f(t_{x_n}\omega , z_{n+1}^\infty ) \exp\{\langle\theta , x_{\tau_l}\rangle-\lambda_a(\theta)\tau_l\}].\end{aligned}\ ] ] _ explanation _ : ( [ leybilbir ] ) follows from the definition of @xmath493 given in ( [ fifi ] ) . for every @xmath494 such that @xmath485 , the random quantity @xmath495 is independent of the ratio @xmath496}{p_o^{t_x\omega}(\beta=\infty)}$ ] appearing in ( [ leybilbir ] ) since the latter is easily seen to be equal to an expectation involving the stopping time @xmath497 ( and nothing beyond that )
. this independence implies ( [ leybiliki ] ) . using ( [ yalebbim ] )
, the @xmath498 in the exponential can be replaced first by @xmath499 and then by @xmath500 .
this gives ( [ leybiluc ] ) .
finally , observe that ( [ leybiluc ] ) agrees with ( [ komurhan ] ) , except that the normalization constant has to be simplified .
however , it is clear that the constant in ( [ komurhan ] ) is correct [ take @xmath501 and apply ( [ yalebbim ] ) ] .
[ zafiyet ] for every @xmath454 bounded such that @xmath455 is @xmath19-measurable , the following convergence takes place : @xmath502 in particular , @xmath503 converges weakly to @xmath282 . for any @xmath443 ,
take a bounded function @xmath504 such that @xmath455 is independent of @xmath446 and @xmath456-measurable .
let @xmath450 .
@xmath505\lim_{n\to\infty}\int f\,{{d}}\hat{\mu}_{n,\xi}^\infty\\ \label{ozgurolcam } & & \qquad = \lim_{n\to\infty}e_o[n<\tau_n , f(t_{x_n}\omega , z_{n+1}^\infty ) \exp\{\langle\theta , x_{\tau_j}\rangle-\lambda_a(\theta)\tau_j\}]\\ & & \qquad\quad { } + \lim_{n\to\infty}\sum_{i=0}^\infty e_o[\tau_{n+i}\leq n<\tau_{n+i+1},\nonumber\\ & & \qquad\hspace*{80.2pt}f(t_{x_n}\omega , z_{n+1}^\infty)\exp\{\langle\theta , x_{\tau_{j+i}}\rangle-\lambda_a(\theta)\tau_{j+i}\}]\nonumber\\ \label{danso } & & \qquad = \lim_{n\to\infty}\sum_{j=0}^n \biggl(\sum_{i=0}^\infty e_o[\tau _ i =
n - j,\exp\{\langle\theta , x_{\tau_i}\rangle-\lambda_a(\theta)\tau_i\ } ] \biggr)\\ & & \qquad\quad\hspace*{37.3pt}{}\times e_o[\tau_n\leq j<\tau_{n+1},\nonumber\\ & & \qquad\hspace*{78.1pt}f(t_{x_j}\omega , z_{j+1}^\infty)\exp\{\langle\theta , x_{\tau_j}\rangle-\lambda_a(\theta ) \tau_j\ } |\beta=\infty]\nonumber\\ & & \qquad = s(\theta)\sum_{j=0}^\infty e_o[\tau_n\leq j<\tau _ { n+1},\nonumber\\ & & \qquad\hspace*{65.1pt}f(t_{x_j}\omega , z_{j+1}^\infty)\exp\{\langle\theta , x_{\tau _ j}\rangle-\lambda_a(\theta)\tau_j\ } |\beta=\infty]\nonumber\\ \label{atesgg } & & \qquad = s(\theta)e_o [ \tau_1\exp\{\langle\theta , x_{\tau _ 1}\rangle- \lambda_a(\theta)\tau_1\ } |\beta=\infty]\int f\,{{d}}\hat{\mu}_\xi^\infty.\end{aligned}\ ] ] _ explanation _ : ( [ ozgurolcam ] ) follows from ( [ komurhan ] ) .
the first term in ( [ ozgurolcam ] ) goes to zero as @xmath506 by the dominated convergence theorem .
the renewal theorem for aperiodic sequences ( see @xcite , theorem 10.8 ) implies that the sum @xmath507)$ ] in ( [ danso ] ) converges to some constant @xmath508 as @xmath506 .
observe that the constants in ( [ atasgg ] ) and ( [ atesgg ] ) have to agree because @xmath509 and @xmath282 are known to be probability measures .
thus far , we have shown that @xmath510 for a separating class of test functions . however
, this is sufficient to conclude that @xmath503 converges weakly to @xmath282 since @xmath127 is compact .
let @xmath511 be the marginal of @xmath512 .
[ kiliyir ] @xmath80 is @xmath79-invariant , that is , @xmath513 for any @xmath514 , define @xmath515 in the usual way : @xmath516 recall ( [ gurbetlik ] ) . for every @xmath320 , @xmath517\}}{\mathbb{e}\{h(\theta,\omega)\}}\\ & = & \frac{\mathbb{e}\ { h(\theta,\omega)e_o^{\theta,\omega}[f(t_{x_{n+1}}\omega)]\}}{\mathbb { e}\{h(\theta,\omega)\}}\\ & = & \int f\,{{d}}\hat{\mu}_{n+1,\xi}^\infty,\end{aligned}\ ] ] by the markov property
let @xmath518 tend to infinity , use lemma [ zafiyet ] and conclude that @xmath519 this is equivalent to the desired result .
[ markovkopek ] @xmath282 induces , via the map @xmath520 an @xmath24-valued stationary markov process with marginal @xmath283 and transition kernel @xmath521 for any @xmath320 , @xmath522 and any two bounded measurable functions @xmath523 and @xmath524 , it follows from ( [ gurbetlik ] ) and the markov property that @xmath525\}\\ & & \qquad = \mathbb{e } \bigl\{h(\theta,\omega)e_o^{\theta,\omega } \bigl[f(t_{x_n}\omega,\ldots , t_{x_{n+k}}\omega)\\ & & \qquad\quad\hspace*{67.5pt}{}\times e_o^{\theta,\omega } [ g(t_{x_{n+k}}\omega , t_{x_{n+k+1}}\omega,\ldots)|x_{n+k } ] \bigr ] \bigr\}\\ & & \qquad = \mathbb{e}\{h(\theta,\omega)\}\\ & & \qquad\quad{}\times\int f(\omega , t_{z_1}\omega,\ldots , t_{z_1+\cdots+z_k}\omega)\\ & & \qquad\quad\hspace*{20.9pt}{}\times e_o^{\theta , t_{z_1+\cdots+z_k}\omega } [ g(t_{z_1+\cdots+z_k}\omega , t_{z_1+\cdots+z_k+x_1}\omega,\ldots ) ] \,{{d}}\hat{\mu}_{n,\xi}^\infty.\end{aligned}\ ] ] let @xmath518 tend to infinity , use lemma [ zafiyet ] and conclude that @xmath282 indeed induces an @xmath24-valued markov process with marginal @xmath80 and transition kernel @xmath526 .
finally , note that the stationarity of this process follows from a straightforward generalization of lemma [ kiliyir ] .
[ yariuzay ] @xmath85 on @xmath527 for every @xmath528 ; see ( [ subsigsub ] ) . for any @xmath529 , take an @xmath514 such that @xmath530 is nonnegative and @xmath456-measurable for some @xmath531 .
let @xmath532 .
it follows from ( [ muyucananan ] ) and the schwarz inequality that @xmath533\int f\,{{d}}\mathbb{q}_\xi\nonumber\\ & & \qquad = e_o \biggl [ \biggl(\sum_{j=\tau_n}^{\tau_{n+1}-1}f(t_{x_j}\omega ) \biggr)\exp\{\langle\theta , x_{\tau_j}\rangle- \lambda_a(\theta)\tau _ j\},\beta=\infty\biggr]\nonumber\\ & & \qquad \leq\sum_{k=1}^\infty e_o \biggl[\tau_{n+1}=k , \biggl(\sum_{|x|\leq k}f(t_x\omega ) \biggr)\exp\{\langle\theta , x_{\tau_j}\rangle- \lambda _ a(\theta)\tau_j\},\beta=\infty\biggr]\nonumber\\ & & \qquad = \sum_{k=1}^\infty\sum_{l=1}^k\mathbb{e } \biggl\ { \biggl(\sum _ { |x|\leq k}f(t_x\omega ) \biggr)\nonumber\\ & & \qquad\quad\hspace*{43.8pt}{}\times e_o^\omega[\tau_{n+1}=k = h_l,\exp\ { \langle\theta , x_{\tau_j}\rangle- \lambda_a(\theta)\tau_j\},\beta=\infty ] \biggr\}\nonumber\\ & & \qquad \leq\sum_{k=1}^\infty\sum_{l=1}^k(2k+1)^d\|f\|_{l^2(\mathbb { p})}\\ & & \qquad\quad\hspace*{27.8pt}{}\times\mathbb{e } \bigl\{e_o^\omega[\tau_{n+1}=k = h_l,\nonumber\\ & & \qquad\quad\hspace*{53.4pt}\hspace*{18.5pt}\exp\{\langle \theta , x_{\tau_j}\rangle- \lambda_a(\theta)\tau_j\},\beta=\infty ] ^2 \bigr\}^{1/2}.\nonumber\end{aligned}\ ] ] there exist constants @xmath534 and @xmath535 such that , for every @xmath377 and @xmath536 , @xmath537 ^ 2 \bigr\}\nonumber\\ \label{seri } & & \qquad \leq\mathbb{e } \bigl\{e_o^\omega[\tau_{n+1}=k = h_l,\exp\ { \langle\theta , x_{\tau_{n+1}}\rangle- \lambda_a(\theta)\tau_{n+1}\ } , \beta=\infty]^2 \bigr\}\\ & & \qquad\quad { } \times\bigl(\inf_{z\in\mathbb{v}_d}p_{o , z}(l_1=0 ) \bigr)^{-1}\nonumber\\ & & \qquad\quad\hspace*{0pt}{}\times\mathbb{e } \bigl\{e_o^\omega[\exp\{\langle\theta , x_{\tau
_ m}\rangle- \lambda_a(\theta)\tau_m\},\beta=\infty]^2 \bigr\ } \nonumber\\ \label{denbil } & & \qquad \leq c_n'{{e}}^{-a_4k}.\end{aligned}\ ] ] _ explanation _ : the first term in ( [ seri ] ) is bounded from above by @xmath538 for some @xmath539 and @xmath535 .
the second term in ( [ seri ] ) is finite by lemma [ nonintersectlemma ] . using the technique in the proof of lemma [ eskil2lemma ] , the third term in ( [ seri ] )
can be shown to be bounded from above by a constant that is independent of @xmath540 .
we leave the details to the reader . plugging ( [ denbil ] ) into ( [ seler ] )
, we see that @xmath541 for some finite constant @xmath542 that is independent of @xmath540 .
since the functions we have considered are dense in @xmath543 , it follows from the riesz representation theorem that @xmath544 combining all of the results in this section , we get the following proof . proof of theorem [ varminvar ] recall that @xmath511 denotes the marginal of @xmath295 which , in turn , is defined in ( [ muyucananan ] ) .
we have seen that : @xmath80 is @xmath79-invariant ( see lemma [ kiliyir ] ) ; @xmath85 on @xmath280 for every @xmath51 ; see lemma [ yariuzay ] .
it follows from lemma [ kozlov ] ( stated below ) that @xmath80 is the unique element of @xmath545 that satisfies these two properties .
we have also proven that @xmath282 induces an @xmath24-valued stationary markov process with marginal @xmath546 and transition kernel @xmath547 ; see lemma [ markovkopek ] .
these results imply part ( a ) of theorem [ varminvar ] .
note that part ( b ) of theorem [ varminvar ] is a special case of theorem [ cizme ] since @xmath548 . in the proof above
, we used the following generalization of a classical homogenization result which is originally due to kozlov @xcite .
[ kozlov ] given any @xmath549 and any environment kernel @xmath550 , define a measure @xmath120 by setting @xmath551 for each @xmath12 .
recall definition [ omurtukendi ] .
if @xmath552 , then the following hold : a. the measures @xmath26 and @xmath64 are , in fact , mutually absolutely continuous on @xmath310 for every @xmath51 ; b. the environment markov chain with kernel @xmath46 and initial distribution @xmath64 is stationary and ergodic ; c. @xmath64 is the unique @xmath46-invariant probability measure on @xmath24 that satisfies @xmath553 on @xmath310 for every @xmath51 ; d. the following lln is satisfied : @xmath554 .
this is theorem 3 of @xcite , except that @xcite is concerned with space
time rwre .
however , the result is obtained directly from theorem [ averagedconditioningsh ] by a standard application of the borel cantelli lemma and chebyshev s inequality . in other words ,
the proof in @xcite makes no use of the space time assumption .
[ the only notational difference is that , in the space time case , @xmath558 is equal to @xmath559 for some explicit function @xmath560 , but this does not play any role in the proof . ] proof of theorem [ rosminros ] take any @xmath5 . recall ( [ axizz ] ) .
if an @xmath561 is not equal to @xmath282 , then @xmath562 for some @xmath453 , @xmath443 and @xmath504 bounded such that @xmath455 is independent of @xmath446 and @xmath456-measurable .
for every @xmath563 and @xmath26-a.e .
@xmath30 , ( the lower bound of ) the quenched level-3 ldp ( i.e. , theorem [ level3ldp ] ) implies that @xmath564 on the other hand , @xmath565 by the quenched level-1 ldp ( i.e. , theorem [ qldpgeneric ] ) .
therefore , @xmath566 by ( [ saglikcan ] ) . in words
, @xmath567 is not a minimizer of ( [ level1ratezz ] ) .
however , since @xmath132 is lower semicontinuous and @xmath568 is compact , there is a minimizer .
we conclude that @xmath282 is the unique minimizer of ( [ level1ratezz ] ) .
fix a sequence of test functions , denoted by @xmath571 , that separate @xmath106 . for
every @xmath572 and @xmath12 , assume that @xmath573 is measurable with respect to @xmath574 .
take any @xmath569 .
for every @xmath427 , @xmath575 is an open set .
recall @xmath191 and the environment kernel @xmath46 corresponding to @xmath114 ; see definition [ omurtukendi ] .
let @xmath576 so that @xmath577 for each @xmath578 .
for every @xmath320 , introduce a new measure @xmath579 by setting @xmath580 where @xmath581 . with this notation , for @xmath64-a.e .
@xmath30 , @xmath582\nonumber\\ & & \qquad = \log p_o^{\hat{\pi},\omega}(\nu_{n , x}\in g_{\mu , n},\beta > n)+\log\int\frac{{{d}}p_o^\omega}{{{d}}p_o^{\hat{\pi},\omega } } \,{{d}}r_{o , n}^{\hat{\pi},\omega}\nonumber\\ & & \qquad \geq\log p_o^{\hat{\pi},\omega}(\nu_{n , x}\in g_{\mu , n},\beta > n)- \int\log\frac{{{d}}p_o^{\hat{\pi},\omega}}{{{d}}p_o^\omega } \,{{d}}r_{o , n}^{\hat{\pi},\omega}\nonumber\\ & & \qquad = \log p_o^{\hat{\pi},\omega}(\nu_{n , x}\in g_{\mu , n},\beta > n)\\ & & \qquad\quad { } -\frac{1}{p_o^{\hat{\pi},\omega}(\nu_{n , x}\in g_{\mu , n},\beta >
n ) } \nonumber\\ & & \qquad\quad\hspace*{10.8pt}{}\times e_o^{\hat{\pi},\omega } \biggl[\nu_{n , x}\in g_{\mu , n},\beta > n , \log\frac{{{d}}p_o^{\hat{\pi},\omega}}{{{d}}p_o^\omega } \biggr],\nonumber\end{aligned}\ ] ] by a change of measure and jensen s inequality .
it follows from lemma [ kozlov ] and the ergodic theorem that @xmath583 and @xmath584 hence , for @xmath64-a.e .
@xmath30 , @xmath585 and @xmath586\nonumber\\[-8pt]\\[-8pt ] & & \qquad\leq p_o^{\hat{\pi},\omega}(\beta=\infty ) h(\mu).\nonumber\end{aligned}\ ] ] here , ( [ kabusbiter ] ) follows from fatou s lemma since @xmath587 by uniform ellipticity ; see ( [ ellipticity ] ) .
it follows from parts ( b ) and ( c ) of definition [ omurtukendi ] that @xmath588 for @xmath26-a.e .
@xmath459 . since @xmath589 is @xmath590-measurable , part ( d ) of definition [ omurtukendi ] implies that @xmath591 combining ( [ evrenak ] ) , ( [ kabusbit ] ) , ( [ kabusbiter ] ) and ( [ kaptis ] ) , we see that @xmath592 for @xmath64-a.e .
@xmath30 . however , since @xmath593 is @xmath594-measurable for every @xmath320 , lemma [ kozlov ] implies that ( [ binbok ] ) holds for @xmath26-a.e .
@xmath30 as well .
therefore , @xmath595 for every @xmath427 and @xmath26-a.e .
@xmath30 , @xmath596 by the quenched level-2 ldp , that is , theorem [ level2ldp ] .
hence , @xmath597 sending @xmath439 to infinity implies ( [ binsolc ] ) since @xmath598 is lower semicontinuous and @xmath571 separates @xmath106 .
proof of theorem [ gobenk ] fix @xmath66 . for any @xmath599 ,
let @xmath46 be the environment kernel given by @xmath600 for each @xmath12 .
it is shown in @xcite that @xmath601 hence , @xmath569 .
[ for part ( c ) of definition [ omurtukendi ] , take any @xmath602 such that @xmath603 . ] in other words , @xmath604 it follows from ( [ level1rate ] ) , ( [ binsolc ] ) , ( [ yavuzz ] ) and ( [ guzelyaziss ] ) that @xmath605 in particular , @xmath606 . proof of theorem [ tezbitti ] fix @xmath5 .
then , @xmath66 .
indeed , by differentiating both sides of ( [ yalebbim ] ) with respect to @xmath237 at @xmath277 , we see that @xmath607}{e_o[\tau_1\exp\{\langle\theta , x_{\tau_1}\rangle-\lambda_a(\theta)\tau_1\}|\beta=\infty]}>0.\ ] ] recall that @xmath608 is the marginal of @xmath295 and @xmath609 for each @xmath12 .
it is clear from theorem [ varminvar ] that @xmath610 .
observe that @xmath611 _ explanation _ : ( [ cinkobir ] ) , ( [ cinkoiki ] ) and ( [ cinkouc ] ) follow from ( [ halitakca ] ) , ( [ azdinyn ] ) and ( [ ximu ] ) , respectively . since @xmath80 is @xmath79-invariant by lemma [ kiliyir ] , it is easy to see that the integral in ( [ cinkouc ] ) is zero .
finally , ( [ cinkodort ] ) is equal to ( [ cinkobes ] ) because @xmath276 and @xmath612 . thus far
, we have shown that @xmath613 .
now , take any @xmath614 .
if @xmath615 , then @xmath616 by corollary [ duzbac ] and ( [ binsolc ] ) .
we conclude that @xmath296 is the unique minimizer of ( [ guzelyazisss ] ) . | we consider large deviations for nearest - neighbor random walk in a uniformly elliptic i.i.d .
environment on @xmath0 .
there exist variational formulae for the quenched and averaged rate functions @xmath1 and @xmath2 , obtained by rosenbluth and varadhan , respectively . @xmath1 and @xmath2 are not identically equal .
however , when @xmath3 and the walk satisfies the so - called ( * t * ) condition of sznitman , they have been previously shown to be equal on an open set @xmath4 . for every @xmath5
, we prove the existence of a positive solution to a laplace - like equation involving @xmath6 and the original transition kernel of the walk .
we then use this solution to define a new transition kernel via the h - transform technique of doob .
this new kernel corresponds to the unique minimizer of varadhan s variational formula at @xmath6 .
it also corresponds to the unique minimizer of rosenbluth s variational formula , provided that the latter is slightly modified . . . |
the narrow - line region ( nlr ) is one of fundamental ingredients of active galactic nuclei ( agns ) , and many efforts have been made up to now to understand its physical , chemical , geometrical , and kinematic properties ( see , e.g. , osterbrock & mathews 1986 ) .
high spatial - resolution imaging observations have shown that nlrs consist of inhomogeneous and discrete gas clouds which are sometimes confined in a biconical structure ( e.g. , tadhunter & tsvetanov 1989 ; evans et al . 1993 ; macchetto et al .
1994 ; boksenberg et al . 1995 ; capetti , axon , & macchetto 1997 ; ferruit et al .
some spectroscopic observations , on the other hand , have suggested that there is a density and/or ionization stratification in nlrs , which has been clarified mainly by correlations between emission - line widths and their ionization potential and/or critical density ( e.g. , filippenko & halpern 1984 ; de robertis & osterbrock 1984 , 1986 ; filippenko 1985 ; veilleux 1991a ; alloin et al . 1992
; see also barth et al .
1999 ; tran , cohen , & villar - martin 2000 ) .
thus it is now recognized that there are many discrete gas clouds with various properties , which are stratified in nlrs ( see , e.g. , ferguson et al . 1997a ; ferguson , korista , & ferland 1997b ) .
this naturally leads to the following question ; how does the density and/or ionization stratification exist in nlrs ? in order to investigate this issue ,
high spatial - resolution spectroscopic observations have been performed recently ( e.g. , kraemer et al . 2000 ; kraemer & crenshaw 2000 ) . however , there are two drawbacks in this approach .
first , this method is applicable only to very nearby agns .
each gas cloud in nlrs of rather distant agns is hard to be resolved spatially even by using hst . and
second , even as for very nearby agns , the innermost region of nlrs ( i.e. , @xmath2 1 pc ) can not be resolved spatially by the current ultraviolet / optical / infrared facilities .
since highly ionized gas clouds may be located in such an innermost region ( e.g. , murayama & taniguchi 1998a , 1998b ; nagao , taniguchi , & murayama 2000 ; nagao , murayama , & taniguchi 2001b , 2001c ; see also barth et al .
1999 ; tran et al .
2000 ) , this drawback seems to be a very serious problem for studies of the structure of ionization and physical properties of nlrs .
one of complementary approaches for the high spatial - resolution observations is high spectral - resolution observations . by using high - dispersion spectra
, we can investigate emission - line flux ratios as a function of a recession velocity , which contains information about the kinematic configuration of ionized gas clouds with various physical properties .
although high - dispersion spectra of only some very strong emission lines such as [ o iii]@xmath35007 were taken to investigate kinematics of nlr clouds ( e.g. , glaspey et al .
1976 ; pelat & alloin 1980 ; pelat , alloin , & fosbury 1981 ; pelat & alloin 1982 ; alloin et al . 1983
; vrtilek & carleton 1985 ) , it is difficult to explore the whole physical properties of nlr clouds solely from the [ o iii]@xmath35007 line . in order to explore the nature of nlrs from the kinematical viewpoint ,
medium - resolution spectra of multi emission lines have been also investigated ( e.g. , de robertis & osterbrock 1984 , 1986 ) .
however , veilleux ( 1991c ) reported that such medium - resolution spectroscopy tends to overestimate the emission - line widths while to underestimate the line asymmetries , owing to the instrumental broadening effect .
multi emission - line spectra with both high spectral resolution and high s / n are thus crucially necessary to understand the nlr clouds .
appenzeller & streicher ( 1988 ) examined profiles of forbidden emission lines including some high - ionization emission lines by using high - resolution spectra ( @xmath4 km s@xmath5 ) .
however , the signal - to - noise ratios of their data are not high enough to study the velocity profiles of weak lines in detail .
veilleux ( 1991b ) presented multi emission - line spectra with both high spectral resolution ( @xmath6 10 km s@xmath5 ) and high s / n for 16 bright seyfert galaxies ( see also veilleux 1991a , 1991c ) . however , the main concern of this work was to explore the statistical properties of nlr kinematics in general and thus the profiles of weak but important emission lines should be investigated further .
thus we have started a program in which optical spectra with a very high spectral resolution ( @xmath7 km s@xmath5 ) and a very high signal - to - noise ratio for some nearby seyfert galaxies are collected by using high - dispersion spectrograph ( hds ; noguchi et al .
2002 ) boarded on the 8.2 m subaru telescope ( kaifu 1998 ) . in this paper , we report the first result of this program on the nearby seyfert 1.5 galaxy ngc 41511 corresponds to a linear scale of 6.4 pc in the plane of the sky . ]
( see ulrich 2000 for a summary of the past observations for ngc 4151 ; see also kaiser et al .
2000 , nelson et al . 2000 and kraemer et al .
2000 for the recent results of a high spatial resolution spectroscopy of this object ) .
lccc [ o iii ] & 4363.21 & @xmath8 & 35.1 & 4711.34 & @xmath9 & 40.7 & 4740.20 & @xmath10 & 40.7 & 4958.92 & @xmath11 & 35.1 & 5006.85 & @xmath11 & 35.1 & 5630.82 & & 75.0 & 5676.96 & & 75.0 & 5721.11 & @xmath12 & 99.1 & 5754.57 & @xmath13 & 14.5 & 6086.92 & @xmath12 & 99.1 & 6300.32 & @xmath14 & 0.0 we carried out spectroscopic observation of the seyfert galaxy ngc 4151 by using hds boarded on the subaru telescope , on the 4th february 2002 ( ut ) . the detector of hds is a mosaic of two 4k @xmath15 2k eev ccd s with 13.5 @xmath16 m pixels .
we used the grating for the red spectra , whose central wavelength is set to 5600 @xmath17 . by this
setting , one ccd covers a wavelength range of @xmath18 ( from the 108th order to the 141st order ) and the other ccd covers a range of @xmath19 ( from the 86th order to the 106th order ) .
a short - wavelength - cutoff filter of kv389 was used to obtain the whole wavelength range written above .
we adopted 2 pixel binning for both wavelength and spatial directions on the chips . as a result of this binning ,
the spatial sampling rate was 0.24 arcsec / binnedpixel . by adopting a 0.80 arcsec ( 0.40 mm ) slit width , we got spectra with a wavelength resolution of @xmath2045,000 .
the atmospheric dispersion compensator ( adc ) was used in order to avoid photon losses caused by the atmospheric dispersion effect .
the image rotator was also used by adopting the red setup . note that the adc and the image rotator do not degrade the spatial resolution .
see noguchi et al .
( 2002 ) for the wavelength dependences of the throughput of the adc and the image rotator .
we carried out four 900 sec exposures for ngc 4151 with the position angle of 77@xmath21 the typical seeing size was @xmath22 during the observation .
we also obtained the spectrum of a spectroscopic standard star , bd+26@xmath21 2606 ( oke & gunn 1983 ) , for flux calibration .
the spectra of a halogen lamp and a thorium - argon lamp were also obtained for the flat field and the wavelength calibration , respectively .
the data were reduced with the iraf package in the standard manner .
since the expected dark current of the ccd s is @xmath201 adu / pixel for an hour ( see noguchi et al .
2002 ) , we neglected the contribution of the dark current in this study .
although the data frames obtained by hds include over - scan regions to be used for the bias subtraction , we did not use the over - scan data for the bias subtraction because the time variation of the bias level estimated by using the over - scan data was negligibly small . instead of using the over - scan data
, we simply subtracted the bias frame taken as the calibration data from the object frames .
the cosmic - ray noise was removed by running the iraf task `` lineclean '' for the object frames divided by the median - filtered object frames .
the flat fielding was performed by dividing the object frames by the halogen lamp frame directly . by this process
, the blaze profiles of the object frames were corrected very roughly ( see section 2.4 and appendix a ) .
the background subtraction was done by the `` apscatter '' routine of iraf .
this background component is thought to be attributed to reflections of the object light at the surface of the ccd s and the field flattener lens just before the ccd s ( see noguchi et al .
2002 for details ) .
the wavelength calibration was carried out by using the thorium - argon spectra obtained before and after the observations of the objects .
the measured rms wavelength error of the thorium - argon lines is less than 0.005 @xmath17 , which corresponds to 0.2 pixels at the bluest region in the spectra . by measuring the fwhm of thorium - argon lines in the wavelength - calibrated thorium - argon spectra , we confirmed the spectral resolution to be @xmath23 .
we extracted the nuclear spectra of the objects adopting the aperture size of 1.44 arcsec ( i.e. , 6 binned pixels ) , by using the iraf task `` apall '' . here
we should mention that the resultant aperture for the extraction of the nuclear spectrum is @xmath24 , which is rather smaller than the spatial extension of the ionized gas clouds around the nucleus of ngc 4151 ( e.g. , heckman & balick 1983 ; unger et al .
1987 ; hutchings et al . 1998 , 1999 ; winge et al . 1999 ; see also kaiser et al .
2000 ; nelson et al .
2000 ; kraemer et al .
although a large part of the nlr emission of ngc 4151 is covered by this adopted aperture since the position angle of the slit is along to the direction of the nlr extension ( 77 deg ) , some ionized gas clouds are missed in our analysis .
we will see how the possible aperture effects can affect our discussion when necessary ( in section 4 ) .
we performed the flux calibration by using the spectrum of a spectroscopic standard star , bd+26@xmath21 2606 , in order to correct the wavelength dependence of the sensitivity .
note that the absolute flux scale is not so accurate since the seeing size was so large that the slit loss of the photons is not negligible .
this ambiguity of the flux scale does not affect the following analysis and discussion , because we do not refer any absolute fluxes of the emission lines and the continuum emission in this study . in order to estimate the sensitivity function for each echelle order ,
the data of the absolute spectral energy distribution ( sed ) of bd+26@xmath21 2606 , sampled by a few @xmath17 , are necessary .
we made the sed data by performing a 3rd polynomial fitting for the data presented by oke & gunn ( 1983 ) .
since bd+26@xmath21 2606 is a typical a - type star ( @xmath25 ) , we can estimate the absolute sed by this fitting , except for the wavelength regions where the absorption feature is significant ( e.g. , the wavelength regions around h@xmath26 and h@xmath27 ) .
we then determined the sensitivity function for each echelle order by comparing the observed spectra ( in adu ) with the estimated absolute sed .
as for the wavelength regions where the absolute sed can not be estimated due to absorption features , we estimated the sensitivity functions by interpolating those of the neighboring echelle orders .
the flux calibration of the spectra of ngc 4151 was then performed by using the estimated sensitivity functions .
note that the terrestrial absorption features are not corrected in this flux calibration manner .
such absorption features do not affect the analysis and the discussion . in general
, spectra taken by echelle spectrographs have a blaze profile similar to a sinc function , which should be corrected in order to investigate some spectral features .
the blaze profile can be in principle corrected by dividing the object frames by the flat frame , if the blaze profile would not change at all .
however , the blaze profile of hds is currently unstable with no systematic trend , i.e. , the blaze profiles are different among the object frames , the standard - star frames , and the flat frames ( see aoki 2002 ) . due to this unstability of the blaze profile ,
significant artificial periodic features appear on the flux - calibrated order - combined spectrum .
since the amplitude of such features reaches up to @xmath28 % of the spectrum itself , it should be corrected in order to analyze the profiles of emission lines appropriately .
therefore we have carried out the blaze profile correction by adopting a method described at appendix a in detail . as a result of the correction ,
the amplitude of the remaining artificial features is @xmath29 % of the continuum level .
lccccccc [ o iii]@xmath34363 & 115 & 240 & 690 & 49 & 54 & 62 & + 0.04 @xmath34712 & 87 & 223 & 430 & 24 & 9 & @xmath300&0.11 @xmath34740 & 64 & 177 & 365 & 27 & 26 & 8 & 0.11 @xmath34959 & 94 & 211 & 504 & 37 & 40 & 74 & + 0.15 @xmath35007 & 94 & 210 & 513 & 36 & 40 & 80 & + 0.17 @xmath35631 & 72 & 191 & 354 & 15 & 23 & 28 & + 0.08 @xmath35677 & 82 & 278 & 574 & 32 & 60 & 106 & + 0.26 @xmath35721 & 128 & 250 & 522 & 38 & 41 & 61 & + 0.09 @xmath35755 & 109 & 258 & & 40 & 42 & & @xmath36087 & 132 & 254 & 497 & 45 & 55 & 75 & + 0.12 @xmath36300 & 106 & 206 & 565 & 36 & 29 & 34 & 0.01 we present the flux - calibrated , blaze - profile - corrected spectrum of ngc 4151 in figure 1 . note that the sky emission is not subtracted and thus the terrestrial [ o i]@xmath16300,6363 airglow emission lines are also seen in this spectrum .
these sky emission lines can be used to check the accuracy of the wavelength calibration .
the measured wavelengths of these two emission lines are 6300.30 @xmath17 and 6363.78 @xmath17 , which are consistent with the theoretical wavelengths within @xmath31 0.01 @xmath17 .
we show the profiles of the continuum - subtracted forbidden emission lines , i.e. , [ o iii]@xmath34363 , [ ar iv]@xmath34712 , [ ar iv]@xmath34740 , [ o iii]@xmath34959 , [ o iii]@xmath35007 , [ fe vi]@xmath35631 , [ fe vi]@xmath35677 , [ fe vii]@xmath35721 , [ n ii]@xmath35755 , [ fe vii]@xmath36087 , and [ o i]@xmath36300 , in figures 2a 2k . the presented profiles are not corrected for the instrumental broadening since the effects of the instrumental broadening are evidently negligible .
indeed the [ o i]@xmath36300 airglow emission seen in figure 2k is much significantly narrower than the velocity features of the observed forbidden emission lines of ngc 4151 .
the origin of the horizontal axis coincides with the recession velocity of the peak of the [ o iii]@xmath35007 emission , which is located at 5022.82 @xmath17 in the obtained spectrum .
see table 1 for the rest wavelengths we adopted for the forbidden lines , which are taken from bowen ( 1960 ) .
the critical density of the transition and the ionization potential of the corresponding ion for each emission line are also given in this table .
the spectral profile in the relative recession - velocity range of 1200 km s@xmath32 km s@xmath5 is shown for each emission line . the continuum level for each emission line
is estimated by interpolating the neighbor wavelength regions of the emission line linearly , except for [ o iii]@xmath34363 .
since this emission line is located on the broad wing of the h@xmath33 emission , we extract the profile of [ o iii]@xmath34363 by interpolating the neighbor wavelength regions adopting the cubic spline fitting . in this figure ,
the flux level of each emission line is normalized by the flux density at the peak of the line , which is measured after performing a 5 pixels smoothing for the spectra .
note that we do not make attempts to extract the velocity profiles of blending forbidden lines , e.g. , [ o i]@xmath36363 , [ fe x]@xmath36374 , [ n ii]@xmath16548,6583 , and [ s ii]@xmath16717,6731 .
this is because we can not deblend the blended lines without some assumptions whose validity is hard to be examined .
because of the same reason , we do not investigate velocity profiles of the narrow component of permitted lines .
here we should mention the contributions of other weak emission lines to the spectra presented in figure 2 .
the emission line seen both at the red side of [ ar iv]@xmath34712 and at the blue side of [ ar iv]@xmath34740 is the blend of the [ ne iv]@xmath14724,4726 doublet ( figures 2b and 2c ) .
this feature may contribute to the red - side tail of the [ ar iv]@xmath34712 emission .
the emission - line feature seen at the blue side of [ fe vi]@xmath35631 is identified as [ ca vii]@xmath35616 . since the wavelengths of the [ fe vii]@xmath35721 and the [ n ii]@xmath35755 emission are close in each other on the spectrum , the red - side tail of the [ fe vii]@xmath35721 emission and the blue - side tail of the [ n ii]@xmath35755 are affected by their neighbor lines (
particularly , the latter one is ) . to show how the [ fe vii]@xmath35721 emission affects the [ n ii]@xmath35755 profile
, we give a larger portion of the spectral region around the [ n ii]@xmath35755 emission in figure 3 .
the weak emission feature seen at the @xmath34 km s@xmath5 of the [ o i]@xmath36300 emission is [ s iii]@xmath36312 .
we do not investigate these faint emission lines since the s / n of the data is not enough high to discuss their kinematic properties .
note that a possible contribution of [ ca v]@xmath36087 to [ fe vii]@xmath36087 has been sometimes suspected for spectra of seyfert nuclei ( e.g. , koski 1978 ; see also appenzeller & streicher 1988 ) . however , simple photoionization models predict that the flux ratio of [ ca v]@xmath36087/[fe vii]@xmath36087 is always @xmath2 0.1 for reasonable parameter ranges , which is confirmed by our calculations @xmath35 and @xmath36 where @xmath37 and @xmath38 are the hydrogen density and the ionization parameter , respectively . ] by using the publicly available code @xmath39 version 94.00 ( ferland 1997 , 2000 )
. we thus neglect the possible contamination of the [ ca v]@xmath36087 emission to the [ fe vii]@xmath36087 emission in the following discussion .
we quantify the observed profiles of the emission lines presented in figure 2 , basically by adopting the manner of heckman et al .
for each emission line , we measure the full line widths at levels of 80% , 50% , and 20% of the peak intensity above the continuum ( @xmath40 , @xmath41 , and @xmath42 , respectively ) .
note again that the effects of the instrumental broadening are not corrected owing to the high wavelength resolution of our observation .
we also measure the velocities of the line centers at the three respective levels ( @xmath43 , @xmath44 , and @xmath45 ) . to quantify the asymmetries of each emission line ,
we measure the widths between @xmath43 and the edges of emission line of both side ( @xmath46 and @xmath47 for the left- and the right - side width from @xmath43 ; see figure 2 of heckman et al .
1981 ) . and following to the manner of heckman et al .
( 1981 ) , we calculate an asymmetry index @xmath48 .
positive figures of this parameter denote the existence of pronounced excess at the blue side of the emission line , and vice versa .
the measured parameters , i.e. , @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 , and @xmath49 , for each emission line , are presented in table 2 .
the measured recession velocities ( @xmath43 , @xmath44 , and @xmath45 ) have once converted into the heliocentric ones , and the presented velocities are relative values to the ( heliocentric ) systemic recession velocity of ngc 4151 determined by the h i 21 cm emission ( @xmath50 km s@xmath5 ; heckman , balick , & sullivan 1978 ) .
we do not measure @xmath42 , @xmath45 , and @xmath49 for [ n ii]@xmath35755 , since the blue wing of this line appears to be significantly affected by the red tail of the [ fe vii]@xmath35721 emission ( see figure 3 ) .
it is apparently shown that the observed forbidden lines are generally blueshifted relative to the systemic recession velocity of the host galaxy , ngc 4151 .
this has been frequently reported for many seyfert galaxies ( e.g. , heckman et al .
1981 ; penston et al .
1984 ; whittle 1985a ; schulz 1987 ; veilleux 1991c ) .
the observed value of @xmath43 for [ o iii]@xmath3 of ngc 4151 is consistent with the previous high - dispersion spectroscopies ( 45 km s@xmath5 , vrtilek & carleton 1985 ; 40 km s@xmath5 , schulz 1987 ) .
the tendency that the asymmetry indices of most of emission lines suggest the blue - wing excess has been also seen for many seyfert galaxies ( e.g. , heckman et al .
1981 ; de robertis & osterbrock 1984 ; vrtilek & carleton 1985 ; whittle 1985a ; veilleux 1991c ; moore et al .
the measured emission - line widths tend to be narrower than the ones reported based on medium - dispersion ( i.e. , @xmath51 km s@xmath5 ) spectroscopic observations of ngc 4151 ( e.g. , heckman et al . 1981 ; de robertis & osterbrock 1984 ; bochkarev , shapovalova , & zhekov 1991 ) .
this is thought to be the effect of the instrumental broadening .
as described by veilleux ( 1991c ) , the effect of the instrumental broadening on medium velocity - resolution spectra can not be corrected completely by the classical quadratic subtraction method .
since the wavelength resolution of our observation is high enough , our results on the emission - line widths are much more reliable than those derived by the previous medium - dispersion spectroscopic observations .
we then investigate how the measured profile parameters of each line depend on or independent of the critical density and the ionization potential of the corresponding transitions .
the full width and the central recession velocity of each level are plotted as a function of the critical density ( figures 4 , 5 ) and of the ionization potential ( figures 6,7 ) . in figure 8
, the asymmetry index @xmath49 is plotted as functions of the critical density and the ionization potential . as shown these figures apparently , clear correlations are seen between the measured profile parameters and the critical density while such correlations are not seen between the profile parameters and the ionization potential . to be more concrete , emission lines with a higher critical density show broader widths and bluer central recession velocities at any intensity level . in order to investigate the statistical significance of these tendencies ,
we apply spearman s rank - order correlation statistical test on the data .
the null hypothesis is that each measured profile parameter is not correlated with the critical density or the ionization potential of the corresponding emission line .
the resultant probabilities are given in table 3 .
it is now statistically confirmed that any of the measured profile parameters are not correlated with the ionization potential of the corresponding ions while the parameters tend to correlate with the critical density of the transitions .
the statistical significances of the correlations concerning the critical density are high for the core properties ( i.e. , @xmath40 and @xmath43 ) while those are statistically not significant for the wing properties ( i.e. , @xmath42 , @xmath45 , and @xmath49 ) .
the difference between the core and the wing may be intrinsic , but we can not exclude the possibility that the measured base properties are affected by the blaze profile of the spectrum described in section 2.4 and appendix a. there is no statistically significant correlations between the asymmetry index @xmath49 and the critical density , and between @xmath49 and the ionization potential .
as described in the last section , the emission - line spectrum of ngc 4151 has the following two remarkable aspects .
( 1 ) the widths of the forbidden emission lines are well correlated with the critical density of the transitions while they are not correlated with the ionization potential of the corresponding ions .
( 2 ) the blueshift amounts of the forbidden emission lines are also well correlated with the critical density while there is no significant correlation between the blueshift amount and the ionization potential .
we interpret these observational results and accordingly discuss the geometrical and kinematical nature of the nlr in ngc 4151 .
however , prior to discuss them , we have to mention the effects of a possible aperture effect on these results .
since the lower - ionization and the lower - critical - density lines tend to arise at the further distant places from the nucleus in general , we may miss the narrow , symmetric components for low - ionization and/or low - critical - density lines .
nevertheless the significant correlations between the critical density and the emission - line profiles are found . since the possible aperture effect should work in an opposite trend
, we can conclude that the observed correlations between the critical density and the emission - line profiles are real .
not only ngc 4151 , many seyfert galaxies also show similar tendencies about the emission - line widths ; i.e. , line widths are correlated with the critical density and/or the ionization potential ( e.g. , filippenko & halpern 1984 ; de robertis & osterbrock 1984 , 1986 ; veilleux 1991a ; alloin et al .
the most simple interpretation for this is that the gas clouds are stratified in nlrs . in this scheme ,
the gas clouds with larger gas densities and with higher ionization degrees are located closer to the nucleus , where the gravitational potential is deeper than the outer part of nlrs and thus the velocity dispersion of gas clouds becomes relatively large .
however , this simple scheme may be difficult to explain the cases that the emission - line widths are correlated only with the critical density or the ionization potential . as for the case of ngc 4151 , one of the high - critical - density forbidden emission lines , [ n ii]@xmath35755 , has a broader velocity width than the other low - critical - density emission lines despite its low - ionization potential , as shown in figures 2 , 3 , and 4 .
a velocity width is thought to be closely related with a depth of the gravitational potential at the effective emitting region of the corresponding emission line .
thus the broadness of the [ n ii]@xmath35755 emission suggests that emission lines with low - ionization potentials can arise at inner nlrs , at least in the nucleus of ngc 4151 . how situation enables to realize such a condition ? to investigate this issue , one of the most powerful approaches would be the locally optimally emitting cloud ( loc ) photoionization model , which was originally constructed to explore the nature of broad - line regions in agns by baldwin et al .
( 1995 ) and which was applied to nlrs later ( ferguson et al .
1997a , 1997b ) . in this model ,
gas clouds with a wide range of physical conditions are present at a wide range in the radial direction , and thus the net emission - line spectra can be calculated by integrating in the parameter space of the gas density and the radius , assuming some distribution functions
. therefore we can know the emissivity distribution of each emission line in the radial direction by integrating the results of this loc models only for the gas density which can be used to discuss the difference of emission - line widths of various forbidden emission lines , assuming that the velocity dispersion of gas clouds is determined by the gravitational potential caused by the central supermassive black hole ( see ferguson et al .
1997a ) . here
we should mention that this assumption is rather inappropriate when gas clouds at outer nlrs are considered , because the gravitational potential caused by the host galaxy is not negligible at such an outer nlr .
however , it seems very difficult to take the contribution of the host galaxy on the kinematics of nlr clouds into account adequately .
this is because the gas motion of nlr clouds tends to be independent of the galactic rotation ( e.g. , heckman et al .
1989 ) , except for clouds at an outer part of very extended nlrs ( see pedlar et al .
1992 for the outer nlr in ngc 4151 ) .
the widths of forbidden lines tend to be correlated with the velocity dispersion of the bulge of host galaxies ( e.g. , nelson & whittle 1996 ; boroson 2003 ) , although such an effect is hard to be quantified for each point in nlrs .
thus , as for the gas clouds at outer nlrs , the moving velocities inferred by the loc model tend to be smaller than the actual situation where the host galaxy contributes to the gravitational potential .
we should be aware of these matters in the following discussion , although we do not take the effects of the host galaxy on the gas kinematics into account in this paper . as shown in figure 1 of ferguson et al .
( 1997a ) , the emission - line strength of each transition reaches its peak where the gas density is comparable to the critical density of the transition and the ionization parameter ( i.e. , the radius ) is preferred by the ionization degree of the corresponding ion . therefore the radius of peak emissivity is in proportion to the square root of the critical density for a given ionization potential , which is apparent when the line - strength distributions of pairs of [ o iii]@xmath14363,5007 , [ s ii]@xmath16717,6731 , or [ o ii]@xmath13727,7325 are compared in each other ( see figure 1 of ferguson et al .
the radius is also in proportion to the square root of the ionization parameter for a given gas density by the definition .
here we focus on the behavior of the emission - line strengths of transitions with a high critical density but with a low ionization potential , such as [ n ii]@xmath35755 and [ o i]@xmath36300 , on the density - radius planes in figure 1 of ferguson et al .
( 1997a ) .
such emission lines can arise from a cloud with a high density and at far from the nucleus .
however , the ionization parameter of such a cloud becomes very low , and thus highly ionized ions can not exist in such circumstances . affected by the existence of such high - density and low - ionization - parameter gas clouds , the peak radius of emission lines with a high critical density and a low ionization potential tends to shift outward .
that is , the radius of emissivity peaks depends more strongly on the ionization potential of the ions than on the critical density of the transition .
see appendix b where this issue is investigated more quantitatively .
de robertis & osterbrock ( 1986 ) also discussed this issue ; i.e. , the emission - line widths would be mainly dependent with the ionization potential of the ions if the density distributions are completely independent of the radius from the nucleus .
they discussed the effects of the radial distance dependences of the gas density as follows .
if the radial distance dependence of the density can be simply described by a power - law form , @xmath52 , then the ionization parameter of a cloud is written as @xmath53 therefore , for the case of @xmath54 , the ionization parameter depends on the radius as @xmath55 and thus the emission - line widths are expected to correlate only with the ionization potential but not to correlate with the critical density .
on the contrary , for the case of @xmath56 , the ionization parameter becomes to be independent of the radius and thus the the emission - line widths are expected not to correlate with the ionization potential but to correlate with the critical density .
therefore , to explain the observed properties of emission - line profiles of ngc 4151 , we should consider the radial distance dependences of the gas density . here
one important concern is that ngc 4151 is a type 1.5 seyfert galaxy and thus the inner regions of the dusty torus are not obscured .
it is known that type 1 agns exhibit statistically stronger high - ionization forbidden emission lines in their spectra than type 2 agns ( e.g. , cohen 1983 ; murayama & taniguchi 1998a ; nagao et al . 2000 , 2001c ) .
this suggests that the high - ionization lines emitting regions ( hereafter `` high - ionization nuclear emission - line regions ( hiners ) '' ; binette 1985 ; murayama , taniguchi , & iwasawa 1998 ) are located very close to the nuclei ( i.e. , inner regions of dusty tori ) and thus can be hidden when we see the nucleus from an edge - on view toward the dusty tori ( e.g. , murayama & taniguchi 1998a ; barth et al .
1999 ; nagao et al .
2000 ; tran et al .
2000 ; nagao et al .
2001c ) like broad - line regions .
murayama & taniguchi ( 1998b ) showed that this hiner component significantly contributes to the observed emission - line spectra of nlrs in type 1 agns ( see also nagao et al .
therefore it seems natural to add this hiner component into the loc model .
if the gas clouds in hiner consist of preferentially high - dense ones , only high - critical - density emission lines are enhanced by introducing the hiner component .
this assumption is consistent with the result of nagao et al .
( 2001c ) that the gas density of this hiner component should be very dense ( i.e. , @xmath57 @xmath35 ) to explain the observed emission - line flux ratios of high - ionization forbidden lines . taking this hiner component into account , the observed correlation between the emission - line width and the critical density of the transitions can be understood .
see appendix b where the effects of adding the hiner component into the loc model are investigated .
we then discuss the origin of the observed correlation between the blueshift amounts of the emission - line center and the critical densities of the transition shown in figure 5 .
it has been often reported that the observed forbidden lines in seyfert galaxies are blueshifted relative to the systemic recession velocity of the host galaxies ( e.g. , heckman et al .
1981 ; penston et al .
1984 ; whittle 1985a ; schulz 1987 ; veilleux 1991c ) .
this has been interpreted that the gas clouds flow radially with some opacity sources , probably dust grains ( e.g. , heckman et al . 1981 ) .
that is , if gas clouds flow outward in the medium with dust grains , outflowing clouds in the far side ( i.e. , in the receding hemisphere ) suffer heavier extinction and thus the net blueward shift of emission lines can occur .
alternatively , infalling gas motions may be the origin of the blueshift if the gas clouds contain plenty of internal dust grains , because such gas clouds are ionized only on the side facing the ionizing continuum source and thus we can see only the receding gas clouds in the far side from an observer . then
, which situation is the case for ngc 4151 ?
we investigate this problem by focusing on the observed correlation between blueshift amounts of the emission - line center and the critical densities of the transitions .
lcc @xmath40 & 4.4@xmath58 & 9.5@xmath59 @xmath41 & 4.7@xmath60 & 4.9@xmath59 @xmath42 & 1.3@xmath59 & 3.9@xmath59 @xmath43 & 2.6@xmath61 & 6.9@xmath59 @xmath44 & 1.8@xmath61 & 6.0@xmath59 @xmath45 & 2.4@xmath59 & 8.7@xmath59 @xmath49 & 3.3@xmath59 & 5.8@xmath59 de robertis & shaw ( 1990 ) examined the behaviors of emission - line profiles for both the outflowing and infalling cases by performing kinematical model calculations .
they showed the following two important results .
( 1 ) as for the outflowing case , a concentration of the dust distribution near the nucleus is required in order to generate more blueward shifts of the lines arising at inner nlrs than those arising at outer nlrs .
( 2 ) as for the infalling case , on the other hand , more blueward shifts of the lines arising at inner nlrs can be caused if high - dense gas clouds at such an inner nlr contains much internal dust grains .
however , neither the former nor the latter situation seems unrealistic , because nagao et al . ( 2003 ) recently reported that iron is not depleted in gas clouds in hiners and thus the innermost part of nlrs is thought to be a dust - free circumstance . here
we it should be mentioned that de robertis & shaw ( 1990 ) pointed out that the existence of a dusty torus around the nucleus may play an important role when the profiles of high - ionization emission lines are discussed since they arise at the innermost part of nlrs , whose spatial extension may comparable to the scale of dusty tori .
taking this point into account , we propose the following situation to explain the observation . that is , in hiners , the gas clouds without dust grains flow outward in the dust - free circumstance .
however , due to the existence of the dusty torus , the receding gas clouds are obscured from an observer . the observed correlation between the blueshift amounts of the emission lines and the critical densities
is then explained if the gas clouds located closer to the nucleus have higher gas densities and larger outflowing velocities .
the schematic view of this idea is shown in figure 9 .
note that erkens , appenzeller , & wagner ( 1997 ) also mentioned that the high - ionization - line emitting clouds flow outward radially , which causes relatively broad widths and blueward velocity shifts of high - ionization forbidden emission lines .
it should be noted that the above model shown in figure 9 requires a special viewing angle from the observer toward the nucleus , since the receding hiner component is hard to be obscured by the torus when seen from a pole - on view toward the torus .
the receding clouds are hidden effectively when our line of sight is near the edge of the torus .
indeed , there are a number of evidence which suggests that the viewing angle toward the nucleus of ngc 4151 is just to be such a special one .
one of them is the type - switching phenomenon seen in ngc 4151 ( e.g. , ayani & maehara 1991 ) ; i.e. , the broad component of balmer lines has been disappeared from the optical spectrum of ngc 4151 ( see also antonucci & cohen 1983 ; penston & perez 1984 ) .
this agn - type switching phenomenon has been sometimes thought to be caused by a time variation of the optical thickness of patchy tori ( e.g. , goodrich 1989 ; tran , osterbrock , & martel 1992 ; loska , czerny , & szczerba 1993 ; aretxega et al .
it has been pointed out that this variation results from the special viewing angle ( i.e. , through near the edge of the torus ) , which is consistent with the model proposed here ( see , e.g. , kriss et al . 1995 ; weymann et al .
the column density of the neutral hydrogen in the line of sight toward the nucleus of ngc 4151 is also remarkable since it is significantly large among type 1 and type 1.5 seyfert galaxies ( @xmath62 @xmath63 ; e.g. , maisack & yaqoob 1991 ; yaqoob & warwick 1991 ; zdziarski , johnson , & magdziarz 1996 ; warwick et al .
1996 ; george et al .
1998 ; yang , wilson , & ferruit 2001 ) .
it is remarkable that the x - ray spectra taken by @xmath64 should consist of both absorbed and non - attenuated radiation ( george et al .
this implies that the absorbing material in the line of sight toward the nucleus has patchy structure and thus only a part of the nuclear emission can reach up to us without the attenuation .
it should be also noted that the column density of the neutral hydrogen toward the nucleus of ngc 4151 is temporarily variable significantly ( yaqoob , warwick , & pounds 1989 ; warwick et al .
1989 ; fiore , perola , & romano 1990 ; yaqoob & warwick 1991 ) . recent high - spatial resolution optical spectroscopic observations have also suggested that the nucleus of ngc 4151 is viewed from the viewing angle near the edge of the torus ( see , e.g. , figure 7 of hutchings et al .
all of the above facts suggests that we see the nucleus of ngc 4151 from the special viewing angle near the edge of the torus and thus the receding gas clouds in hiners is effectively obscured by the torus , which is completely consistent with the model proposed for ngc 4151 in this paper .
we carried out high - dispersion spectroscopy ( @xmath65 ) of the nlr in ngc 4151 by using subaru / hds .
the main findings by examining the obtained emission - line spectrum are as follows .
* the emission - line centers are generally blueshifted relative to the systemic recession velocity of the host galaxy . * the widths and the blueshift amounts of emission lines
are well correlated with the critical densities of the transitions . *
the widths and the blueshift amounts of emission lines show no significant correlations with the ionization potentials of the ions . to interpret the origin of these properties , we refer the results of the photoionization model calculations performed by ferguson et al .
( 1997a ) , i.e. , the loc model .
it is suggested that simple power - law distribution of gas density which is independent of the radius from the nucleus can not explain the observed correlation between the emission - line widths and the critical densities of the transitions .
however , taking the dense gas clouds expected to exist in hiners into account , high - critical - density emission lines can arise at relatively inner regions of nlrs even if their ionization potential is low .
we note that the observed correlation between the blueshift amounts of emission lines is also explained if such dense gas clouds located closer to the nucleus have larger outflowing velocities .
lcc [ o i]@xmath36300 & 10@xmath66 & 10@xmath67 [ ar iv]@xmath34712 & 10@xmath68 & 10@xmath69 [ ar iv]@xmath34740 & 10@xmath70 & 10@xmath71 [ o iii]@xmath35007 & 10@xmath72 & 10@xmath73 [ o iii]@xmath34363 & 10@xmath74 & 10@xmath75 [ n ii]@xmath35755 & 10@xmath76 & 10@xmath69 [ fe vii]@xmath36087 & 10@xmath77 & 10@xmath71 we thank w. aoki , a. tajitsu , and s. s. fujita for assisting our observation .
we are grateful to all the staffs of the subaru telescope , especially to the hds instrument team .
we acknowledge p. a. m. van hoof for giving us useful comments on the rest - frame wavelengths of weak forbidden transitions , and the anonymous referee for useful suggestions .
we also thank g. ferland for providing his code @xmath39 to the public .
tn acknowledges financial support from the japan society for the promotion of science ( jsps ) through jsps research fellowship for young scientists .
a part of this work was financially supported by grants - in - aid for the scientific research ( 10044052 , 10304013 , and 13740122 ) of the japanese ministry of education , culture , sports , science , and technology .
in this appendix , we describe our method of the blaze profile correction in detail ( see also section 2.4 ) . in figure 10
, we give the spectra of the standard star bd+26@xmath21 2606 which are not flux - calibrated and are not corrected for the blaze profile .
the atmospheric absorption features and intrinsic metallic absorption lines are removed from the spectra , and a 15-pixel smoothing is performed . in this figure ,
the spectra have a curvature in some level , though they are expected to be intrinsically linear except for the region around a balmer absorption feature ( recall that bd+26@xmath21 2606 is a typical a - type star ) .
this is just the effect of the unstability of the blaze profile pattern , because the blaze profile pattern should be canceled when the standard - star frame has been divided by the flat flame , if the blaze profile pattern does not change .
although the blaze profile patterns are different among the orders , the difference is very gradual as a function of the echelle order .
since we know the intrinsic sed of bd+26@xmath21 2606 , we can estimate the correction function for the remaining blaze profile patterns and can perform the blaze profile correction , only as for bd+26@xmath21 2606 itself .
however , the estimated correction function about this standard star can not be used to carry out the blaze profile correction for ngc 4151 , because the blaze profile is not the same between the standard - star frame and the object frame .
therefore we should make the blaze profile correction for ngc 4151 by using the spectrum of ngc 4151 itself .
being different from a - type stars , spectra of agns are far from featureless ; i.e. , various emission and absorption features are exhibited . however , as a first - order approximation , the power - law continuum emission component can be regarded to be linear in the wavelength range covered by one echelle order ( @xmath78 ) . therefore , in the same way as for bd+26@xmath21 2606 , the blaze profile of the spectrum of ngc 4151 can be corrected by using the power - law component of ngc 4151 itself , in principle . however unfortunately , the emission - line widths in the spectra of agns are so broad that we can not extract the power - law component of ngc 4151 easily , except for some spectral regions without strong emission / absorption features
. there are some roughly featureless wavelength regions in the spectrum of ngc 4151 around 4600@xmath17 , 5400@xmath17 , and 6000@xmath17 , where the correction function for the blaze profile patterns can be estimated . since the blaze profile pattern depends on the echelle order very smoothly ( see figure 10 )
, we can perform the blaze profile correction for the whole range of the spectrum of ngc 4151 by using the three estimated correction functions .
note that there are weak emission / absorption features in the above three spectral regions .
such features should be removed carefully by comparing the neighboring echelle orders , when the correction functions are estimated .
as for the spectrum covered by the red ccd ( i.e. , from the 86th order to the 106th order ) , we used the correction function estimated at @xmath79 ( the 99th and the 100th order ) . as for the spectrum covered by the blue ccd ( i.e. , from the 108th to the 141st order ) , we used the two correction functions , estimated at @xmath80 ( from the 128th to the 134th order ) and at @xmath81 ( from the 108th to the 114 th order ) .
the former correction function is adopted for correcting the blaze profile of the echelle orders from the 108th to the 114th , while the latter one is adopted for the orders from the 128th to the 141st . as for the orders from the 115th to the 127th , the interpolated correction functions , which are made by using the above two correction functions , are used .
the resultant blaze - profile - corrected spectrum of ngc 4151 are shown in figure 1 .
here we present the results of our calculations of the loc photoionization model .
this presentation mainly aims to show the behaviors of emissivities of weak forbidden transitions such as [ ar iv]@xmath14712,4740 and [ n ii]@xmath35755 on the loc model which were not presented by ferguson et al .
( 1997a ) , and to visualize the effect of taking the contribution of the additional hiner component into account .
for simplification , we assume dust - free , plane - parallel , and constant density gas clouds with the chemical abundances of the solar ones , which are taken from grevesse & anders ( 1989 ) with extensions by grevesse & noels ( 1993 ) .
as for the spectral energy distribution ( sed ) of the input ionizing continuum , we adopt the empirically constructed sed for typical seyfert galaxies ( nagao , murayama , & taniguchi 2001a ) .
this sed is described by the following function ; @xmath82 ( see ferland 1997 ; nagao et al .
2001a ) . here
the following parameter set is adopted ( nagao et al .
2001a ) : ( i ) the infrared cutoff of the big blue bump component , @xmath83 = 0.01 ryd , ( ii ) the slope of the low - energy side of the big blue bump , @xmath84 = 0.5 , ( iii ) the uv to x - ray spectral index , @xmath85 = 1.35 , ( iv ) the slope of the x - ray power - law continuum , @xmath86 = 0.85 , and ( v ) the characteristic temperature of the big blue bump , @xmath87 = 490,000 k. note that the parameter @xmath88 in the equation ( 1 ) is determined from the adopted value of @xmath85 .
the last term in equation ( 1 ) is not extrapolated below 1.36 ev or above 100 kev .
below 1.36 ev the last term is simply set to zero . above 100 kev
the continuum is assumed to fall off as @xmath89 .
the luminosity of the ionizing continuum radiation is set to @xmath90 ergs s@xmath5 , which is a typical value for seyfert galaxies .
we investigate the output emission - line spectra from a gas cloud with a hydrogen density of @xmath91 @xmath35 and a distance from the ionizing radiation of @xmath92 cm .
the ionization parameters of gas clouds are determined when a pair of hydrogen density and distance from the nucleus is given .
note that this distance range scales with the luminosity as @xmath93 and thus does not mean an absolute one .
the calculation are stopped when one of the following three conditions are met .
first is when the electron temperature falls to 3000 k , below which the gas does not contribute significantly to the observed optical emission - line spectra .
second is when the cloud thickness exceeds 10% of its distance from the central source in order to keep gas clouds on the plane - parallel condition . and
third is when the column density of gas clouds exceeds 10@xmath94 @xmath95 , beyond which the gas cloud becomes thomson thick .
see ferguson et al .
( 1997a , 1997b ) for details of the stopping criteria .
we show the results of photoionization model calculations in figure 11 , in which the contour maps of logarithm equivalent widths referred to the incident continuum at 4860@xmath17 are presented .
the contours of calculated equivalent widths of the lines appear to be roughly parallel to lines with a slope of @xmath96 log @xmath97 / @xmath96 log @xmath37 = 0.5 , which corresponds to the condition of a constant ionization parameter .
our results are almost the same as those of ferguson et al .
( 1997a ) with small differences , which do not affect the following discussion significantly .
those differences are thought to be due to the difference in the adopted shape of the input sed and in the version of the code . in table 4 , we give the pair of a hydrogen density and a radius from the nucleus which produces maximum equivalent width of each emission line . here
we do not take the radial distribution function of gas clouds into account , i.e. , the values given in table 4 mean the peak loci in the contour maps in figure 11 .
note that the density condition for the maximum equivalent width of [ o i]@xmath36300 , @xmath98 @xmath35 , is a lower limit value , since our calculations examine the line emissivities only in the range of @xmath99 @xmath35 . by the calculations of ferguson
et al . ( 1997a ) , it is shown that the hydrogen density which produces the maximum equivalent width of [ o i]@xmath36300 is @xmath100 cm@xmath5 ( see figure 1 of ferguson et al .
as clearly presented in figure 11 and table 4 , the radius which produces the maximum equivalent width of each emission line does not correspond to the critical density of each transition .
for instance , the radii of the maximum equivalent widths of [ o i]@xmath36300 and [ n ii]@xmath35755 are larger than those of [ ar iv]@xmath34740 and [ o iii]@xmath35007 .
however , this does not necessarily suggest that the effective line - emitting radius for emission lines is independent of the critical density of the transitions .
this is because the effective line - emitting radius for each emission line is determined by how the gas clouds distribute in the density . in order to examine the radial distribution of the strengths of the forbidden emission lines ,
the calculated emission - line fluxes should be integrated taking the distribution function of the gas clouds in density into account .
the radial distribution of emission - line strength is thus given by @xmath102 where @xmath103 is the emission - line flux of a single cloud with a radius @xmath97 and a density @xmath37 , and @xmath104 is the adopted distribution function of gas clouds on the density .
first , for the most simplified case , we adopt a simple power - law form as the distribution function of gas clouds on the density , i.e. , @xmath105 .
this is the assumption adopted by ferguson et al .
( 1997a ) ( see also baldwin et al .
1995 ; ferguson et al .
the density range of the integration is from @xmath106 @xmath35 to @xmath107 @xmath35 .
note that ferguson et al .
( 1997a ) reported that there is no significant difference on the results of integrations if gas clouds with @xmath108 @xmath35 are included . in figure 12 , we show the results of the density integrations of the emission - line fluxes in the cases of @xmath27 = 0.9 and 1.2 . in both cases ,
the effectively line - emitting radius of the emission lines is mainly dependent with the ionization potential of the ions but is not well dependent with the critical density of the transitions ; e.g. , the effectively line - emitting radii of the [ o i]@xmath36300 emission and the [ n ii]@xmath35755 emission is relatively large comparing to other emission lines .
that is , the resultant radial dependences of emission lines are not to be expected for ngc 4151 .
then , we consider the density integration taking the contribution of the hiner component into account . we adopt the following function as a density distribution ; @xmath109 where @xmath110 is an arbitrary constant , and the two parameters , @xmath111 and @xmath112 , characterize the hiner component in the loc model . again
the density range for the integration is from @xmath113 @xmath35 to @xmath98 @xmath35 .
we perform the density integration by equation ( 2 ) assuming @xmath114 cm and @xmath115 @xmath35 . in figure 13
, we show the results of the density integrations of the emission - line fluxes in the cases of @xmath27 = 0.9 and 1.2 .
as shown in this figure clearly , emission lines with high critical densities are enhanced in the hiners while emission lines with low critical densities are scarcely affected by introducing the hiner component .
in other words , the dependences of emission - line widths on the critical density of the transitions can be explained by introducing the hiner component . | we report on a study of forbidden emission - line spectrum of nearby seyfert 1.5 galaxy ngc 4151 based on the high - resolution ( @xmath0 ) optical spectrum obtained by using the high dispersion spectrograph boarded on the subaru telescope .
the profile parameters such as the emission - line widths , the velocity shifts from the recession velocity of the host galaxy , and the asymmetry indices , for emission lines including very faint ones such as [ ar iv]@xmath14712,4740 and [ fe vi]@xmath15631,5677 , are investigated .
statistically significant correlations between the measured profile parameters and the critical densities of transitions are found while there are no meaningful correlations between the profile parameters and the ionization potentials of ions . by comparing the results with photoionization model calculations
, we remark that a simple power - law distribution of the gas density which is independent of the radius from the nucleus can not explain the observed correlation between the emission - line widths and the critical densities of the transitions . taking the additional dense gas component expected to exist at the innermost of the narrow - line region into account ,
the observed correlations between the emission - line width and the critical density of the transitions can be understood since high - critical - density emission lines can arise at such relatively inner regions even if their ionization potentials are low .
the observed correlation between the blueshift amounts of emission lines and the critical densities of the ions is also explained if such dense gas clouds located closer to the nucleus have larger outflowing velocities . |
polymer brushes are highly tunable systems that have recently attracted considerable attention in the scientific community .
this is mainly due to the already large number of technological applications in which they are used , but also to the promising role these materials hold for the future .
apart from their well known role in the stabilization of colloidal particles @xcite , polymer brushes are also used as lubricants , in chromatographic devices and in adhesives , and their use has recently been proposed in a variety of biotechnological applications including drug delivery and drug - biocompatibility enhancers @xcite . generally speaking ,
they offer an ideal platform that provides control over the physical and chemical properties of solid and fluid surfaces .
polymer brushes are basically dense systems of polymer chains having one end tethered to a non - adsorbing surface , and they have been thoroughly studied theoretically and numerically on different geometries ( see @xcite and references therein ) .
the equilibrium properties of these systems typically depend on the molecular weight of the chains , their grafting density , and the quality of the solvent .
although we have a good qualitative understanding of homogeneous and single components brushes , for which scaling arguments have been successfully put forward , the problem becomes very complex as soon as we deal with non - homogeneous systems with multi - component or nonlinear chains.@xcite of particular relevance for the present paper is the work done by stellacci and collaborators @xcite where basically a two - component brush of immiscible ligands having different length has been shown to phase separate into striped phases of different widths when anchored to spherical and cylindrical templates .
what is surprising is not the phase separation in itself , but rather the formation of the striped phase and its not well understood dependence on the mismatch in polymer length .
microphase separation is also expected in more traditional mismatched two - component polymer - brushes @xcite , where potential applications in stimuli - responsive systems and nanotemplating have recently attracted great attention to the field ( see ref @xcite and references therein ) .
being able to control pattern formation at the nanoscale is a core element in the production of novel materials via the process of self - assembly .
direct coarse grained molecular dynamics simulations by glotzer et al .
have clearly indicated the driving forces behind the formation of these phases as the difference between the chains molecular weight , and the overall height of the brush itself @xcite .
very recently , a coarse - grained model for two - component polymer mixtures of different length has also been put forward , and the formation of different phases as a function of length mismatch on a planar geometry has been analyzed , revealing the presence of striped phases in these systems as well @xcite . unlike other numerical work that focused on the details of the de - mixing transition , in this paper we focus exclusively on the origin of the microphase separation .
we use numerical simulations to compute free energies of an inhomogeneous two - component polymer brush to understand how molecular weight and overall brush height affect the stability and width of the striped phases . specifically , we focus on the case of a cylindrical solid template where stripes have been shown to form promptly both numerically and experimentally @xcite , and we compute , at constant grafting density and template radius , how the system free energy changes when imposing stripes of different width and height on the template .
our results indicate that what limits macrophase separation into a two - phase region , thus leading to the stabilization of the striped phase , is the elastic strain that builds up in the polymer brush formed by the mismatched ( exposed ) polymer segments .
to ensure that our data are not affected by the lateral tension due to the immiscibility of the two polymer types , and that we are indeed measuring exclusively the free energy difference arising from the chains length mismatch and overall brush morphology , we simulate systems in which the only difference between the two polymer types is their length .
this is equivalent to setting the line tension between the stripe s boundaries that comes directly from the immiscibility term in the pair potential between the two polymer types equal to zero . as a consequence
, we are required to lock in place the position of the anchoring monomers to prevent the different chains from trivially mixing .
polymers are modeled as sequences of spherical beads of radius @xmath2 linearly connected via a harmonic potential @xmath3 , with @xmath4 .
any two monomers in the system interact via the soft and purely repulsive dissapative particle dynamics ( dpd ) simulation potential : @xmath5 where @xmath6 and @xmath7 indicate the identity of the polymer @xmath8 and @xmath9 refers to the identity of the monomer . here
@xmath10 is the length of a polymer of type @xmath6 .
all the results that follow use @xmath11 .
each polymer is grafted to a fixed point on the outer surface of a cylindrical template of radius @xmath12 via the same harmonic potential tethering the consecutive monomers in a chain . in all our simulations we considered a total number of polymers @xmath13 arranged in a homogenous grid having square symmetry . the polymer identity is finally selected to generate alternating stripes of width @xmath14 . for a lateral grafting density of the polymers equal to @xmath15 ,
our system contains @xmath16 one - polymer - wide rings , making the possible values of @xmath17 , when equal number of polymer types are considered @xmath18 .
figure [ snap ] shows snapshots of typical initial configurations for @xmath19 and @xmath20 .
this particular geometry is selected because we find that test unconstrained simulations of immiscible chains always lead to stripe formation perpendicular to the cylindrical axis .
this result has also been observed in @xcite .
given the size of the system we carried out our simulations using the molecular dynamics package lammps @xcite with langevin dynamics in the @xmath21 ensemble .
dimensionless units are used throughout this paper .
the timestep size was set to @xmath22 ( @xmath23 is the dimensionless time ) .
our goal is to compute the free energy difference @xmath24 between systems having different stripe width @xmath14 , while keeping everything else unaltered , as a function of the difference between the lengths @xmath0 and @xmath1 . to compute the system free energies we use the thermodynamic integration method @xcite
the idea is to introduce a fictitious potential @xmath25 which acts between any two monomers in the system . for @xmath26
the chains are ideal and the system free energy is independent of the specific grouping of the polymer types ; as @xmath27 we recover the system of interest .
the free energy of the full system can then be extracted by performing the following integral @xmath28 in practice , we perform simulations for several values of @xmath29 , and numerically compute the integral above .
to get insight into the problem we first consider the case in which @xmath30 ; that is , only one chain type is present in the system .
figure [ fvslp ] shows our numerical data for the free energy cost @xmath31 associated with the partitioning of the two polymer species into alternate stripes of different @xmath14 on a cylinder of radius @xmath32 and lateral grafting density @xmath15 , as a function of @xmath14 and for polymer lengths @xmath33 $ ] and @xmath34 .
this specific lateral grafting density was selected because unconstrained simulations under this condition lead to prompt microphase separation for moderate values of polymer immiscibility .
furthermore , a sufficiently large value of lateral grafting density guarantees large enough differences in free energies as a function length mismatch .
+ + our data show that the most favorable state corresponds to that having the largest number of stripes ( i.e. the smallest value of @xmath14 ) .
the inset on the same figure shows how the free energy difference @xmath35 grows with the length of the chain @xmath0 .
these results can be qualitatively understood by referring back to the theory of polymer brushes on flat surfaces and its extension to spherical and cylindrical surfaces @xcite .
the key point is that the main contribution to the free energy cost per chain associated with a polymer brush has an inverse dependence on the the lateral chain - to - chain grafting distance @xmath36 .
@xmath37 \label{f_cyl}\ ] ] where @xmath38 is the height of a brush made with polymers of length @xmath39 grafted on a planar surface .
although in our system the location of the grafting lattice on the cylindrical surface is independent of @xmath14 and therefore the total grafting density of the system is kept constant at all times , different distributions of the chains on the lattice result in different local densities .
crucially , when considering the free energy contribution due to the different lateral organization of the chains , one can consider our system with @xmath40 as a polymer brush with twice the lateral grafting distance of that relative to @xmath20 . clearly the latter system can not be straightforwardly described with eq .
[ f_cyl ] due to the free boundaries of the brush , nevertheless we expect that for @xmath41 the role of the boundary should become negligible .
if we now introduce the dimensionless parameter @xmath42 defined as @xmath43 the argument above leads to clear upper - bounds on the value of @xmath44 .
specifically , we expect @xmath45 in the flat limit , and @xmath46 for @xmath47 , independently of chain length .
our numerical results are indeed within these bounds .
figure [ fvsn ] shows @xmath44 plotted as a function of @xmath14 for different values of @xmath0 .
@xmath44 has a nontrivial dependence on @xmath14 ; nevertheless , simple geometrical arguments can be used to understand at least qualitatively the overall behavior of these curves .
figure [ geom ] shows a sketch of the expected chain distributions for @xmath48 and @xmath49 for the simpler case of a flat surface . for small values of @xmath14
the problem is dominated by boundary effects , and we can imagine the chains equally sharing the overall space @xmath50 per stripe available to them .
+ it is immediately obvious that the free energy difference between @xmath40 and @xmath51 per chain should be very small indeed . in both cases the chains
can be accommodated into space as a sequence of blobs of diameter equal to twice the grafting distance @xmath36 .
a similar argument can be made for @xmath52 .
the main extra cost in free energy per polymer for a growing @xmath14 should come from the weak entropic stretching of the polymers next to the grafting surface and should weakly grow as @xmath53 . when @xmath14 becomes large ,
the free energy of the system is dominated by chains in the bulk of the stripe . in this regime
the blob size of the outer chains is larger than that of the inner chains . assuming a generic size profile for the blobs of the form @xmath54 where @xmath55 $ ] and @xmath56 is fixed by the constraint @xmath57 , one can estimate the free energy per stripe as @xmath58 .
given that there are @xmath59 particles in each stripe , the free energy per particle on a flat interface reads as @xmath60 it is difficult to extract @xmath61 from our simulations , but from visual inspection of the brush profiles one should expect @xmath62 , leading to functional forms for @xmath63 having the typical saturating behavior shown in our data .
in the current form the data do not collapse onto a universal master curve .
in fact , the free energy difference of shorter chains tends to grow faster than that of the longer ones .
this result is quite revealing as it directly reflects on the nontrivial interactions between neighboring regions of grafted polymers .
such interactions will cease as soon as the mutual distance between two neighboring regions is larger than twice the side spread of the polymers at the boundaries .
clearly this will happen sooner for the short polymer than for the longer ones
. we should be able to account for this difference by rescaling of @xmath14 with @xmath64 .
we find the best collapse for @xmath65 .
this result suggests a length - scale , @xmath66 , controlling the extent of the side interactions that scales as @xmath67 .
figure [ fvslpscaled ] shows the data collapse into a universal master curve @xmath68 . vs. the rescaled @xmath69 .
the black line is a guide to the eye.,scaledwidth=40.0% ] let s now turn our attention to the case in which @xmath70
. it would be desirable to develop a simple way of extending our results to the more general @xmath71 case .
in particular , we are interested in understanding whether it is possible to map a system having different polymer lengths @xmath72 to a system of a single polymer type of length @xmath73 ( and @xmath74 . ) vs. the rescaled @xmath69 for different values of @xmath1 .
the lines are guides to the eye.,scaledwidth=40.0% ] figure [ fvslpscaleddiff ] shows @xmath44 as a function of @xmath69 for different values of @xmath0 at @xmath75 and @xmath76 . the results for @xmath30 are also shown as a reference .
as the cause for the free energy gap between the different configurations rests in the difference @xmath77 , to properly compare the data coming from systems having different values of @xmath1 , when computing @xmath44 we subtracted from the measured free energies @xmath31 the constant core free energy of the underlying full brush of height @xmath1 , @xmath78 . when @xmath79 , we define @xmath44 as @xmath80 .
it is comforting to report that indeed it is possible to systematically collapse all of the data corresponding to different values of @xmath0 at a given ( fixed ) @xmath1 into unique curves ; however , we can not collapse data coming from different values of @xmath1 .
this result clearly points to the fact that the universal curve previously described for @xmath30 also has a nontrivial dependence on @xmath1 : @xmath81 . in principle
, this should nt come as a surprise for brushes grafted on a curved surface , in fact , the effect of the `` interior '' layer of @xmath1 monomers is to increase the effective radius of the cylinder to which the reduced system would be grafted onto , and to decrease the grafting density of the effective brush of length @xmath77 ; both are variables the free energy depends on ( see eq .
[ f_cyl ] ) .
simple geometrical arguments can be used to estimate the change in lateral grafting distance . following @xcite
, we can write the change in lateral grafting spacing as we radially move away from the cylinder surface of an amount @xmath82 as @xmath83 . for short polymers and at the relatively high densities of our systems we have @xmath84 .
it follows that @xmath85 .
simultaneously , the effective polymer of length @xmath77 would be grafted onto a cylinder of radius @xmath86 .
the ratios @xmath87 appearing in the brush free energy should then be adjusted to @xmath88 introducing a predictable dependence on @xmath1 in the problem .
although this argument suggests an overall increase in free energy for the system at a given @xmath14 , it is nt obvious how it will affect the dimensionless ratio @xmath42 .
clearly this correction should factor out in the large @xmath89 limit or for @xmath90 .
our numerical data show that @xmath44 is quite sensitive to @xmath1 .
specifically , @xmath91 tends to flatten as @xmath1 increases .
this is most likely explained by the fact that the base of the effective chain of length @xmath77 is not really grafted in place , but the layer of polymers below allow for significant reorganization of the base due , for instance , to their compressibility .
because of the blob - size dependence on the distance from the surface , on a convex template the chains become systematically more laterally compressible as @xmath1 is increased .
the net result is a flattening of the free energy difference as a function of @xmath14 .
unfortunately , we find that our data for different values of @xmath1 can not be collapsed by a simply rescaling of the free energy with a power of @xmath1 , indicating a subtle interplay between @xmath31 and the height of the bottom layer .
in summary , we analyzed the relative stability of the striped phases arising from the de - mixing of immiscible polymers on a cylindrical template as a function of the width of the stripe , the length mismatch between the chains , and the overall length of the brush .
we set numerical bounds for the free energy gap between an alternating ( many thin stripes ) and a fully de - mixed ( two wide stripes ) phase , and found that indeed the former becomes more and more favorable as the mismatch between the chains lengths , @xmath92 is increased .
we also found that when appropriately normalized the free energies as a function of stripe width for different values of @xmath92 can be collapsed into unique master curves that only depend on the overall length of the lower brush @xmath1 .
finally we discussed the possibility of reducing a two - component system of chains having lengths @xmath93 into a simpler single component system of chains of length @xmath92 .
our results are fully consistent with previous molecular dynamics simulations on this problem in identifying the key parameters setting the width of the stripes in the polymer length mismatch @xmath77 , and the overall length of the shorter brush @xmath1 together with the degree of chain immiscibility . in all our data the direct line tension @xmath94 coming from the pair potential between chains of different types
has been set to zero .
this term would add to the free energy balance a predictable contribution of the type @xmath95 , where @xmath96 is the height of the polymer brush formed by the lower chains and @xmath97 is the number of phase boundaries .
clearly @xmath98 has a minimum for @xmath41 and is large for small @xmath14 .
the balance between the line tension and the configurational free energy computed in this paper should therefore set the width of the stripe .
more work in this direction is currently underway .
this work was supported by the american chemical society under prf grant no . 50221-dni6 . | we use computer simulations to investigate the stability of a two - component polymer brush de - mixing on a curved template into phases of different morphological properties .
it has been previously shown via molecular dynamics simulations that immiscible chains having different length and anchored to a cylindrical template will phase separate into stripes of different widths oriented perpendicularly to the cylindrical axis . we calculate free energy differences for a variety of stripe widths , and extract simple relationships between the sizes of the two polymers , @xmath0 and @xmath1 , and the free energy dependence on the stripe width .
we explain these relationships using simple physical arguments based upon previous theoretical work on the free energy of polymer brushes . |
particle detectors based on condensed noble gases have found wide application in high energy physics , astro - particle physics , and nuclear physics .
noble liquids are attractive candidates for particle detectors due to their ease of purification , good charge transport properties , high scintillation efficiency , and in the case of xenon , high density and short radiation length .
examples of recent rare - event searches based on cryogenic noble gases include clean / deap @xcite , xmass @xcite , zeplin @xcite , xenon @xcite , lux @xcite , warp @xcite , ardm @xcite , exo @xcite , icarus @xcite , and meg @xcite .
the experimental methods of liquid noble gas detectors have been developed in small prototype instruments whose liquid volumes range from a few cubic centimeters to tens of liters . in these prototypes ,
the system is typically cooled either by liquid nitrogen or by directly coupling the storage vessel to the cold head of a refrigerator .
these techniques have several attractive features , including simplicity , robustness , and the availability of great cooling power . however , both of these cooling strategies typically require a significant amount of space above or below the storage vessel be devoted to the cryogenics .
in contrast , room temperature detector technologies , such as gas proportional counters or plastic scintillator , are not burdened by a cryogenic system , which is particularly advantageous for prototyping work where the freedom to make maximal use of the space around the detector provides valuable flexibility . in this article
we describe a system for condensing and storing a noble gas ( xenon ) where the cooling system is spatially separate from the liquid storage vessel , leaving only the cryostat in the space surrounding the vessel .
this configuration facilitates many common laboratory operations , particularly those which require access to the upper or lower flanges of the storage vessel .
examples include the installation of detector structures and the insertion and removal of material samples and radioactive sources .
this setup also provides direct optical access to the interior of the vessel for viewing or for laser injection , and it simplifies the construction of a lead shield . our primary motivation for pursuing the remote cooling method discussed here is to allow the space above the liquid xenon vessel to be used for ion tagging and retrieval experiments in the context of the exo double beta decay search .
exo has proposed to eliminate radioactive backgrounds by identifying the barium ion produced in the double beta decay of @xmath0xe @xcite .
the identification method may require that a device be inserted into the active volume of the double beta decay detector to retrieve the final state nucleus .
the condenser and liquid xenon vessel described in this article will allow both a barium ion calibration source and an insertion and retrieval device to be coupled to the xenon vessel from above . in the last decade
pulse tube cryocoolers have attracted attention as convenient and reliable means to liquefy noble gases .
for example , technology has been developed for the meg and xenon experiments using a modified cryocooler integrated into a liquid xenon storage vessel @xcite .
several other recent articles have reported on the development of a small - scale helium condenser based on a pulse tube cryocooler where the condenser is located directly in the neck of a liquid helium storage dewar @xcite@xcite@xcite . in our system
, we also employ a pulse tube cryocooler , and our condenser is conceptually similar to the helium condenser described in references @xcite-@xcite , although in our case the remote storage vessel represents an additional complication .
note that we report here the results of cooling and liquefaction tests carried out with xenon .
similar results could likely be obtained with other heavy noble gases having lower saturation temperature , such as argon or krypton , provided that the heat leaks in the system are minimized .
the system consists of two units : a helical copper condenser and a stainless steel liquid xenon ( lxe ) storage vessel .
a system drawing and plumbing schematic are shown in figures [ fig : system ] and [ fig : schematic ] , respectively .
the condenser is located above the vessel , and condensed liquid flows downward through a 1/4 " stainless steel ( ss ) tube to the top of the vessel .
the condenser is cooled by the cold head of a cryocooler , while the vessel is cooled only by the xenon .
the entire system is enclosed in a vacuum - insulated cryostat composed of two ss cylinders connected by a bellows .
the two cylinders and bellows form one vacuum volume for pump - out purposes .
the radiative heat leak is reduced by wrapping the system components in super - insulation consisting of 10 - 15 alternating layers of aluminized mylar and fabric .
the condenser is a helical coil of 1/4 `` diameter oxygen free high conductivity copper ( ofhc ) tube , which was chosen for its purity and thermal conductivity .
the tubing is partially annealed ; non - annealed tube was found to kink when coiled .
the coil is brazed to a 2 '' diameter cylindrical shank of ofhc copper , which is mechanically attached to the coldhead of a pulse tube cryocooler @xcite .
the total length of the cooled portion of the tube is 30 inches ( five turns ) .
as shown in figure [ fig : condenser ] , the cold head has two stages , with base temperatures of 22 k and 8 k for the first and second stages , respectively .
these base temperatures are far below what is required to liquefy xenon ( triple point of 161 k ) , but should allow the system to condense the lighter noble gases . to ensure temperature uniformity in the condenser , and to achieve good thermal control , each end of the condenser is coupled to one of the two stages of the cold head .
we refer to the upper half of the condenser ( coupled to the first stage of the cold head ) as the pre - cooler " , and the lower half ( second stage ) as the post - cooler " .
the function of the pre - cooler is to cool the room temperature xenon gas to the saturation temperature , while the function of the post - cooler is to remove the latent heat of vaporization , thereby effecting the phase change .
the pre - cooler and the post - cooler are independently temperature controlled by trim heaters .
the heaters are driven by a common 77 w regulated dc power supply , and each heater circuit is controlled by a pid temperature feedback unit @xcite .
the controllers adjust the current in each heater via the gate voltage on two power fets .
the pre - cooler trim heaters are mounted on flats in the cylindrical shank of the condenser .
these flats make a narrow neck " between the coil and the upper cold stage , thus reducing the cooling power and allowing for more precise temperature control .
a stainless steel shim between the condenser shank and the cold head further reduces the cooling power of the pre - cooler , which was found to be excessive for our purposes .
the post - cooler trim heaters are attached to a plate on the bottom of the shank , which is itself attached to the second stage of the cold head via a flexible copper braid .
the braid was chosen to provide a flexible thermal bridge so that the condenser is mechanically constrained at its upper end only .
the control temperatures for the two pid feedback loops are measured by thermocouples located on the neck and on the plate for the pre - cooler and post - cooler , respectively . in typical operation
, the pre - cooler set point temperature is chosen to be 179.5 k , and the post - cooler temperature is chosen to be a few degrees cooler . in practice , however , the post - cooler temperature tends to stabilize around 193 k during condensation , and thus the post - cooler heater does not power on .. ] this reflects the fact that the thermal coupling between the post - cooler and the second stage of the cold head has too much thermal resistance .
it is likely that this resistance limits the power and efficiency of the condenser , so we intend to modify this arrangement in the near future .
currently , the condensation that does occur is sufficient for our purposes .
the trim heaters on the pre - cooler allow the condenser to adjust for the effects of changes in the gas flow rate . at high flow rates ,
significant heat is delivered to the condenser by incoming room temperature xenon gas , so the pre - cooler trim heater reduces its power output to maintain the set point temperature . in some cases ,
the incoming heat was found to be sufficient to warm the condenser to one or two degrees above the pid set point . at zero flow or low flow rates , the heat delivered to the condenser by
the gas is small , and the pre - cooler trim heater provides compensation to keep the condenser temperature above the freezing point of xenon . consequently , one benefit of the pre - cooler pid feedback loop is that it prevents over - cooling in the event that a large gas flow suddenly stops .
we present some measurements of the cooling power of the second stage of the pt805 near lxe temperature .
( cryomech , the cryocooler manufacturer , does not report this data at such high temperatures . )
we attached a known thermal resistance to the second stage of the cold head , applied heat from the opposite end , and measured the temperature at each end .
once thermal equilibrium is reached , the cooling power is equal to the power delivered by the heater , and can also be inferred from the temperature gradient across the known thermal resistance .
these two methods give results in agreement with each other .
we find that the cooling power is 25 w at 65 k , 26.2 w at 75 k , and 32.2 w at 152 k. the xenon vessel , seen in figure [ fig : vessel ] , is a 6 `` x 6 '' od cylinder constructed from stainless steel , which was chosen for its purity and thermal mass .
this volume is sufficient to hold 10 kg , or about 3 liters , of lxe , as well as a particle detector .
the top and bottom of the vessel are 8 `` conflat flanges .
the vessel is suspended from a vertical 3 '' od stainless steel tube , 6 `` in length , which penetrates its top flange . at its upper end ,
the 3 '' tube is welded to a 4 5/8 `` conflat flange and a large stainless steel plate which provides mechanical support .
the plate also acts as part of the cryostat .
the 4 5/8 '' flange and tube allows direct access to the interior of the xenon vessel from the laboratory . with a glass viewport attached to this flange
, the lxe can be seen inside the vessel .
this access port can also be used to introduce detectors , materials , or radioactive sources .
a smaller , fused - silica viewport is welded into the bottom of the vessel ; it is paired with an identical viewport in the cryostat to admit laser light for use in lxe purity tests .
the vessel has three plumbing connections for xenon flow : a 1/4 `` ss liquid supply line that enters the vessel at the top , a 3/8 '' liquid return line that drains the vessel from the bottom , and a 1/4 " gas return line that exits the vessel at the top ; see figures [ fig : schematic ] and [ fig : vessel ] .
lxe from the condenser runs down the liquid supply line and collects in the bottom of the vessel and in the liquid return line .
a heater on the liquid return line is used to boil the liquid for re - circulation or for recovery .
the gas return line at the top of the vessel allows gas to circulate freely through the vessel , especially when the vessel is filled with liquid . as discussed in section [ sec : gasflow ] below , free recirculation of gas is important for system operation .
all three plumbing lines have vcr fittings so they can be disconnected from the rest of the system , thus allowing the condenser to be removed for servicing . an external , custom gas recirculation pump is used to force xenon flow through the condenser and the xenon vessel . it can achieve controllable xenon flow rates of up to 10 slpm .
the pump is a bellows - type , made entirely of stainless steel , except for a teflon sleeve .
the pump is driven by a 1/3-hp , three - phase motor @xcite ; the motor itself is controlled by an inverter @xcite , which allows flow control via adjustment of the repetition rate of the pump .
a capacitive level sensor is integrated into the liquid return line .
it has a co - axial cylindrical geometry formed by suspending an 11.5 `` x 1/4 '' od stainless steel tube inside the 3/8 " od liquid return line .
the inner conductor is vented to allow the liquid to flow unimpeded .
it is wrapped at the top and bottom with a small amount of kapton tape to prevent electrical shorting .
the sensor has a capacitance of 104 @xmath1 3 pf in vacuum ( 224 @xmath1 3 pf including cabling ) . because the level meter is in the liquid return line , rather than the vessel itself
, it will only accurately read the liquid level in the vessel when the pressures are equal .
this can be ensured by keeping the gas return valve in figure [ fig : schematic ] open .
changes in capacitance are measured with a custom circuit .
the level sensor is in series with a resistor , forming a low - pass rc filter .
an ac voltage of amplitude 0.15 v and frequency 8 khz is input to the filter , and the voltage across the capacitor is amplified and rectified for computer readout by a data acquisition board @xcite . to maximize sensitivity and dynamic range ,
the resistor in the filter was chosen such that 8 khz is near the knee frequency .
the response of the circuit output was calibrated with a set of known capacitances . a quadratic fit of this data
is used to interpolate for future measurements .
the capacitance can then be converted to a liquid level measurement based on the known dielectric constant of lxe .
changes in capacitance as small as 1 pf can be measured , corresponding to a height sensitivity of about 1 mm .
we have constructed an alarm system to notify lab personnel in the event of a serious system failure , such as a power outage .
three alarm conditions are considered : loss of electrical power to the cryocooler , loss of electrical power to the trim heater power supply and/or pid controllers , and an overpressure alarm .
each alarm is represented by a switch which closes if the alarm condition is present .
the switch then activates a commerical pager unit @xcite , which dials a list of phone numbers until the alarm is acknowledged .
the pager unit derives power from a ups to ensure that it remains active in the event that electrical power is lost throughout the lab .
test alarms can be generated with a push button to confirm that the system is active .
during normal operation , the recirculation pump is used to force gas into the condenser , where some fraction of it condenses and travels with the remaining gas flow down into the vessel .
the liquid either cools the vessel through evaporation or collects in the vessel , depending on the vessel temperature .
the gas usually exits the system through the gas return line , although it can return through the liquid return line as well if no liquid is present .
cryo - pumping can also be used as a source of gas flow , but this is less convenient due to the consumption of liquid nitrogen .
we find that forced gas flow significantly increases the cooling power delivered from the condenser to the vessel .
one possible explanation for this effect is that gas flow is necessary to transport the xenon dew " from the cold surfaces of the condenser to the remote storage vessel .
regardless of the origin of the effect , however , the importance of gas flow for cooling purposes magnifies the role of the gas return line in the system . during the initial cool down
, it makes no difference if the gas return valve is open or closed , because the liquid return line provides an alternate return path .
but once liquid has collected , the valve must remain open , or else the liquid will block the flow of gas , preventing cooling power from being delivered to the vessel . starting with the system at room temperature and under vacuum ,
i.e. , with no xenon gas present , the vessel temperature drops by only a few degrees when the cryocooler is activated , demonstrating that conductive cooling through the ss plumbing is negligible .
once xenon gas is introduced , the xenon vessel cools through convective heat exchange with the condenser , but the vessel temperature levels out at roughly 200 k in the absence of forced gas flow .
we find that a minimum gas flow rate of about 1.2 slpm is typically necessary to cool the xenon vessel from room temperature to the saturation temperature . at higher flow rates , from 2 to 4 slpm
, the vessel cools more quickly , as shown in figure [ fig : temp_v_flow ] .
the effects of increasing the flow rate even further are unclear . in one instance , increasing the flow rate to 5 slpm decreased the vessel cooling rate , since the large amount of incoming warm gas heated the condenser above lxe temperature . however
, in another instance , flows of 6.8 slpm led to greater cooling rates in the vessel , so there may be other effects , such as pressure , that play a role .
cooling rates are discussed further in section [ sec : coolingrates ] . forced gas flow may only be critical during the initial phase of vessel cooling .
during one cool - down , gas flow and condensation were established , but the recirculation pump was unexpectedly halted , interrupting flow for @xmath21 hour .
condensation continued during this time . upon restarting the pump , the xenon gas pressure rose and would not stabilize until the pump was shut off again .
however , condensation continued and the liquid level in the vessel rose without the forced flow .
it is therefore not clear if the 1.2 slpm flow rate is an absolute limit for vessel cooling ; it may only be required to initialize condensation .
once the vessel is cold and filled with liquid , it is easy to maintain at constant temperature and pressure for indefinite periods of time by forcing gas flow through the condenser and liquid vessel , using the gas return line as the exhaust . in this arrangement
the gas circulates in a closed loop .
this is generally our default configuration during liquid maintenance , and it allows for continuous purification of the xenon gas with a gas phase purifier .
however , we have also studied the possibility of maintaining the liquid at constant temperature and pressure with the recirculation pump turned off .
this situation could be important , for example , if the pump were to lose power unexpectedly .
as detailed below , we find that it is possible , but more difficult , to achieve stability with the recirculation pump turned off .
note that the xenon will thermodynamically recirculate at a small rate ( a few sccm ) if the external gas plumbing allows gas to flow from the system output to the condenser input .
this thermal recirculation " is driven by the system heat leak , and it can be augmented by warming the liquid return line with its heater .
we achieve temperature and pressure stability most easily in the absence of forced gas flow by closing a valve in the external plumbing which prevents thermal recirculation . under these conditions
, we expect that no gas will enter the condenser input from the external system .
our flowmeter confirms this expectation .
nevertheless , we see through the large viewport that a steady stream of liquid drops falls into the vessel from above .
this indicates that gas is counter - flowing up the liquid supply line from the vessel , liquefying , and falling back down , creating a closed heat exchange loop .
this behavior is rather sensitive to the condenser temperature .
for example , raising the pre - cooler temperature from 179.5 k to 180.5 k is enough to disturb the establishment of this heat exchange loop .
if we turn off the external recirculation pump without preventing thermal recirculation , then we usually find that condensing slows or stops , and that the system temperature and pressure slowly rise .
since the previous tests show that the system is able to condense the cold counter - flowing gas from the lxe vessel , this behavior could indicate that the extra heat load from the room temperature gas is too large .
it is possible that this situation could be remedied by improving the cooling power of the condenser , particularly that of the post - cooler .
also note that if we encourage thermal recirculation with the heater , this makes the situation worse , increasing the rate of temperature and pressure rise .
on the other hand , if we establish the internal heat exchange loop by preventing external thermal recirculation , and we allow the system to run in this mode for several hours , we find that opening the external valve to allow thermal recirculation does not disturb the system stability .
that is , the system behaves as if the valve were still closed .
the operation of the system is divided into four stages : preparation , vessel cool - down and filling , liquid maintenance , and recovery . to remove impurities in the system , the condenser and vessel
are evacuated to @xmath210@xmath3 torr using a turbomolecular pump backed by a dry scroll pump .
the cryostat is evacuated to @xmath210@xmath4 torr using a similar configuration , and it is pumped continuously while the cryocooler is running . a xenon gas supply bottle with a regulator is opened , and the regulator is set to introduce gas at a pressure between 1000 and 1400 torr . the recirculation pump is turned on and set to a flow rate of at least 1.2 slpm .
the gas circulates though the condenser , vessel , and external plumbing , and optionally through a zirconium getter for purification @xcite .
the pressure regulator on the xenon gas supply bottle is left open to allow additional gas to enter the system as needed .
the pre - cooler set point temperature is set to 179.5 k. once this is done , the system is prepared for cool - down . during the vessel cool - down and filling phase
, the gas continues to circulate through the condenser , vessel , and external plumbing .
as the gas cools and becomes more dense , and later as the gas liquefies , additional gas is delivered from the supply bottle to the system as needed to maintain a constant pressure .
the total amount of xenon in the liquid system can be monitored by measuring the remaining pressure in the supply bottle .
figure [ fig : cooldown ] shows a temperature history of a typical cooldown as recorded by thermocouples at the output of the condenser and at the liquid return line near the bottom of the vessel .
the cryocooler is activated to begin cooling the system , and the condenser temperature immediately drops .
when the temperature reaches 260 k , the trim heaters turn on for the first time , momentarily warming the condenser .
( the heaters are programmed to stay off when the condenser is above this temperature .
this avoids overheating the cold head ) .
after @xmath21 hour , the temperature of the output of the condenser falls sharply , indicating lxe has formed and is pouring down into the vessel . after this sharp temperature decrease ,
drops of lxe can be seen via the viewport falling onto the bottom of the vessel and boiling away , and the temperature of the lxe return line decreases at about six times the previous rate .
note that this indicates that most of the cooling power is delivered to the vessel in the form of liquid xenon , rather than gaseous xenon .
the vessel cools this way for some time , typically about 8 hours , but it can be more or less depending on flow and pressure conditions . when the vessel temperature is low enough , liquid begins to collect in the bottom of the vessel .
when enough liquid has collected , it overflows a small lip at the drain of the vessel that leads to the liquid return line .
this `` splash '' brings the cold liquid into direct contact with the return line , causing the temperature there to quickly drop , as can be seen in figure [ fig : cooldown ] around 16:00 on 12/29 .
this drop is correlated with a quick rise in the measured liquid level as the level meter is filled for the first time .
after the `` splash '' of liquid fills the return line , lxe continues to collect in the vessel . at this stage
, the gas return bypass valve must be open to ensure that a high rate of gas flow can continue .
( we typically keep this valve open through the entire process , from initial preparation through xenon recovery . )
condensing 1 kg of xenon takes 2 - 3 hours , depending on pressure and flow conditions .
liquid can be maintained indefinitely in the vessel at constant temperature and pressure by keeping the pre - cooler set point temperature at 179.5 k , and by maintaining a nominal gas flow rate of 1 - 2 slpm with the recirculation pump . as described in section [ sec : gasflowliquidmaintenance ] , the recirculation pump can also be turned off under certain conditions .
the xenon is recovered by cryopumping .
the room temperature gas supply bottle , which is made of aluminum , is placed in a liquid nitrogen bath , freezing any xenon inside .
the supply bottle pressure regulator is fully opened to allow xenon gas to flow backwards through it .
in addition , a bypass valve which is connected in parallel with the pressure regulator can be opened to reduce the pumping impedance further , but this is usually not necessary . finally the cryocooler is turned off and the lxe is allowed to warm , raising the vapor pressure .
a hand valve between the liquid system and the cold supply bottle is used to manually regulate the gas flow rate .
care should be taken to avoid forming xenon ice by pumping too quickly .
recovery can be made quicker by heating the liquid return line directly with its integrated heater , or by filling the cryostat insulating vacuum with a dry gas such as nitrogen or helium to provide a thermal connection to the room temperature lab .
figure [ fig : level ] shows a sample plot of the height of lxe in the vessel during a cool - down .
the gas pressure was 1000 torr .
the recirculation pump forced a constant flow rate of 2.8 slpm from 11:00 to 18:45 pm , after which it was turned off , and the xenon supply bottle was valved off . the initial rapid rise from -2.7 cm to 0.6
cm corresponds to the `` splash '' when liquid first overflows from the vessel into the return line .
after that , a roughly linear increase in the height of the xenon can be seen as the storage vessel fills with liquid .
a surprising feature is that from 6:45 pm until 9:00 am the next day , the liquid level is seen to be rising slowly .
this could indicate that the level sensor circuit has a slowly drifting systematic error .
the cooling rate of the stainless steel vessel was measured on two separate trials . during the first run
, the flow rate was maintained at about 1.4 slpm , with a xenon pressure of 1500 torr .
the vessel cooled at 5.6 k / hr . during another run ,
the vessel cooled at 6.5 k / hr , with a much higher gas flow rate of 6.8 slpm and pressure of 1500 torr . using a specific heat for stainless steel of 500
j / kg / k , and estimating the vessel weight at 20 kg , we find that these cooling rates correspond to 15.5 w and 18.9 w of cooling power transferred to the vessel .
two trials are presented in which the condensation rate is measured at various pressure and flow parameters .
condensation rate is measured during vessel filling in two ways . 1 . )
the mass of xenon remaining in the gas supply bottle is calculated from the known volume of the bottle , the measured pressure , and the density of xenon gas at that pressure .
( the density of xenon gas has a non - linear relationship to the pressure at typical bottle pressures , and this dependence must be taken into account . )
the rate of decrease of source mass was then calculated using central differencing and equated to the rate of condensation in the liquid system .
the height of xenon in the vessel is measured , and converted into a volume and mass using the known cylindrical geometry of the system and the density of lxe .
the rate of change in liquid height then equates to the condensation rate .
these two methods give results in good agreement with each other , although the liquid height method is complicated by the volume displacement of the irregularly shaped particle detector in the lxe vessel . in the following , we quote results based on the bottle pressure method . in the first trial ,
the xenon gas pressure was set to 1550 torr , and the gas flow rate was varied .
the condensation rates were 0.54
kg / hr for a flow of 2.65 slpm and 0.61
kg / hr for a flow 3.89 slpm . in a second trial ,
the gas pressure was set to 1000 torr .
the recirculation flow rate was set to 2.8 slpm for the bulk of the trial .
condensation rates between 0.36 and 0.40 kg / hr were observed . the largest rate of condensation , 0.61
kg / hr , was observed at a high pressure and large recirculation rate , 3.89 slpm and 1550 torr .
this implies that 44% of the circulating gas is condensed in a single pass . a condensation fraction of 58%
was also achieved , but only at the cost of a lower condensation rate of 0.54 kg / hr .
tests at a lower pressure of 1000 torr indicate both lower condensation rate , @xmath2 0.36
kg / hr , and a lower efficiency of 36% .
thus , the condensation fraction depends moderately on pressure and flow . using the latent heat of vaporization of xenon , 12.64 kj / mol , and the specific heat of xenon gas , 20.8 j / mol / k , we can calculate the cooling power implied by our condensing rates . for the largest condensation rate , 0.61 kg / hr
, we find a cooling requirement of 3.5 w and a latent heat removal of 16.4 w , for a total of 19.9 w. this is similar to the 18 w of cooling power we estimated is delivered to the stainless steel vessel during cooling .
we find that for reliable operation of our condenser it is essential to control the temperature at both the top and bottom .
initial tests in which the condenser was cooled only at the top showed that a large temperature gradient would appear along its length . in this arrangement , the top of the condenser must be over - cooled to allow liquefaction to occur in the lower portion of the condenser .
this leaves the condenser prone to ice formation , particularly if the gas flow rate suddenly decreases , removing a heat source . with a dual control mechanism ,
both ends of the condenser are cooled , ensuring that temperature gradients are small .
in addition , the two thermodynamic functions of the condenser ( cooling warm gas and liquefaction ) are separated spatially in the condenser , allowing for semi - independent temperature regulation with a pre - cooler and a post - cooler .
this gives the condenser the flexibility to adjust to changing conditions , such as a change in the gas flow rate . in practice ,
our post - cooler temperature is usually above its set point temperature during condensation , and therefore its trim heater plays little role .
nevertheless , the additional cooling provided by the second stage of the cold head through the post - cooler improves temperature uniformity in the condenser , leading to more robust operation . to achieve greater control in the future
, we intend to increase the cooling power of the post - cooler by reducing its thermal resistance .
xenon ice formation is a dangerous problem for external condensers such as the one described here .
ice can block the flow of liquid and gas to the xenon vessel .
since the xenon flow is the only cooling mechanism for the vessel , its interruption can lead to a dangerous rise in system temperature and pressure .
our design is particularly sensitive to this problem , because the helical coil of the condenser has a cross - sectional outer diameter of only 1/4 " .
therefore even a small amount of ice can lead to flow blockage .
a condenser coil made from larger diameter tubing may improve the situation , or perhaps a condenser with an altogether different geometry may be better . for our purposes , however , the current design has proved to be adequate with appropriate safeguards . gas flow is essential for transferring the cooling power from the condenser to the xenon vessel , and a xenon gas return line from the top of the vessel greatly improves the effectiveness .
there is a minimum flow rate necessary to cool the vessel to lxe temperature , and the condensation rate increases , albeit only slightly , with increasing flow .
cryopumping can serve as the method for forcing flow , but this is an awkward process that consumes large amounts of liquid nitrogen , and flow must be interrupted frequently to warm and cool the supply and recovery bottles
. therefore we find it is very useful to have a recirculation pump .
using the pump , gas flow can be easily maintained for days at a time , and with an inverter controller , the flow rate can be dialed to a desired value for easy testing .
the main drawback to the pump is a loss of purity : our custom pump contains a teflon sleeve which is a source of outgassing and teflon debris in the system .
these problems can be solved with purifiers and filters , however .
we have described a system for condensing and storing xenon where the source of cooling power has been removed from the vicinity of the liquid storage vessel , facilitating the introduction of instruments and materials to the vessel .
condensation rates as high as 0.61 kg / hr were achieved , after an initial cool - down period of 8 - 10 hours .
this corresponds to a condensation fraction of 44% and a cooling power of about 20 w. changes in condenser design may be able to improve the condensation fraction and cooling power .
our design includes a two - stage cooling system for improved temperature uniformity and control .
we find that a nominal gas flow rate is important for delivering cooling power to the vessel , and that a dedicated gas return line is useful for maintaining this flow when the vessel is filled with liquid .
we thank john carriker for his many contributions to the system described in this article . this work was supported by award number 0652690 from the national science foundation .
p. benetti , _ et .
( warp collaboration ) , `` first results from a dark matter search with liquid argon at 87 k in the gran sasso underground laboratory '' , astropart . phys . * 28*:495 - 507 ( 2008 ) ,
arxiv : astro - ph/0701286v2 .
c. hall , `` searching for double beta decay with the enriched xenon observatory '' , proceedings of the 9th conference on the intersections of particle and nuclear physics ( cipanp 2006 ) , aip conf .
proc . 870:532 - 535 , 2006 . | we describe the design and operation of a system for xenon liquefaction in which the condenser is separated from the liquid storage vessel .
the condenser is cooled by a pulse tube cryocooler , while the vessel is cooled only by the liquid xenon itself .
this arrangement facilitates liquid particle detector research by allowing easy access to the upper and lower flanges of the vessel .
we find that an external xenon gas pump is useful for increasing the rate at which cooling power is delivered to the vessel , and we present measurements of the power and efficiency of the apparatus .
xenon , condenser , recirculation pump |
it is generally agreed that the broad absorption line ( bal ) systems seen in the spectra of many quasars are intrinsic to the quasar , arising from high velocity outflow of gas directly from the accretion disk region @xcite .
narrow absorption lines in quasar spectra , however , arise from a variety of sources .
narrow absorption lines with small velocities relative to the quasar emission redshift could be caused by galaxies along the line of sight , clouds in the interstellar medium of the host galaxy , or even smaller scale gas flows within a few parsecs of the black hole .
on the other hand , most of the narrow absorption systems that have large velocities with respect to the emission redshift of the quasar are thought to be intervening @xcite ; however , there is increasing evidence that even some fraction of these may be intrinsic . in many bal quasars ,
the absorption troughs have considerable structure , suggesting a connection to narrow absorption line systems .
we discuss the spectrum of a newly discovered quasar , first j074711.2 + 273904 ( hereafter first 0747 + 2739 ) , with an overdensity of absorbers in its spectrum .
it may be an example of an object whose bal component has been caught in a transitional state that could be related to orientation , time , or density .
first 0747 + 2739 was discovered in the early stages of an ongoing project to search for extremely red quasars @xcite .
candidates are found by matching point sources from the the 2-micron all - sky survey ( 2mass ; @xcite ) with first radio sources , and then further matching to an optical catalog such as the apm @xcite .
this object also has been discovered independently by @xcite . first 0747 + 2739 has @xmath6 , not particularly red , but it is undetected on the blue palomar sky survey plate , yielding @xmath7 .
the initial spectrum was obtained using the kast spectrograph on the lick observatory 3 m telescope .
a higher signal - to - noise spectrum was obtained at keck observatory using the low resolution imaging spectrograph ( lris ; @xcite ) , revealing a strikingly rich system of absorption lines redward of ly@xmath1 .
first 0747 + 2739 is located at @xmath8 ( j2000 ) .
there is a 2mass source within @xmath9 with @xmath10 , @xmath11 , and @xmath12 magnitudes of @xmath13 , @xmath14 , and @xmath15 , respectively .
the peak @xmath16 flux density as measured by first @xcite is @xmath17 ; the integrated first flux density is @xmath18 .
motivated by the lris spectrum , we obtained a high - dispersion spectrum of first 0747 + 2739 using the echelle spectrograph and imager ( esi ; @xcite ) on the keck ii telescope on the night of 2000 april 6 .
the night was photometric with @xmath19 seeing .
a 900s high - resolution spectrum was taken through a @xmath20 slit in the echellette mode of esi . in this mode , the spectral range of @xmath21 to @xmath22 is covered in 10 spectral orders with a nearly constant dispersion of @xmath23 .
the signal - to - noise of the spectrum ranges from 20 to 40 per pixel .
wavelength calibrations were performed with observations of a cuar lamp .
the spectrophotometric standard g191-b2b @xcite was observed for flux calibration .
all observations were carried out at the parallactic angle .
the data were reduced using a tailored set of iraf and idl routines developed by us specifically for esi data .
the overdensity of absorption features in the spectrum of first 0747 + 2739 is immediately apparent in the keck esi data ( figure [ fig : fig1 ] ) .
also conspicuous are a weak trough - like feature near @xmath24 and two damped ly@xmath1 systems .
the region longward of ly@xmath1 emission is plotted on an expanded scale in the lower panels of figure [ fig : fig1 ] .
we used an automated routine to search this spectral region for absorption lines with significance greater than @xmath25 .
a total of 42 absorption line systems were identified , 34 being doublets , including a bal - like trough .
the systems are given in table [ tab : tab1 ] , which lists the system identification code , the redshift , the rest equivalent width of the blue and red components of the doublet ( if detected ) , the `` ejection '' velocity of the system , and other absorption species belonging to each system .
the velocities are with respect to the quasar redshift and are given by @xmath26 , where @xmath27 .
the individual systems are merged into 18 `` poisson '' systems ( 14 of which are ) by combining absorbers within @xmath28 @xcite .
how unusual is the abundance of absorbers seen towards first 0747 + 2739 ? at high - redshift , @xcite found that ( based on 11 quasars ) the expected number of absorbers per unit redshift is approximately @xmath29 at @xmath30 for rest equivalent width larger than @xmath31 and @xmath32 .
if we sum the equivalent widths of the each of the systems within @xmath33 , and ignore those three systems with @xmath34 , we find seven combined systems with rest equivalent widths larger than @xmath31 for both lines in the doublet . the path length searched for between ly@xmath1 emission and emission
is @xmath2 .
thus , @xmath35 is @xmath36 for this line of sight an excess of 12 standard deviations from the @xcite measurement . for weak lines ,
@xcite find @xmath37 for @xmath38 and @xmath39 ( based on four quasars ) .
if we impose constraints on the @xcite data to match our more conservative @xmath25 significance limit and combine systems within @xmath40 , then the density reported by @xcite is reduced to @xmath41 . in our keck esi spectrum of first 0747 + 2739
we find , using the same criteria , at least 9 distinct high - velocity systems . over the path length
observed , that yields @xmath42 , which is a @xmath43 excess as compared to @xcite ; thus there is evidence that the overdensity of absorption may extend to weaker systems . since @xcite and @xcite found that the density of absorption lines decreases with increasing redshift ( for strong lines ) and
since the absorption systems found in first 0747 + 2739 are at higher redshifts than either the @xcite or @xcite samples , our observed density of lines may be even more significant .
what can explain the large overdensity of absorption toward first 0747 + 2739 ?
all possible explanations can be classified as intervening or intrinsic . in the intervening case ,
the absorption lines arise from material associated with galaxies along the line of sight that have no relation to the quasar . in the intrinsic case ,
the absorption is caused by high velocity ejecta from the quasar itself .
the number density of strong lines towards this quasar is much larger than is normally observed .
an examination of the spectra and tables from @xcite and @xcite along with a query of the @xcite catalog of quasar absorption lines reveals that such a concentration of absorption over such a long path length is extremely rare .
none of the 11 high - redshift quasars from @xcite has a density of absorption higher than first 0747 + 2739 .
two of the 55 lower redshift quasars from @xcite do have absorption line densities higher then first 0747 + 2739 ; q0013 - 004 ( um 224 ; @xmath44 ) and q0854 + 191 ( @xmath45 ) have six and seven strong ( @xmath46 ) `` poisson '' systems over path lengths of @xmath47 , respectively .
quasars that have relatively high densities of absorption have created much interest in the search for signs of superclustering at high redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , most of these quasars have absorption systems where the overdensity is very limited in its redshift range , such as cso 118 @xcite and the well - studied pks 0237@xmath48233 @xcite .
first 074711.2 + 273904 is different in the sense that the overdensity of absorption is spread over a large redshift range ( @xmath5 ) much too large to be explained by a single supercluster .
a search of the nasa extragalactic database ( ned ) to look for other quasars to test the superclustering hypothesis found no other known quasars within 30 arcminutes , although there are five other radio sources within @xmath49 of first 0747 + 2739 .
one possible way to reconcile the intervening hypothesis with the absorption line overdensity in first 0747 + 2739 is if the quasar is being gravitationally magnified by the intervening material .
such magnification could occur without producing multiple images , and might help to explain why a quasar with @xmath50 is so bright .
the possibility of compound microlensing of quasars by their absorption line systems is not a new concept @xcite .
this object , however , is a point source in our nirc k - band image from keck observatory , obtained in 04 seeing . finally , it is possible that the apparent overdensity of absorption systems in first 0747 + 2739 is either just chance coincidence , or is not actually all that uncommon .
the number of high - redshift quasars that are well - studied for absorption is still quite small and additional investigations may yet turn up more cases like first 0747 + 2739 .
such a finding would not necessarily be evidence for either the intrinsic or the intervening hypothesis , however .
for now , we conclude that the nature of first 0747 + 2739 is highly unusual and next to impossible to reconcile with any picture where a significant number of the absorbing systems can be attributed to intervening objects unassociated with the quasar . strong circumstantial evidence to support the intrinsic interpretation for the plethora of narrow , high - velocity absorption systems in first
0747 + 2739 is the presence of similar , almost certainly intrinsic , absorption at low velocities relative to the quasar emission line frame .
system s is typical of intrinsic absorption line systems seen in other quasars , exhibiting absorption and a small blueshift relative to the quasar s emission redshift .
system q appears to be a weak bal - like outflow ; system p is also broad , but the blue and red components of the troughs are resolved .
finally , system r , with a velocity of only @xmath51 km s@xmath52 , may also be an intrinsic absorption system .
there is ample evidence in the literature for the intrinsic interpretation of high - velocity non - bal outflows in quasars . @xcite and
@xcite found weak , high - velocity , trough - like outflows ; @xcite showed that the outflows in q 2345 + 125 are intrinsic because the absorption profiles varied in @xmath53 months in the quasar rest frame . @xcite and
@xcite give statistical evidence that some narrow , high - velocity absorption line systems may be intrinsic in nature .
finally , @xcite discuss an overdensity of absorption in cso 118 .
first 0747 + 2739 is similar to one of the quasars that comprise the so - called `` tololo pair '' ( e.g. , * ? ? ?
* ) , namely tol 1038@xmath482712 , which also has weak bal - like troughs .
this well - studied pair is a favorite for superclustering studies ( e.g. , * ? ? ?
* ) since the two quasars appear to have common absorption line systems despite the rather large ( @xmath54 ) separation between the quasars .
other authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ) still favor the hypothesis that both are balqsos and the absorption is from intrinsic high velocity outflow .
the velocity separation of the lines in the tololo pair is reminiscent of the of the double - trough bal structure found by @xcite for balqsos in general , lending credence to the idea that the absorption lines in the tololo pair are intrinsic . in any case , the similarity of tol 1038@xmath482712 and first 0747 + 2739 may help lead to an understanding of the nature of the absorption in both systems .
q1303 + 308 provides another interesting comparison ; because of its complex absorption structure , it has been classified as a balqso @xcite ; however , the absorption lines are all very narrow and do not meet the traditional bal criteria set forth largely by the same authors a few years later @xcite .
one must then wonder , how we would classify an object with a slightly lower incidence of absorption than q1303 + 308 .
such an object might simply be classified as a quasar with an overdensity of absorption line systems much like first 0747 + 2739
. the fashionable method of determining whether a particular absorption system is intrinsic is either to determine the `` coverage fraction '' of the quasar or to show that the absorption line system is variable @xcite . lacking the data to test for variability
, we have analyzed the coverage fractions for each of our systems as a function of velocity .
many of the high - velocity systems do indeed have coverage fractions that are consistent with the intrinsic hypothesis ; however , the majority of the systems are resolved into multiple components or blended , which means that the coverage fraction results are not entirely reliable .
thus , on the basis of our covering fraction analysis alone , we are not able to confirm ( or exclude ) the possibility that the overdensity of absorption lines in this quasar spectrum is due to intrinsic absorption .
regardless of the lack of conclusive evidence from variability or coverage fractions , the presence of trough - like absorption at moderate velocities and the large redshift span of the overdensity of the absorption points towards an intrinsic origin of at least some of the relatively narrow , high - velocity absorption systems .
we suggest that first 0747 + 2739 is a quasar in transition from ( to ) a quasi - bal object to ( from ) a standard quasar .
the lack of strong , broad absorption can be explained as being the result of one or more effects .
as time goes on , it may be natural for the lines to become narrow as the clouds dissipate down to their small , dense cores ; or bal features might be a superposition of many narrow features and first 0747 + 2739 has a relatively low density of them ; or perhaps the many narrow absorbers can somehow be attributed to orientation effects .
in summary , we find that that overdensity of absorption lines seen along the line of sight to first 0747 + 2739 is sufficiently high that it is likely that at least some of these systems are high - velocity , intrinsic absorption systems .
the presence of weak trough - like absorption features suggests that first 0747 + 2739 may be a transitional bal quasar that is being observed at a brief phase in its transition to ( or from ) a standard quasar , or is being viewed at a highly specific orientation , or that it is a bal with a very low absorber density .
if the absorption is intrinsic , monitoring with high resolution spectroscopy over the course of @xmath55 months may reveal variability of the absorbers , a signature of intrinsic absorption .
confirmation ( or exclusion ) of the intrinsic hypothesis for the unusually large number of high - velocity systems in first 0747 + 2739 would have interesting consequences for absorption line studies , particularly high - redshift superclustering studies .
gtr acknowledges support from nsf grant ast99 - 00703 .
rhb and mdg acknowledge support from the nsf ( grants ast-98 - 02791 and ast-98 - 02732 ) and the institute of geophysics and planetary physics ( operated under the auspices of the u.s .
department of energy by the university of california lawrence livermore national laboratory under contract no .
w-7405-eng-48 ) .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
we thank rajib ganguly and jane charlton for useful discussions .
lrccrl a1a & 1.3762 & & & 193330 & mgii + a1b & 1.3766 & & & 193300 & mgii , mgi + a2 & 1.3783 & & & 193170 & mgii , mgi + b & 2.0704 & & & 140840 & mgii , mgi , feii 2344 , + & & & & & 2374,2383,2587,2600 + c & 2.6083 & & & 100370 & feii 2344,2374,2383
+ d & 3.0741 & 0.18 & 0.11 & 66830 & + e1a & 3.1556 & 0.09 & 0.05 & 61150 & + e1b & 3.1565 & 0.14 & 0.08 & 61090 & + e1c & 3.1574 & 0.24 & 0.18 & 61030 & + e2 & 3.1585 & 0.22 & 0.17 & 60950 & + f1 & 3.1893 & 0.09 & 0.06 & 58830 & + f2 & 3.1939 & 0.21 & 0.12 & 58510 & + g1 & 3.4212 & 0.28 & 0.24 & 43140 & dla , siii , feii , alii + g2 & 3.4222 & 0.15 & 0.08 & 43070 & dla + g3 & 3.4231 & 0.09 & 0.04 & 43020 & dla , siii , feii , alii + h1 & 3.4534 & 0.06 & 0.03 & 41000 & + h2 & 3.4546 & 0.04 & 0.04 & 40920 & + ja & 3.5344 & 0.10 & 0.05 & 35680 & siiv?,alii ?
jb & 3.5353 & 0.21 & 0.14 & 35620 & siiv?,alii ?
+ k1 & 3.6248 & 0.06 & 0.03 & 29830 & + k2a & 3.6260 & 0.11 & 0.06 & 29750 & + k2b & 3.6268 & 0.04 & 0.03 & 29700 & + l & 3.7380 & 0.24 & 0.15 & 22630 & + m & 3.8428 & 0.11 & 0.05 & 16100 & + n1 & 3.8972 & & & 12750 & dla , oi , siii , cii , + & & & & & , + n2 & 3.8985 & & & 12670 & dla , oi , siii , cii , + & & & & & , + n3 & 3.8996 & ? & ? & 12610 & dla , oi , siii , cii , + & & & & & feii , alii , siiv ? + o1 & 3.9008 & 0.06 & 0.07 & 12530 &
siiv + o2 & 3.9036 & 0.20 & 0.11 & 12360 & siiv + o3 & 3.9050 & 0.14 & 0.10 & 12280 & siiv + o4 & 3.9065 & 0.15 & 0.11 & 12190 & siiv + p1 & 3.9289 & 0.06 & 0.04 & 10820 & + p2 & 3.9297 & 0.08 & 0.03 & 10770 & + p3 & 3.9306 & 0.11 & 0.05 & 10720 & + p4 & 3.9315 & 0.08 & 0.04 & 10660 & + q & 4.0449 & bal ?
& bal ? & 3850 & + r & 4.0825 & 0.06 & 0.03 & 1620 & + s1a & 4.1164 & 0.05 & 0.04 & -380 & nv + s1b & 4.1174 & 0.11 & 0.07 & -430 & nv + s1c & 4.1182 & 0.07 & 0.05 & -480 & nv + s2a & 4.1200 & 0.11 & 0.09 & -590 & nv + s2b & 4.1208 & 0.23 & 0.22 & -630 & nv + | we present a keck esi spectrum of first 074711.2 + 273904 , a @xmath0 quasar with redshift 4.11 that is detected by both first and 2mass .
the spectrum contains at least 14 independent absorption systems longward of the ly@xmath1 forest .
these systems are found over a path length of @xmath2 , constituting one of the highest densities per unit redshift of absorption ever observed .
one of the systems is trough - like and resembles a weak bal - type outflow .
two of the are `` associated '' absorption systems with @xmath3 .
of the 11 remaining systems with @xmath4 , eight are either resolved or require multiple discrete systems to fit the line profiles .
in addition to absorption , there are two low - ionization absorption systems along with two damped ly@xmath1 systems , at least one of which may be a system .
the overdensity of absorption spans a redshift range of @xmath5 .
superclusters along the line of sight are unlikely to cause an overdensity stretching over such a long redshift path , thus the absorption may be an example of narrow , high - velocity , intrinsic absorption that originates from the quasar .
we suggest that this quasar is a member of a transitional class of bal quasars where we are just barely seeing the spatial , density , or temporal edge of the bal - producing region ( or period ) ; the multiple high - velocity absorption systems may be the remnants ( or precursors ) of a stronger bal outflow .
if correct , then some simpler absorption line complexes in other quasars may also be due to outflowing rather than intervening material . |
chiral perturbation theory ( chpt ) has been proven as an extremely successful effective theory for low energy hadron dynamics .
nevertheless , the knowledge of higher order effects is rather scarse at present .
the decays @xmath0 and @xmath1 are well suited for investigation of higher order contributions . in both cases
the lowest order @xmath5 vanishes@xcite and the next order @xmath6 can precisely been calculated , since no counter - terms exist@xcite ( fig . [ fig : klpigg_feynman ] ( left ) ) . the @xmath6 prediction of the @xmath0 branching ratio is @xmath7 to an accuracy of about @xmath8@xcite . calculations of the next order @xmath2 do not exist up to now .
the measured value of @xmath9@xcite is in agreement with this prediction , however , the experimental errors were still too large to allow an accurate comparison . for @xmath1
it is known that @xmath6 alone underestimates the observed branching fraction by about a factor of three@xcite . at @xmath2
the rate can be reproduced by adding a vector meson exchange contribution@xcite ( fig .
[ fig : klpigg_feynman ] ( right ) ) via the coupling constant @xmath10@xcite .
however , the parameter @xmath10 has to be measured experimentally .
additional interest in measuring @xmath1 arises , since it can constrain the cp conserving amplitude of the direct cp violating decay @xmath11 .
the vmd mechanism could enhance the size of this amplitude , depending on the value of @xmath10@xcite .
with ktev and na48 mainly two experiments have investigated neutral kaon decays into neutral final states in the recent years .
both experiments were built to perform precise measurements on the parameter @xmath12 of direct cp violation . both are fixed target experiments with the neutral kaon beams being produced by high - energetic proton beams .
the na48 experiment , located at the sps accelerator at cern , has been taking data for the @xmath13 measurement in the years @xmath14 and 2001 .
in addition to the regular @xmath12 data taking , several runs with a high intensity @xmath15 beam have been taken , where the @xmath15 intensity has been increased by more than a factor of 200 with respect to the @xmath12 runs . in the year 2000
, the spectrometer has not been available . in this year half of the data taking took place under @xmath12 conditions while the second half of the run period was performed with a high intensity @xmath15 beam .
the ktev experiment is located at the tevatron at fermilab .
it has been taking data in the years 1996 , 1997 , and 1999 with runs dedicated to determine @xmath13 and runs for measuring rare @xmath16 and hyperon decays .
the data from the last year of data taking are mostly still being analyzed . for detecting the photons in the final states of the decays discussed here , na48 and ktev use a quasi - homogeneous liquid krypton and , respectively , a csi crystal electromagnetic calorimeter .
by assuming a kaon decay , the @xmath17 position ( along the beam pipe ) of the decay vertex can be calculated to @xmath18 with the shower energies @xmath19 , distances @xmath20 , and the nominal kaon mass @xmath21 .
if photons are lost , the missing energy shifts the vertex position down - stream towards the calorimeter .
if the decay contains one or more intermediate @xmath22 , background can be suppressed by requiring the @xmath22 decay vertex to be consistent with the kaon decay vertex .
for a decay rate measurement of @xmath0 in a fixed - target experiment an irreducible background of @xmath3 events has to be subtracted .
therefore , a precise determination of @xmath3 is necessary while the current world average on @xmath24 has a relative error of about @xmath25@xcite .
for this reason the na48 experiment has used a different method for subtracting @xmath3 events : in using the @xmath16 target run in 2000 with a very similar experimental set - up the relative rate @xmath26 is measured . by also measuring the @xmath27 rate in the high intensity @xmath15 run 2000 , the number of background @xmath3 events can then be accurately subtracted .
the na48 @xmath16 target run of the year 2000 has provided very clean conditions to measure @xmath3 decays .
backgrounds from @xmath28 or other @xmath16 decays are completely negligible , in particular @xmath29 dalitz decays are swept out by the spectrometer magnet .
the only remaining background source is hadronic activity , e.g. from @xmath30 decay products , which in rare cases might enter the decay volume via the @xmath16 beam line . to estimate this background the sidebands in the shower width and center - of - gravity distributions
are evaluated ( fig .
[ fig : klgg_bkg ] ) . by using both methods ,
the hadronic background is determined to @xmath31 , where the error reflects the difference of the two estimations .
the normalization channel @xmath27 , which needs three good @xmath32 combinations to be selected , is virtually background - free .
both , signal and normalization channel have trigger efficiencies larger than @xmath33 . for the analyis only @xmath34 of the @xmath16 target run data
were used which already provides sufficient statistics .
evaluating the event numbers in the vertex region @xmath35 m @xmath36 m ( measured from the @xmath15 collimator ) and taking the acceptance ratio from monte carlo simulation , na48 finds @xmath37 the systematic error is dominated by uncertainties in the @xmath38 acceptance ratio .
this result improves the current world average by about a factor of four .
in addition to the irreducible @xmath3 decays more background sources have to be taken into account when selecting @xmath0 candidates : @xmath39 with lost and/or overlapping photons may fake a good @xmath40 event . since in these cases energy
is lost , the neutral vertex is shifted down - stream .
background from @xmath39 can therefore be efficiently rejected by restricting the allowed vertex range to be within @xmath35 and 5 m behind the final collimator .
doing so , the remaining background from @xmath39 is estimated to @xmath41 , where the uncertainty is arising from the shower overlap probabilities being different in the simulation as in the data .
further background sources are hadronic events ( originating e.g. from scattering at the collimator ) or from accidentally overlapping events . both are determined by a sideband subtraction in the center - of - gravity distribution of the lkr calorimeter ( fig .
[ fig : ksgg_bkg ] ) . from this , the total background from hadronic and accidental activity is estimated to @xmath42 .
finally , dalitz decays @xmath43 and @xmath44 have to be taken into account . in the high intensity
@xmath15 running in 2000 , the na48 experiment had no magnetic field in the detector .
therefore , many dalitz @xmath45 pairs did not separate and were overlapping in the calorimeter . from monte carlo simulation ,
the probability of a dalitz decay misidentified as @xmath40 pair was estimated to about @xmath46 .
with dalitz decay probabilities assumed to be equal for @xmath47 and @xmath32 , the effect is twice as big for the @xmath39 normalization than in the @xmath47 events and results in a relative correction to the @xmath0 branching ratio of @xmath48 . the decay vertex distribution of the na48 @xmath40 candidates is shown in fig .
[ fig : ksgg_zvertex ] together with the estimated contributions of @xmath0 , @xmath3 and @xmath39 background .
the @xmath39 background contribution has been normalized to the absolute @xmath15 flux . in the fiducial region between @xmath35 and 5 m about 20000 @xmath40 candidates
were found . subtracting all background contributions and normalizing to fully reconstructed @xmath39 events , the branching fraction was determined to @xmath49 the systematic uncertainty is dominated by the branching fraction of the @xmath39 normalization ( @xmath50 ) , the estimation of the hadronic and accidental background ( @xmath51 ) and the monte carlo statistics ( @xmath52 ) .
this new result significantly improves the previous measurements and is in clear discrepancy with the @xmath6 based theoretical predictions .
with four photons in the final state the decay @xmath1 has almost the same signature as the cp - violating decay @xmath28 , which is one of the signal decays for the @xmath13 measurement of the ktev and na48 experiments .
therefore the decay @xmath1 can be taken in parallel with regular @xmath12 data taking . moreover ,
both signal and normalization ( @xmath28 ) have practically identical trigger efficiencies .
however , the analysis of @xmath1 events has to fight strong backgrounds , in particular @xmath27 events with lost and/or overlapping photons , badly reconstructed @xmath28 events , and accidentally overlapping events .
the na48 analysis@xcite , which is described more detailed in the following , is based on the na48 data of the years 1998 and 1999 .
the suppression of background from @xmath27 with overlapping or lost photons is done by cutting on the shower width in the calorimeter . as shown in fig .
[ fig : klpigg_bkg ] ( left ) , overlapping photons produce on average a much wider shower and can clearly be separated from the signal . to suppress the @xmath27 background even further ,
a variable @xmath53 is constructed as estimate of the real kaon vertex under the assumption of a misidentified @xmath27 event . for real background events
this variable should be in the physical region down - stream of the collimator ( fig .
[ fig : klpigg_bkg ] ( right ) ) . by rejecting events with @xmath54 m
, the @xmath27 background is reduced to @xmath55 .
however , the signal acceptance is also reduced by @xmath56 .
a second source of background are misidentified @xmath28 events .
they are rejected by requiring the invariant mass @xmath57 of the photons not coming from the @xmath22 to be outside a window of @xmath58 mev/@xmath59 around the nominal @xmath22 mass . by using @xmath39 events from both @xmath12 and high intensity @xmath15 runs , the remaining background from @xmath28
is estimated to @xmath60 .
finally , good @xmath61 candidates may be mimicked by accidentally overlapping events .
this background is estimated by extrapolating events with high transverse momentum into the signal region to @xmath62 .
the invariant mass @xmath63 of the two photons originating from the @xmath22 is shown in fig .
[ fig : klpigg_m12_m34_y ] ( left ) . na48 finds 2558 @xmath1 candidates with an estimated background of @xmath64 events , which are used for the branching ratio measurement . for the fit of the vmd parameter @xmath10 , 345 ambigous events which allow
two possible @xmath32 assignments are excluded .
the parameter @xmath10 is fitted using the distributions of both kinematic decay variables @xmath57 and @xmath65 ( fig .
[ fig : klpigg_m12_m34_y ] ) .
the result of the fit is @xmath66 with the systematic error being dominated by uncertainties of the acceptance evaluation ( @xmath67 ) and the parametrization of the @xmath27 vertex@xcite ( @xmath68 ) . for low @xmath57 the distributions of the @xmath22 signal @xmath63 are shown in fig .
[ fig : klpigg_av ] .
there is no significant signal of @xmath1 events for @xmath69 mev/@xmath59 , as expected for @xmath70 due to cancellation effects .
nevertheless , in the region @xmath71
mev/@xmath59 a clear signal shows up , which is evidence for a sizeable @xmath2 contribution to the @xmath1 amplitude .
however , an analysis of the ktev experiment using the data of the years 1996 and 1997 , comes to a different result@xcite . performing a similar analysis
as described above , ktev finds 884 signal candidates with estimated @xmath72 background events .
fitting both the @xmath57 and @xmath73 distributions ( fig .
[ fig : klpigg_ktev_m34_y ] ) the measured value of @xmath10 is @xmath74 also measuring a large @xmath2 contribution , but about @xmath75 standard deviations different from the na48 result .
the systematics are dominated by the knowledge of the @xmath76 background . while na48 has finished the analysis on @xmath1 , the ktev collaboration plans to additionally analyze the data from the 1999 data taking period , which might help to understand the different results of the two experiments .
the two experiments also arrive at different measurements of the @xmath1 branching fraction , which can partially be explained by the strong dependence of the detector acceptances on @xmath57 : @xmath77 { \rm ktev } \ ; ( 1999 ) : \quad { { \rm br}}({{k_{{\mbox{\tiny \rm l}}}}\to
\pi^0 \gamma \gamma } ) & = & ( 1.68 \ , \pm \ , 0.07_{{\mbox{\tiny \rm stat } } } \ , \pm \ , 0.08_{{\mbox{\tiny \rm syst } } } ) \times 10^{-6}\end{aligned}\ ] ] the na48 @xmath10 measurement indicates a negligible cp - conserving amplitude of the ( yet to be observed ) direct cp - violating decay @xmath11@xcite .
however , this does not hold for the @xmath10 value measured by ktev .
the decay @xmath4 is dominated by the pion pole in @xmath39.to be able to distinguish this decay from the dominating @xmath39 , a minimum two - gamma invariant mass of @xmath78 is required .
one theoretical investigation based on chpt exists@xcite , predicting the shape of the @xmath57 distribution together with a branching fraction of @xmath79 .
no experimental observation or limit has been published so far . using the data of a two - day high intensity @xmath15 run in 1999
, the na48 experiment has performed a search for the decay @xmath4 . in this run about @xmath80 @xmath15 decays took place in the fiducial detector volume . because of the smallness of the expected branching fraction , the main experimental problem is the background suppression .
any non-@xmath81 activity in the photon anti - counters , drift chambers , and hadron calorimeter is vetoed . to suppress @xmath39 events , the @xmath82 of the event for being a @xmath39 decay is required to be larger than 2000 . in this way ,
only @xmath83 mismeasured @xmath39 events are expected , as determined from monte carlo simulation .
the main background then comes from @xmath39 decays with one lost and one accidentally in - time photon .
this background is estimated by simulated and randomly triggered events to @xmath84 events .
the irreducible background coming from @xmath1 decays accounts for another @xmath85 expected background event . the @xmath82 distribution of the remaining events is shown in fig .
[ fig : kspigg_chi2 ] . only two events above @xmath86 survive the selection , compatible with the background expectation of @xmath87 events . using @xmath39 decays for normalization , the limit on the branching fraction set by na48 is @xmath88 this is the first limit set on the decay width of @xmath4 .
however , it is still about one order of magnitude above the theoretical expectation .
barr _ et al _
( na31 collab . ) , phys .
* b 242 * ( 1990 ) 523 ; v. papadimitriou _ et al _ ( e731 collab . ) , phys . rev . * d 284 * ( 1991 ) 573 ; g.d .
barr _ et al _
( na31 collab . ) , phys .
* b 284 * ( 1992 ) 440 .
cohen , g. ecker , and a. pich , phys .
* b 304 * ( 1993 ) 417 ; l. cappiello , g. dambrosio , and m. miragliuolo , phys .
* b 298 * ( 1993 ) 423 ; j. kambor and b. holstein , phys .
* d 49 * ( 1994 ) 2346 . | the neutral kaon decays @xmath0 and @xmath1 are very sensitive to higher order loop effects of chiral perturbation theory ( chpt ) . new measurements of the na48 experiment show that chpt contributions of @xmath2 can not be neglected in these modes .
in addition a new measurement of the related decay @xmath3 and an upper limit of the rate of @xmath4 are presented .
= 11.6pt |
unlike electromagnetic radiation from astrophysical sources , distant magnetic fields are difficult to detect . nonetheless , recent measurements have begun to reveal that such fields exist at significant strengths , and on surprisingly large scales , in the extragalactic universe .
observations have shown that magnetic fields are ubiquitous in cluster atmospheres , playing a critical role in the cluster formation and evolution , in determining the energy balance in cluster gas through their effect on heat conduction , and in some cases , perhaps even becoming dynamically important .
our knowledge of the magnetic field properties in galaxy clusters and of how they relate to other cluster properties is limited by the sensitivity and resolution of current instruments .
in particular , several questions are still unanswered : are the fields filamentary , what are the coherence scales , to what extent do the thermal and non - thermal plasmas mix in cluster atmospheres , what is the radial dependence of the field strength , how does the field depend on cluster parameters such as the gas temperature , metallicity , mass , substructure and density profile , how do the fields evolve with cosmic time , how do the fields extend and finally how were these fields generated ? magnetic fields associated with the intracluster medium ( icm ) in galaxy clusters are investigated in the radio band through studies of the rotation measure and of the synchrotron emission of both individual radio galaxies and radio halos and relics .
other techniques include x - ray studies of the inverse compton emission and of cold fronts , and hydrodynamic simulations .
the studies in the radio band are , however , the most relevant and provide the most accurate field estimates . the ska s high sensitivity , high resolution , multifrequency capability , and polarization purity will be crucial to these studies .
cluster surveys of the faraday rotation measures of polarized radio sources both within and behind clusters provide an important probe of the existence of intracluster magnetic fields , as the radio waves traversing the magnetized intracluster medium show depolarization and rotation of the polarization position angle as a function of wavelength . ]
the rm values derived from multifrequency polarimetric observations of background or embedded cluster sources are of the order of tens to thousands rad m@xmath0 .
these have to be combined with measurements of the thermal gas density , @xmath1 , to estimate the cluster magnetic field along the line of sight .
the observing strategy to derive information on the magnetic field intensity and structure is twofold : i ) obtain the average value of the rm of sources located at different impact parameters of the cluster , ii ) derive maps of the rm of extended radio sources , to evaluate the @xmath2 of the rm distribution .
part of the difficulty of investigating cluster magnetic fields through faraday rotation is that at present such a study may only be undertaken statistically over a large number of clusters .
this is due , in part , to the lack of available rms which are limited by current instrument sensitivity and by the vectorial nature of the rms themselves . since every measured rm is the vectorial addition of all contributing faraday screens along the line of sight it is impossible to disentangle the cluster rotation measure components from either internal rotation in the source , or a galactic rotation measure component without a sufficient number of rms available .
nevertheless , studies on both statistical samples and individual clusters have been carried out ( see the review @xcite and references therein ) .
kim et al.@xcite analyzed the rm of 53 radio sources in and behind clusters and 99 sources in a control sample .
this study contains the largest cluster sample to date .
it demonstrated that @xmath3 g level fields are widespread in the icm , regardless of whether there is the direct evidence for the existence of magnetic fields from the presence of diffuse radio emission . in a more recent statistical study ,
clarke et al.@xcite analyzed the rms for a representative sample of 27 cluster sources , plus a control sample , and found a statistically significant broadening of the rm distribution in the cluster sample , and a clear increase in the width of the rm distribution toward smaller impact parameter ( see fig .
[ rmstat ] ) .
they derived that the icm is permeated with a high filling factor by magnetic fields at levels of 4 - 8 @xmath3 g and with a correlation length of @xmath415 kpc , up to @xmath40.75 mpc from the cluster center .
the first detailed studies of rm have been performed on cooling core clusters , owing to the extremely high rms of the powerful radio galaxies at their centers ( e.g. , hydra a @xcite & 3c295 @xcite ) .
high values of the magnetic fields , up to tens of @xmath3 g , have been obtained , but they only refer to the innermost cluster regions . studies on larger areas of clusters have been carried out e.g. for coma @xcite , a119 @xcite , a514 @xcite and 3c129 @xcite . however , because of the limited sensitivity of current instruments , reliable maps of the rm can be obtained only for the strongest sources ( total flux 50 mjy ) and only in the regions of high radio surface brightness ( see e.g. fig .
[ rmsource ] ) .
the number of targeted sources per cluster is on average 1 - 2 , with a maximum of 3 - 5 in a few clusters . ]
overall , the data are consistent with cluster atmospheres containing @xmath3 g fields , with perhaps an order of magnitude scatter in field strength between clusters , or within a given cluster , and with extreme field values in cluster cooling cores .
the rm distribution is generally patchy , indicating that large scale magnetic fields are not regularly ordered on cluster scales , but have coherence scales between 1 and 10 kpc .
the estimates of the magnetic field strength crucially depend on the magnetic field structure and geometry , thus accurate maps of rm are needed .
we note that in the coma cluster the presence of a weaker field component of 0.1 - 0.2 @xmath3 g , ordered on a scale of about one cluster core radius , was inferred in addition to the stronger tangled magnetic field component @xcite .
a caveat in the interpretation of rm data is the possible existence of local rm enhancements , produced by the compression of the icm fields by a radio galaxy . in this case
the rm would not be indicative of the cluster magnetic field , leading to overestimates of the icm magnetic field strength @xcite .
this difficulty can be overcome by sensitive observations at very high resolution , which can not be obtained with the presently available instruments .
the observations are often interpreted in terms of the simplest possible model , i.e. in this case a constant field throughout the whole cluster . however , a decline with radius is expected if the intensity of the magnetic field results from the compression of the thermal plasma during the cluster gravitational collapse . according to this model , a correlation between observable parameters , the rm and the x - ray surface brightness ,
is expected to reflect the correlation between the physical quantities , magnetic field and gas density .
the application of this approach has been possible so far only in a119 , giving the radial profile of the magnetic field as @xmath5 @xcite in this cluster .
in addition , beck et al.@xcite pointed out that field estimates derived from rm may be too large in the case of a turbulent medium where small - scale fluctuations in the magnetic field and the electron density are highly correlated .
new generation instruments are rather promising and establish a clear connection between radio astronomical techniques and the improvement in the knowledge of the x - ray sky .
there are various satellite missions which will map the x - ray sky at low energies in the next years
. these will provide a more precise knowledge of the x - ray surface brightness of clusters , i.e. of their thermal gas density , allowing a more accurate and correct interpretation of the sensitive rm measurements .
the accurate experimental determination of large scale magnetic fields in the intracluster medium will thus be possible .
the presence of magnetic fields in clusters is directly demonstrated by the existence of the radio halos and relics , i.e. diffuse cluster - wide synchrotron radio sources , as revealed in coma ( fig .
[ comaclus ] ) and some other clusters @xcite . under the assumption that the energy density within radio sources is minimum ( equipartition condition ) , magnetic field values in the range 0.1 - 1 @xmath3 g are derived for the radio emitting regions , i.e. on scales as large as @xmath6 mpc .
these calculations typically assume equal energy in relativistic protons and electrons , a magnetic field entirely filling the radio source volume , a low frequency cut - off of 10 mhz , and a high frequency cut - off of 10 ghz .
the magnetic field values derived in this way are consistent with those suggested from the recent detection of inverse compton hard x - ray emission in clusters with halos or relics @xcite .
the number of clusters presently known to host halos and relics is around 50 , i.e. @xmath4 10% of rich clusters .
indeed , the typically low surface brightness of cluster radio halos and their steep spectrum makes it difficult to image them accurately with the current resources .
further , at lower resolution , where beam averaging enhances the detectability of extended radio emission , true diffuse emission is sometimes difficult to distinguish from a blend of weak , discrete radio sources . the observations of clusters with the ska will allow a dramatic improvement of the knowledge of halos and relics . it will be possible to detect new halos and relics , and study these sources in great detail ( see detailed discussion feretti , burigana , & ensslin , this volume ) .
in particular , polarimetric studies will be of crucial importance , to give direct information on the magnetic field orientation , degree of ordering , and overall structure .
@xmath7 125@xmath8 ( hpbw , ra @xmath7 dec ) , showing the halo source coma c at the cluster center and the relic source 1253 + 275 at the cluster periphery .
[ comaclus ] ] the detection of synchrotron radiation at the lowest possible levels will allow the measurement of magnetic fields in even more rarefied regions of the intergalactic space , and investigations of the relation between the formation of magnetic fields and the formation of the large scale structure in the universe . [
cols="<,^,^,^",options="header " , ] [ tab : est ]
the cluster magnetic field values obtained from rm arguments are about an order of magnitude higher than the estimates , typically of 0.2 to 1 @xmath3 g , derived from both the synchrotron diffuse radio emission @xcite and the inverse compton hard x - ray emission @xcite .
the discrepancy can be alleviated by taking into account that i ) the values deduced from radio synchrotron emission and from inverse compton refer to averages over large volumes , whereas the rm estimates give a weighted average of the field along the line of sight ; ii ) the magnetic field intensity is likely to decline with the distance from the cluster center ; iii ) the magnetic field may show complex structure , as filamentation and/or substructure with a range of coherence scales . therefore , the rm data should be interpreted using realistic models of the cluster magnetic fields , as shown by a recent investigation performed using a numerical approach @xcite . additionally , evidence suggests that the magnetic field strength will vary depending on the dynamical history and location within the cluster .
a striking example of the variation of magnetic field strength estimates for various methods and in various locations throughout the cluster is given in table 1 for the post merger cluster a3667 ( at z = 0.055 ) .
the results of each estimation , are consistent with a typical 12 @xmath3 g field , tangled on scales of 10 to 100 kpc , pervading the cluster s central region .
this field has been further enhanced in the region of the observed central x - ray cold front to a level of 716 @xmath3 g @xcite and to around 35 @xmath3 g in the region of the mpc - scaled radio relic in the northern part of the cluster @xcite . as the relic emission is currently thought to be the result of a cluster merger it is likely that the central field would be compressed and elevated by a factor of 34 in the region of the shock accelerated relic . as shown in table 1 the observational data needed to obtain sufficiently detailed information about the cluster magnetic field strength and structure
can only be achieved by a new generation instrument .
although several interesting results have been obtained in recent years about the cluster magnetic fields , such studies are still limited to works on a few clusters , and on a few radio sources , as reported in the previous sections .
magnetic field strengths of the order of @xmath4 1 @xmath3 g are found to be common in clusters . however , estimates obtained with different approaches may differ ( sec .
4 ) , thus detailed information on the cluster field is still needed .
@xmath9 distinguish the external faraday rotation measure from that arising internally in a radio source , in order to get careful information on the cluster magnetic field ( and on the radio source too ) ; @xmath9 a frequency range 1 - 10 ghz with a large number of channels@xcite , to get reliable information on the rm , solving the ambiguity related to its computation and disentangling various contributions ; the all sky rm survey ( ska key project on magnetic fields @xcite ) , will provide the rm for @xmath10 compact polarized extragalactic sources , expanding the sample of rm measurements by five orders of magnitude over current data sets . when combined with redshift and x - ray information from future instruments , such a data - base will allow a statistical analysis of the rm of sources within or behind clusters , leading to a great improvement of the knowledge of the strength of the magnetic field in clusters .
for example in the case of a3667 illustrated above only 3 reliable rms were obtained after over 100 hours of observations on present day instruments . with the all sky rm survey , it will be possible to obtain about 1000 rm measurements through lines of sight in this cluster , and generally in clusters at similar distances .
additionally , we will detect at least 20 rms through clusters up to redshifts of 3.5 giving the first opportunity to perform detailed analysis of the evolution of cluster magnetic fields .
deep multifrequency surveys , targeting individual clusters , will allow the investigation of the intensity and structure of the icm magnetic fields .
low surface brightness radio features such as relics and halos should be detected in their thousands @xcite allowing us to explore the role of dynamically important magnetic fields in merging clusters and providing vital clues to the origin of cluster magnetic fields . using the described techniques , magnetic fields
can also be observed and studied in the jets and lobes of radio galaxies . the largely improved sensitivity and resolution of ska will allow the study of faint objects and of the low brightness components of extended radio sources , the distinction of local features , the discrimination between internal and external faraday dispersion , the connection between magnetic fields within the radio galaxies and the cosmological magnetic fields . | an important area of study of cosmic magnetic fields is on the largest scales , those of clusters of galaxies . in the last decade
it has become clear that the intra - cluster medium ( icm ) in clusters of galaxies is magnetized and that magnetic fields play a critical role in the cluster formation and evolution .
the observational evidence for the existence of cluster magnetic fields is obtained by the diffuse cluster - wide synchrotron radio emission and from rotation measure ( rm ) studies of extragalactic radio sources located within or behind the clusters . a significant breakthrough in the knowledge of the cluster magnetic fields
will be reached through the ska , owing to its capabilities , in particular the deep sensitivity and the polarization purity . |
at epochs corresponding to @xmath1 the intergalactic medium is expected to recombine and remain neutral until sources of radiation and heat develop that are capable of reionizing it .
the detection of transmitted flux shortward of the wavelength in the spectra of sources at @xmath2 implies that the hydrogen component of this igm was ionized at even higher redshifts .
it appears then that substantial sources of ultraviolet photons and mechanical energy were already present when the universe was less than 6% of its current age , perhaps quasars and/or young star - forming galaxies : an episode of pregalactic star formation may provide a possible explanation for the widespread existence of heavy elements ( like carbon , oxygen , and silicon ) in the igm , while the integrated radiation emitted from quasars is likely responsible for the reionization of intergalactic helium at later times .
popular cosmological models predict that some time beyond a redshift of 15 the gas within halos with virial temperatures @xmath3k [ or , equivalently , with masses @xmath4 comparable to present - day dwarf ellipticals ] cooled rapidly due to the excitation of hydrogen ly@xmath5by the maxwellian tail of the electron distribution , and fragmented .
massive stars formed with some initial mass function ( imf ) , synthesized heavy elements , and exploded as type ii supernovae ( sne ) after a few @xmath6yr , enriching and heating the surrounding medium . whilst collisional excitation of molecular hydrogen may have allowed the gas in even smaller systems [ virial temperatures of only a few hundred k , corresponding to masses around @xmath7 to cool and form stars at earlier times ( abel , ciardi , and ferrara , this volume ) , h@xmath8 molecules are efficiently photo - dissociated by stellar uv radiation , and such negative ` feedback ' is likely to have suppressed molecular cooling and further star formation inside very small halos .
one should note that while numerical n - body@xmath9hydrodynamical simulations have convincingly shown that the igm is expected to fragment into structures at early times in cdm cosmogonies ( e.g. cen et al .
1994 ; zhang et al .
1995 ) , the same simulations are much less able to predict the efficiency with which the first gravitationally collapsed objects lit up the universe at the end of the ` dark age ' .
the reionization scenario that has been the subject of the most theoretical study is one in which intergalactic hydrogen is photoionized by the uv radiation emitted either by quasars or by stars with masses @xmath10 , rather than ionized by collisions with electrons heated up by , e.g. , supernova - driven winds from early pregalactic objects . in the former case a high degree of ionization requires about @xmath11ev per hydrogen atom , where @xmath12 is the volume - averaged hydrogen recombination timescale , the ratio @xmath13 being much greater than unity already at @xmath14 ( haiman and gnedin , this volume ) .
collisional ionization to a neutral fraction of only few parts in @xmath15 requires a comparable energy input , i.e. an igm temperature close to @xmath16k or about @xmath17ev per atom .
massive stars will deposit both radiative and mechanical energy into the interstellar medium of subgalactic fragments .
a complex network of ` feedback ' mechanisms is likely at work in these systems , as the gas in shallow potential is more easily blown away thereby quenching further star formation , and the blastwaves produced by supernova explosions reheat the surrounding intergalactic gas and enrich it with newly formed heavy elements and dust .
it is therefore unclear at this stage whether an early input of mechanical energy will actually play a major role in determining the thermal and ionization state of the igm on large scales ( tegmark et al .
what can be easily shown is that , during the evolution of a a ` typical ' stellar population , more energy is lost in ultraviolet radiation than in mechanical form .
this is because in nuclear burning from zero to solar metallicity ( @xmath18 ) , the energy radiated per baryon is @xmath19 ; about one third of it goes into h - ionizing photons .
the same massive stars that dominate the uv light also explode as sne , returning most of the metals to the interstellar medium and injecting about @xmath20ergs per event in kinetic energy . for a salpeter imf
, one has about one sn every @xmath21 of baryons that forms stars .
the mass fraction in mechanical energy is then approximately @xmath22 , ten times lower than the fraction released in photons above 1 ryd . the relative importance of photoionization versus shock ionization will depend , however , on the efficiency with which radiation and mechanical energy actually escape into the igm .
consider , in particular , the case of an early generation of halos with circular speed @xmath23 , corresponding in top - hat spherical collapse to a virial temperature @xmath24k and halo mass @xmath25^{-3/2 } h^{-1}\ , \msun$ ] . in these systems
rapid cooling by atomic hydrogen can take place and a significant fraction , @xmath26 , of their total mass may be converted into stars over a dynamical timescale ( here @xmath27 is the baryon density parameter ) . in a flat cosmology with @xmath28 , @xmath29 , and
rms mass fluctuation normalized at present to @xmath30 on spheres of @xmath31mpc , halos with @xmath32 would be collapsing at @xmath33 from 2 @xmath34 fluctuations . at this epoch
more massive halos with @xmath35 , while able to cool rapidly , would be collapsing from 3 @xmath34 peaks and be too rare to produce significant amounts of heavy elements and uv radiation ( in a gaussian theory the 3 @xmath34 peaks contain only 5% as much mass as the 2 @xmath34 peaks at a given epoch ) , unless they were able somehow to form stars more efficiently than lower - mass objects .
halos from 1 @xmath34 fluctuations would be more numerous and contain most of the mass , but with virial temperatures of only a few hundred degrees they would most likely be unable to cool via h@xmath8 before reionization actually occurs at @xmath36 . for @xmath37 , @xmath38 , and @xmath29 , the explosive output of @xmath39 sne would inject a total energy @xmath40ergs .
correlated multi - sn explosions will create large holes in the ism of these pregalactic halos , perhaps enlarging by far pre - existing ones due to winds from their progenitors stars .
the energy from sne will impart enough momentum to the interstellar medium that large portions of it will become unbound and leave the parent halo , taking the metal - enriched stellar debris along ( maclow & ferrara 1999 ) .
a large fraction perhaps as much as 90% of the energy injected in these sn - driven bubbles could be lost to radiation in the halo before the bubbles expand out into the surrounding igm ( madau et al .
eventually the hot gas will escape its host , shock the igm , and form a cosmological blast wave .
if the explosion occurs at cosmic time @xmath41yr ( corresponding in the adopted cosmology to @xmath33 ) , at time @xmath42 after the event we can use the standard sedov - taylor self - similar ( adiabatic ) solution to give a rough estimate of the proper radius of the shock , @xmath43^{1/5}t^{2/5}\delta t^{2/5 } \approx 10\,{\rm kpc}. \eqno(1)\ ] ] at this instant the shock velocity relative to the hubble flow is @xmath44 lower than the escape velocity from the halo .
the gas temperature just behind the shock front is @xmath45k , more than enough to efficiently ionize all the incoming hydrogen . at these redshifts , it is the onset of compton cooling off cosmic microwave background photons that ends the adiabatic stage of blast wave propagation . according to the press - schechter formalism ,
the comoving abundance of collapsed halos with mass @xmath46 at @xmath33 is @xmath47mpc@xmath48 , corresponding to a mean proper distance between neighboring halos of @xmath49kpc , and to a total mass density parameter of order @xmath50 . with the assumed star formation efficiency only a small fraction ,
about one percent , of the stars seen today would form at these early epochs .
still , our simplistic analysis ( see madau et al .
2000 for a more detailed treatment ) conveys the interesting idea that the blast waves from such a population of pregalactic objects may actually fill a significant fraction of the hubble volume , and drive the intergalalactic medium at early epochs to a higher adiabat , @xmath51k , than expected from photoionization , so as to inhibit the formation of further protogalaxies by raising the jeans mass .
this effect could perhaps be responsible for providing a global negative feedback to self - regulate the early stellar birthrate , effectively causing a pause in the cosmic history of star formation .
after this epoch of pregalactic outflows , the igm would be polluted to a mean metallicity @xmath52 ( madau et al .
. a lower density of sources which would therefore have to originate from higher amplitude peaks would suffice if the typical efficiency of star formation were larger than assumed here .
in this section we will focus our attention to the photoionization of the igm , i.e. we will assume that uv photons from an early generation of stars and/or quasars are the main source of energy for the reionization and reheating of the universe , and that star formation and quasar activity occurs in collapsed galaxy halos .
the process then begins as individual sources start to generate expanding regions in the surrounding igm ; throughout an region , h is ionized and he is either singly or doubly ionized . as more and more sources of ultraviolet radiation switch on , the ionized volume grows in size while the neutral phase shrinks .
reionization is completed when the regions overlap , and every point in the intergalactic space gets exposed for the first time to a nearly uniform lyman - continuum ( lyc ) background . in the presence of a population of ionizing sources , the transition from a neutral igm to one
that is almost fully ionized can be statistically described by the evolution with redshift of the volume filling factor ( or porosity ) @xmath53 of , , and regions .
the radiation emitted by spatially clustered stellar - like and quasar - like sources the number densities and luminosities of which may change rapidly as a function of redshift coupled with absorption processes in a medium that becomes more and more clumpy owing to the non - linear collapse of structures ( figure 1 ) , all determine the complex topology of neutral and ionized zones in the universe ( gnedin , this volume ; ciardi et al .
2000 ; abel et al .
when @xmath54 and the radiation sources are randomly distributed , the ionized regions are spatially isolated , every uv photon is absorbed somewhere in the igm , and the uv radiation field is highly inhomogeneous .
as @xmath53 grows , the crossing of ionization fronts becomes more and more common , until percolation occurs at @xmath55 .
since the mean free path of lyc radiation is always much smaller than the horizon ( this is also true after ` overlapping ' because of the residual still present in the forest clouds and the lyman - limit systems ) , the filling factor of cosmological regions is equal at any given time @xmath56 to the total number of ionizing photons emitted per hydrogen atom by all radiation sources present at earlier epochs , @xmath57 , minus the total number of radiative recombinations per atom , @xmath58 .
this statement reflects the simple fact that every ultraviolet photon that is emitted is either absorbed by a newly ionized hydrogen atom or by a recombining one .
differentiating one gets @xmath59 it is this differential equation and its equivalent for expanding helium zones that statistically describes the transition from a neutral universe to a fully ionized one ( madau et al .
initially , when the filling factor is @xmath60 , recombinations can be neglected and the ionized volume increases at a rate fixed solely by the ratio @xmath61 . as time goes on and more and more lyc photons
are emitted , radiative recombinations become important and slow down the growth of the ionized volume , until @xmath53 reaches unity , the recombination term saturates , and reionization is finally completed ( except for the high density regions far from any source which are only gradually eaten away , miralda - escud , haehnelt , & rees 2000 ) . in the limit of a fast recombining igm ( @xmath62 ) , one can neglect the derivative on the left - hand side of equation ( 3 ) and derive @xmath63 i.e. the volume filling factor of ionized bubbles must be less ( or equal ) to the number of lyc photons emitted per hydrogen atom in one recombination time . in other words , because of radiative recombinations , only a fraction @xmath64 of the photons emitted above 1 ryd is actually used to ionize new igm material .
the universe is completely reionized when @xmath65 i.e. when the emission rate of ultraviolet photons exceeds the mean rate of recombinations .
the simplest way to treat reionization in a inhomogeneous medium is in terms of a clumping factor that increases the effective gas recombination rate . in this case
the volume - averaged recombination time is @xmath66^{-1}=0.06\ , { \rm gyr } \left(\frac{\omega_b h^2}{0.019}\right)^{-1}\left(\frac{1+z}{10}\right)^{-3 } \frac{\overline n_\nh}{\overline n_p}~c_{10}^{-1 } , \eqno(6)\ ] ] where @xmath67 is the recombination coefficient to the excited states of hydrogen ( at an assumed gas temperature of @xmath68k ) , @xmath69 the helium to hydrogen abundance ratio , and the factor @xmath70 takes into account the degree of clumpiness of photoionized regions ( hereafter @xmath71 ) .
if ionized gas with density @xmath72 filled uniformly a fraction @xmath73 of the available volume , the rest being empty space , the mean square density would be @xmath74 .
more in general , if @xmath75 is the fraction of baryonic mass in photoionized gas at an overdensity @xmath76 relative to the mean , and the remaining ( underdense ) medium is distributed uniformly , then the fractional volume occupied by the denser component is @xmath77 the density of the diffuse component is @xmath78 and the recombination rate is larger than that of a homogeneous universe by the factor @xmath79 it is difficult to estimate the clumping factor accurately .
according to hydrodynamics simulations of structure formation in the igm ( within the framework of cdm - dominated cosmologies ) , ly@xmath5forest clouds with moderate overdensities , @xmath80 , occupy a fraction of the available volume which is too small for them to dominate the clumping at high redshifts ( e.g. theuns et al .
1998 ) . in hierarchical clustering models , it is the virialized gas with @xmath81 in the earliest non - linear objects with @xmath82k ( here @xmath83 is the virial temperature corresponding to the cosmological jean mass ) which will boost the recombination rate by large factors as soon as the collapsed mass fraction exceeds @xmath84% ( haiman , this volume ) .
the importance of photoevaporating minihalos as sinks of ionizing photons during cosmological reionization has been recently discussed by haiman et al . (
note that such minihalos are not yet resolved in large - scale three - dimensional cosmological simulations .
halos or halo cores which are dense and thick enough to be self - shielded from uv radiation will stay neutral and will not contribute to the recombination rate .
this is also true of gas in more massive halos , which will be virialized to higher temperatures and ionized by collisions with thermal electrons .
prior to complete reionization , sources of ultraviolet radiation will be seen behind a large column of intervening gas that is still neutral . in this case , because of scattering off the line - of - sight due to the diffuse neutral igm , the spectrum of a source at @xmath85 should show the red damping wing of the gunn - peterson absorption trough at wavelengths longer than the local ly@xmath5resonance , @xmath86 , where @xmath87a . at @xmath88 ,
this characteristic feature extends for more than @xmath89 to the red of the resonance , and may significantly suppress the ly@xmath5 emission line in the spectra of the first generation of objects in the universe . measuring the shape of the absorption profile of the damping wing could provide a determination of the density of the neutral igm near the source ( miralda - escud 1998 ) .
cen & haiman ( 2000 ) and madau & rees ( 2000 ) have recently focused on the width of the red damping wing
related to the expected strength of the ly@xmath5 emission line in the spectra of very distant qsos as a flag of the observation of the igm before reionization .
they have assessed , in particular , the impact of the photoionized , mpc - size regions which will surround individual luminous sources of lyc radiation on the transmission of photons redward of the ly@xmath5resonance , and shown that the damping wing of the gunn - peterson trough may nearly completely disappear because of the lack of neutral hydrogen in the vicinity of a bright qso . if the quasar lifetime is shorter than the expansion and gas recombination timescales , the volume ionized will be proportional to the total number of photons produced above 13.6 ev : the effect of this local photoionization is to greatly reduce the scattering opacity between the redshift of the quasar and the boundary of its region .
the transmission on the red side of the ly@xmath5resonance is always greater than 50% for sources radiating a total of @xmath90 ionizing photons into the igm ( figure 2 ) .
the detection of a strong ly@xmath5emission line in the spectra of bright qsos shining for @xmath91yr can not then be used , by itself , as a constraint on the reionization epoch .
the first signs of an object radiating prior to the transition from a neutral to an ionized universe may be best searched for in the spectra of luminous sources with a small escape fraction of lyman - continuum photons into the igm , or sources with a short duty cycle . | in currently popular cosmological scenarios all variants of the cold dark matter ( cdm ) cosmogony some time beyond a redshift of 15 , stars within the numerous small halos that condense with virial temperatures @xmath0k created the first heavy elements ; these protogalactic systems , together perhaps with an early population of mini - quasars , generated the ultraviolet radiation and mechanical energy that reheated and reionized the cosmos . the history of the universe during and soon after these crucial formative stages is recorded in the thermal , ionization , and chemical state of the all - pervading intergalactic medium ( igm ) , which contains most of the ordinary baryonic material left over from the big bang . throughout the epoch of structure formation
, the igm becomes clumpy under the influence of gravity , and acts as a source for the gas that gets accreted , cools , and forms stars within subgalactic fragments , and as a sink for the metal enriched material , energy , and radiation which they eject .
cm^-2cm^-2 cm^-3cm^-3 ergsgpc^-3s^-1hz^-1ergss^-1hz^-1mpc^-3 _ em _ |
it is known @xcite that the zero modes of lattice dirac operators are strongly localised .
we show that not only zero modes but all low - lying modes of the hermitian wilson - dirac operator @xmath1 are strongly localised for bare quark masses @xmath2 .
furthermore , we show that this localization is strongly correlated to topological objects , including instantons .
a similar examination has been done for the overlap formalism @xcite .
we are interested in examining the extent to which similar properties are already manifested at the kernel level of the overlap formalism .
our wilson - dirac operator is defined in the standard way by @xmath3 ( x ) & = & \psi(x)-\kappa \sum_\mu [ ( 1-\gamma_{\mu})u_{\mu}(x)\psi(x+\mu)\nonumber\\ & & + ( 1+\gamma_{\mu})u_{\mu}^{\dagger}(x-\mu)\psi(x-\mu ) ] , \label{eq : dwilson}\end{aligned}\ ] ] where the hopping parameter @xmath4 is related to the bare mass by @xmath5 therefore the above mentioned bare mass range is equivalent to @xmath6 at tree level .
we examine eigenmodes in and just outside the region @xmath7 , where @xmath8 is the `` critical - mass '' which is 0 at tree - level but for non - trivial gauge fields shifts away from 0 .
this is the range of the mass parameter @xmath9 needed for use in the overlap formalism .
modes of @xmath10 crossing zero in this region are accompanied by the abrupt appearance of exact zero modes of the overlap - dirac operator at the corresponding @xmath9 value @xcite .
we solve the eigenvalue problem @xmath11 for the first four low - lying eigenvalues .
we are interested in calculating eigenmodes in the physical region of the overlap formalism with some further points just at the edge of this region .
this corresponds to a @xmath4-region starting slightly below the critical value of @xmath12 and extending slightly beyond the point where doublers appear in the continuum limit of the overlap formalism .
we consider @xmath13 at tree level .
since the critical @xmath4-value shifts from its free field value of @xmath14 to some higher value for non - zero gauge coupling , we have to adjust our @xmath4-range accordingly .
we look for any change in behaviour of the eigenmodes at the border of the region of physical interest .
eigenmodes are found by an accelerated conjugate gradient routine @xcite which is further improved by using dynamic state renormalisation .
the major advantage of a conjugate gradient algorithm besides its almost perfect parallel structure is that it yields not only eigenvalues with appropriate degeneracies , but eigenvectors , as well . for selected @xmath4-values
we also calculate up to 20 eigenmodes . in the following ,
the phrase low - lying eigenmodes should be understood to mean eigenmodes corresponding to the low - lying eigenvalues . in order to examine localisations of calculated eigenmodes
we plot the probability density @xmath15 for three dimensional cuts through the lattice . for comparison we plot the action density as well as the topological charge density for the appropriate configuration in the same way .
our calculations are made on @xmath16 , @xmath17 and @xmath18 lattices with anti - periodic fermion boundary conditions in the time direction .
on the @xmath16 lattice we test the correlation of low - lying eigenmodes with a single instanton configuration and standard monte - carlo generated su(3 ) background fields . for the latter background fields we consider hot , 5-sweep and 12-sweep cooled configurations .
calculations on the larger lattices are to verify the conclusions we obtain from the smaller lattice . in order to quantify our results
further we fit an instanton model to the obtained data .
we distinguish between the model for action or charge densities and the model for the zero - mode density @xcite , @xmath19 @xmath20 where @xmath21 is the distance from the instanton peak at @xmath22 to the calculated densities .
the normalisation factor @xmath23 allows us to fit to the instanton shape and prevents the fit from being dominated by the maximal value of the fitted object which is affected by periodic images due to the finite volume of the 4-torus .
both models are continuum results derived from the standard t hooft ansatz .
( [ eq : instactmodel ] ) is the action density and is used to fit action and charge densities .
( [ eq : instzeromodel ] ) is the density of the fermion field in the zero - mode and is used to fit eigenmode densities .
this allows us to compare localisation sizes and positions quantitatively .
this paper is organised as follows : after a discussion in sec . [
sect : lattice ] of lattice techniques and general properties , we describe results for a single instanton configuration on a @xmath17 lattice in sec . [
sect : inst ] . in sec .
[ sect : su3 ] we investigate localisations of eigenmodes of the hermitian wilson - dirac operator and the correlations those eigenmodes have with topology on standard monte - carlo generated su(3 ) configurations on a @xmath18 lattice .
we do this for both hot and 12-sweep - cooled configurations .
we quantify those localisations and their correlations to topology and compare sizes of the eigenmode localization with sizes of the corresponding topological object in sec .
[ sect : quant ] .
conclusions are presented in sec .
[ sect : concl ] .
we carry out our studies on three different lattices .
single instanton configurations are created on an @xmath16 lattice with @xmath24 and on a @xmath17 lattice with @xmath25 .
the latter configuration is cooled for 2 sweeps using the standard wilson action to minimize boundary effects .
we consider three standard monte - carlo su(3 ) configurations generated on an @xmath16 lattice with @xmath26 , which corresponds to a lattice spacing of 0.165(2 ) fm .
the second lattice size used is @xmath27 with @xmath28 which corresponds to a lattice spacing of 0.125(2 ) fm .
these configurations are generated with a plaquette plus rectangle improved action with mean - field improved coefficients . for the smaller exploratory lattice
the standard wilson action and standard plaquette topological charge are used to minimize boundary effects . for the bigger lattice , a three - loop improved action and three - loop improved topological charge operator @xcite are used .
the topological charge density is given by @xmath29 for the standard plaquette topological charge we use the traceless definition of @xmath30 extracted from the consideration of @xmath31 plaquettes alone @xcite @xmath32_{\rm traceless}\ ] ] where @xmath33 is the clover - sum of four @xmath31 wilson loops lying in the @xmath34 plane .
for the three - loop improved topological charge operator , we employ an @xmath35 improved definition of @xmath30 in which the standard clover - sum of four @xmath36 wilson loops lying in the @xmath34 plane is combined with @xmath37 and @xmath38 wilson loop clovers .
bilson - thompson _
_ @xcite find @xmath39_{\rm traceless}\ ] ] where @xmath40 is the clover - sum of four @xmath41 wilson loops and where @xmath30 is made traceless by subtracting @xmath42 of the trace from each diagonal element of the @xmath38 colour matrix .
this definition reproduces the continuum limit with @xmath43 errors . on approximately self - dual configurations , this operator produces integer topological charge to better than 4 parts in @xmath44 .
we use periodic boundary conditions in space and anti - periodic in the time direction for fermions .
each of the hot configurations is also cooled for 12 sweeps using a cabibbo marinari based algorithm in which the three diagonal su(2 ) subgroups of su(3 ) are looped over twice @xcite using a three - loop @xmath35 improved action @xcite .
twelve sweeps of cooling is just enough to see clear structure on the topological charge and action densities , but it still preserves much of the original topology . to see correlations for some eigenmode densities we also need to use the topological charge density of 5-sweep cooled configurations .
these topological charge densities are much rougher , but are closer to the original configuration , with topological objects moving less than in the 12-sweep cooled configurations .
the visualisations of the eigenmode probability densities are treated consistently in order to allow direct comparison .
this means they are all normalised with respect to the maximum value of each eigenmode . in this way
, an isosurface at half the peak height will reflect the size of the object .
we generate a single instanton background on an @xmath45 and on a @xmath17 lattice by performing the path integration of @xmath46 with @xmath47 where @xmath48 is an instanton and @xmath49 an anti - instanton , to create the link variable .
we find in the regular gauge @xmath50,\ ] ] @xmath51 @xmath52 in the singular gauge we find @xmath53,\ ] ] @xmath54 the singular gauge instanton is clearly recognisable on the volume rendered action density plot as seen in fig .
[ fig : install ] ( a ) .
the outer surface shown is half the peak height .
the results seen on the smaller exploratory lattice are also found on the @xmath17 lattice .
the instanton on the latter lattice is cooled for two sweeps to minimise boundary effects .
eigenmodes of this configuration are calculated for @xmath55 with an increment of 0.01 between values .
the first four eigenvalues of the spectrum are shown in fig .
[ fig : instspectr ] .
we then evaluate the localisation of @xmath56 for the first three low - lying eigenmodes for each @xmath4 .
this is done by plotting @xmath56 as seen in fig.[fig : install ] for selected eigenmodes . for eigenmodes with @xmath57
we find a rather uniform distribution of @xmath56 whereas eigenmodes for all @xmath58 are strongly localised .
for the lowest eigenmode this localisation strongly corresponds to the localisation of the instanton in the action density plot as seen by comparing fig .
[ fig : install ] ( a ) and fig .
[ fig : install ] ( b ) , ( c ) .
we therefore say the eigenmode displays strong correlation with the instanton .
the size and shape of this correlation varies from broad with some wall like structures for @xmath59 to very small for @xmath60 and is getting broader again for larger values of @xmath4 . for higher eigenmodes
this localization gets broader and less correlated with the instanton .
some higher eigenmodes show no correlation with the instanton but just wall like structure as seen in fig .
[ fig : install ] ( f ) .
others show prolate - like correlations as seen in fig .
[ fig : install ] ( d ) .
some eigenmodes also show a wall - like structure and a prolate - like correlation as seen in fig . [
fig : install ] ( e ) .
it is useful to note that plane wave behaviour would display uniform behaviour in the density plot .
after seeing that the low - lying eigenmodes of the wilson - dirac operator are strongly correlated to the instanton on a one - instanton configuration , we next carry out investigations on standard monte - carlo generated su(3 ) background fields .
we investigate three hot and three cooled configurations on both @xmath16 ( @xmath61 ) and @xmath27 ( @xmath62 ) lattices .
three - loop improved cooling @xcite is used to cool the configurations for 12 sweeps .
this is just enough to get clear structure in the action and topological charge . as a reference for comparison of the eigenmode density we use the action and topological charge density of the cooled configuration .
we do this because those densities are too rough for the hot configuration and no accurate comparison would be possible .
as there is more structure observed on the topological charge density plot , we use this as our preferred reference .
see fig.[fig : su3all ] ( a ) for a typical topological charge density plot of a @xmath27 configuration .
however the action density plot is also useful for guidance .
we say an eigenmode is correlated to a topological object if @xmath56 has a peak within one lattice site of the peak topological charge density .
we calculate the first four eigenmodes for values of @xmath63 for cooled configurations and @xmath64 for hot configurations in steps of 0.01 .
this range is from approximately 0.02 smaller than @xmath65 to a region where doublers appear in the overlap formalism .
we find that the behaviour described in the following is general for all configurations : each of the eigenmodes is localised .
this was already observed for the lowest eigenmode @xcite . for @xmath66
this localization weakens and the density , @xmath67 , broadens quickly .
an exponential decay of the density @xmath56 , as previously observed @xcite , seems likely to occur .
all low - lying eigenmodes for @xmath68 are correlated to topological objects .
we are able to track a correlation along the modes for all @xmath68 and for one step smaller than @xmath65 .
thus we can label a mode by its correlation .
[ fig : modes ] shows eigenmodes of a hot configuration , where the symbol used denotes which topological object the localised low - lying eigenmode is correlated with .
careful inspection of fig .
[ fig : modes ] reveals the presence of an eigenmode , correlated to one topological object , but with a spectral flow containing two zero crossings .
the size of the correlated object , obtained in the following section , suggests that this object is a lattice artifact . around @xmath65 and towards the upper end of the analysed spectrum ,
localisations are weak . however , the correlation of positions is strong . at these @xmath4
we tend to get correlations to more than one topological object .
the localisations get stronger and sharper for increasing values of @xmath4 until a maximal localization is reached for a value @xmath4 we will define to be @xmath69 . for @xmath70
the localisations get weaker again . fig .
[ fig : su3all ] shows a visualisation of this behaviour .
additional figures can be found elsewhere @xcite .
we take a closer look at this behaviour in the next section . for the lower eigenmodes , which are separated by a gap from the higher eigenmodes , we find correlation to one topological object per eigenmode . for higher eigenmodes , which are closer together with some degenerate modes , we find correlations to more than one topological object . in general
we find that the closer the eigenvalues and the weaker the actual localisations , the more likely it is to get correlation with more than one topological object in the corresponding eigenmodes .
calculations of 20 eigenmodes for selected values of @xmath4 on hot configurations show only little broadening in the localisations and persistent correlations with topology .
this suggests that this behaviour will not change quickly and such correlations will persist for even higher eigenmodes .
as mentioned above , we use the topological charge density of the 12-sweep cooled configuration for comparison .
the position of eigenmodes on cooled fields agree perfectly well with the position of topological objects seen in this density .
a typical distance between such positions being 0.05 lattice units . for hot configurations
some of the correlations are slightly offset compared to the structure in the smoothed configurations whereas other structure in the eigenmodes has no corresponding structure in the smoothed topological charge .
but in those cases a comparison with a less cooled configuration reveals correlations between eigenmodes and topological objects which are moved or destroyed by further cooling .
this behaviour is expected and understood , as topological objects are known to move under cooling as instantons and anti - instantons attract each other and annihilate when they are close enough together .
this clear correlation between @xmath56 of eigenmodes on hot configurations and topological objects suggests that it is possible to identify areas through the noise of a hot configuration with significant topology .
we can `` see through '' the noise by using eigenmodes of the hermitian wilson - dirac operator .
the spectral plot of the first four eigenvalues of a 12-sweep cooled configuration looks different compared to the same plot on a hot configuration . comparing fig .
[ fig : modes ] and fig .
[ fig : coolspectr ] we notice the rhomboid shape with an area without any eigenvalues in the spectral plot of the cooled configuration .
this is expected for such smoothed configurations @xcite .
eigenmodes on the right - hand side of this rhomboid behave different than eigenmodes on the left - hand side of the rhomboid .
the localization of the eigenmodes on the left - hand side is approximately the same as the localisations in the hot configuration .
but the eigenmodes on the right - hand side are much weaker in localization .
those weaker localised eigenmodes are related to very high eigenmodes in the hot configuration and the process of cooling brings them into an area where we can observe them as low eigenmodes .
these eigenmodes must be very high eigenmodes in a hot configuration for they are much weaker localised as localisations we have seen for up to the 20th eigenmode in a hot configuration .
although the behaviour in localization strength for eigenmodes of cooled configurations is different from that of eigenmodes of hot configurations the correlations with topological objects still exist for all eigenmodes .
we will take a closer look at localization strength and quantify it in the next section .
as described earlier in this paper , the localization of the eigenmodes change shape and size with changing of @xmath4 . in order to quantify this behaviour
we have two methods .
the first one is fairly simple .
we have already seen that most eigenmodes are localized at one topological object .
as the eigenmodes are all normalized , the maximum , or peak , value of the eigenmode density is an indicator of how strong this localization is .
[ fig : maxhot ] shows the plot of such peak values for four eigenmodes each of three hot @xmath71 configurations . this plot shows a smooth behaviour with a maximum for @xmath72 for the lowest eigenmodes .
this suggests that for hot su(3 ) configurations the strongest localisation occurs in that @xmath4 region .
[ fig : maxcool ] shows the same plot for four eigenmode densities each for three 12-sweep cooled @xmath71 configurations .
this plot shows a different behaviour .
again there is a clear peak , but the values drop to almost zero immediately after the peak .
we expect this to happen after our qualitative observations in the previous section .
these low maximum values correspond to weakly localized modes , which are on the right - hand side of the spectrum shown in fig .
[ fig : coolspectr ] . in this case
we can not determine @xmath69 , for it is not clear how the localisations are going to develop if we follow the stronger localised eigenmodes up to higher values of @xmath4 .
the one eigenmode we could follow suggests that @xmath73 . this reduction of @xmath69 is about the same size as the reduction of @xmath65 for going from the hot to the 12-sweep cooled configuration . to get more information about the shape of the eigenmodes
, we fit the instanton model , eq . ( [ eq : instzeromodel ] ) , to the eigenmodes .
this model gives us very good fits as @xmath74 where @xmath75 is the 6 parameter fit of @xmath76 points of @xmath77 centered about the peak of @xmath56 .
this value is about 1000 times smaller than the peak value and @xmath78 is the worst case with most of the fits of order @xmath79 to @xmath80 .
the fit parameter @xmath81 is then a good measurement for the size of the localization . in fig .
[ fig : rhohot ] we plot @xmath82 for four eigenmodes calculated on all hot configurations and in fig . [
fig : rhocool ] for four eigenmodes calculated on all 12-sweep cooled configurations .
we see a behaviour which corresponds to the behaviour described above for the peak values of the eigenmodes .
@xmath82 for the hot configurations shows a smooth behaviour with a minimum around @xmath83 .
this means the eigenmodes are maximally localised for this value of @xmath4 and are less localized for both ends of the spectrum .
the @xmath69 found this way varies slightly from @xmath69 found by just taking the peak values of the eigenmode densities .
it is about 0.03 larger at 0.26 for the lowest eigenmode .
as we mentioned earlier the eigenmodes for low and high values of @xmath4 are localized on more than one topological object .
therefore fig .
[ fig : rhohot ] reports more local maxima than modes at these @xmath4 .
however in the range @xmath84 only one local maximum is found per mode .
we have established that low - lying eigenmodes are correlated to a single topological object when eigenmodes are non degenerate .
it is also established that an instanton gives rise to a zero crossing in the spectral flow with the sign of the slope equal to the sign of the topological charge @xcite . however , as mentioned above , fig .
[ fig : modes ] reveals an eigenmode correlated to one topological object , but with two zero crossings .
the size of the eigenmode varies from @xmath85 to @xmath86 as @xmath4 varies from 0.21 to 0.27 .
the size of the correlated object on the topological charge density , which can only be seen on the 5-sweep - cooled configuration , is @xmath87 .
we have been able to reproduce similar spectral flows on single instanton configurations , which have been cooled with the wilson action to the point where the topological charge is @xmath88 , well below 1 .
hence the double zero crossing of the spectral flow in fig .
[ fig : modes ] suggests the presence of a lattice artifact . ideally an improved fermion action should act to remove the zero crossings associated with this artifact .
the behaviour of @xmath82 of the cooled configurations shows a jump at @xmath89 , where the weakly localised modes set in as described above .
some of these weakly localised modes are not sufficiently localised to allow a fit to the instanton model .
a fit would result in values for @xmath90 , which are not reasonable for the instanton model and therefore neglected .
for those eigenmodes where we can do a fit , we get large values of @xmath81 compared to the strongly localised eigenmodes .
we find that @xmath81 is about 2 for strongly localised eigenmodes and about 6 for weakly localised eigenmodes . for @xmath91
, @xmath82 decreases smoothly , but for @xmath92 the values of @xmath82 are higher than at the lower end of the spectrum and do not show a smooth behaviour .
again it is hard to extract @xmath69 , but it seems that if we could follow the stronger localised modes further it would be around 0.22 .
this is about 0.02 larger than the @xmath69 , which is extracted using the peak values .
the fitting of the models to the eigenmode densities , using eq .
( [ eq : instzeromodel ] ) , as well as to the topological charge densities , using eq .
( [ eq : instactmodel ] ) , allows us to compare the sizes for the eigenmode localisations with the sizes of the actual topological objects . in order to do this
, we find the topological object which is located closest to the position of the eigenmode .
we can fit the instanton model only to topological charge densities of cooled configurations , as only those are smooth enough .
therefore we only compare the sizes of eigenmodes of cooled configurations with sizes of actual topological objects . due to the different localization strength for eigenmodes on the left - hand and on the right - hand side of the rhomboid spectrum of a cooled configuration ,
we look at those eigenmodes separately . we find that the fitted positions for the strongly localised eigenmodes agree very well .
with the fitted positions of the correlated topological objects lie within a fifth of a lattice spacing , and most of the times even better .
the fitted positions of the weakly localised modes to the right of the spectrum agree only within one lattice spacing with their correlated topological objects .
strongly localised eigenmodes are correlated to smaller objects in the topological density .
such eigenmodes on the left - hand side of the eigenvalue spectrum , except those for @xmath93 , have a size between @xmath94 and @xmath95 .
the correlated topological objects have a size between @xmath96 and @xmath97 .
the eigenmodes are larger than their correlated topological objects for smaller values of @xmath4 , but as they shrink with growing @xmath4 they get smaller than their correlated topological objects . all followed modes reach the size of the correlated topological object for @xmath98 . fig .
[ fig : sizecomp ] shows an example of this behaviour .
the weakly localised eigenmodes on the right - hand side of the eigenvalue spectrum are larger than the strongly localised eigenmodes on the left - hand side of the spectrum .
they turn out to have a size between @xmath99 and @xmath100 .
but the topological charge density correlated with those eigenmodes are themselves bigger than the topological charge density correlated with the stronger localised modes .
the sizes of the topological charge density objects lie between @xmath97 and @xmath101 .
the weaker localised eigenmodes are always bigger than the correlated topological objects with @xmath102 .
for @xmath12 the size of the eigenmodes is between @xmath103 and @xmath104 with a relation to the size of the correlated topological objects of @xmath105 .
sizes of strongly localised eigenmodes of hot configurations are of a comparable , but slightly smaller size than that on cooled configurations . in order to understand why modes on the right - hand side of the rhomboid are just weakly localised
we calculate spectra for one configuration with different amounts of cooling .
we find that the value of @xmath4 where the jump in the localisation size occurs becomes smaller with cooling .
compare fig .
[ fig : rhocool ] and fig .
[ fig : rhocool4 ] .
it is known @xcite that zero crossings for larger @xmath4 correspond to smaller topological objects .
as we see localisation sizes of calculated eigenmodes shrink with growing @xmath4 as well , we generally think of larger values of @xmath4 being associated with smaller objects .
cooling removes smaller objects first , therefore eigenmodes for large @xmath4 become `` unassociated '' with small objects .
the eigenmodes show a behaviour of weak localisation which can be seen in fig . [
fig : rhocool ] and fig .
[ fig : rhocool4 ] after the jump .
further cooling removes larger objects and therefore the value of @xmath4 , where the change of behaviour sets in , becomes smaller .
not only zero modes , but all low - lying eigenmodes of the hermitian wilson - dirac operator @xmath106 are strongly correlated to topological objects for @xmath68 .
these objects can be instantons for which @xmath107 or topological fluctuations with @xmath108 .
one eigenmode is correlated to at least one topological object with correlations to more than one object becoming more likely as eigenvalues become degenerate and eigenmodes become broader in size . for
@xmath66 the correlations broaden very quickly and are lost for values smaller than about @xmath109 . for @xmath110
the correlations become sharper until @xmath111 for which the correlations are strongest . for @xmath70 the correlations
broaden again .
eigenmodes of 12-sweep cooled configurations show a different behaviour depending on whether they belong to eigenvalues on the left or right - hand side of the rhomboid - shaped eigenvalue spectrum of a cooled configuration as seen in fig .
[ fig : coolspectr ] .
eigenmodes belonging to the left - hand side of the eigenvalue spectrum are strongly localised and show the same behaviour as eigenmodes of hot configurations .
eigenmodes belonging to the right - hand side of the eigenvalue spectrum are very weakly localised , but are still correlated to topology .
this suggests that those eigenmodes correspond to very high eigenmodes in a hot configuration , which are lowered by cooling . the value of @xmath4 where the weakly localised eigenmodes set in becomes smaller with cooling
this supports the idea of high values of @xmath4 corresponding to localisations on small topological objects .
small topological objects are removed first under improved cooling thus eigenmodes for high values of @xmath4 are the first ones to loose the strong localisation . when an instanton model is fitted to the eigenmode density , using eq .
( [ eq : instzeromodel ] ) , and to the topological charge density , using eq .
( [ eq : instactmodel ] ) , strongly localised eigenmodes have about the same size for @xmath81 as the correlated topological objects . for @xmath69 eigenmodes
are slightly smaller than the correlated topological and for @xmath4 smaller than @xmath112 eigenmodes are slightly larger . on a single instanton configuration
the correlation to the instanton persists strongly only for the lowest eigenmode and is then gradually lost for higher eigenmodes . on su(3 )
background fields we see correlation for the 20 lowest eigenmodes .
there is only little broadening which suggests that the correlation will persist for eigenmodes higher than 20 .
future work will examine the manner in which the eigenvalue spectrum is modified under improved fermion actions .
in particular we plan to examine the effects of using ape - smeared fat - links in the irrelevant operators of fermion actions .
we would like to thank the australian national computing facility for lattice gauge theory for time on its supercomputer orion and paul coddington , ramona adorjan and francis vaughan for their technical support of this work .
we would like to thank paul rakow and alex kalloniatis for useful discussions .
djk wants to thank the baden - w@xmath113rttemberg - south australia exchange program for supporting his stay in adelaide .
this research was supported by the australian research council . | the probability density of low - lying eigenvectors of the hermitian wilson - dirac operator @xmath0 is examined .
comparisons in position and size between eigenvectors , topological charge and action density are made .
we do this for standard monte - carlo generated su(3 ) background fields and for single instanton background fields .
both hot and cooled su(3 ) background fields are considered .
an instanton model is fitted to eigenmodes and topological charge density and the sizes and positions of these are compared .
adp-01 - 46/t478 + , , , and lattice , eigenmodes , wilson - dirac operator , topology , instantons |
the impressive active and passive mechanical performance of a cell acting as the smallest biological unit able to move , survive and replicate independently has been of much interest .
most importantly , the motility of malignant cells is an important prognosticator in cancer@xmath0}$ ] .
the cells movement and response to external mechanical stimuli is a complex process mainly determined by its cytoskeleton .
the cytoskeleton is a protein network extending from the cell membrane to the nucleus and filling large parts of the cytoplasm .
it mostly consists of three different types of protein fibres : actin filaments , microtubuli and intermediate filaments@xmath1}$ ] .
actin filaments , with a persistence length of about 15 pm , are responsible for the movement of the cell , whereas the stiffer microtubuli with a persistence length of several millimetres@xmath2}$ ] are the transport pathways in the cell .
intermediate filaments are more flexible with a persistence length between 500 nm and 1000 nm@xmath3}$ ] ( keratin 8/18 @xmath4 , unpublished data ) , their function appears to be to provide the mechanical stiffness and stability of the cell .
thus they are hypothesized to be the `` stress - buffering - system''@xmath5}$ ] of cells . whereas actin filaments and microtubuli have been investigated in much detail@xmath6}$ ] , little is known about intermediate filaments .
we have recently demonstrated that the intermediate filament network in pancreatic cancer cells , mostly consisting of keratins 8 and 18 , is a crucial determinator for cancer cell motility@xmath7}$ ] . to further improve the understanding of the role of keratin networks in this disease we are now focusing on the nanomechanical properties of the keratin cytoskeleton .
this paper presents two different approaches : in vivo measurements of the mechanical properties of the extracted cytoskeleton and measurements of in vitro assembled protein networks .
both approaches have different advantages that shall be discussed briefly . as a top down
approach it is useful to extract the keratin cytoskeleton from whole cells@xmath7}$ ] . here
the original architecture of the keratin network is conserved while other parts of the cell , like membrane , organels , actin , microtubuli etc .
are removed .
the advantage is obvious : the mechanical properties of the original keratin cytoskeleton can be investigated without any environmental influence while the local network topology is easily measured by sem afterwards@xmath8}$ ] .
but there is one major problem .
the extraction does not remove all other proteins and some linker molecules , i.e. plectin , still remain in situ , especially in the crosslinking points of the network . for this purpose
it is reasonable to have a look from the opposite point of view .
the bottom up approach of in vitro assembled protein networks is not as unnatural as it may sound .
especially the cytoskeleton is a part of the cell with several components fulfilling all their special tasks independently@xmath0}$ ] .
hence it is quite useful to analyse the mechanical properties of parts of the cytoskeleton as isolated functional modules .
additionally it is possible to create well defined networks in vitro without any undesirable components . determining the viscoelastic and mechanical properties of these in vitro assembled networks should show comparable results to extracted cell networks .
microrheology is a suitable tool for measurements of the mechanical properties of both the extracted cytoskeleton and of in vitro assembled keratin networks .
the possibility to measure over an extended range only by observing thermal fluctuation is an advantage over the traditional measurements where mechanical stress is applied and the frequency range is limited@xmath9}$ ] . in our model micron - sized spherical particles , act as probes , which scan their environment driven by the thermal forces ( brownian motion ) dependent on the viscoelastic behaviour of the surrounding network . by means of particle tracking the brownian motion of several beads
can be detected over a wide range of frequencies ( up to 5000 hz with a high - speed ccd - camera ) with a high accuracy in resolution ( down to 5 nm ) .
this allows to calculate the complex shear moduli which give an insight to the properties of the network and enables to show the role of the intermediate filaments in the cell . in that way
general statement about network dynamics are possible@xmath10}$ ] .
typical afm@xmath11}$ ] pictures of the in vitro assembled network and the extracted cytoskeleton are shown in figure 1 .
one can clearly see structural differences between the two kind of networks . in this work the shear modulus of the extracted keratin cytoskeleton ( k8/18 ) from panc 1 cells
is compared to the shear modulus of in vitro assembled keratin 8/18 networks .
both data sets from microrheology measurements are evaluated with the model relating stochastic movement to mechanical frequency dependent moduli .
the derivation of the complex shear modulus from the average motion of beads was first described by mason in the late 90s@xmath9}$ ] .
the model is based on the generalized stokes - einstein equation and provides a good way to determine the linear viscoelasticity of complex fluids . to derive the moduli we assume that the filament network is treated as a viscoelastic , incompressible , isotropic fluid .
it therefore is assumed to be a continuum and the beads inertia was neglected , too . around the beads surface
there are noslip boundary conditions which are a good approximation in the frequency range covered by our experiment .
a brief sketch of the derivation of the shear moduli is given below@xmath12}$ ] .
the motion of a single bead with radius @xmath13 in a viscoelastic medium can be described by a generalized langevin equation@xmath14}$ ] : @xmath15 with @xmath16 and @xmath17 being the mass of the bead and its acceleration .
the memory function @xmath18 describes the response of the incompressible complex fluid .
the function @xmath19 contains the stochastic forces of the viscous fluid .
causality and the usage of the equipartition theorem of thermal energy relates the local memory function to the velocity of the bead . the complex viscosity @xmath20 can be related to the memory function @xmath18 by using the stokes relation : @xmath21 with @xmath22 the complex shear modulus can be calculated in terms of a unilateral fourier transform @xmath23 this equation represents a generalization of the stokes - einstein equation in the fourier domain .
the evaluation of the fourier term can be done by an estimation of the transforms by expanding @xmath24 , the mean square displacement , locally around the frequency of interest using the power law expansion .
this evaluation leads to a relation which is suitable for analytic computation .
hence , we find @xmath25}\ ] ] with @xmath26 the power law exponent which depends on the logarithmic slope of the mean square displacement ( msd ) @xmath27 .
since @xmath28 in general is a complex function , it can be split into the real and the imaginary part .
they are related by the kramers - kronig - relation .
@xmath29 , the so called storage modulus , is the real part and describes the dissipation - free spring - like behaviour .
@xmath30 , the loss modulus , is the imaginary part and gives information about the liquid - like dissipative behaviour of the material . for the shear moduli
we achieve @xmath31 in simple viscous fluids the observed msd becomes proportional to @xmath32 and the complex modulus purely imaginary with @xmath33 proportional to the macroscopic shear viscosity . for a simple elastic solid
the msd is lag - time independent .
a viscoelastic material shows an intermediate form , where the storage modulus dominates at low frequencies and the loss modulus at high ones .
the human keratin 8/18 network was analysed in vitro to compare the mechanical properties with those from the extracted keratin cytoskleton of a cell .
figure 2 shows storage and loss moduli for a fixed frequency @xmath34 .
one can clearly see the dependency between the distance rim of the nucleus - bead and the moduli .
the farther away the bead is from the nucleus rim the weaker is the surrounding network . compared to the extracted cytoskeleton the storage modulus of the in vitro assembled network is about one order of magnitude lower whereas the loss modulus of the in vitro assembled network is higher than that of the extracted keratin cytoskeleton .
possible reasons are mainly the differences in the construction of the two networks and the differences in the bead interaction .
on the one hand the extracted cytoskeleton is a dense crosslinked network with chemical bondings at the branching points as displayed in figure 1 on the right side . on the other hand
the in vitro polymerised network is entangled with at most few intersections ( figure 1 left side ) .
the force connecting filaments to an entanglement originates from friction forces and not from chemical bonding and leads therefore to a weaker connection point .
this difference can be clearly seen in figure 3 .
two point clouds from typical measurements are displayed as iso surface representations of the 3d position distribution .
figure 3a shows the density cloud of a bead embedded in the extracted cytoskeleton .
the bead is fluctuating almost homogeneously in all directions .
by contrast figure 3b shows the iso surface representation of the 3d position distribution of a bead in the in vitro assembled network .
the bead is fluctuating more freely .
it can circle a filament ( small picture ) or jump from one mesh to another ( supporting data ) . from this structural difference the characteristics of the moduli can be explained with a possible reptation - like behaviour in the entangled network@xmath35}$ ] .
this leads to a higher loss modulus and a lower storage modulus because of the dissipation in the network .
by contrast in the crosslinked network the storage modulus is higher and the dissipation is lower because reptation like movements are suppressed .
the different behaviour might be due to differences in the assembly process in cells and in vitro .
in the cell a directed assembly process takes place whereas the in vitro polymerisation is a self assembly process starting everywhere in the volume .
the network in the cell is continuously constructed with the help of many players like helper proteins or linker proteins@xmath36}$ ] .
additionally there may be a preferred direction of assembly because of cell movement or a prestress from the membrane .
all these factors indicate a different topology compared to the in vitro polymerised network . here
the assembly is a stochastic temperature driven process .
there is no influence from the surrounding environment or a directed development of the network .
just two filament ends meeting coincidentally lead to a junction and two crossing filaments can lead to an entanglement .
the interaction of the bead with the network is strongly dependent on the polymerisation process . during the self assembly of the in vitro polymerised network
the bead is continuously fluctuating .
this movement leads to the formation of a cavity around the bead where no polymerisation takes place .
thus the bead is not as strongly embedded in the network as in case of the cytoskeleton .
there the bead is integrated in the network by strong connections of single filaments with the bead .
another aspect is the keratin concentration of 1 mg / ml ( see experimental section ) in the network .
this leads to an estimated mesh size of 400 nm in the in vitro polymerised network and is comparable to the mesh size of the extracted cytoskeleton@xmath8}$ ] .
our experiments compare the dynamic shear moduli of the extracted keratin cytoskleton with those of in vitro assembled keratin networks .
it was shown that the dynamic shear modulus of the keratin cytoskeleton is decreasing significantly with increasing distance to the nucleus . in comparison
the storage modulus of the in vitro assembled network is about one order of magnitude lower whereas the loss modulus of the in vitro assembled network is higher than that of the extracted keratin cytoskeleton .
one possible explanation is based on the different topology of the two types of networks .
the in vitro assembled networks is dominated by entanglements .
the movement of the filaments is probably reptation - like causing dissipation in the network .
this leads to a higher loss modulus and a lower storage modulus .
the extracted keratin cytoskeleton is a crosslinked network , where reptation - like movements are suppressed .
that leads to an increasing storage modulus and , because of the lower dissipation , to a lower loss modulus .
additionally the presented work shows that rheology measurements from 0.1 hz up to 5 khz are possible and @xmath37 and @xmath38 can be determined in a range from @xmath39 to @xmath40 . here
microrheology is a suitable tool for investigation of the link between structure and function .
the comparison of measurements of in vitro assembled networks with the extracted keratin cytoskeleton of living cells shows that the bottom up approach is a promising way for further investigations .
indeed it is very important to take the structural differences into account .
human keratin 8 and keratin 18 were expressed and purified as described in harald herrmann et al 1999@xmath41}$ ] .
the assembly protocol was as follows@xmath42}$ ] : keratin 8 and keratin 18 were mixed in 8 m urea in 1:1 ratio .
afterwards the urea concentration was lowered stepwise by dialysis first into buffer1 ( 4 mm urea , 10 mm tris , 2 mm dtt , ph 8.5 ) then buffer2 ( 2 mm urea , 10 mm tris , 2 mm dtt , ph 8.5 ) and finally buffer3 ( 0 mm urea , 10 mm tris , 2 mm dtt , ph 8.5 ) at room temperature .
dialysis was continued overnight into dialysis buffer ( 2 mm tris , 1 mm dtt , ph 9.0 ) at @xmath43c . for starting the assembly an equal volume of 2 x assemblypuffer (
20 mm tris , ph 7.0 ) was added .
the assembly concentration was 1 mg / ml . for the sample preparation of the afm measurements of the in vitro assembled network the assembly was diluted to a concentration of 0.005 mg / ml .
@xmath44l of the keratin assembly was applied to a freshly cleaved mica surface , fixed with a fixation buffer ( 0.2 % glutaralduhyde ) and carefully washed with filament buffer ( equal volumes of dialysis buffer and 2 x assemblybuffer ) .
the extracted cytoskeleton was fixed with 4% glutaraldehyde and gradually dehydrated and subjected to critically point drying .
afm measurements were done with a scan frequency of 0.5 hz using cantilevers for tapping mode from olympus ( omcl - ac240ts - w2 ) with a nominal spring constant of 2 n / m .
the pictures were recorded using an mfp-3d from asylum research and processed with wsxm 4.0 develop 12.21@xmath45}$ ] .
the protein concentration of the assembly for the microrheology measurements was 1 mg / ml . additionally 1@xmath46 m spheres were added in a concentration of 1 % by weight in millipore water to @xmath47l protein solution before starting the assembly . for the measurement of the brownian motion beads of a diameter of 500 nm and @xmath48 m were used in an extracted keratin network .
living carcinoma cells incorporate the beads and embed them into their cytoskeleton .
after the extraction of the cell remains the keratin network , the nucleus and the incorporated beads .
for further details on cell extraction we refer to beil et al .
2003@xmath7}$ ] .
a microscope setup with additional high - speed camera was used to track the embedded beads with a frequency of 5000 hz at a time window of 3.2 sec and a resolution of 256x256 pixels .
the usage of the camera allows the tracking of several beads simultaneously in three dimensions ( manuscript in preparation ) and therefore has the advantage of the measurement of beads at the same conditions .
we thank the dfg sfb518 ( tp ) and project ma1297/10 - 1 ( al ) for financial support , andreas huler , carlo di giambattista and sarah pomiersky for discussion and measurements and all other members of the institute of experimental physics , ulm univerity , paul walther and the ze elektronenmikroskopie , ulm university , and the members of the division of molecular genetics from the dkfz heidelberg .
special thanks to prof .
adler and the department of internal medicine i , university hospital ulm , and to prof .
mizaikoff and the institute of analytical and bioanalytical chemistry , ulm university , for sharing laboratories during constructions .
* references * 1
. a. r. bausch , k. kroy , nature physics 2006 , 2 , 231 - 238 ; 2 . h. herrmann , u. aebi , annu .
biochem.2004 , 73 , 749 - 789 ; 3 .
h. herrmann , t. wedig , r. m. porter , e. b. lane , u. aebi , j. struct .
2002 , 137 , 82 - 96 ; 4 .
e. fuchs , k. weber , annu .
1994 , 63 , 345 - 382 ; 5 .
t. yanagida , m. nakase , k. nishiyama , f. oosawa , nature 1984 , 301 , 58 ; 6 .
f. gittes , b. mickey , j. nettleton , j. howard , j. cell biol .
1993 , 120 , 923 ; 7 .
t. g. mason , rheol acta 2000 , 39 , 371 - 378 ; 8 .
f. gittes , b. schnurr , p.d .
olmsted , f.c .
mackintosh , c. f. schmidt , phys .
rev . letter .
1997 , 79 , 3286 - 3289 ; 9 .
t. g. mason , d. a. weitz , phys .
rev . letter .
1995 , 74 , 1250 - 1253 ; 10 .
r. b. bird , r. c. armstrong , o. hassager , dynamics of polymer liquids 1977 , wiley , new york ; 11 .
p hansen , i. r. mcdonald . ,
theory of simple liquids 1986 , academic press , london ; 12 .
n. mcke , l. kreplak , r. kirmse , t wedig h. herrmann , u. aebi , j. langowski , j. mol .
2004 , 335 , 1241 - 1250 ; 13
. m. beil , a. micoulet , g. v. wichert , s. paschke , p. walther , m. bishr omary , p.p .
van veldhoven , u. gern , e. wolff - hieber , j. eggermann , j. waltenberger , g. adler , j. spatz , t. seufferlein , nat .
cell biol .
2003 , 5 , 803 - 811 ; 14 .
i. horcas , r. fernandez , j. m. gomez - rodriguez , j. colchero , j. gomez - herrero , a. m. baro , rev .
2007 , 78 , 013705 ; 15 .
m. beil , h. braxmeier , f. fleischer , v. schmidt , p. walther , journal of microscopy 2005 , 220 , 84 - 95 ; 16
. p. g. de gennes , j. chem . phys .
1971 , 55 , 572 - 579 ; 17 .
h. herrmann , m. hner , m. brettel , n. ku , u. aebi , j. mol .
1999 , 286 , 1403 - 1420 ; 18 .
r. windoffer , s. wll , p. strnad , r. e. leube , mol .
biol . cell .
2004 , 15 , 2436 - 2448 ; 19 . j. c. crocker , b. d. hoffman , meth .
cell biol .
2007 , 83 , 141 - 178 ; 20 . for an overview
see b. bhushan , o. marti , handbook of nanotechnology , 2007 , springer , new york , and references there in ;
\a ) afm picture of the in vitro assembled keratin network , 7.5 @xmath49 m x 7.5 @xmath49 m topography b ) afm picture of the extracted cytoskeleton , @xmath50 m x 2 @xmath49 m topography .
2 . @xmath37 , @xmath38 as function of distance to the rim of the nucleus for a fixed frequency @xmath51 hz .
@xmath37 ( empty squares ) and @xmath38 ( empty circles ) of the extracted cytoskeleton and @xmath37 ( full squares ) and @xmath38 ( full circles ) of the in vitro assembled network . in this sketch the shear moduli of the in vitro assembled network are constant , because of the lack of a nucleus .
\a ) iso surface representation of 3d position distribution of a bead in the in vitro assembled network .
b ) iso surface representation of 3d position distribution of a bead in the extracted cytoskeleton
. iso surface representation of 3d position distribution shows all point with an equal probability to find the particle . | in our work we compare the mechanical properties of the extracted keratin cytoskeleton of pancreatic carcinoma cells with the mechanical properties of in vitro assembled keratin 8/18 networks . for this purpose
we use microrheology measurements with embedded tracer beads .
this method is a suitable tool , because the size of the beads compared to the meshsize of the network allows us to treat the network as a continuum . observing the beads motion with a ccd - high - speed - camera then leads to the dynamic shear modulus .
our measurements show lower storage moduli with increasing distance between the rim of the nucleus and the bead , but no clear tendency for the loss modulus .
the same measurement method applied to in vitro assembled keratin 8/18 networks shows different characteristics of storage and loss moduli .
the storage modulus is one order of magnitude lower than that of the extracted cytoskeleton and the loss modulus is higher .
we draw conclusions on the network topology of both keratin network types based on the mechanical behaviour .
* keywords * : afm , cytoskeleton , dynamic shear modulus , keratin 8/18 , particle tracking microrheology |
the main result of the present study is a resolution of a puzzle which existed since the middle of 80 s and has been one of the major obstacles towards our understanding of the behavior of qcd at finite baryon density .
the contradiction is simple and acute .
the quark chemical potential @xmath0 in qcd shifts the energy of a given quantum state by @xmath4 times this state s net baryon number .
a transition at @xmath5 should occur when @xmath6 : the vacuum state without baryons is no longer the lowest energy state .
it is believed that the chiral symmetry is restored in the new phase . testing this prediction by monte carlo simulation is difficult since the determinant of the fermion matrix becomes complex at nonzero @xmath0 .
nevertheless one can use quenched approximation in this case .
such monte carlo studies @xcite give consistently the critical value of @xmath7 a clearly unphysical result , since the pion does not carry baryon number .
the present study is motivated by a desire to clarify the problem using random matrix approach . on an exactly solvable model of chiral symmetry breaking in qcd
we demonstrate explicitly a general phenomenon related to the nature of the quenched limit .
namely , one can view the quenched theory as a somewhat abstract limit @xmath1 of qcd with variable number @xmath2 of quark flavors .
then if the limit is smooth the quenched theory is a reasonable approximation to the theory at finite @xmath2 .
in fact , this limit is not always smooth and can be different from the quenched theory as , e.g. , in a theory of spin glasses .
we show that the quenched limit is not smooth in qcd at finite @xmath0 .
quenched theory is therefore not a good approximation to full qcd at finite @xmath0 .
this resolves the old contradiction .
moreover , we can describe a theory to which quenched qcd _ is _ a good approximation .
it is a theory ( suggested in @xcite ) in which each quark has a conjugate partner , essentially , with opposite baryon number assignment .
we study analytical properties of the quark condensate @xmath8 the order parameter of the chiral symmetry breaking as a function of the bare quark mass @xmath10 which we allow to go into the complex plane : @xmath11 . integrating out the quark degrees of freedom we can express @xmath8 through the average of the inverse dirac operator @xmath12 : @xmath13 . in the infinite volume the value of the @xmath8 in the chiral limit @xmath14 can be related to the density of small eigenvalues @xmath15 by the banks - casher formula : @xmath16 . to generalize this well - known relation to our case we must consider the density per unit _ area _ of the complex plane : @xmath17 .
the relation between @xmath9 and @xmath18 is straightforward : @xmath19 . viewing @xmath9 as a density of charges and @xmath20 , @xmath21
as components of the electric field strength @xmath22 one easily inverts this relation : @xmath23 the main point here is that @xmath9 vanishes where @xmath18 depends analytically on @xmath24 , the bare quark mass .
for example , when @xmath25 all eigenvalues of @xmath12 lie on the imaginary axis .
this means that @xmath18 has a cut as on fig .
[ fig : n]a .
the discontinuity across the cut is the signature of the chiral symmetry breaking and is given precisely by the banks - casher formula .
now we would like to calculate @xmath18 and @xmath9 .
since we are interested in the density of _ small _ eigenvalues of the dirac operator one can expect that some simplification can be made . indeed , it is by now well understood that these small eigenvalues are related to the zero modes of the dirac operator in the instanton field background .
the behavior of these small eigenvalues is rather universal and can be described by a simple random matrix model @xcite .
we choose chiral representation of the dirac gamma - matrices .
then the matrix of the dirac operator in a certain basis has the block - diagonal form : @xmath26 nonzero entries @xmath27 are approximated by a complex random @xmath28 matrix with a gaussian distribution : @xmath29 . to adapt the model to our case
we add a chemical potential : @xmath30 .
to find @xmath18 one calculates the quenched free energy : @xmath31 and differentiates it with respect to the bare quark mass @xmath24 : @xmath32 .
then @xmath9 is given by : @xmath33 a standard way to deal with the logarithm inside the averaging in ( [ v ] ) is the ` replica trick ' .
one does the calculation for arbitrary number of fermion species @xmath2 : @xmath34 and then takes @xmath35 hoping that the limit is smooth . doing this calculation
we find that , first of all , @xmath18 is a holomorphic function of @xmath24 .
it is given by a solution of a cubic equation : @xmath36 .
it is analytic except for branch points , where 2 solutions coincide , connected by cuts .
the cuts follow lines where the value of @xmath37 $ ] on two solutions coincide .
the physical sheet is determined by an asymptotic condition at @xmath38 : @xmath39 , from the normalization of @xmath9 .
the singularities of @xmath18 are shown in fig .
[ fig : n ] . for @xmath40
the cut is on the imaginary axis . at @xmath41
the branch points bifurcate and the cuts go into the complex plane . above a certain value of @xmath42
the cut no longer goes through @xmath43 there is no discontinuity across the imaginary axis , i.e. , the chiral symmetry is restored .
-6 mm -4 mm
one can now compare this replica trick calculation to the actual distribution of eigenvalues of the random matrix @xmath12 in the quenched theory .
it is obtained numerically , by plotting the eigenvalues for an ensemble of random matrices ( [ d ] ) on fig .
[ fig : q ] .
first of all , the eigenvalues do not stay on the imaginary axis for any nonzero @xmath0 !
they spread in a blob of the @xmath44-width growing @xmath45 at small @xmath0 .
exactly the same behavior is observed in the quenched qcd calculations @xcite and leads to the puzzling conclusion that @xmath7 and vanishes in the chiral limit @xmath46 because there is no discontinuity in @xmath18 across the imaginary axis when @xmath47 .
now , however , we are able to find the explicit solution of the model and see the nature of this quenched blob .
first of all , we must learn that the limit @xmath1 of qcd with @xmath2 quarks is not smooth and the quenched theory is not a good approximation to the full theory at @xmath47 .
second , we want to know what theory has quenched qcd as its @xmath1 limit .
we find that it is a theory where each quark has a partner with the conjugate dirac operator .
the free energy reads : @xmath48 this statement can be made rigorous even beyond the random matrix approximation using equations ( [ rhov ] ) and : @xmath49 naively , when @xmath1 the conjugate quarks decouple : @xmath50 , and @xmath32 is holomorphic . in fact , the quarks @xmath51 and conjugate quarks @xmath52 mix for some values of parameters @xmath24 and @xmath0 , i.e. , precisely inside the blob
. a condensate @xmath53 develops nonzero vacuum expectation value .
this condensate carries nonzero baryon number .
it was indeed observed in a monte carlo simulation of the @xmath54 theory _ with _ quarks @xcite , which are self - conjugate in that case .
we learn now that similar phenomenon occurs in the quenched theory . calculating ( [ vnn ] )
we find @xcite for the density of eigenvalues inside of the blob : @xmath55 which one can check is in perfect agreement with the numerical data fig .
[ fig : q ] . -6 mm -4 mm
using a simple random matrix model of chiral symmetry breaking we demonstrate a general phenomenon that at _ nonzero _ @xmath0 the limit @xmath1 of qcd with @xmath2 quarks is not smooth and does not coincide with the quenched theory . on the other hand ,
the quenched theory is a smooth limit of a theory where each quark has a conjugate partner .
another way of putting this is to notice that the difference between such a theory and qcd with @xmath3 quarks is that the phase of the fermion determinant is omitted in the former .
we learn that the phase of the determinant is extremely important for a simulation of full qcd at finite @xmath0 . without this phase
the result @xmath56 is natural , as happens in quenched qcd .
this is due to the formation of the baryonic condensate , which breaks the ( replica ) chiral symmetry of quarks and conjugate quarks .
the goldstone boson associated with this breaking is the so - called baryonic pion a bound state of a quark and a conjugate antiquark .
the presence of this particle explains the result @xmath57 .
9 i. barbour _ et al _ , nucl . phys .
b * 275 * [ fs17 ] ( 1986 ) 296 ; j. b. kogut , m. p. lombardo and d. k. sinclair , phys .
d * 51 * ( 1995 ) 1282 .
a. gocksch , phys .
d * 37 * ( 1988 ) 1014 .
e. v. shuryak and j. j. m. verbaarschot , nucl .
a * 560 * ( 1993 ) 306 .
e. dagotto , f. karsch , and a. moreo , phys .
b * 169 * ( 1986 ) 421 .
m. a. stephanov , phys .
rev . lett . * 76 * ( 1996 ) 4472 . | the behavior of quenched qcd at nonzero chemical potential @xmath0 has been a long - standing puzzle .
an explicit solution is found using the random matrix approach to chiral symmetry breaking . at nonzero
@xmath0 the quenched qcd is not a simple @xmath1 limit of a theory with @xmath2 quarks : a naive ` replica trick ' fails .
a limit that leads to the quenched qcd is that of a theory with @xmath3 quarks : @xmath2 quarks with original action and @xmath2 quarks with conjugate action . |
the evolution of the baryonic component in galaxies is the hardest part to model in galaxy formation theories .
how gas is accreted onto dark matter halos , its thermal history , how it is turned into stars , and if , how and when such star formation is quenched can not be reliably predicted from first principles .
the physical processes that are involved are too complex and non linear , with hydrodynamic simulations failing by a large margin to cover the extremely wide dynamical range that would be required to describe phenomena ranging from the formation of stars and supermassive black holes to the behaviour of multiphase gas on megaparsec scales . where it is hard to progress with pure theory alone , observations can help and lead to further advances in our understanding of how galaxies form and evolve . indeed , over the last two decades
a wealth of multiwavelength observations from both ground and space facilities have provided us with a rich vision of the galaxy populations in the high redshift universe .
once correctly interpreted , such photometric , spectroscopic , and high spatial resolution data can provide us with estimates of stellar mass ( @xmath4 ) , star formation rate ( sfr ) and star formation history ( sfh ) , structure , dynamics and nuclear activity , for a very large number of galaxies , and do so as a function of redshift and environment . in particular
, the accurate measurement of stellar mass and sfr is critical for trying to establish the evolutionary links connecting galaxy populations at one redshift with those at another redshift . setting constraints on the previous sfh of individual galaxies is also relevant in this context .
all this is currently obtained by fitting the spectral energy distribution ( sed ) of synthetic stellar populations to the sed of galaxies from their multi - band photometry . in practice , a large set of template synthetic seds is constructed with different sfhs , extinctions and metallicities , and a best fit is sought by picking the sfh that minimises the @xmath5 . in this paper
we focus on star forming galaxies at @xmath6 , hence covering the epoch of major star formation activity , and explore a wide set of sfhs showing advantages and disadvantages of different parameterizations of them .
the case of passively evolving galaxies at @xmath7 was already addressed in maraston et al .
( 2006 ) .
it is well known that at @xmath2 galaxies with sfrs as high as a few @xmath8 are quite common ( e.g. , daddi et al . 2005 ) , and by analogy with the rare objects at @xmath9 with similar sfrs ( like ultra luminous infra - red galaxies , ulirgs ) , it was widely believed that such galaxies were caught in a merging - driven starburst . however ,
integral - field near - infrared spectroscopy has revealed that at least some of these galaxies have ordered , rotating velocity fields with no kinematic evidence for ongoing merging ( genzel et al .
still , the disk was shown to harbour several star - forming clumps and to have high velocity dispersion and gas fraction .
all this makes such disk quite different from local disk galaxies , and it is now well documented that the same properties apply to many similar objects at @xmath2 ( frster - schreiber et al .
2009 ) .
that high sfrs in @xmath2 galaxies do not necessarily imply _ starburst _ activity became clear from a study of galaxies in the goods fields ( daddi et al .
indeed , for starforming galaxies at @xmath6 the sfr tightly correlates with stellar mass ( with sfr @xmath10 ) , with small dispersion ( @xmath11 dex ) , as shown in figure [ sfrm ] .
only a few galaxies lie far away from the correlation : a relatively small number of passive galaxies ( with undetectable sfr , not shown in the figure ) , and sub - mm galaxies ( smg ) with much higher sfrs , which may indeed be the result of gas - rich major mergers . among starforming galaxies
, the small dispersion of the sfr for given @xmath4 demonstrates that these objects can not have been caught in a special , starburst moment of their existence .
rather , they must sustain such high sfrs for a major fraction of the time interval between @xmath12 and @xmath13 , i.e. for some @xmath14 gyr , instead of the order of one dynamical time ( @xmath15 yr ) typical of starbursts .
similar correlations have also been found at lower redshifts , notably at @xmath16 ( elbaz et al .
2007 ) , @xmath17 ( noeske et al .
2007 ) , and @xmath18 ( brinchmann et al . 2004 ) .
-selected galaxies in the goods - south field ( from daddi et al .
. the sub - sample with spectroscopic redshifts is indicated , and represents the set of galaxies for which various sed fits are attempted in this paper .
the sfrs are in @xmath19/yr and masses in @xmath19 units .
[ sfrm],scaledwidth=49.0% ] in a recent study , pannella et al .
( 2009 ) have measured the average sfr vs. stellar mass for @xmath6 galaxies using the 1.4 ghz flux from the vla coverage of the cosmos field ( schinnerer et al .
2007 ) . when combined also with data at lower redshifts ( see references above ) ,
pannella et al .
derive the following relation for the average sfr as a function of galaxy mass and time : @xmath20 where the sfr is in @xmath19/yr units and @xmath21 is the cosmic time , i.e. , the time since the big bang .
beyond @xmath22 ( @xmath23 gyr ) the specific sfr (= sfr/@xmath4 ) appears to flatten out and remain constant all the way to very high redshifts ( daddi et al .
2009 ; stark et al . 2009 ; gonzalez et al .
2009 ) . in parallel with these observational evidences
, theorists are shifting their interests from ( major ) mergers as the main mechanism to grow galaxies , to continuous _ cold stream _ accretion of baryons , that are then turned into stars in a quasi - steady fashion ( e.g. , dekel et al .
clearly , a continuous , albeit fluctuating sfr such as in these cold stream models provides a far better match to the observed tight sfr@xmath24 relation , compared to a scenario in which star formation proceeds through a series of short starbursts interleaved by long periods of reduced activity .
this is not to say that major mergers do not play a role .
they certainly exist , and can lead to real giant starbursts bringing galaxies to sfrs as high as @xmath25 , currently identified with smgs ( e.g. tacconi et al .
2008 ; menendez - delmestre et al .
2009 ) .
deriving sfrs , ages , stellar masses , etc . from sed
fitting requires making assumptions on the previous sfh of galaxies .
a widespread approach is to fit the sed of galaxies at low as well as high redshifts with so called @xmath0-models " , i.e. , synthetic seds in which the sfh is described by an exponentially declining sfr , starting at cosmic time @xmath26 , i.e. : @xmath27 where @xmath28sfr(@xmath29 ) . for a galaxy at cosmic time @xmath21 a @xmath30 fit then gives the age ( i.e. , @xmath31 , the time elapsed since the beginning of star formation ) , the sfr e - folding time @xmath0 , the reddening @xmath32 , the metallicity @xmath33 and finally the stellar mass @xmath4 via the scale factor @xmath34 .
the sfr then follows from equation ( 2 ) , where @xmath21 is the cosmic time corresponding to the observed redshift of each galaxy . of course , the reliability of the results depends on the extent to which the actual sfh is well represented by a declining exponential .
it is worth recalling that the first applications of @xmath0-models were to figure out the ages of local elliptical galaxies , and the typical result was that the age is of the order of one hubble time , and @xmath0 of the order of 1 gyr or less ( e.g. , bruzual 1983 ) .
this approach confirmed that the bulk of stars in local ellipticals are very old , hence formed within a short time interval compared to the present hubble time .
later , the usage of @xmath0-models was widely extended also to actively star forming galaxies at virtually all redshifts , to the extent that it became the default assumption in this kind of studies ( e.g. , papovich et al . 2001 ; shapley et al .
2005 ; lee et al .
2009 ; pozzetti et al . 2009 ; fster - schreiber et al .
2009 ; wuyts et al .
. such an assumed sfh may give reasonable results for local spirals , as their sf activity has been secularly declining for an order of one hubble time ( e.g. , kennicutt 1986 ) , but we shall argue that it may be a rather poor representation of the sfh of high-@xmath35 galaxies , and may lead to quite unphysical results .
cimatti et al .
( 2008 ) noted that the age of elliptical galaxies at @xmath36 turns out to be @xmath37 gyr both when using only the rest - frame uv part of the sed , and when using the whole optical - to - near - ir sed in conjunction with @xmath0-models .
however , they also noted that the former `` age '' is actually the age of the population formed in the last significant episode of star formation , while the latter `` age '' corresponds to the time elapsed since the beginning of star formation .
the near equality of these two ages suggests that the sfr peaked shortly before being quenched , rather than having peaked at an earlier time and having declined ever since .
moreover , using @xmath0 models one implicitly assumes that galaxies are all caught at their minimum sfr , which is possibly justified for local ellipticals and spirals , but not necessarily for star - forming galaxies at high redshifts that may actually be at the peak of their sf activity .
indeed , an integration of @xmath38 where @xmath39 is given by equation ( 1 ) shows that the sfr can _ increase _ quasi - exponentially with time before the effect of the declining term @xmath40 takes over , or star formation is suddenly quenched and the galaxy turns passive ( renzini 2009 ) .
mass and sfr formally increase exponentially when sfr @xmath41 , independent of time , as appears to be the case for @xmath42 ( gonzales et al .
thus , the observations of both passive and star - forming galaxies at @xmath6 suggest that the sfrs of these galaxies may well have increased with time , rather than decreased . for these reasons , in this paper we use both _
direct_-@xmath0 models , with the sfr given by equation ( 2 ) , as well as _ inverted_-@xmath0 models in which the sfr increases exponentially with time , i.e. : @xmath43 thus , direct-@xmath0 models assume that galaxies are caught at their minimum sfr and had their maximum sfr at the beginning , whereas inverted-@xmath0 models assume that galaxies are caught at their maximum sfr and had their minimum sfr at the beginning .
these two extreme assumptions may to some extent _ bracket _ the actual sfhs of real galaxies , or at least of the majority of them which , because of the tight sfr@xmath24 relation , must have a relatively smooth sfh . here
we explore which of the two assumptions gives the better fit to the sed of @xmath2 galaxies , and discuss the astrophysical plausibility of the relative results . besides these exponential sfhs we also consider the case of constant sfrs . in principle ,
other , differently motivated sfhs could also be explored , but in this paper we restrict the comparison to these three simple options , with sfrs increasing with time , decreasing , or constant .
the paper is organised as follows .
section 2 provides information on the galaxy data base that is used for our sed fitting experiments .
section 3 illustrates the procedure of sed fitting and describes the models .
section 4 presents the results we obtain for the different star formation history templates and their comparisons .
section 5 is devoted to a general discussion and presents our conclusions .
finally , we adopt a cosmology with @xmath44 , @xmath45 and + @xmath46 equal to 0.7 , 0.3 and 0.75 , respectively , for consistency with most previous works .
the age of the best - fit model is required to be lower than the age of the universe at the given spectroscopic redshift .
the sfr and masses are always in @xmath19/yr and @xmath19 units , respectively .
we use a sample of 96 galaxies in the goods - south field from daddi et al .
( 2007a ) .
these galaxies were selected via the @xmath47 diagram ( daddi et al . 2004 ) to be star forming , as confirmed by the detection in deep spitzer+mips data at @xmath48 m for @xmath49% of the galaxies in the sample .
we include only objects with accurately determined spectroscopic redshifts at @xmath50 , which were derived from a variety of surveys ( see daddi et al .
2007a for references ) , including notably ultra - deep spectroscopy from the gmass project(j .
kurk et al .
2009 , in preparation ) and from the goods survey at the very large telescope ( vanzella et al . 2008 ; popesso et al .
galaxies in the resulting sample lie in the range @xmath51 .
the multicolour photometry was obtained using the 4 bands hst+acs data in the optical ( giavalisco et al .
2004 ) , the jhk bands in the near - ir from vlt+isaac observations ( retzlaff et al . 2010 ) and from spitzer+irac ( dickinson et al .
, _ in preparation _ ) .
we used psf - matched images to build photometric catalogues from @xmath52 to @xmath53 .
the irac photometry was measured over 4@xmath54 diameter apertures and corrected to total magnitudes using corrections appropriate for point sources , and matched to the @xmath53 band using total @xmath53-band magnitudes as a comparison .
this is the same procedure that was used for the goods - s catalogues presented by cimatti et al .
( 2008 ) and daddi et al .
( 2007a ) . in summary ,
all sed fits make use of the following bands : @xmath55 plus the spitzer / irac channels 1 , 2 and 3 . besides satisfying the @xmath47 criterion ,
the sample of galaxies used in this paper is subject to the additional selection imposed by the various spectroscopic surveys mentioned above .
figure [ histo3 ] shows the distribution functions of the mass , sfr , and reddening , as derived by daddi et al .
( 2007a ) , separately for the full @xmath56 sample in the goods - south field , and for the sub - sample of 96 objects with spectroscopic redshifts .
the sample used here is somewhat biased towards higher masses , sfrs , and extinctions , but by and large for all three quantities it covers a major fraction of the range exhibited by the full sample .
this can also be appreciated by inspecting figure [ sfrm ] , where the 96 galaxies used in the present study are compared to the whole sample in daddi et al .
( 2007a ) .
therefore , we consider that the conclusions drawn from this spectroscopic sample should hold for the full sample , perhaps with the exclusion of some low mass , low sfr objects .
this database allows the sampling of galaxy seds up to the rest - frame @xmath53 band .
the seds over the whole wavelength range from the rest - frame @xmath57 to the @xmath53 band will be analysed in the next sections .
the method we adopt for the sed fitting is similar to the one used in maraston et al .
we construct composite population templates based on the stellar population models of maraston ( 2005 ) and we use an adapted version of the code _ hyper - z _ ( bolzonella et al .
2000 ) , in which the sed fitting is performed at fixed spectroscopic redshift .
we use an updated version of hyper-@xmath33 , kindly provided to us by m. bolzonella , in which 221 ages ( or @xmath0 s ) are used for each kind of sfh , instead of the 51 used in earlier versions .
the use of denser grids tends to give somewhat different results , an effect that is explored in detail in a parallel paper ( j. pforr et al . in preparation ) .
it is important to note that the code does not interpolate on the template grids , hence the template set must be densely populated .
the fitting procedure is based on maximum - likelihood algorithms and the goodness of the fit is quantified via the @xmath5 statistics .
the code computes @xmath5 for a large number of templates , which differ for sfhs , and finds the best - fitting template among them , having the reddening @xmath32 as an additional free parameter .
it is important to note that the code does not interpolate on the template grids , hence the template set must be densely populated .
the extinction @xmath58 is allowed to vary from 0 to 3 in steps of 0.2 , which corresponds to @xmath32 from 0 to 0.74 according to the reddening law of calzetti et al .
( 2000 ) , that we adopt for all fits .
by doing so we implicitly assume that the dust composition is the same in all examined galaxies and that there are no major galaxy to galaxy differences in the relative distribution of dust and young , hot stars .
differences in the extinction curve have been detected among @xmath2 galaxies , with some galaxies exhibiting the 2175 uv bump , while others not showing it ( noll et al .
2009 ) , but such differences appear to have only minor effects on the derived sfr .
extinction is unlikely to be uniform across the surface of galaxies , particularly in extremely dusty ones ( e.g. , serjeant , gruppioni & oliver 2002 ) .
however , the tightness of the sfr - mass relation for @xmath2 galaxies ( cf .
1 ) , together with the agreement of their uv - derived sfrs with those derived from the radio ( pannella et al .
2009 ) argues for such _ average extinction _ approximation to be a fairly good one , at least for the majority of the galaxies at these redshifts . for all models we considered only solar metallicity .
this is different from the approach adopted by maraston et al .
( 2006 ) in fitting the near - passive galaxies at @xmath59 , for which we considered four metallicities and several possible reddening laws .
indeed , we have noticed that varying the reddening law has only a very mild influence on the derived properties of star - forning galaxies , hence for economy we decided to stick to the reddening law that gives the best - fit in most cases , which is the calzetti law .
as for the metallicity , it is known that super - solar metallicities tend to give younger ages and higher masses and vice - versa for sub - solar metallicities ( e.g. , maraston 2005 ) .
however , metallicity effects do not change the main results of the present investigation .
indeed , our focus is on exploring the effects of adopting different functional forms for the sfhs and we restrict the main analysis at fixed , solar metallicity . on the other hand , metallicity effects over the sed of star forming galaxies are generally less important than age effects . the main difference with respect to maraston et al . ( 2006 ) , which was focused on nearly - passive galaxies , consists in the composite population templates that are used in the fits . in particular , besides a constant star formation and direct-@xmath0 models , we explore inverted-@xmath0 models for various setups of ages and @xmath0 s .
the latter models are quite a novelty in this kind of studies , and we describe them in more detail below . as mentioned above , besides constant sfr models , in this paper we consider two main sets of sfhs , namely direct-@xmath0 and inverted-@xmath0 models , where the evolution of the sfr is given by equation ( 2 ) and ( 3 ) , respectively . for both types of sfhs
, we calculate three model flavours which mostly differ with respect to how the parameter age is treated , namely : \a ) leaving the age as a free parameter ; \b ) fixing the age by fixing the formation redshift and varying only @xmath0 ; \c ) constraining the age to be larger than @xmath0 .
the age of composite models is the time elapsed since the beginning of star formation , i.e. is the age of the oldest stars @xmath60 . in the age free case
, we simply compute exponentially increasing / decreasing models for various @xmath0 s , and 221 ages for each @xmath0 .
the sed fit with these age - free models releases for each galaxy a value of age @xmath61 , @xmath32 , @xmath4 and of @xmath0 . in the case of fixed age
( case b ) , we assume that all galaxies started to form stars at the same redshift @xmath62 , and therefore the age of a galaxy follows from the cosmic time difference between its individual spectroscopic redshift and @xmath62 .
thus , in this case the best fit releases a value of @xmath0 , @xmath32 and @xmath4 . in this experiment
we take @xmath63 for the formation redshift , which implies that all galaxies are older than @xmath64 1 gyr ( ages range between 1 and 3 gyr , depending on their redshift ) . in case
c ) the age can still vary freely , but only within values that exceed the corresponding @xmath0 , so e.g. when @xmath0 is 0.5 gyr , the allowed ages for the fit must be @xmath65 gyr .
in practice , among all the fits attempted for case a ) those with age @xmath66 are not considered .
age is a default free parameter in hyper-@xmath33 , and therefore the procedure had to be modified to cope with case b ) , for which age is no longer a free parameter .
thus , for each age we have calculated a grid of models for 221 different @xmath0 s ranging from 0.05 to 10.3 gyr in step of 0.045 gyr , although the cases with @xmath67 gyr will be discussed separately .
the best - fit then finds the preferred @xmath0 .
if we were to consider all kinds of galaxies , including passive galaxies and low - redshift ones , the exponentially - increasing models should be quenched at some point , which could be done either by sfr truncation , or adding a further exponential decline .
we did not quench the models here as the galaxies we focus on are all actively starforming at the epoch of observation .
it is worth emphasising that the set - up with fixed age clearly has one degree of freedom less than those with age free , which will affect their @xmath5 values .
the quality of the fits is measured , as usual , by their reduced @xmath30 , but we believe that the plausibility of the fits can be assessed only considering the broadest possible astrophysical context . before presenting our results it is worth adding some final comments on the derived stellar masses and sfrs . in this work
we have used templates adopting a straight salpeter imf down to 0.1 @xmath68 .
this is not the optimal choice as it is generally agreed that an imf with a flatter slope or cut - off at low masses , like those proposed by kroupa ( 2001 ) or chabrier ( 2003 ) , is more appropriate for deriving an absolute value of the stellar mass or of the sfr .
however , the focus of this work is in comparing the results obtained with different sfhs , hence the slope of the imf below @xmath69 is irrelevant .
note also that the reported stellar mass @xmath70 is the mass that went into stars by the age of the galaxy .
this overestimates the true stellar mass , as stars die leaving remnants whose mass is smaller than the initial one , and the galaxy mass decrement in case of extended star formation histories is @xmath71 ( see e.g. , maraston 1998 ; maraston et al .
in this section we compare the stellar population properties derived under the three different assumptions for the sfh , namely : direct- and inverted-@xmath0 models and constant star formation , both leaving age as a free parameter , and also assuming age ( i.e. , formation redshift ) as a prior , as described above .
a first set of best fits was performed allowing the procedure to select the preferred galaxy age .
the results are reported in figures 3 to 8 , showing the histograms for the resulting ages , @xmath0 s , stellar masses , reddening , sfrs and reduced @xmath30 , respectively .
model ( middle and lower panels , respectively ) .
the age is left free in all three models .
[ ages],scaledwidth=49.0% ] ( in gyr ) .
[ taus],scaledwidth=49.0% ] , scaledwidth=49.0% ] .
[ ebv],scaledwidth=49.0% ] , scaledwidth=49.0% ] .
[ chihisto],scaledwidth=49.0% ] models with age as a free parameter .
similar plots could be shown using ages derived from direct-@xmath0 models or models with sfr = const .
[ zzform],scaledwidth=49.0% ] surprisingly , all histograms look quite similar , irrespective of the adopted sfh . in particular , the distributions of the reduced @xmath30 values are almost identical , indicating that no adopted functional shape of the sfh gives substantially better fits than another .
however , it would be premature to conclude that the resulting physical quantities ( ages , masses , etc . )
have been reliably determined .
most of the derived ages are indeed very short in all three cases , with the majority of them being less than @xmath72 yr , with several galaxies appearing as young as just @xmath73 yr ( see figure 3 ) .
we believe that these ages - intended to indicate the epoch at which the star formation started - can not be trusted as such , for the following reasons .
first , as we shall show later and as also known in the literature , the latest episode of star formation outshines older stars even when composite models are considered .
in addition , cosmological arguments render this interpretation suspicious , as we are going to discuss .
figure [ zzform ] shows the redshift of each individual galaxy as a function of its formation redshift , the latter being derived by combing the redshift and age of each galaxy .
note that with the derived ages most galaxies would have started to form stars just shortly before we happen to observe them .
if this were true , the implied cosmic sfr should rapidly vanish by @xmath74 , while this is far from happening : the cosmic sfr at @xmath74 is nearly as high as it is at @xmath2 ( e.g. , hopkins & beacom 2006 ) , and the specific sfr at a given mass stays at the same high level all the way to @xmath75 ( gonzalez et al .
2010 ) .
we consider far more likely that most of the descendants of galaxies responsible for the bulk of cosmic star formation at @xmath76 are to be found among the still most actively star forming galaxies at @xmath2 , rather than a scenario in which the former galaxies would have faded to unobservability by @xmath2 , and those we see at @xmath2 would have no counterpart at @xmath76 .
in other words , massive star forming galaxies at @xmath2 must have started to form stars at redshifts well beyond @xmath77 .
note that some star forming galaxy at @xmath76 may have turned passive by @xmath2 , but the number and mass density of passive galaxies at @xmath2 is much lower than that of @xmath76 star forming galaxies of similar mass ( e.g. kong et al .
2006 ; fontana et al . 2009 ; williams et al . 2009 ; wuyts et al .
2009b ) .
thus , the question is why does the best fit procedure choose such short ages ?
clearly , the result is not completely unrealistic , as the bulk of the light must indeed come from very young stars .
this is illustrated by the example shown in figure [ sedfit ] , for a typical case in which very short ages are returned .
the main age - sensitive feature in the spectrum is the balmer break , which for this galaxy as for most others in the sample , is rather weak or absent . with the whole optical - to - near - ir sed being dominated by the stars having formed in the recent past
, the spectrum does not convey much age information at all .
figure [ sedfit ] , lower / left panel , shows that if one forces age to be as large as 1.5 gyr ( and fix the formation redshift accordingly ) , then the balmer break deepens , and the @xmath5 worsens .
figure [ sedfit457 ] shows the sed and relative best fit spectra for one of the few galaxies for which the procedure indicates a large age ( @xmath37 gyr ) .
clearly , this larger age follows from the much stronger balmer break present in this galaxy .
note that for both direct- and inverted-@xmath0 models , in most cases @xmath78 , i.e. , the e - folding time of sfr is ( much ) longer than the age , i.e. , the sfr does not change much within the time interval @xmath31 .
this explains why both kind of models give results so similar to those in which sfr is assumed to be constant .
we shall show in the next subsections that other fits may actually result in more plausible physical solutions , even if they have a _ worse _ @xmath79 .
the extremely young ages and the insensitivity of the fit to different star formation histories when the age is left free should not be so surprising .
they are the consequence of the fact that the very young , massive stars just formed during the last @xmath80 few @xmath81 yr dominate the light at virtually all wavelengths in these actively star forming galaxies , even if they represent a relatively small fraction of the stellar mass .
the capability of a very young population to outshine previous stellar generations is illustrated in figure [ overshine ] .
synthetic spectra are shown for a composite stellar population that has formed stars at constant rate for 1 gyr , and then separately for the contributions of the stars formed during the first half and the second half of this time interval .
the contribution of the young component clearly outshines the old component at all wavelengths , making it difficult to assess the presence and contribution of the latter one , even if it represents half of the total stellar mass .
this plot demonstrates why it is not appropriate to interpret the `` age '' resulting from the previous fits as the time elapsed since the beginning of star formation , as it is formally meant to be in the fitting procedure and as might be suggested by a naive interpretation of the fitting results .
it would instead be better interpreted as the age of the stars producing the bulk of the light . in the next subsection we experiment with a different approach that may circumvent this intrinsic problem and get more robust properties of ( high - redshift ) star - forming galaxies .
clearly massive star - forming galaxies at @xmath2 can not have started to form stars just shortly before we observe them .
instead , they must have started long before , as indeed demonstrated by the fact that starforming galaxies are found to much higher redshifts . in this section
we arbitrarily assume that all our galaxies have started to form stars at the same cosmic epoch , corresponding to @xmath82 , i.e. when the universe was @xmath37 gyr old .
this implies an age for the galaxy at redshift @xmath35 which is given by @xmath83 . at first sight
such an assumption may seem a very strong one .
actually this is not the case .
we have already argued that our galaxies must have started to form stars at a redshift much higher than that at which they are observed , i.e. , @xmath84 .
therefore , the prior age changes only marginally ( few percent ) if instead of @xmath85 we had chosen @xmath86 , 7 , or more , and so does the result of the fit , i.e. , such result is fairly insensitive to the precise value of @xmath62 , provided it is well in excess of @xmath87 .
obviously , the effect is especially strong for exponentially increasing sfrs .
incidentally , we note that exponentially decreasing sfrs have been assumed also for galaxies at very high redshifts ( e.g. , @xmath88 ; stark et al .
2009 ) .
models ) with those obtained assuming exponentially increasing star formation rates ( inverted-@xmath0 models ) .
the beginning of star formation is set at the cosmic epoch corresponding to @xmath82 for all models .
[ masszf5],scaledwidth=40.0% ] for both the exponentially declining and increasing @xmath0 models , we identify the best - fitting sfhs , first allowing only values of @xmath89 gyr .
the results are shown in figures [ masszf5]-[ebvzf5 ] , in which the parameters of the best - fitting models for these two sfhs are compared . , but for a comparison of the derived star formation rates .
[ sfrzf5],scaledwidth=40.0% ] , but for a comparison of the derived @xmath0 values .
[ tauzf5],scaledwidth=40.0% ] , but for a comparison of the derived values of the reddening @xmath32 .[ebvzf5],scaledwidth=40.0% ] , but for a comparison of the reduced @xmath30 values of the best fit solutions .
[ chizf5],scaledwidth=40.0% ] figure [ masszf5 ] compares the stellar masses obtained with the two sfhs , showing that those obtained with direct-@xmath0 models are systematically larger by @xmath90 dex with respect to stellar masses derived using the inverted-@xmath0 models .
figure [ sfrzf5 ] shows the comparison of the sfrs , with the sfr from inverted-@xmath0 models being systematically higher by @xmath91 dex , and therefore the specific sfr turns out to be systematically higher by @xmath92 dex .
figure [ tauzf5 ] shows how different the @xmath0 values derived using the two sfhs are .
direct-@xmath0 models prefer very large @xmath0 s , up to @xmath93 gyr , which is to say that they prefer nearly constant sfrs . on the contrary ,
inverted-@xmath0 models prefer short @xmath0 s , typically shorter than @xmath91 gyr , hence a sfr that is rapidly increasing with time .
figure [ tauzf5 ] gives the key to understand the physical origin of the mass and sfr offsets between the two families of models .
being forced to put more mass at early times , direct-@xmath0 models pick very large @xmath0 s trying to find the best compromise for mass and sfr , and , compared to inverted-@xmath0 models , they overproduce mass at early times and not enough star formation at late times . on the other hand ,
these latter models by construction put very little mass at early times , and most of it is formed at late times . in order to compromise mass and sfr
, they may overestimate the current sfr , and may be forced to hide part of it demanding more extinction .
indeed , figure [ ebvzf5 ] compares the reddening obtained with the two opposite sfh assumptions , showing that to obtain a good fit in the rest - frame uv the higher ongoing sfr derived from inverted-@xmath0 models needs to be more dust - obscured than in the case of direct-@xmath0 models .
finally , figure [ chizf5 ] compares the reduced @xmath30 s obtained with the two sets of models . those relative to inverted-@xmath0 models are not very good , but those with direct-@xmath0 models are definitely much worse .
still , a robust choice for the sfh can not rely only on this relatively marginal advantage of inverted-@xmath0 models . in the next subsections
we try to gather independent evidence that may help favouring one or the other option .
the rest - frame uv part of the explored spectrum ( from observed @xmath52 band up to spitzer / irac channel 3 , i.e. , 5.8 @xmath94 m ) is the one which most directly depends on the rate of ongoing star formation . on the other hand , when the sfr is estimated with an sed - fitting procedure as in the previous sections , the derived sfr results from the best - possible compromise with all the free parameters , given the adopted templates .
in other words , the resulting sfr is _ compromised _ relative to the other free parameters of the fit and the adopted sfhs .
sfrs from the uv flux , corrected for extinction using the uv slope ( plus a reddening law , such as the calzetti law , calzetti et al .
2000 ) are widely derived in the literature for high redshift galaxies , and shown to be in very good agreement with independent estimates from the radio flux at 1.4 ghz ( e.g. , reddy et al .
2004 ; daddi et al .
2007a , b ; pannella et al .
2009 ) , and also from the mid - ir ( 24 @xmath94 m ) and soft x - ray fluxes , for those galaxies with no mid - ir excess ( daddi et al .
thus , the sfr derived only from the uv flux represents an _ uncompromised _ template ( in the sense that it has no covariance with other parameters of the fit ) , and therefore with respect to which we may gauge the physical plausibility of a _ whole optical - near - ir sed _ fit .
figure [ sfrebv ] compares the @xmath32 vs. sfr relations that are obtained from sed fittings with various sfhs , to those obtained using only the @xmath57 part of the spectrum .
the latter ones were derived by daddi et al .
( 2004 , 2007a ) using the bruzual & charlot ( 2003 ) models for converting @xmath95 into sfr , following madau et al ( 1998 ) .
@xmath32 is derived by mapping the @xmath96 colour into @xmath32 using the calzetti et al .
note that there are no appreciable differences in the uv between the models of bruzual & charlot ( 2003 ) and those of maraston ( 2005 ) that are used in this paper .
as clearly shown , the best agreement is between uv - derived sfrs and those obtained from sed fitting with inverted-@xmath0 models and fixed formation redshift .
notice that when leaving age as a free parameter several galaxies are found with exceedingly large sfrs and @xmath32 values ( see also section 4.1 ) , much at variance with the more robust values derived from the uv flux .
direct-@xmath0 models with fixed formation redshift are in better agreement with the uv - derived sfrs , yet they systematically underestimate the sfr as already noticed in section 4.2 .
this is further illustrated in figure [ sfrcomp ] and figure [ ebvcomp ] which show the sfr and the reddening @xmath32 from the sed fits directly compared to those from the uv . among all explored sfhs ,
the inverted-@xmath0 models clearly are in best agreement with the values derived from the uv flux and slope .
we interpret this as an indication that the sfhs of inverted-@xmath0 models models are closer to those of real galaxies , compared to the sfhs of direct-@xmath0 models .
it is easy to realise that in galaxies that follow a sfr such as that given by equation ( 1 ) the sfr tends to increase exponentially with time ( pannella et al .
2009 ; renzini 2009 ) .
more precisely , if one ignores the @xmath21 term in this equation the exponential increase proceeds with a @xmath98 gyr .
thus , in this section we present the results of performing a new set of best fits , this time assuming a fixed @xmath0 ( namely @xmath0=0.5 and 1 gyr , that bracket the empirical value ) , leaving age as a free parameter but constraining it to be larger than @xmath0 .
we also consider the case of direct-@xmath0 models with the same parameters and restrictions .
this choice to set limits on age may appear rather artificial , and indeed it is purely meant to avoid the extremely small ages that are found when age is left completely free . in practice , this exercise consists in considering only the cases with age @xmath97 among those already explored in section 4.1 .
the comparison of the sfrs derived from direct- and inverted-@xmath0 models ( figure [ sfr_agelargertau ] ) is qualitatively similar to the case of fixed formation redshift ( cf .
figure [ sfrzf5 ] ) , and the same comments apply also here .
the sfr and @xmath32 values that are obtained with these further models are then compared to those obtained from the uv in figures [ sfrcomp ] and [ ebvcomp ] .
it is apparent that inverted-@xmath0 models with @xmath0=0.5 gyr and ages larger than @xmath0 give results nearly as good as those derived for the case of fixed formation redshift . also
the case of sfr = constant and age @xmath99 0.1 gyr sfrs results in reasonable agreement with those derived from the uv .
all other explored sfhs give results at variance from those obtained from the uv , that we regard as the most robust method to estimate the sfr and reddening .
figure [ mass_agelargertau ] shows the derived stellar masses . in this parameterisation
there is less difference between the masses that are derived with the age @xmath97 constraint with respect to the case of fixing the formation redshift .
this is due to the fact that direct-@xmath0 models indicate lower masses compared to those in the case of fixed formation redshift , because of their lower age , hence shorter duration of the star formation activity .
note , however , that in most cases the smallest @xmath5 are obtained for the smallest possible age , i.e. , age@xmath100 , or @xmath91 gyr , as shown in figure [ age_agelargertau ] , again a consequence of the outshining effect .
the ability of an adopted shape of the sfh to recover the basic properties of a composite stellar population can be tested on mock galaxies with known sfhs . in a parallel project ( j. pforr et al .
, _ in preparation _ )
we use synthetic galaxies from semi - analytic models ( galics , hatton et al .
2003 ) in which the input ssps are the maraston ( 2005 ) models in the rendition of tonini et al .
( 2009 , 2010 ) .
their observed - frame magnitudes at the various redshifts are then calculated , fed into hyper-@xmath33 just as if they were relative to real observed galaxies , and best fits are sought using various template composite stellar populations .
models to derive masses and sfrs of mock galaxies at @xmath101 are also investigated by lee et al .
2009 ) . ] here we focus on an experiment involving the templates used in this paper , namely inverted- and direct-@xmath0 models and constant sf models , applied to mock star - forming galaxies at redshift 2 .
the model spectra are reddened according to the ongoing sfr , using the empirical evidence that the @xmath32 is proportional to the sfr ( e.g. , daddi et al .
2007a ) of each individual galaxy , with @xmath102 .
extinction as a function of wavelength is then applied following the law of calzetti et al .
note that we do not take such mock galaxies as representative of real galaxies .
the goal of the experiment is to show what happens if the sed of a galaxy with a certain sfh is used to derive its sfr and stellar mass using a different sfh .
figures [ mock1 ] and [ mock2 ] show the results .
the upper left panel of figure [ mock1 ] shows the _ input _ values , i.e. the distribution of sfrs and masses of the semi - analytic models at redshift 2 ( black points ) .
the other panels show the _ output _ values , i.e. the sfrs and masses that are obtained by fitting the sed of the mock galaxies with the various templates , as indicated , over - plotted to the input values .
the inverted-@xmath0 models with fixed formation redshift can recover the input quantities strikingly well ( middle panels ) . towards the low - mass end ( @xmath1031.5pt@xmath104 ) the sfr
is somewhat underestimated , suggesting that the assumed start of sf is too early .
note also , as shown by the two middle panels in figure [ mock1 ] , that the result is almost independent of the assumed formation redshift , as already argued in section 4.2 .
none of the other sfh templates do as well . with the free age mode
a substantial number of outliers with very high sfrs are obtained . the green line in figure [ mock1 ] and
figure [ mock2 ] is a fit to the outliers found among the real goods galaxies from the direct-@xmath0 models ( the blue objects in figure [ sfrcomp ] and figure [ ebvcomp ] ) , and it is quite instructive to note that when the fit produces wrong values , those follow the same relation of fake values in mock galaxies .
finally , figure [ mock2 ] , analogous to figure [ mock1 ] , shows the result of the same experiment with mock galaxies , but now using the template sfhs discussed in the previous subsection , i.e. , with fixed @xmath0 and age @xmath97 .
the inverted-@xmath0 models with age @xmath105 , and @xmath106 gyr give fairly good fits to the actual values in the mock galaxies , but not as good as those with a prior on the formation redshift ( see figure [ mock1 ] , middle panels ) .
finally , note that some outliers among the mock galaxies , a few with low sfr for their masses ( hence almost passive ) , are poorly reproduced by the explored sfhs , as all assume that star formation is still going on .
models ) with those obtained assuming exponentially increasing star formation rates ( inverted-@xmath0 models ) . in both models ,
@xmath0 has been fixed to two values of 0.5 and 1 gyr , and the age is constrained to be larger than @xmath0 .
[ sfr_agelargertau],scaledwidth=49.0% ] , but for a comparison of the derived stellar masses .
[ mass_agelargertau],scaledwidth=49.0% ] and ages constrained to be larger than @xmath0 , for @xmath106 gyr ( black histogram ) and @xmath107 gyr ( shaded histogram ) .
[ age_agelargertau],scaledwidth=49.0% ] the reason why inverted-@xmath0 models fit better is very simple .
most mock galaxies constructed from semi - analytic models exhibit secularly increasing sfrs , as do the cold - stream hydrodynamical models of dekel et al .
( 2009 ) as one expects from the empirical sfr - mass relation ( e.g. , renzini 2009 ) .
therefore , we believe that quite enough arguments militate in favour of preferring exponentially increasing sfrs , as opposed to decreasing or constant . as one can notice from figure [ tauzf5 ] , using inverted-@xmath0 models the vast majority of the best fits tend to cluster close to the minimum allowed value of @xmath0 , i.e. , 0.3 gyr , a limit that was imposed based on the value of @xmath0 implied by the empirical sfr@xmath24 relation , as discussed in section 4.4 .
but what happens if this limit is removed , and the best fit procedure is allowed to chose any value of @xmath0 down to 0.05 gyr ?
the results are shown in figure [ smalltau ] : several galaxies are now `` best fitted '' with very small @xmath0 s , implying sfhs in which most stars formed just shortly before the time at which the galaxy is observed .
this is to say that leaving the procedure free to choose very small values of @xmath0 results in best fits that closely resemble those obtained with age as a free parameter , that were discussed ( and rejected ) in section 4.1 .
figure [ smalltau ] shows that also in this case sfrs well in excess of those derived from the uv are derived for several galaxies .
therefore , also inverted-@xmath0 models with unconstrained @xmath0 s suffer from the outshining phenomenon that plagues models with age as a fully unconstrained free parameter .
the lesson we learn from these experiments is that only by setting a prior on age one derives astrophysically more acceptable solutions .
depending on the adopted shape of the sfh , this prior can be a minimum age ( for constant sfr models ) , or setting age@xmath97 ( for direct-@xmath0 models ) , or finally setting a minimum acceptable value for @xmath0 ( for inverted-@xmath0 models ) .
this kind of fix looks quite artificial , and admittedly the resulting procedure is far from being elegant .
it is a simple way of avoiding the outshining effect , and obtain sfrs that are in agreement with those derived from other direct methods , such as from uv , radio , or mid - ir ( e.g. , daddi et al .
2007a , b ; pannella et al 2009 ) . while sfr and reddening can be best evaluated by just using the uv part of a galaxy s spectrum , the determination of the stellar mass , such a crucial quantity for understanding galaxy evolution , requires the full sed fitting .
hence , it requires exploring a variety of assumed sfhs .
we have shown that there are some adopted sfhs that , constrained to avoid the outshining problem , give sfrs and @xmath32 values in excellent agreement with those derived from the uv .
now , how different are the derived stellar masses when using these viable options for the sfh ? and , ultimately , how much can we trust such derived masses ? figure [ masse ] - first and second panel - compares the stellar masses that are derived for the real galaxies studied in this paper using these viable sfhs , namely inverted-@xmath0 with high formation redshift , inverted-@xmath0 with age constrained , and constant sf with age constrained .
the corrensponding masses agree with each other quite well , systematic differences are quite small ( @xmath108 dex ) , and also the scatter is quite modest , of the order of 0.1 dex .
can this consistency suffice to conclude that the adopted procedures give a robust estimate of stellar masses ? in this respect , the fairly accurate recovery of the masses of mock galaxies should increase our confidence . yet , the problem of outshining remains .
what we consider viable options are sfhs described by extremely simple functions , i.e. , either constant or exponential .
nature certainly realises far more complex sfhs .
for example , an early burst of star formation , followed by a continuous sfr , could be easily missed by the best fit procedure , by remaining out - shined by the successive and ongoing star formation .
thus , we suspect that derived masses may underestimate the actual stellar mass , at least for those galaxies with early massive starbursts .
we note that figure [ masse ] , third panel compares also the masses derived from direct-@xmath0 models with unconstrained age , to those derived from the viable sfhs .
clearly , such direct-@xmath0 models can substantially underestimate the stellar mass , while they overestimate the sfr ( cf .
figure [ sfrcomp ] ) . to conclude this section ,
figure [ massefinali ] shows the _ absolute _ values of the stellar mass of star - forming galaxies at redshift @xmath109 that are obtained using our preferred star formation history .
we have tried different priors for the sfh of @xmath2 galaxies , and used a @xmath30 best fit approach to derive the basic properties of such galaxies : sfr , stellar mass , reddening , age , etc .
we do not expect that any of the simple mathematical forms adopted for the sfh ( constant sfr , exponentially decreasing , exponentially increasing ) is strictly followed by real galaxies .
we expect , however , that the two opposite choices for the exponential case are sufficiently extreme to encompass the real behavior of galaxies , or at least those having experienced a quasi - steady sfr over most of their lifetime .
we try various tests to distinguish which of the various options is the most acceptable one ( or the least unacceptable ) for giving the astrophysically most plausible values for the basic properties of star forming galaxies in the redshift range between @xmath110 and @xmath111 .
we find that by leaving age as a free parameter the best fit procedure delivers implausibly short ages , no matter which sfh template is adopted .
this is so because the sed is dominated by the stars that have formed most recently , and they outshine the older stellar populations that may inhabit these galaxies .
thus , introducing a prior on the beginning of star formation appears to be necessary . assuming that star formation started at high redshift ( i.e. , @xmath112
, the precise value being almost irrelevant ) provides a more credible framework , with models with exponentially increasing sfr ( that we call inverted-@xmath0 models ) giving systematically higher sfrs for the @xmath2 galaxies in the test sample , and lower stellar masses , compared to models with exponentially decreasing sfr .
these two systematic differences together imply a specific sfr that is a factor of @xmath113 higher in inverted-@xmath0 models , compared to direct-@xmath0 models , when star formation is assumed to start at high redshift ( e.g. , @xmath82 ) in both cases .
on the other hand , inverted-@xmath0 models offer various notable advantages : 1 ) they indicate sfrs and extinctions in excellent agreement with those derived from the rest - frame uv part of the sed , which most directly relates to the ongoing star formation and reddening , 2 ) they fairly accurately recover the sfrs and masses of mock galaxies constructed from semi - analytic models , in which sfr are indeed secularly increasing in most cases , and 3 ) have systematically better reduced @xmath30 values compared to models with exponentially decreasing sfrs , though one can not consider them fully satisfactory .
these advantages are somewhat reduced if the best fit procedure is allowed to pick very small values of @xmath0 , down to 0.05 gyr , as in such case derived sfrs tend to be somewhat overestimated compared to those derived from the uv .
we have also explored star formation histories ( such as constant sfr , or exponentially decreasing ) in which age is a priori constrained to avoid very small values ( e.g. , age @xmath114 gyr , or age @xmath97 , or fixed @xmath0 ) .
adopting such priors , though it may seem quite artificial , gives more plausible solutions with respect to cases in which age is left fully free .
this is due to the mentioned outshining effect of very young stellar populations over the older ones .
still , models with exponentially increasing sfrs tend to give systematically better results , as judged from the consistency with the uv derived sfrs .
besides the above arguments , there are other reasons to prefer stellar population models constructed with exponentially increasing sfrs , including : \1 ) the almost linear relation between sfr and stellar mass ( sfr @xmath10 ) , and the small scatter about it , that are empirically established for redshift @xmath87 galaxies implies that they can indeed experience a quasi - exponential growth at these cosmic epochs .
later , some process may quench their star formation entirely and turn them into passively evolving _ ellipticals _ , or keep growing until the secular decrease of the specific star formation rate ( the @xmath40 term in equation ( 1 ) ) takes over , and galaxies ( such as spirals ) continue forming stars at a slowly decreasing rate all the way to the present epoch ( renzini 2009 ) .
\2 ) the direct observation of the evolution of the luminosity function in rest - frame uv ( which is directly related to sfr ) shows that the characteristic luminosity at 1600 ( @xmath115 ) brightens by over one magnitude between @xmath116 and @xmath117 ( bouwens et al .
since extinction is likely to increase with time following metal enrichment , the increase of the rest - frame uv luminosity can only be due to an increased sfr in individual galaxies .
the aim of this paper is to explore which assumptions and procedures are likely to give the most robust estimates of the basic stellar population parameters , when using data that extend from the rest - frame uv to the near - ir .
we argue that ongoing sfrs are best derived at wavelengths below about one micron by using the uv part of the sed of galaxies , and inverted-@xmath0 models give sfrs in excellent agreement with them .
however , the fit to the full optical to near - infrared sed is required to derive the mass in stars .
we show that the sfhs , with a prior on age or @xmath0 to reduce the outshining effect , give results that agree quite well with each other , giving some confidence on the reliability of the derived masses . however , it is still possible that the mass may be somewhat underestimated for those galaxies in which a substantial fraction of the stellar mass was produced in a strong burst , early in their evolution . goods star - forming galaxies as derived from inverted-@xmath0 models , with @xmath118 and @xmath89 gyr .
[ massefinali],scaledwidth=49.0% ] at first sight the higher sfrs and lower stellar masses indicated by inverted-@xmath0 models ( compared to direct-@xmath0 models , both with fixed formation redshift ) may exacerbate an apparent mismatch between the _ cosmic _ ( volume averaged ) sfr history ( @xmath119 ) and the empirical stellar mass density ( @xmath120 ) , such that the integration of sfr@xmath121 tends to overproduce the stellar mass density at most / all redshifts ( e.g. , wilkins , trentham & hopkins 2008 ) .
however , sfrs are more often derived from direct sfr indicators , such as the rest - frame uv , radio flux , h@xmath122 flux , or mid / far - infrared , rather than from sed fits .
therefore the 0.5 dex effect on the sfrs relative to the direct-@xmath0 models should not affect this discrepancy .
there may remain the 0.2 dex effect on the masses , but inverted-@xmath0 models , by construction minimising star formation at early times , may systematically underestimate the stellar mass of galaxies , in particular if part of it was formed in early bursts .
thus , with this caveat in mind , the use of inverted-@xmath0 models with fixed formation redshift should not appreciably exacerbate the mismatch problem mentioned above . instead
, direct-@xmath0 models with unconstrained age most certainly worsen this problem as they underestimate the stellar mass ( see figure [ masse ] , right panel ) , and overestimate the sfr ( see figure [ sfrcomp ] ) .
we conclude that the use of synthetic stellar populations with an exponentially declining sfr should be avoided in the case of star forming galaxies at redshift beyond @xmath37 . exponentially increasing sfrs , with e - folding times of @xmath123 gyr , and a high starting redshift
appear to provide astrophysically more plausible results , and while nature may not closely follow such a simple functional behavior , they should be preferred in deriving sfrs and stellar masses of high redshift galaxies .
in any event , we believe that maximum likelihood parameters are not necessarily unbiased estimators of star formation rates and masses of high - redshift galaxies , but the broadest possible astrophysical context should be considered .
we thank the referee stephen serjeant for a prompt and useful report .
we are grateful to andi burkert for valuable comments on a very early draft of the paper , and to laura greggio and daniel thomas for useful discussions .
we acknowledge micol bolzonella for her constructive help with the _ hyper - z _ code .
cm , jp , and ct acknowledge the marie - curie excellence team grant `` unimass '' , ref .
mext - ct-2006 - 042754 of the training and mobility of researchers programme financed by the european community .
ed acknowledges the funding support of the erc - stg grant upgal-240039 , anr-07-blan-0228 and anr-08-jcjc-0008 .
ar is grateful for the hospitality of the icg during the final rush for the completion of this paper .
some of the data used here are part of the goods spitzer space telescope legacy science program , that is supported by nasa through contract number 1224666 issued by the jpl , caltech , under nasa contract 1407 . | fitting synthetic spectral energy distributions ( sed ) to the multi - band photometry of galaxies to derive their star formation rates ( sfr ) , stellar masses , ages , etc .
requires making a priori assumptions about their star formation histories ( sfh ) .
a widely adopted parameterization of the sfh , the so - called @xmath0-models where sfr @xmath1 is shown to lead to unrealistically low ages when applied to a sample of actively star forming galaxies at @xmath2 , a problem shared by other sfhs when the age is left as a free parameter in the fitting procedure .
this happens because the sed of such galaxies , at all wavelengths , is dominated by their youngest stellar populations , which outshine the older ones .
thus , the sed of such galaxies conveys little information on the beginning of star formation , i.e. , on the age of their oldest stellar populations . to cope with this problem , besides @xmath0-models ( hereafter called direct-@xmath0 models ) , we explore a variety of sfhs , such as constant sfr and inverted-@xmath0 models ( with sfr @xmath3 ) , along with various priors on age , including assuming that star formation started at high redshift in all the galaxies in the test sample .
we find that inverted-@xmath0 models with such latter assumption give sfrs and extinctions in excellent agreement with the values derived using only the uv part of the sed , which is the one most sensitive to ongoing star formation and reddening .
these models are also shown to accurately recover the sfrs and masses of mock galaxies at @xmath2 constructed from semi - analytic models , which we use as a further test .
all other explored sfh templates do not fulfil these two test as well as inverted-@xmath0 models do .
in particular , direct-@xmath0 models with unconstrained age in the fitting procedure overstimate sfrs and underestimate stellar mass , and would exacerbate an apparent mismatch between the cosmic evolution of the volume densities of sfr and stellar mass . we conclude that for high - redshift star forming galaxies an exponentially increasing sfr with a high formation redshift is preferable to other forms of the sfh so far adopted in the literature .
[ firstpage ] galaxies : evolution galaxies : starbursts galaxies : high - redshift |
the recently - discovered soft x ray transient ( sxt ) xte j1118 + 480 ( = kv uma ; remillard et al 2000 ) lies at high galactic latitude , close to the lockman hole in the local ism .
the exceptionally low interstellar absorption permits unprecedented wavelength coverage ( mauche et al 2000 , hynes et al 2000 , mcclintock et al 2001b , chaty et al 2002 ) .
a low amplitude photometric modulation with period 4.1 hr was reported by cook et al ( 2000 ) and uemura et al ( 2000 ) .
dubus et al ( 2000 ) present spectroscopy showing an emission line radial velocity ` s - wave ' modulation at a period close to this .
this period was recently confirmed as orbital , and a mass function of @xmath3 was measured by mcclintock et al ( 2001a ) and wagner et al ( 2001 ) .
therefore the accretor clearly has to be a black hole .
we present the uv line spectrum of , which we compare with that of another recent sxt , xtej1859 + 226 .
full analyses of these hst spectra will be presented elsewhere ( hynes et al 2002a , b ; chaty et al 2002 ) .
here we focus on the prior evolution of .
we obtained _ hst _ spectra of xtej1859 + 226 near the peak of outburst on 1999 october 18 and 27 and november 6 . at each epoch
far - uv spectra were obtained using the low - resolution g140l grating and wide ( 0.5arcsec ) slit , yielding wavelength coverage 11501730 and resolution @xmath4 .
we used the standard pipeline data products except that the spectral extraction was done by hand using the iraf implementation of optimal extraction .
this gave a significantly cleaner removal of geocoronal emission lines ( principally ly @xmath5 and oi 1302 ) .
after extracting the one - dimensional spectra ( with the pipeline wavelength and flux calibrations ) , we resampled them onto a common wavelength grid and took an exposure - time weighted average . for xtej1118 + 480
, we obtained high - resolution ( echelle ) spectra using the e140 m grating and 0.2arcsec square aperture on 2000 april 8 , 18 and 29 , may 28 , june 24 , and july 8 .
the pipeline extraction was adequate , although a few high pixel values had to be replaced by hand .
the extracted spectra were rebinned into 0.2 bins and averaged with exposure - time weighting .
the process is clearly not perfect . in particular , there are a number of abrupt steps in the spectrum at longer wavelengths .
these artefacts , at @xmath6per cent level , arise from inconsistencies in the relative flux calibration of adjacent echelle orders .
figure 1 gives the uv spectra of and xte j1859 + 226 .
both show interstellar absorption features . in these
appear sharper because the spectral resolution was higher .
the difference in the equivalent widths of the emission lines in the two systems is striking ( fig . 1 , table 1 ; haswell et al 2000 ) .
in xtej1859 + 226 civ is the strongest line , with nv , ciii , ov , siiv , and heii also strongly present .
these emission line strengths are roughly as expected for an x - ray bright lmxb ( cf sco x-1 , kallman , boroson , & vrtilek 1998 ) . compared to cataclysmic variables ( cvs ) , the nv / civ ratio is relatively large , while siiv / civ is small , as expected for a high ionisation parameter ( mauche , lee & kallman 1997 ) .
in contrast , for xtej1118 + 480 , the carbon and oxygen lines are undetectable , while the nv emission appears enhanced .
a similar anomalous emission line spectrum is seen in the magnetic cv ae aqr , in which the nv is much stronger than the civ line ( jameson et al 1980 , eracleous et al 1994 ) . [
cols="<,<,>,>,>",options="header " , ] the niv 1718@xmath7 line would add a valuable additional constraint on the interpretation of the line spectrum in terms of the photo - ionisation conditions , so we carefully assessed our spectra to determine whether it was present .
unfortunately our well - exposed e140 m echelle spectrum stops at 1710@xmath7 and our e230 m echelle spectrum suffers from shorter total exposure and the low sensitivity of the nuv mama in this region .
1718@xmath7 falls right on the boundary between the two stis mama detectors , and consequently neither is optimised at this wavelength .
we find no strong ( @xmath8 ) line in the spectrum , but this upper limit provides a weak constraint due to the poor quality of our data at this wavelength . the ly@xmath2 absorption in ( fig .
1 ) clearly has a sharp core , which is probably interstellar , with very broad wings .
the continuum slope is well constrained by the spectrum longwards of @xmath9 .
extrapolating this continuum is suggestive of broad ly@xmath2 extending from @xmath10 to @xmath11 .
the combination of noise , nv emission , and absorption features hinder measurement of the ly@xmath2 wings , but they definitely extend at least from @xmath12 to @xmath13 .
the broad absorption may be damped ly@xmath2 from the hi in our galaxy ( bowen , priv.comm . ) .
if we assume , instead , that it is due to absorbing gas executing keplerian motion around the black hole , and take the largest width suggested by our data , the full width implies velocities of @xmath14 and absorbing material at distances as close as @xmath15 .
taking the securely determined line width would imply absorption from keplerian material at @xmath16 .
an independent line of argument based on the shape of the spectral energy distribution , and dependent on the value of the neutral hydrogen column density , @xmath17 , has been used to estimate the disc inner radius ( hynes et al 2000 , mcclintock et al 2001 , chaty et al 2002 ) .
hynes et al ( 2000 ) estimate @xmath18 , a choice which suggests the disc is terminated at @xmath19 .
this is inconsistent with the keplerian interpretation of the broad ly@xmath2 absorption .
alternatively , informed by the _ chandra _ data , mcclintock et al ( 2001 ) suggest that @xmath17 could be as high as @xmath20 , with @xmath21 .
chaty et al ( 2002 ) adopt @xmath22 , which leads to @xmath23 .
the latter two values are consistent with the keplerian interpretation of the broad ly@xmath2 absorption .
the feature near @xmath24 does not have an obvious identification and we believe it may be spurious , although it does not correspond to any known instrumental artefacts ( sahu 2001 priv .
this issue is under investigation .
the civ 1550@xmath25 and the nv 1240@xmath7 lines are both resonance lines of lithium - like ions , and are produced under essentially the same physical conditions .
kallman and mccray ( 1982 , hereafter km82 ) present theoretical models of compact x - ray sources predicting the ionisation structure expected in a wide variety of astrophysical situations , including galactic x - ray binaries . km82
present 8 distinct models ; in each one the source luminosity , @xmath26 , spectral shape , and gas density , @xmath27 , are fixed . for each model the output includes the dependence on ionisation parameter , @xmath28 ( where @xmath29 is the distance from the central source ) of the abundances ( as a fraction of the total elemental abundance ) of the ions of common elements .
km82 consider ( i ) three species detected in both and xtej1859 + 226 : heii , siiv , and nv ; and ( ii ) three species detected only in the spectrum of xtej1859 + 226 : civ , ciii , and ov . strikingly similar ionisation parameters were required for the two sets of lines in all models . in almost all cases the presence of nv places the tightest constraint on @xmath30 , and almost invariably this is encompassed within the range producing civ . in the minority of cases where there exists an ionisation parameter which produces nv and not civ , this range also produces ov . consequently , km82 give no set of photoionisation parameters which would predict production of nv in while suppressing the civ and ov .
the x - ray spectrum of is clearly different from that of xtej1859 + 226 : the former is an extended power - law ( hynes et al 2000 , mcclintock et al 2001 , frontera et al 2001 , chaty et al 2002 ) with a cut - off at @xmath31 kev indicative of the low - hard state , and often attributed to an advection - dominated accretion flow ( adaf ; e.g. esin et al 2001 ) ; while in the early stages of the outburst the latter was dominated by the soft thermal black - body disk spectrum which turned over by @xmath32 kev ( hynes et al 2002a ) , typical of the high - soft state .
ho et al ( 2000 ) note that the characteristically harder ionising spectrum of an adaf lowers the effective ionisation parameter , and hence favours the production of lower ionisation species .
hence , if anything , the differences in the x - ray spectra should cause civ to be relatively prominent ( compared with nv ) in , rather than being suppressed as we observe .
hamann and ferland ( 1992 ) used the observed civ 1550@xmath25 and nv 1240@xmath7 lines to estimate the n / c abundance ratio in high redshift qsos .
their photionisation calculations show that for a broad range of ionisation parameters the nv 1240@xmath7 / civ 1550@xmath25 line ratio is lower at lower metalicities even when the n / c abundance ratio is kept constant .
this means that if anything we should expect the nv 1240@xmath7 line to be suppressed relative to civ 1550@xmath25 in which is a halo object , and consequently might be expected to have a lower metalicity than that of xtej1859 + 226 .
a definitive abundance analysis for would require exact knowledge of the geometry , densities , and ionising spectrum in the regions emitting the uv lines . while hynes et al ( 2000 , 2002b ) gives some information on the geometry , and chaty et al .
( 2002 ) gives good coverage of the x - ray spectrum , the euv spectrum remains open to significant uncertainty ( compare the range of dereddened euv spectra in chaty et al 2002 , mcclintock et al 2001 , esin et al 2001 , hynes et al 2000 , merloni et al 2000 ) .
furthermore the range of densities present in the uv line emitting gas is difficult to determine .
consequently a quantitative abundance determination of the type carried out in stellar photospheres , where the physical conditions are well - known , is not possible .
hence it is impossible to rule out definitively a contrived spectrum and gas distribution which would produce the set ( i ) lines while suppressing the set ( ii ) lines .
however as * set ( i ) and set ( ii ) ions require essentially identical ranges of @xmath30 * the ionising spectrum of might be expected to favour civ relative to nv * the lower metallicity expected for might be expected to boost the civ1550@xmath25 / nv 1240@xmath7 line ratio by far the simplest explanation of our observed uv line spectra is that a substantial underabundance of carbon is present in the surface layers of the mass donor star in . in xtej1118 + 480 ( as in ae aqr ) ,
the underabundance of carbon compared to nitrogen strongly suggests that the material in the accretion flow is substantially cno processed ( clayton 1983 ) .
if the cno bi cycle achieves equilibrium , it converts most cno nuclei into @xmath33n . at the lower temperatures typical in @xmath34 ms stars ( which we shall suggest as the progenitor of the companion in xtej1118 + 480 ) this conversion is much more at the expense of @xmath35c than @xmath36o , and in fact @xmath37o increases .
hence the oxygen abundance is not expected to decrease by much in xtej1118 + 480 and this argument does not of itself explain the observed weakness of ov 1371@xmath25 .
however civ1550@xmath25 , nv 1240@xmath7 are resonance lines , whose ratio must be regarded as a robust indicator of conditions in the line emitting gas , whereas ov 1371@xmath38 is a subordinate line , which may not be formed efficiently in a low density photoionized gas .
we thus regard the spectrum of xtej1118 + 480 as indicating cno processing .
the spectrum shown in figure 1 implies that the companion star in must be partially nuclear evolved and have lost its outer layers , exposing inner layers which have been mixed with cno processed material from the central nuclear
burning region .
mass transfer must therefore have been initiated from a somewhat evolved and sufficiently massive donor of @xmath39 , and thus with an initial period @xmath40 hr . the main difficulty in understanding the current status of is to explain its observed short orbital period ; in general , significant nuclear evolution is associated with a period _ increase _ , rather than the inferred decrease .
there are two ways in which this decrease could have occurred : \(i ) the system came into contact with initial secondary / primary mass ratio @xmath41 and underwent a phase of thermal timescale mass transfer until @xmath42 . at the end of this phase
the period would have either decreased or not increased greatly .
subsequent orbital angular momentum loss could then pull the system in to the observed @xmath43 hr . this case is similar to the likely evolution of ae aqr ( schenker et al . , 2002a ) .
\(ii ) the system came into contact with @xmath44 already , and subsequent evolution was driven by angular momentum loss towards shorter periods .
this case resembles that proposed for a062000 by de kool et al ( 1983 ) , with a severe additional constraint : the donor must be sufficiently evolved to provide the observed surface abundance ratios .
this requires a near equality of the nuclear and orbital angular momentum loss timescales @xmath45 ( cf fig
. 2 ) . put another way , @xmath46 must have been very close to the ` bifurcation period ' defined by pylyser & savonije ( 1988 ) . case ( i ) above can be ruled out by the current system parameters . at the end of thermal timescale mass transfer
we must have @xmath42 , which would require a donor mass @xmath47 and an orbital period @xmath1 hr .
but for such a system @xmath48 ( cf fig .
2 ) , implying evolution to _ longer _ periods , in stark contrast to the current 4.1 hr . put another way , there is an upper limit on the binary mass @xmath49 for case ( i ) evolution to give nuclear evolved donors at short orbital periods .
the near equality @xmath50 can only realistically hold if magnetic braking dominates angular momentum loss , so we require @xmath51 at the end of the thermal timescale episode .
but since @xmath52 there , we must have total binary mass @xmath53 in case ( i ) evolutions .
we are therefore left only with case ( ii ) above .
the secondary mass when the system came into contact is now constrained to be close to @xmath0 : cno processing excludes significantly lower masses , while the requirement for @xmath54 excludes significantly higher masses .
, 4 show the evolution of such a system : a @xmath55 ms star was allowed to evolve and fill its roche lobe in a 15 hr binary with a @xmath56 bh .
the predicted mass transfer rate throughout the evolution , including its current value @xmath57 , shows that the system would indeed have appeared as a soft x
ray transient , according to the irradiated disc criterion for black hole systems given by king , kolb & szuszkiewicz ( 1997 ) .
it should be noted that the change in c / n is mostly due to a drastic depletion of @xmath35c , supplemented by a more modest increase in @xmath33n .
the total o abundance at the surface has only decreased very little ; in fact the o / n ratio is reduced by only a factor of 3 to 5 at most , predominantly due to the increase of n. interestingly , this model also passes through @xmath58 at around @xmath59 , showing much weaker n enhancements ( i.e. almost normal abundances , cf fig .
thus it can also be considered a reasonable model for a062000 . a subsequent paper ( schenker et al . , 2002b )
will explore this and related evolutions in detail .
finally we can compare further properties of the secondary in our model to observations : figure 4 shows the evolution of effective temperature with orbital period , together with a simple mapping to spectral types .
the procedure is based on a conversion of observed colours to spectral types ( beuermann et al . ,
1998 ) and a set of non - grey stellar atmospheres ( hauschildt et al . , 1999 )
providing the colours for each set of stellar boundary conditions . for solar metallicity
the resulting spt turns out to be only a function of effective temperature with a very weak dependence on surface gravity .
however this mapping should only be considered a first estimate , as the evolutionary code in its current form still uses grey atmospheres .
furthermore the whole conversion is based on an observed set of unevolved stars and theoretical zams models , i.e. no fully self - consistent application to a partially evolved ms star such as shown in fig .
4 is possible for the time being .
allowing for the uncertainties described above , our model may be slightly too cool ( m2 rather than k5-m1 as mentioned in mcclintock et al ( 2001 ) , or k7-m0 by wagner et al ( 2000 ) .
similar evolutionary tracks of strongly evolved cvs are known to get hotter again ( baraffe & kolb , 2000 ) at short periods , so a confirmed secondary spectral type would provide insight as to the state of nuclear evolution in the donor star of .
figure 4 shows that in any case that the donor will currently appear to be close to the zams , and thus that the distance estimate derived by mcclintock et al ( 2000 ) on this basis is likely to be quite good .
we have compared the uv spectra of and xtej1859 + 226 .
the former shows strong evidence of cno processing , which tightly constrains the evolution of the system .
must have first reached contact with the donor star having a significantly nuclear evolved core , at a period where nuclear and angular momentum loss timescales were comparable .
this in turn constrains the end point of the earlier common
envelope phase : immediately after the common
envelope phase the system had a wide enough separation that significant nuclear evolution could occur before contact was achieved . in a future paper we will investigate such evolutions systematically .
cah and rih gratefully acknowledge the superb support of tony roman , kailash sahu , and all involved in the implementation of our time - critical _ hst _ observations .
support for proposals go-08245 and go-08647 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 , inc .
, under nasa contract nas5 - 26555 .
this work was supported by the leverhulme trust f/00 - 180/a .
theoretical astrophysics research at leicester is supported by a pparc rolling grant .
we thank christian knigge for discussion about the nature of the @xmath24 feature , and an anonymous referee for useful comments . | we compare uv spectra of the recent soft x ray transients and xtej1859 + 226 .
the emission line strengths in strongly suggest that the accreting material has been cno processed .
we show that this system must have come into contact with a secondary star of about @xmath0 , and an orbital period @xmath1 hr , very close to the bifurcation value at which nuclear and angular momentum loss timescales are similar .
subsequent evolution to the current period of 4.1 hr was driven by angular momentum loss . in passing through a period of 7.75
hr the secondary star would have shown essentially normal surface abundances .
could thus represent a slightly later evolutionary stage of a062000 .
we briefly discuss the broad ly@xmath2 absorption wings in .
* key words : * accretion , accretion discs - stars : individual xtej1118 + 480 - x rays : stars .
+ = = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] |
the notion of extra dimensions and that they can be probed by gravitons and eventually non - standard matter has been the focus of attention in recent years .
these models usually yield the correct newtonian @xmath3-potential at large distances because the gravitational field is quenched on sub - millimeter transverse scales .
this quenching appears either due to finite extension of the transverse dimensions @xcite or due to sub - millimeter transverse curvature scales induced by negative cosmological constants @xcite .
a common feature of both of these types of models and also of the old kaluza - klein type models is the prediction of deviations from four - dimensional einstein gravity at short distances .
if the transverse length scale is not too small , this implies the possibility to generate bulk gravitons in accelerators or stars @xcite .
the model of dvali , gabadadze and porrati ( dgp ) @xcite , see @xcite for extensions , is different , predicting that 4-dimensional einstein gravity is a short - distance phenomenon with deviations showing up at large distances .
the transition between four and higher - dimensional gravitational potentials in the dgp model arises as a consequence of the presence of both the brane and bulk einstein terms in the action .
further , it was observed in @xcite that the dgp model allows for an embedding of the standard friedmann cosmology in the sense that the cosmological evolution of the background metric on the brane can entirely be described by the standard friedmann equation plus energy conservation on the brane .
this was later generalized to arbitrary number of transverse dimensions in @xcite .
for a recent and comprehensive review of the phenomenology of dgp cosmology , the reader is referred to @xcite .
an interesting observation made a few years ago is that the expansion of our universe is currently undergoing a period of acceleration which is directly measured from the light - curves of several hundred type ia supernovae @xcite and independently from observations of the cosmic microwave background ( cmb ) by the wmap satellite @xcite and other cmb experiments @xcite . however , the mechanism responsible for this acceleration is not well understood and many authors introduce a mysterious cosmic fluid , the so called dark energy , to explain this effect @xcite .
recently , it has been shown that such an accelerated expansion could be the result of a modification to the einstein - hilbert action @xcite . a scenario where the issue of cosmic acceleration in the framework of higher order theories of gravity in @xmath4 is addressed
can be found in @xcite .
one of the first proposals in this regards was suggested in @xcite where a term of the form @xmath5 was added to the usual einstein - hilbert action .
it was then shown that this term could give rise to accelerating solutions of the field equations without dark energy . in @xcite a dgp brane model with a scalar field on the brane
was proposed , predicting that for the gravitational potential , the mass density should be modified by the addition of the mass density of the scalar filed on the brane .
such a scalar field has its origin in the conformal transformation used to transform the action in the jordan frame to the einstein frame in the usual @xmath4 modified gravity . for non - minimally coupled scalar field scenarios
see @xcite . in this paper , we focus attention on the dgp brane model where the action contains an arbitrary function of the ricci scalar , @xmath0 , and obtain the evolution of the metric on the space - time .
we concentrate on a specific form for @xmath6 and solve the resulting dynamical equations , predicting a power - law acceleration on the brane .
the components of the metric in the gaussian normal coordinates are then calculated and presented .
finally , we show that there exists two classes of solutions close to the usual frw cosmology for small hubble radii in the model presented here .
we start by writing the action for the dgp model with an arbitrary function of the scalar curvature on the brane part of the action , that is @xmath8,\ ] ] where the first term in ( [ eq1 ] ) corresponds to the einstein - hilbert action in @xmath9 for the 5-dimensional bulk metric @xmath10 , with the ricci scalar denoted by @xmath11 .
similarly , the second term is the modified einstein - hilbert action with @xmath7 corresponding to the induced metric @xmath12 on the brane , where @xmath7 is some arbitrary function of the @xmath4 scalar curvature and @xmath13 and @xmath14 are reduced planck masses in four and five dimensions respectively with @xmath15 being the matter action on the brane with a matter field denoted by @xmath16 .
the induced metric @xmath12 is defined as usual from the bulk metric @xmath10 by @xmath17 it would now be possible to write the field equations resulting from this action , yielding , in @xmath18 spatial dimensions @xmath19 where @xmath20 and @xmath21 is the energy - momentum tensor in the matter frame , and @xmath22+{\cal l'}(r)^{;\alpha\beta}(q_{\mu\alpha}q_{\nu\beta}-q_{\mu\nu}q_{\alpha\beta})\right\}.\ ] ] a prime here denotes differentiation with respect to @xmath23 . note that @xmath23 lives on the brane .
the corresponding junction conditions , relating the extrinsic curvature to the energy - momentum tensor , become @xmath24^{y=+\epsilon}_{y=-\epsilon } & = & \left.\frac{{\cal l}'(r)}{m^{3}_{4}}\left(\hat{t}_{\mu\nu}+ t^{({\it curv})}_{\mu\nu}-\frac{1}{d-1}q_{\mu\nu}q^{\alpha\beta}\left(\hat{t}_{\alpha\beta}+ t^{({\it curv})}_{\alpha\beta}\right)\right)\right|_{y=0}\nonumber\\ & -&\left.\frac{{\cal l}'(r)m^{2}_{3}}{m^{3}_{4}}\left(r^{(d-1)}_{\mu\nu}- \frac{1}{2(d-1)}q_{\mu\nu}q^{\alpha\beta}r^{(d-1)}_{\alpha\beta}\right)\right|_{y=0}.\end{aligned}\ ] ] it should be noted that @xmath25 is a function of the brane parameters .
from the fact that the dgp model predicts deviations only at large distances , one might hope it could be ruled out from cosmological observations . however
, as we shall see , it could account for cosmological equations of motion at any distance scale on the brane with any function of the ricci scalar .
brane cosmology usually starts from the line element @xmath26 where @xmath27 is a maximally symmetric 3-dimensional metric with @xmath28 being the usual parameters denoting the spatial curvatures . building on the results of @xcite , the cosmological evolution equations of a 3-brane in a @xmath9 bulk resulting from equations ( [ eq3 ] ) and ( [ eq6 ] )
were presented in the first two references in @xcite . here
we will follow @xcite and give the results for a brane of dimension @xmath29 . adopting the gaussian normal system gauge @xmath30 the field equations on the brane for metric ( [ eq15 ] ) and @xmath31 spatial dimensions are @xmath32 @xmath33 the junction conditions ( [ eq6 ] ) for an ideal fluid on the brane , given by @xmath34 read @xmath35^{y=+\epsilon}_{y=-\epsilon}=\frac{n{\cal l'}(r)}{\nu m^{\nu}_{\nu+1}}\left[(\nu-1)\rho^{({\it tot})}\left.+\nu p^{({\it tot})}\right]\right|_{y=0}\nonumber\\ + \frac{{\cal l'}(r)m^{\nu-1}_{\nu}}{m^{\nu}_{\nu+1}}(\nu-1)n\left[-\frac{\dot{n}\dot{a}}{n^{3}a}+ \frac{\ddot{a}}{n^{2}a } \left.-\frac{\dot{a}^{2}}{2n^{2}a^{2}}- \frac{k}{2a^{2}}\right]\right|_{y=0},\end{aligned}\ ] ] @xmath36^{y=+\epsilon}_{y=-\epsilon}= \left.\frac{{\cal l'}(r)m^{\nu-1}_{\nu}}{2m^{\nu}_{\nu+1}}(\nu-1)\left[\frac{\dot{a}^{2}}{n^{2}a}+ \frac{k}{a}\right]\right|_{y=0 } -\left.\frac{{\cal l'}(r)\rho^{({\it tot})}a}{\nu m^{\nu}_{\nu+1}}\right|_{y=0},\ ] ] where @xmath37 @xmath38 and @xmath39 and @xmath40 are the energy density and pressure in the matter frame associated with @xmath41 respectively . energy conservation on the brane follows from the vanishing of @xmath42 .
we obtain @xmath43 and in particular @xmath44^{y=+\epsilon}_{y=-\epsilon}= \lim_{\epsilon\rightarrow+0}\left[\frac{\dot{a}'}{\dot{a}}\right]^{y=+\epsilon}_{y=-\epsilon}.\ ] ] insertion of ( [ eq25 ] ) and ( [ eq26 ] ) into this equation yields the equation of conservation @xmath45 also , insertion of ( [ eq29 ] ) into the bulk equations @xmath46 and @xmath47 for @xmath48 yields a @xmath49-dimensional version of the integral of @xcite , that is @xmath50 and @xmath51 this means that if we define the quantities @xmath52 and @xmath53 by taking the factor @xmath54 out of the right hand sides of equations ( [ eq32 ] ) and ( [ eq33 ] ) , that is @xmath55 @xmath56 then @xmath52 and @xmath53 are constants with the property that @xmath57 if @xmath58 we can now simplify the previous equations by further restricting the gauge @xmath59 and by simply performing the transformation @xmath60 of the time coordinate .
this gauge is convenient because it gives the usual cosmological time on the brane . using equations ( [ eq29 ] ) and ( [ eq37 ] )
, we find that our basic dynamical variable is @xmath61 with @xmath62 given by @xmath63 the basic set of cosmological equations in the present setting for any function of the ricci scalar on the brane in the dgp model without a cosmological constant in the bulk are thus equations ( [ eq26 ] ) , ( [ eq31 ] ) , ( [ eq34 ] ) and ( [ eq35 ] ) which have to be amended with dispersion relations ( or the corresponding evolution equations ) for the ideal fluid components on the brane , that is @xmath64^{y= + \epsilon}_{y=-\epsilon}(t)&=&\frac{{\cal l'}(r)m^{\nu-1}_{\nu}}{2m^{\nu}_{\nu+1}}(\nu-1 ) \left[\frac{\dot{a}^{2}(0,t)}{a(0,t)}+ \left.\frac{k}{a(0,t)}\right]\right|_{y=0 } \nonumber\\ & -&\left.\frac{{\cal l'}(r)(\hat{\rho}+\rho^{({\it curv})})a(0,t)}{\nu m^{\nu}_{\nu+1}}\right|_{y=0},\end{aligned}\ ] ] @xmath65a^{\nu-1}(y , t)\right|_{y>0},\ ] ] @xmath66a^{\nu-1}(y , t)\right|_{y<0},\ ] ] @xmath67 - 3\left(\frac{\dot{a}(0,t)}{a(0,t)}\right)\dot{r}{\cal l''}(r)\right\},\ ] ] @xmath68\right\},\ ] ] @xmath69 let us now discuss the cosmology in the dgp model by taking @xmath70 .
the cosmological equations in this framework for a @xmath71-dimensional space are given by @xmath72 @xmath73a^{\nu-1}(y , t),\ ] ] @xmath74 the evolution of the background geometry of the observable universe according to the friedmann equation can thus be embedded in the dgp model , with the behavior of @xmath75 off the brane determined solely by the integral @xmath76 and the boundary condition @xmath77 from the friedmann equation .
this embedding will be asymmetric in all realistic cases , because the requirement that the friedmann equation holds on the brane is equivalent to the smoothness condition @xmath78 this could yield a symmetric embedding only for @xmath79 , but this is incompatible with the time independence of the integral @xmath76 apart from the case @xmath80 , @xmath81 . for
a discussion of this point the reader may consult @xcite . in the present calculations we will choose the sign of @xmath82 in the direction of the increasing scale factor , @xmath83 .
the possibility of a direct embedding of friedmann cosmology is a consequence of the fact that the evolution of the background geometry ( [ eq15 ] ) and the source terms @xmath84 and @xmath85 are supposed to depend only on @xmath86 and @xmath82 .
for @xmath87 we obtain @xmath88a^{2}(y , t),\ ] ] and from the equation for @xmath89 , the solutions for the metric components off the brane in terms of the metric on the brane ( assuming @xmath83 ) are @xmath90 @xmath91\frac{1}{a(y , t)}.\ ] ] this embedding of the friedmann cosmology on the brane becomes particularly simple for @xmath92 , that is @xmath93 @xmath94
to progress further , the form of @xmath7 should be specified . for ease of exposition and clarity ,
let us focus attention on theories where a @xmath1 term is present in the action and write @xmath95 let us also assume a power law solution form for the brane scale factor as @xmath96 the interesting cases are for @xmath97 which give rise to acceleration . to proceed , we consider the evolution of the scale factor with time on the brane .
using equation ( [ eq45 ] ) for the spatially flat frw metric and setting @xmath87 , we write @xmath98 now , use of the conservation equation leads to @xmath99.\ ] ] we must now solve the system of equations ( [ eq56 ] ) and ( [ eq57 ] ) with @xmath100 . substituting equations ( [ eq55 ] ) and ( [ eqq55 ] ) into the above equations describing the dynamical system for the brane
, we obtain an algebraic system for the parameters @xmath2 and @xmath101 @xmath102=0\nonumber , \\
\alpha[\alpha(n-2)+(n^{2}-n+1)]=n(2n-1)(n-1),\end{aligned}\ ] ] from which the allowed solutions @xmath103 follow
. the solutions with @xmath104 are not interesting since they provide static cosmologies with a non - evolving scale factor on the brane , matching cosmological models resulting from the solutions of the einstein equations without matter and @xmath105 .
this particular value of @xmath2 warrants a further discussion .
it is well known that in ordinary @xmath4 gravity an action of the form @xmath106 results in a constant scale factor in a cosmological setting without any ordinary matter and a singular equation of state for @xmath105 @xcite .
it is therefore not surprising to expect the same behavior in dgp models when the action on the brane is taken as that mentioned above .
however , if one takes @xmath107 , one is lead to ordinary general relativity as the low energy limit of the theory and one finds that this theory is equivalent to scalar quintessence models through conformal transformations @xcite .
the corresponding result for @xmath108 , @xmath109 and @xmath110 is also spelled out in @xcite . using equations ( [ eq43 ] ) and ( [ eqq43 ] ) we can deduce the equation of state for the family of solutions @xmath111 . we then have @xmath112 which clearly behaves as @xmath113 for @xmath114 , playing the role of a cosmological constant . for @xmath115 which corresponds to @xmath116
we have a power - law acceleration on the brane without having to introduce dark energy .
this result is consistent with the observational results similar to dark energy with the equation of state parameter @xmath117 @xcite . using equations ( [ eqq55 ] ) and ( [ eq60 ] )
we obtain the deceleration parameter @xmath118 it is clear that for @xmath119 we have @xmath120 .
this means that the universe ( brane ) is continuously expanding at an ever - increasing rate .
clearly , for @xmath121 or @xmath116 we obtain @xmath122 .
it is worth noting again , as was mentioned before , that in the normal dgp model , that is when @xmath105 and with no ordinary matter present , we can not define the equation of state and deceleration parameter because the scale factor on the brane is constant , see equations ( [ eq56 ] ) and ( [ eq57 ] ) .
we therefore expect the same behavior in our model and see that these equations diverge for @xmath105 .
figure 1 shows the behavior of @xmath123 as a function of @xmath2 . as can be seen , for @xmath124 and @xmath125 , @xmath126 .
let us now find the evolution of @xmath61 and @xmath62 everywhere in space - time .
substituting equations ( [ eqq55 ] ) and ( [ eq60 ] ) into equations ( [ eq51 ] ) and ( [ eq52 ] ) , one finds @xmath127 + 2\sqrt{a_{0}^{4}\alpha^{2}\left(\frac{t}{t_{0}}\right)^{4\alpha}-i}y,\ ] ] and @xmath128\frac{1}{a(y , t)}.\ ] ] in the particular case @xmath129 , we obtain @xmath130,\ ] ] and @xmath131.\ ] ] the scale factor on the full space - time , @xmath61 , is plotted in figure 2 for the special case @xmath129 and @xmath116 or @xmath121 .
note that for @xmath132 , equations ( [ eq63 ] ) and ( [ eq64 ] ) reduce to ( [ eqq55 ] ) and @xmath133 respectively .
there appears coordinate singularities on the space - like hypercone @xmath134 .
this is presumably a consequence of the fact that the orthogonal geodesics emerging from the brane ( which we used to set up our gaussian normal system , @xmath135 ) do not cover the full five - dimensional space - time .
in this section we derive the friedman equations for the brane metric in the presence of @xmath7 in the brane part of dgp action .
thus , for @xmath129 in equation ( [ eq50 ] ) and using equation ( [ eq37 ] ) we can write @xmath136a^{2}(y , t)=0.\ ] ] note that we assume @xmath137 and the bulk cosmological constant is zero . taking the symmetry @xmath138 @xmath139 for simplicity , equation ( [ eq40 ] ) can be used to compute @xmath140 on the two sides of the brane .
we have in this case @xmath141=2a'(0^{+},t)$ ] . by continuity when @xmath142 , equation ( [ eq65 ] ) yields the generalized ( first ) friedman equation @xmath143 where the hubble parameter is @xmath144 , @xmath145 , @xmath146 and @xmath147 is the sign of @xmath141 $ ] .
now , let us discuss the solutions of the friedman equation ( [ eq66 ] ) together with ( [ eq31 ] ) when the bulk cosmological constant vanishes .
it is now apparent from ( [ eq66 ] ) that the standard cosmology , namely the usual @xmath4 friedman equation ( [ eq56 ] ) is recovered whenever the last term on the left hand side of ( [ eq66 ] ) is subdominant with respect to the first term , namely when @xmath148 thus using ansatzs ( [ eq55 ] ) and ( [ eqq55 ] ) , the above equation in terms of the hubble radius @xmath149 can be written @xmath150 where @xmath151 .
for @xmath105 this matches the scale @xmath152 found in @xcite , setting the crossover between the @xmath4 and @xmath9 gravity regimes . for the special case @xmath121 , or @xmath116 for which a power - law acceleration consistent with @xmath153 is obtained we have @xmath154 also , for @xmath155 @xmath156 one
has @xmath157 thus , the hubble radius @xmath149 in this model depends on @xmath2 . for the hubble radius when equation ( [ eq68 ] ) is not satisfied ,
there are two distinct behaviors depending on @xmath158 .
equation ( [ eq66 ] ) can indeed be rewritten as @xmath159 let us start by examining the case where @xmath160 . using condition @xmath161 , one
may then expand ( [ eq66 ] ) to obtain @xmath162 or @xmath163 which is the full @xmath9 regime with @xmath7 gravity on the brane .
equation ( [ eq74 ] ) for @xmath1 becomes @xmath164 note that @xmath84 is a function of @xmath23 .
one has thus a transition from a @xmath4 to a @xmath9 regime .
if @xmath165 , however , @xmath166 is always larger than @xmath167 given by @xmath168 where @xmath169 . for @xmath1
we find @xmath170 and the expansion will never enter into a fully @xmath9 regime .
thus for @xmath137 or @xmath171 the hubbel parameter is bounded from below by equation ( [ eq76 ] ) for @xmath1 on the brane .
an interesting feature of the normal dgp models is the existence of ghost - like excitations @xcite
. it would therefore be interesting to briefly mention recent results relevant to the present work .
in @xcite , the author has shown , in the context of normal dgp models , that if we introduce a positive cosmological constant on the brane @xmath172 , then the spin-2 graviton will have a mass in the range @xmath173 and that there is a normalizable brane fluctuation mode with mass @xmath174 .
although the brane fluctuation mode is healthy , the spin-2 graviton has a helicity-0 excitation which is known as a ghost .
if we allow a negative cosmological constant on the brane , the brane bending mode becomes a ghost for @xmath175 .
this confirms the results obtained by the boundary effective action that there exists a scalar ghost mode for @xmath176 . in a self - accelerating universe@xmath177
, the spin-2 graviton has mass @xmath174 , which is known to be a special case for massive gravitons in de sitter spacetime where the graviton has no helicity-0 excitation and so no ghost .
however , in dgp models , there exists a brane fluctuation mode with the same mass and there arises a mixing between the brane fluctuation mode and the spin-2 graviton .
thus , according to this scenario , our model may be prone to having ghosts and this could be the subject of a separate investigation .
brane models generally provide an interesting extension of our parameter space for gravitational theories . in this work
we have discussed the dgp model with an arbitrary function of the ricci scalar in the brane part of action .
the cosmological evolution of this model was studied by solving the relevant dynamical equations .
the components of the metric on the space - time was obtained for the @xmath1 term in @xmath4 gravity .
the evolution of the universe in such a scenario was shown to be consistent with the present observations , predicting an accelerated expansion .
finally we have shown that there exists two classes of solutions , close to the usual frw cosmology for small enough hubble radii for @xmath7 gravity .
9 n. arkani - hamed , s. dimopoulos , g. dvali , phys .
b 429 ( 1998 ) 263 , [ hep - ph/9803315 ] , + n. arkani - hamed , s. dimopoulos , g. dvali , phys . rev .
d 59 ( 1999 ) 086004 , [ hep - th/9807344 ] .
i. antoniadis , n. arkani - hamed , s. dimopoulos , g. dvali , phys .
b 436 ( 1998 ) 257 , [ hep - ph/9804398 ] .
l. randall , r. sundrum , phys .
83 ( 1999 ) 4690 , [ hep - th/0906064 ] .
w. mck , k. s. viswanathan , i. v. volovich , phys .
d 62 ( 2000 ) 105019 , [ hep - th/0004017 ] .
r. gregory , v. a. rubakov , s. m. sibiryakov , class .
17 ( 2000 ) 4437 , [ hep - th/0003109 ] .
arefeva , m. g. ivanov , w. mck , k. s. viswanathan , i. v.volovich , nucl .
b 590 ( 2000 ) 273 , [ hep - th/0004114 ] .
m. cveti@xmath178 , m. j. duff , j.t .
liu , h. lu , c. n. pope , k. s. stelle , nucl .
b 605 ( 2001 ) 141 , [ peh - th/0011167 ] .
b. abdesselam , n. mohammedi , phys .
d 65 ( 2002 ) 084018 , [ hep - th/0110143 ] .
e. a. mirabelli , m. perelstein , m. e. peskin , phys .
82 ( 1999 ) 2236 , [ hep - ph/981133 ] , + t. appelquist , h. c. cheng , b. a. dobrescu , phys . rev .
d 64 ( 2001 ) 035002 , [ hep - ph/0201131 ] , + t. g. rizzo , phys . rev .
d 64 ( 2001 ) 095010 , [ hep - ph/0106336 ] , + s. cullen , m. perelstein , phys . rev .
83 ( 1999 ) 268 , [ hep - ph/9903422 ] , + v. barger , t. han , c. kao , r. j. zhang , phys . lett .
b 461 ( 1999 ) 34 , [ hep - ph/9905474 ] , + s. cassisi , v. castellani , s. deglinnocenti , g. fiorentini , b. ricci , phys . lett .
b 481 ( 2000 ) 323 , [ astro - ph/0002182 ] , + m. biesiada , b. malec , phys . rev .
d 65 ( 2002 ) 043008 , [ astro - ph/0109545 ] , + s. hannestad , g. g. raffelt , phys .
88 ( 2002 ) 071301 , [ hep - ph/0110067 ] .
g. dvali , g. gabadadze , m. porrati , phys .
b 485 ( 2000 ) 208 , [ hep - th/0005016 ] .
g. dvali , g. gabadadze , phys .
d 63 ( 2001 ) 065007 , [ hep - th/0008054 ] , + g. dvali , g. gabadadze , m. kolanovi@xmath179 , f. nitti , phys . rev .
d 65 ( 2002 ) 024031 , [ hep - th/0106058 ] .
r. dick , class .
18 ( 2001 ) r1 , [ hep - th/0105320 ] .
r. dick , actaphys .
b 32 ( 2001 ) 3669 , [ hep - th/0110162 ] .
r. cordero , a. vilenkin , phys .
d 65 ( 2002 ) 083519 , [ hep - th/0107175 ] .
a. lue , phys .
423 ( 2006 ) 1 , [ astro - ph/0510068 ] .
a. g. riess _ et .
[ supernova search team collaboration ] , astron .
j. 116 ( 1998 ) 1006 , [ astro - ph/9805201 ] .
s. perlmutter _ et .
j. 517 ( 1999 ) 565 , [ astro - ph/9812133 ] , + d. n. spergel _ et .
j. suppl . 148
( 2003 ) 175 , [ astro - ph/0302209 ] . c. l. bennett _ et .
j. suppl . 148
( 2003 ) 1 , [ astro - ph/0302207 ] .
c. b. netterfiled _ et .
al . _ , astrophys .
j. 571 ( 2002 ) 604 , [ astro - ph/0104460 ] , + n. w. halverson _ et . al .
_ , astrophys
. j. 568 ( 2002 ) 38 , [ astro - ph/0104489 ] .
s. m. carroll , living rev .
rel . 4 ( 2001 ) 1 , [ astro - ph/0004075 ] . c. deffayet , phys .
b 502 ( 2001 ) 199 , [ hep - th/0010186 ] , + c. deffayet , g. r. dvali and g. gabadadze , phys . rev .
d 65 ( 2002 ) 044023 , [ astro - ph/0105068 ] , + c. deffayet and s. j. landau , j. raux , m. zaldarriaga and p. astier , phys .
d 66 ( 2002 ) 024019 , [ astro - ph/0201164 ] , + j. s. alcaniz , phys .
d 65 ( 2002 ) 123514 , [ astro - ph/0202492 ] , + d. jain , a. dev and j. s. alcaniz , phys .
d 66 ( 2002 ) 083511 , [ astro - ph/0206224 ] , + a. lue , r. scoccimarro , g. starkman , phys .
d 69 ( 2004 ) 044005 , [ astro - ph/0307034 ] .
s. capozziello , int . j. mod
d 11 ( 2002 ) 483 , [ astro - ph/0201033 ] , + s. capozziello , int .
d 12 ( 2003 ) 1969 , [ astro - ph/0307018 ] .
s. m. carroll , v. duvvuri , m. trodden , m. turner , phys .
d 70 ( 2004 ) 043528 , [ astro - ph/0306438 ] .
k. atazadeh and h. r. sepangi , phys .
b 643 ( 2006 ) 76 , [ gr - qc/0610107 ] .
k. nozari , phys .
b 652 ( 2007 ) 159 , [ hep - th/0707.0719 ] , + k. nozari , jcap 09 ( 2007 ) 003 , [ hep - th/0708.1611 ] .
p. bin@xmath180truy , c. deffayet , d. langlois , nucl .
b 565 ( 2000 ) 269 , [ hep - th/9905012 ] .
p. bin@xmath180truy , c. deffayet , u. ellwanger , d. langlois , phys .
b 477 ( 2000 ) 285 , [ hep - th/9910219 ] . p. p. avelino , c. j. a. p. martins , astrophys.j . 565
( 2002 ) 661 , [ astro - ph/0106274 ] .
c. deffayet , g. dvali , g. gabadadze , [ astro - ph/0106449 ] . t. padmanabhan , phys . rept .
380 ( 2003 ) 235 , [ hep - th/ 0212290 ] .
m. a. luty , m. porrati and r. rattazzi , jhep 0309 ( 2003 ) 029 , [ hep - th/ 0303116 ] .
a. nicolis and r. rattazzi , jhep 0406 ( 2004 ) 059 , [ hep - th/0404159 ] .
k. koyama , phys.rev .
d 72 ( 2005 ) 123511 , [ hep - th/0503191 ] . | we consider a dgp inspired brane scenario where the action on the brane is augmented by a function of the ricci scalar , @xmath0 .
the cosmological implications that such a scenario entails are examined for @xmath1 and shown to be consistent with a universe expanding with power - law acceleration .
it is shown that two classes of solutions exist for the usual frw metric and small hubble radii .
when the hubble radius becomes larger , we either have a transition to a fully 5d regime or to a self - inflationary solution which produces a late accelerated expansion such that the radius becomes a function of @xmath2 . -2.5 cm |
a random graph is a set of vertices that are connected by random links @xcite .
random graphs underlie numerous natural phenomena ranging from polymerization @xcite to the spread of infectious diseases @xcite , and they also have applications to transportation systems , electrical distribution systems , the internet , the world - wide web , social networks , etc . @xcite . in random graph models ,
links are usually treated as undirected . in a growing number of applications
, however , directionality plays a prominent role .
one example is modeling of the web growth @xcite .
in modeling of food webs directionality ( reflecting predation ) is even more crucial .
food webs are directed graphs with vertexes @xmath4 labeling different species .
the presence of the directed link @xmath5 indicates that species @xmath6 is eaten by species @xmath7 . usually in food web only links @xmath5 with @xmath8
are allowed .
( loops @xmath9 which would account cannibalism are ignored ; the directed link @xmath5 with @xmath10 could e.g. represent predation on the young of the ` stronger ' species @xmath6 by adults of species @xmath7 , but such links are also disregarded in most models . )
the simplest cascade model @xcite generates a food web at random , namely for each pair of species @xmath6 and @xmath7 with @xmath8 the directed link @xmath5 is drawn at random with a certain predation probability @xmath11 .
a number of questions , particularly those related to the maximal length of food chains , have been investigated in the framework of the this cascade model .
for instance , what is the length ( the number of links ) of the longest direct path starting from the _ basal _ species ( vertex 0 ) ? a dual question concerns the length of the longest path finishing at the _ top _ species ( vertex @xmath12 ) .
one can also ask about the length @xmath13 of the longest path irrespectively on the first and last species . the simplest cascade model is a kind of ` standard model ' in the subject , and it had been widely used to interpret ecological data on community food webs @xcite .
the standard cascade model provides a very natural mechanism for generating directed random graphs and the same model has been suggested in other contexts , e.g. as a model of parallel computation @xcite in which the presence of the directed link @xmath5 with @xmath8 indicates that task @xmath6 must be performed before task @xmath7 . for a parallel computation
in which each task takes a unit of time , the processing time will be equal @xmath14 ( where @xmath13 is the length of the the longest path ) .
food webs typically involve a huge number of species , while the average predation per species is usually not too large .
hence it is interesting to investigate large food webs with small predation probability , more precisely the scaling limit @xmath15 this suggests to study a continuum cascade model where the vertex set is the interval @xmath0 $ ] .
for each species @xmath16 , the number of predator species is random , such species are chosen at random from the interval @xmath2 $ ] according to the poisson distribution with unit density .
the poisson distribution immediately follows from the binomial distribution ( characterizing the discrete cascade model ) in the scaling limit .
this cascade model is the minimalist continuum model of directed random graphs .
simple models tend to arise in various unrelated subjects and they are interesting on purely intellectual grounds . nevertheless , for concreteness in the following exposition we shall often use the language of food webs .
the rest of this article is organized as follows . in sec .
[ model ] we define the model , discuss its simplest properties , and derive a recurrence for the longest directed path starting from the origin .
the asymptotic behavior of the solution to that recurrence is analyzed in the following sections [ elem ] and [ t_wave ] . in section [ size_tree ]
we discuss the total number of vertices in a cascade tree with the root at the origin ; on the language of food webs it counts the basal species and species feeding on it , both directly and indirectly .
the vertex set of our random graph is the interval @xmath0 $ ] . in the illustrative picture below we draw only the vertex set and links from the cascade subgraph initiating at the origin ( the open circle on the picture ) .
namely , we draw all links emanating from the origin indicating direct predation on the basal species ( there are 3 such predators in the picture ) ; then we draw all the links from these direct predators ( 4 such predators in the picture ) ; etc .
links are drawn in a cascade manner thereby explaining the name of the model .
( 100,120)(20,10 ) ( 20,50)(420,0 ) 420 ( 100,50)(160,50 ) [ t ] ( 150,50)(260,80 ) [ t ] ( 200,50)(360,120 ) [ t ] ( 210,50)(60,20 ) [ b ] ( 260,50)(160,50 ) [ b ] ( 300,50)(240,70 ) [ b ] ( 275,50)(70,25 ) [ t ] ( 300,50)(120,40 ) [ t ] ( 340,50)(120,40 ) [ b ] ( 375,50)(70,25 ) [ t ] ( 20,50 ) ( 310,50 ) ( 360,50 ) ( 380,50 ) ( 400,50 ) ( 410,50 ) ( 420,50 ) ( 435,35)*x * overall , in the above illustrative picture the cascade subgraph is a tree with 10 links and 11 vertices .
six of these vertices ( closed circles on the picture ) are terminal , that is , there are no links emanating from them .
every cascade subgraph is a tree ; the size and the number of terminal vertices in cascade trees fluctuate from realization to realization .
terminal vertices represent top predators on the language of food webs .
it is easy to compute the fraction of top predators : @xmath17 the fraction of bottom preys , that is , species who do not eat other species , is the same .
the overlap of the sets of top predators and bottom preys ( one can call them neutral species ) is non - empty , the fraction of neutral species is @xmath18 we now turn to more subtle properties of the continuum cascade model which are related to the cascade tree .
this tree is finite and it varies from realization to realization ; accordingly , the properties of the cascade tree are probabilistic . to define these properties
it is convenient to utilize a more traditional way of plotting trees ; the cascade tree pictured above is presented on fig .
[ fig : food_web ] .
this figure resembles binary search trees and both the relevant properties of binary search trees and the methods used in analysis of binary search trees @xcite are useful in our situation .
for instance , the height of the binary search trees has attracted a lot of attention , and a traveling wave analysis @xcite has provided a very efficient way of tackling the asymptotic ( in number of vertices of the tree ) behavior of the height . in the present problem ,
the height is indeed an interesting quantity , namely it is the length of longest chain from the basal species to the bottom of the cascade tree , and the traveling wave analysis will be helpful as well .
we now establish a recurrence relation for the height distribution .
the height is a non - negative integer .
it is convenient to work with the cumulative distribution @xmath19 the basal species is the terminal vertex with probability @xmath20 , and therefore @xmath21 for @xmath22 , @xmath23 where the first line accounts for any possible number @xmath24 of links emanating from the origin and finishing at all possible points @xmath25 .
there is no ` interaction ' between different branches of the cascade tree , so we merely must assure that all cascade trees originating at @xmath25 have heights not exceeding @xmath26 . computing the sum in the above equation we arrive at our main recurrence @xmath27\ ] ] starting with
we find @xmath28.\ ] ] one can recursively determine @xmath29 , then @xmath30 ; analytical expressions for @xmath31 become very cumbersome as @xmath32 increases .
fortunately , in the large ` time ' limit , @xmath33 , the behavior greatly simplifies , namely the solution acquires a traveling wave form [ see fig . [
fig : tw ] ] , @xmath34 with the front position growing linearly with ` velocity ' equal to @xmath35 : @xmath36 versus @xmath37 obtained by iteration the recurrence . shown is @xmath38 for @xmath39 ( left to right ) .
iterations were performed using _
mathematica_. the observed velocity of the traveling wave is in a good agreement with the theoretical prediction . ]
the traveling wave profile @xmath40 decreases monotonically from 1 to 0 as @xmath41 increases from @xmath42 to @xmath43 .
more precisely , @xmath44 and @xmath45 in the next section we give an elementary argument which allows one to understand . a more comprehensive traveling wave analysis that leads to above results
is presented in section [ t_wave ] .
let us begin with the behavior of @xmath38 for small @xmath37 . expanding @xmath46 and @xmath47 , see equation
, we obtain @xmath48 using equation we recurrently determine the expansions of the following @xmath31 to yield @xmath49 this result is easy to prove by induction
. one can continue this expansion , e.g. , @xmath50 is valid for all @xmath22 ; this is also easily proven by induction .
let us now estimate the front position @xmath51 from the criterion @xmath52 .
keeping only two terms as in equation we obtain @xmath53 , or @xmath54 in the leading order .
what will happen if we keep e.g. four terms in the expansion ? using the same criterion @xmath52 in conjunction with equation we get @xmath55=\frac{1}{2}.\end{aligned}\ ] ] in the leading order we recover the previous prediction @xmath54 .
this does not prove equation , but at least it shows its consistency with the series .
we want to understand the behavior of the recurrence @xmath56\ ] ] when @xmath57 .
we assume the convergence to traveling wave solution and the validity of the traveling wave ansatz .
numerical results strongly support this assumption [ see fig . [
fig : tw ] ] and show that the convergence is rather fast , that is , the asymptotic shape emerges already for not to large @xmath32 .
we also assume that @xmath58 ( for large @xmath32 ) , but we do not specify @xmath59 . the left - hand side of equation becomes @xmath60 while the right - hand side of equation turns into @xmath61 , \qquad \eta = y - nv\ ] ] we will allow @xmath62 to be large , but we will always assume that @xmath63 .
in this situation , the integral in equation can be simplified as follows : @xmath64 + \int_0^\xi d\eta\,\pi(\eta ) \nonumber\\ & = & nv + \int_0^\xi d\eta\,\pi(\eta ) -
l + l_n , \end{aligned}\ ] ] where we have used the shorthand notation @xmath65 , \quad l_n = \int_{-\infty}^{-nv } d\eta\,[1-\pi(\eta)].\ ] ] since @xmath66 quickly approaches to 1 as @xmath67 , we drop @xmath68 from ; we shall justify this step _ a posteriori_. combining and we see that the right - hand side of becomes @xmath69.\ ] ] equating and we arrive at @xmath70.\ ] ] this is the governing equation for @xmath40 . since @xmath40 quickly approaches to zero as @xmath72 , equation gives @xmath73 , where @xmath74 .
thus we confirm ; more precisely we get @xmath75 it is more convenient to work with @xmath77 rather than @xmath40 . in terms of @xmath78 , equation becomes @xmath79.\ ] ] from the definition we see that @xmath80 therefore we can re - write as @xmath81.\ ] ] equation is equivalent to eq .
, we have not made any approximation . turning to the @xmath76 limit
we note that in this regime @xmath82 and hence we can expand the exponent on the right - hand side of equation . keeping only two terms we simplify equation to @xmath83 from , or from equation @xmath84 obtained by differentiating of eq .
, we see that the solution has an exponential form @xmath85 plugging into we arrive at the dispersion relation @xmath86 an elementary analysis of this equation indicates that solutions exist only when @xmath87 .
we now invoke the selection principle which asserts that the extremal value , @xmath88 in our case , is realized .
traveling wave solutions have been investigated in the context of partial differential equations .
a few partial differential equations admitting traveling wave solutions have been deeply studied .
one such equation is the celebrated fisher - kpp equation @xcite for which the selection principle had been proven @xcite for sufficiently steep initial conditions .
( for more recent work see e.g. @xcite .
a very comprehensive review of traveling wave solutions of non - linear partial differential equations has been given by van saarloos @xcite , a lighter expositions appear in books @xcite . )
more recently , traveling wave solutions have been investigated in the context of nonlinear recurrences arising in the analyses of binary search algorithms @xcite , kinetic theory @xcite , and other problems @xcite ; see @xcite for a review of the applications of traveling wave techniques to recurrences .
asymptotically , the wave front advances at a constant velocity @xmath88 .
the approach to this asymptotic value is rather slow , namely there is a @xmath89 correction in the leading order , resulting in a logarithmic correction to the front position .
this correction was first established by bramson @xcite for the fisher - kpp equation ; it was subsequently generalized @xcite to more general partial differential equations and to recurrences @xcite .
this correction generally has the form @xmath90 . for the selected velocity @xmath88 ,
the decay amplitude @xmath91 is implied by dispersion relation .
taking into account this logarithmic correction we get @xmath92 it was convenient to think about @xmath37 and @xmath32 as space and time coordinates , so that the front of the traveling wave was advancing and we determined @xmath93 . in the original problem ,
the parameter @xmath37 is fixed and we are interested in the height @xmath94 of the cascade tree .
the height is essentially the inverse to @xmath93 which is taken when @xmath95 .
thus @xmath96 the height is of course a random quantity . equation gives the average height . in the @xmath3 limit
, the average provides a faithful description as it is a growing quantity while the variance remains finite .
we havent proved this assertion , but at least on the physical level of rigor it is obvious : the probability distribution @xmath97 has asymptotically a traveling wave shape with the width of the front remaining finite , and this is essentially equivalent to the finite width of the height distribution .
the size @xmath98 of the cascade tree , that is , the total number of vertices in the tree , is a random variable .
let us compute the average size @xmath99 . from the definition of the continuum cascade model we deduce @xmath100 \\\nonumber & = & 1 + \sum _ { k=1}^{\infty } \frac{x^k } { k!}\,e^{-x } k\int _
0^x \frac{dy}{x}\,\langle s(y)\rangle \\ & = & 1 + \int _ 0^x dy\,\langle s(y)\rangle \label{eq : size}\end{aligned}\ ] ] differentiating we obtain @xmath101 , from which @xmath102 a similar line of reasoning leads to an integral equation for the second moment @xmath103 using we simplify above integral equation to @xmath104 which is solved to yield @xmath105 one can continue and compute @xmath106 and a few higher moments @xmath107 , but results quickly become very cumbersome .
the explicit results show that , in contrast to the height , the size of the cascade tree is the random quantity whose limiting distribution ( in the @xmath3 limit ) remains broad .
more precisely , in the scaling limit @xmath108 the size distribution becomes @xmath109 = e^{-x } f(\sigma)\ ] ] with the limiting distribution being different from the delta function , @xmath110 .
the normalization requirement together with and
similar equations for higher moments @xmath107 show that the moments @xmath111 of the limiting distribution are @xmath112 etc .
we proposed a minimalist model of infinite directed random graphs .
the model is a continuum version of a model of finite directed random graphs , known as the cascade model , which has been investigated in the context of food webs and parallel computation .
our model presumes a total order on the set of vertices .
we chose the simplest such set , an interval of length @xmath37 . from each @xmath1 $ ] , directed links to points @xmath113 are drawn at random according to the poisson distribution , that is , the points @xmath114 $ ] are chosen independently from each other and uniformly with density one .
the analysis of this continuum cascade model is actually simpler than the analysis of the discrete cascade model .
this is demonstrated by studying the distribution of the length of the longest directed paths starting at the origin ( equivalently , the height of the cascade tree with the root at the origin ) .
we employed traveling wave techniques to extract the asymptotic behavior of the length of the longest directed paths in the @xmath3 limit .
it will be interesting to understand the limiting distribution of the size of the cascade tree with the root at the origin as well as other properties of the continuum cascade model .
* acknowledgment * yi thanks joel e. cohen for helpful comments and discussions .
yi is supported in part by us national science foundation grant dms 0443803 to rockefeller university and by jsps grant - in - aid for scientific research 23540177 .
99 gelenbe e , nelson r , philips t and tantawi a 1986 an approximation of the processing time for a random graph model of parallel computation , acm 86 proceedings of 1986 acm fall joint computer conference ieee computer society press , los alamos , 691697 | we study a class of directed random graphs . in these graphs ,
the interval @xmath0 $ ] is the vertex set , and from each @xmath1 $ ] , directed links are drawn to points in the interval @xmath2 $ ] which are chosen uniformly with density one .
we analyze the length of the longest directed path starting from the origin . in the @xmath3 limit , we employ traveling wave techniques to extract the asymptotic behavior of this quantity .
we also study the size of a cascade tree composed of vertices which can be reached via directed paths starting at the origin . |
it has long been known that globular cluster stars present some striking anomalies in their content in light elements whereas their heavy elements ( i.e. , fe - group , @xmath0-elements ) remain fairly constant from star to star ( with the notable exception of @xmath1 cen ) .
while in all the galactic globular clusters studied so far one finds `` normal '' stars with detailed chemical compositions similar to those of field stars of same metallicity ( i.e. , same [ fe / h ] ) , one also observes numerous `` anomalous '' main sequence and red giant stars that are simultaneously deficient ( to various degrees ) in c , o , and mg , as well as enriched in n , na , and al ( for reviews see @xcite ) . these abundance variations are expected to result from h - burning nucleosynthesis at high temperatures around @xmath2 k @xcite .
such temperatures are not reached in the low - mass main sequence and rgb stars that are chemically peculiar , meaning that the stars inherited their abundance anomalies at stellar birth . here
we follow the work of @xcite and @xcite who propose that abundance anomalies are build up by fast rotating , fast evolving massive stars . during their main sequence evolution , rotationally - induced mixing transports elements synthesised in the convective h - burning core to the stellar surface . for stars heavier than 20 @xmath3 , the surface reaches break - up at the equator ( i.e. , the centrifugal equatorial force balances gravity ) , providing their initial rotational velocity is high enough .
in such a situation , a slow mechanical wind develops at the equator and forms a disc around the stars similar to what happens to be stars @xcite .
matter in discs is strongly enriched in h - burning products and has a slow escape velocity that allows it to stay in the potential well of the cluster . on the contrary ,
matter released later through radiativelly - driven winds during most of the he - burning phase and then through sn explosion has a very high velocity and is lost by the cluster .
therefore , new stars can form only from the matter available in discs with the abundance patterns we observe today .
thus globular clusters can contain two populations of low - mass stars : a first generation which has the chemical composition of the material out of which the cluster formed ( similar to field stars with similar metallicity ) ; and a second generation of stars harbouring the abundance anomalies born from the ejecta of fast rotating massive stars .
this scenario is sketched in fig .
[ fig : selfenrich ] .
based on the determination of the composition of giant stars in ngc 2808 by @xcite , @xcite determined that around 70% of stars present abundance anomalies in this specific cluster today .
@xcite find similar results for ngc 6752 with their analysis of the data of @xcite : around 85% of the cluster stars ( of the sample of 120 stars ) present abundance anomalies .
therefore most stars still evolving in globular clusters seem to be second generation stars . how to produce such a high fraction of chemically peculiar stars ?
the main problem is that assuming a @xcite imf for the polluters , the accumulated mass of the slow winds ejected by the fast rotating massive stars would only provide 10% of the total number of low - mass stars . to match the observations
thus requires either ( a ) a flat imf with a slope of 0.55 instead of 1.35 ( salpeter s value ) , or ( b ) that 95% of the first generation stars have escaped the cluster @xcite . here
we first verify whether such a high loss of stars is possible , and which are the main processes that could drive it .
first we assume that the globular clusters display primordial mass segregation so that the massive stars are located at their center .
since we expect that the formation of the second generation of low - mass stars happens locally around individual massive stars ( see @xcite for more details ) , the second generation of stars will also be initially more centrally concentrated than the first generation .
in such a situation , two competitive processes act in the clusters : the loss of stars in the outer cluster parts will first reduce the number of bound first generation stars ; and the dynamical spread of the initially more concentrated second generation stars will stop this differential loss when the two populations are dynamically mixed . our analysis , based on the n - body models computed by @xcite with the collisional aarseth n - body code nbody4 @xcite , is presented in detail in @xcite . as these models have been computed for a single stellar population , we apply the following process to mimic the formation of a cluster with two dynamically distinct populations : we sort all the low - mass stars ( @xmath4 @xmath3 ) according to their specific energy ( i.e. , their energy per unit mass ) .
we define the second stellar generation as the stars with lowest specific energies , ( i.e. , those which are most tightly bound to the cluster due to their small central distance and low velocity ) .
the number of second generation stars is given by having their total number representing 10% of the total number of low - mass stars . in fig .
[ fig : distr ] one can see the radial distribution of the two populations at various epochs .
initially , first generation stars show an extended distribution up to 40 pc whereas the second generation stars ( with low specific energy ) are concentrated within 6 pc around the centre .
progressively the second generation stars spread out due to dynamical encounters so their radial distribution extends .
however this process operates on long timescale : even after 5 gyr of evolution the two populations still have different distributions .
the bottom panel of fig .
[ fig : distr ] shows that after nearly 12 gyr of evolution ( slightly less than 3 passed relaxation times ) the two populations have similar radial distributions and can no longer be distinguished owing to their dynamical properties .
as previously seen , the effect of the external potential of the galaxy on the cluster is to strip away stars lying in the outer part of the cluster .
initially , as only stars of the first generation populate the outer part of the cluster owing to their high specific energy , only these first generation stars are lost in the early times .
this lasts until the second generation stars migrate towards the outer part of the cluster .
depending on the cluster mass , it takes between 1 to 4 gyr to start losing second generation stars .
due to the time - delay to lose second generation stars , their remaining fraction in the cluster is always higher than that of the first generation stars except during the final stage of cluster dissolution . fig . [
fig : distr ] ( right panel ) quantifies this point by showing the time evolution of the number ratio of second to first generation stars . as a direct consequence of our selection procedure ,
the initial ratio is 0.1 ; it then increases gradually with time and it tends to stay nearly constant as soon as the two distributions are similar . finally , at the time of cluster dissolution ( i.e. , when the cluster has lost 95% of its initial mass , indicated by the label `` dis '' in fig .
[ fig : distr ] ) , large variations occur due to the low number of low - mass stars present in the cluster . in fig.[fig : distr ] ( right panel ) we have also indicated the number of passed relaxation times , showing that the increase of the number ratio lasts only 3 relaxation times .
the fraction of second generation stars relative to first generation ones increases by a factor of 2.5 over the cluster history .
therefore , these second generations stars can account for 25% of the low - mass stars present in the clusters .
compared to the observed ratios ( 70% and 85% in ngc 2808 and ngc 6752 respectively ) the internal dynamical evolution and the dissolution due to the tidal forces of the host galaxy are not efficient enough .
an additional mechanism is thus needed to expel the first generation stars more effectively . and
@xmath5.,scaledwidth=48.0% ] as it operates early in the cluster history ( a few million years after cluster formation at the latest ) , initial gas expulsion by supernovae is an ideal candidate for such a process . as the gas still present after the star formation is quickly removed , it ensues a strong lowering of the potential well of the cluster so that the outer parts of the cluster can become unbound .
@xcite computed a grid of n - body models to study this process and its influence on cluster evolution by varying the free parameters : star formation efficiency , sfe , ratio between the half - mass and tidal radius , @xmath6 , and the ratio between the timescale for gas expulsion to the crossing time , @xmath7 .
they show in particular that , in some extreme cases , the complete disruption of the cluster can be induced by gas expulsion .
this process has also been used successfully by @xcite to explain the challenging correlation between the central concentration and the mass function of globular clusters as found by @xcite .
we have applied the same method as the one we used in 2.2 to the models of @xcite .
[ fig : ge ] shows that in the case of a cluster which loses around 90% of its stars , the ejection of stars from the cluster mostly concerns first generation ones . at the end of the computation only 5% of first generation stars
remain bound to the cluster along with around half of second generation stars .
therefore the number ratio between the second to first generation stars increases by a factor of 10 : half of the population of low - mass stars still populating the cluster are second generation stars . besides , the initial radial distribution is not totally erased by this mechanism as the second generation stars are still more centrally distributed .
we can expect that this ratio will continue to increase in the long - term evolution of the cluster ( see decressin et al .
, in preparation ) .
thus if globular clusters are born mass segregated , dynamical processes ( gas expulsion , tidal stripping and two - body relaxation ) can explain the number fraction of second generation stars with abundance anomalies .
similar conclusions have been reached by @xcite .
as abundance variations in light elements are expected to be due to h - burning whose direct product is he , we expect that second generation stars are also enriched in he to some degree .
unfortunately he abundance can not be directly measured in globular cluster stars and we have only indirect evidence for an overabundance in he .
the globular clusters @xmath1 cen and ngc 2808 display multiple main sequences @xcite ; a double sub - giant branch is also found in ngc 1851 @xcite .
such features can be understood if the stars present various he contents .
he enrichment is also a possible explanation for the appearance of extreme horizontal branches seen for several globular clusters @xcite . ) and initial he value ( 0.245 - 0.72 , mass fraction ) .
white area indicates stars still on the main sequence after 15 gyr of evolution.,title="fig:",scaledwidth=48.0% ] ) and initial he value ( 0.245 - 0.72 , mass fraction ) .
white area indicates stars still on the main sequence after 15 gyr of evolution.,title="fig:",scaledwidth=48.0% ] as explained in @xcite , matter stored in the discs around massive stars is heavily enriched with he .
[ fig : herich ] ( left panel ) gives our expected theoretical distribution function of the he - value in low - mass stars in ngc 6752 .
a main peak is present at @xmath8 and it extends up to 0.4 . however a long tail toward higher y - values is also present with around 12% of the stars with initial he value between 0.4 and 0.72 . to assert the implications for globular clusters induced by this population of he - rich stars
we have computed a grid of low - mass stellar models from 0.2 to 0.9 @xmath3 at a metallicity of @xmath9 ( similar to the metal - poor globular cluster ngc 6752 ) for initial he mass fraction between 0.245 and 0.72 with the stellar evolution code v2.92 ( see * ? ? ?
* ; * ? ? ?
* for more details ) .
these models have been computed without any kind of mixing except for an instantaneous mixing in convection zones .
the adopted mass - loss rate follows @xcite prescription ( with a parameter @xmath10 ) with a @xmath11 dependence .
all models have been computed from the pre - main sequence to the end of the central he - burning phase .
( left panel ) .
_ right : _
isochrone of 12 gyr and initial he of 0.25 ( dashed lines ) and isochrones of 10 gyr and initial he of 0.35 without ( full lines ) and with ( dotted lines ) additional reddening.,title="fig:",scaledwidth=48.0% ] ( left panel ) .
_ right : _
isochrone of 12 gyr and initial he of 0.25 ( dashed lines ) and isochrones of 10 gyr and initial he of 0.35 without ( full lines ) and with ( dotted lines ) additional reddening.,title="fig:",scaledwidth=48.0% ] for a given stellar mass , he - rich stars evolve faster on the main - sequence due to their lower initial h - content and to their higher luminosity .
figure [ fig : herich ] illustrates this point showing the turn - off age as a function of the initial mass and he mass fraction of stars .
after 12 gyr stars of 0.85 @xmath3 with standard helium ( @xmath12 ) as well as he - rich stars of 0.4 @xmath3 ( @xmath13 ) are leaving the main sequence .
figure [ fig : cmd ] ( left panel ) shows a synthetic colour - magnitude diagram ( cmd ) of a 12 gyr old globular cluster with the initial spread in he as given by fig .
[ fig : herich ] ( left panel ) .
this cmd has been computed with a modified version of the program used by @xcite to investigate supergiant populations .
the spread in he converts into a spread in mass at the turn - off .
the luminosity increase of he - rich stars is mainly compensated by their shorter lifetime so that the turn - off luminosity is almost constant .
besides due to their differences in opacity and to their compactness they are also hotter . thus the he - rich main - sequence and rgb stars
are shifted to the left side of the cmd .
they also occupy the blue part of the hb down to the extreme - hb location .
if we compare our theoretical cmd with the one observed by @xcite for ngc 6752 we note some discrepancies .
first the theoretical width of the main sequence at the turn - off is too large for this cluster .
additionally ngc 6752 shows on extended blue hb with no stars in the red part .
this last discrepancy could be attributed to the low mass - loss rate used in the stellar models which do not remove enough mass along the rgb and hence produce too cool stars on the hb . as the theoretical spread of the initial he is strongly affected by the dilution of the disc ejected by fast rotating massive stars and the ism , we plan to constrain this dilution with the observed sequences to check whether we are able to consistently reproduce both the abundance anomalies and the he - value inferred in globular clusters ( see decressin et al .
, in preparation ) . the uncertainties pertaining the ages of globular clusters are manifold . among them
, photometric uncertainties widen both sides of the main - sequence , unresolved binaries extend main - sequence towards cooler effective temperature , an increase of metallicity ( as seen in @xmath1 cen ) induces redder sequences .
the presence of he - rich stars can induce some additional uncertainties . let us note that they are the only physical parameter which broadens the main - sequence only to its left ( i.e. , blue ) side . in fig .
[ fig : cmd ] ( right panel ) we evaluate uncertainties related to he - rich stars : we try to reproduce a he - normal isochrone ( @xmath14 ) with a he - rich ( @xmath15 ) one .
this could be done with an isochrone 2 gyr younger along with an increase of the reddening of 0.04 magnitudes .
the differences between the he - rich with reddening and he - normal isochrones are small around the to , the main - sequence and the subgiant - branch .
discrepancies appear along the rgb and at the level of the hb , where the normal he - rich isochrone is much less extended toward the blue .
thus the age uncertainties due to a population of he - rich stars can be of the order of 12 gyr . | a significant fraction of stars in globular clusters ( about 70%-85% ) exhibit peculiar chemical patterns with strong abundance variations in light elements along with constant abundances in heavy elements .
these abundance anomalies can be created in the h - burning core of a first generation of fast rotating massive stars and the corresponding elements are convoyed to the stellar surface thanks to rotational induced mixing .
if the rotation of the stars is fast enough this matter is ejected at low velocity through a mechanical wind at the equator .
it then pollutes the ism from which a second generation of chemically anomalous stars can be formed .
the proportion of anomalous to normal star observed today depends on at least two quantities : ( 1 ) the number of polluter stars ; ( 2 ) the dynamical history of the cluster which may lose during its lifetime first and second generation stars in different proportions .
here we estimate these proportions based on dynamical models for globular clusters . when internal dynamical evolution and dissolution due to tidal forces are accounted for
, starting from an initial fraction of anomalous stars of 10% produces a present day fraction of about 25% , still too small with respect to the observed 70 - 85% . in case gas expulsion by supernovae
is accounted for , much higher fraction is expected to be produced . in this paper
we also address the question of the evolution of the second generation stars that are he - rich , and deduce consequences for the age determination of globular clusters . |
atomic hydrogen ( h i ) emission , as traced by the @xmath3 cm line , is ubiquitous throughout the milky way disk . as such it is an excellent tracer of gas kinematics and dynamics in the disk .
there is a long and fruitful history of using measurements of the h i line to measure the rotation curve of the milky way and to infer the galaxy s structure ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the departures from circular rotation , observed as the various bumps and wiggles in the rotation curve , have been used to infer the mass structure of the galaxy using spiral density - wave theories @xcite or more recently through comparison with simulations @xcite and mass discrepancy acceleration modelling @xcite .
fundamentally , accurate knowledge of the rotation curve is required for any attempt to model the surface density profile of a galaxy and the milky way is no exception .
in addition , the milky way rotation curve is used extensively for kinematic distance estimates based on the observed radial velocity of an object or associated gas .
much of the work on the rotation curve interior to the solar circle ( @xmath4 ) is based on measurements of the h i or co terminal - velocity . inside the solar circle
each line of sight has a location where it is tangent to a circular orbit about the galactic center . at this location , known as the tangent point , the projection of galactic rotational velocity , @xmath5 onto the local standard of rest ( lsr ) is
greatest , and the measured lsr velocity is called the terminal velocity . at the tangent
point all radial or vertical velocities are projected perpendicular to the lsr and the distance from the galactic center is geometrically defined at @xmath6 as @xmath7 . assuming that azimuthal streaming motions are not too large ( see however , chemin et al.2015 ) , the maximum velocity of h i emission can be equated to the terminal velocity , @xmath8 , and assigned to a galactocentric radius , @xmath9 .
this is the second paper in a series about the kinematics of the inner milky way . in the first paper , @xcite , hereafter paper i
, we developed a new technique for measuring the h i terminal velocity curve . in that paper
we discussed the technique extensively and produced a table of h i terminal velocities sampled every four arcminutes for the fourth galactic quadrant using h i data from the southern galactic plane survey ( sgps ; * ? ? ? * ) . in this companion paper
we employ the same technique to measure the h i terminal velocities for the first galactic quadrant from the vla galactic plane survey ( vgps ; * ? ? ?
together these terminal velocity curves give a uniform , densely sampled measurement of the h i rotation curve of the milky way between @xmath10 .
this paper is organized as follows : in [ sec : measure ] we review the key aspects of the technique described in paper i and apply this technique to the vgps data . in [ sec : termvel ] we present the first quadrant terminal velocity curve , comparing with the co terminal velocities from @xcite and with the fourth quadrant curve . in [ sec : rotcurve ] we show both the first and fourth quadrant rotation curves together with a variety of standard rotation curve fits and discuss the curves as an ensemble .
the h i data used here are from the vla galactic plane survey ( vgps ; * ? ? ?
the vgps is a survey of 21 cm continuum and h i emission and absorption obtained with the very large array in its d - array configuration and combined with single dish data from the nrao green bank telescope .
the survey covers galactic longitudes @xmath11 to @xmath12 with a varying latitude range from @xmath13 to @xmath14 .
the h i spectral line data have an angular resolution of @xmath15 with rms noise of 2 k of brightness temperature , @xmath16 , in each @xmath17 velocity channel . here
we use the h i data , as well as the continuum data , in the latitude range @xmath18 .
we compare the first quadrant terminal velocity data with terminal velocity data of the fourth quadrant from the southern galactic plane survey , as presented in paper i. the noise and resolution characteristics of the sgps and vgps are comparable , leading to a very uniform dataset across the first and fourth quadrants .
for all lines of sight near the galactic plane the circular rotation of the milky way projects onto the local standard of rest ( lsr ) as : @xmath19 where @xmath20 is the angular velocity of galactic rotation at the point of interest , and @xmath21 is the corresponding angular velocity of galactic rotation at the solar circle ( @xmath22 ) . in general
we discuss the circular rotation speed , @xmath23 .
the iau recommended values for the distance to the galactic center , @xmath24 , and the circular rotation speed at the sun , @xmath25 , are @xmath26 kpc and @xmath27 , respectively . although there is strong evidence that these should be varied , i.e. @xcite , we use the iau values here for consistency with previous work , except where noted . by measuring the terminal velocity , @xmath28 , at the tangent point , @xmath29 we can derive the galactic rotation curve , @xmath30 , interior to the solar circle . in practice , extracting the terminal velocities close to the solar circle is almost impossible because the circular velocity projected onto the lsr has nearly zero radial component .
we therefore restrict our measurements to about @xmath31 ( @xmath32 ) . as described in an extensive body of past work on the milky way h irotation curve ( e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and summarized in paper i , if the milky way were filled with cold gas in perfectly circular orbits the h i spectrum would drop abruptly to zero at the terminal velocity . in reality , thermal velocity motions , random gas motions and bulk motions all contribute to extend the h ispectrum beyond the terminal velocity .
the magnitude of thermal gas motions and turbulent widths are determined from the spectral width beyond the terminal velocity .
in paper i we built on the analysis by @xcite and @xcite to cast the drop - off of the h i profile beyond the terminal velocity in terms of an error function .
we will not reproduce the full derivation , but the conceptional point is that for optically thin gas beyond the terminal velocity , the brightness temperature , @xmath16 , as a function of velocity , @xmath33 , in each velocity interval @xmath34 is : @xmath35 where @xmath36 is the local gas number density in units of @xmath37 ; @xmath38 is the familiar constant converting velocity integrated brightness temperature to column density , @xmath39 ; @xmath40 is the velocity distribution of hi atoms ; and @xmath41 is the magnitude of the velocity gradient at the position , @xmath42 , along the line of sight , where the radial velocity is @xmath43 .
@xcite showed that for a smooth rotation curve @xmath44 is small and directly proportional to @xmath42 , taking values of @xmath45 for galactic longitudes @xmath46 .
we can therefore move the gradient term outside of the integral and if the velocity distribution function of h i atoms , @xmath47 , is gaussian , then the drop - off in the h i profile beyond the terminal velocity as shown in equation ( [ eq : tb ] ) is an error function .
in fact , we showed in paper i that the data were best fit by a sum of two or three error functions of widths , @xmath48 : @xmath49 where @xmath50 is the number of error functions . here the term @xmath51 is the velocity offset between the terminal velocity and a brightness temperature threshold , which we use to define a starting place for fitting the h i spectrum at the terminal velocity . to find the terminal velocity
some authors have defined a brightness temperature threshold ( e.g. * ? ? ? * ) or a relative brightness temperature threshold such as half of the intensity of the most extreme peak ( e.g. * ? ? ?
* ) , whereas @xcite and @xcite fit a functional form , similar to the error function described above .
any threshold method is subject to errors because both the local density , @xmath36 , and the path length , @xmath52 , contribute to the h i brightness temperature ( as shown in eq .
[ eq : tb ] ) .
in paper i we found best results with a functional fitting method that was initiated from a threshold .
we showed that the threshold was useful to determine a starting point for the functional fitting , but that by fitting the spectrum with a sum of two error functions we were able to largely remove ambiguities caused by the choice of threshold ( see 3.3 of paper i ) .
an advantage of using a survey like the vgps or sgps for measuring terminal velocities is that the effects of h i absorption of continuum emission can be removed from the data .
absorption can alter inferred terminal velocity by reducing the brightness temperature of velocities near the terminal velocity .
the net effect is a reduction in the terminal velocity at positions where there is significant absorption .
a clear example was shown in figure 1 of paper i. in our technique we use the vgps continuum images to define a mask of pixels to exclude from our analysis .
masking at a level of 40 k in the continuum images we reduce the total number of pixels by @xmath53% but also decrease the scatter in the terminal velocity measurements by @xmath54% .
our steps to finding the terminal velocity from the vgps were therefore : 1 .
mask h i spectra towards pixels containing continuum emission in excess of 40 k , as measured in vgps continuum images . 2 .
find the velocity of the 20 k @xmath16 threshold on latitude averages over the range @xmath18 .
we chose a threshold value of 20 k , as opposed to 10 k as used in paper i , because the noise in the vgps h i data is slightly higher , such that the threshold is at @xmath55 .
3 . shift all profiles to place the threshold crossing velocity , @xmath56 , at a common velocity of @xmath57 @xmath58 to simplify binning with longitude and the functional fitting .
average the shifted spectra in longitude by 3 to improve the signal - to - noise of a given longitude .
5 . fit each averaged spectrum with a sum of two error functions as defined by equation [ eq : errfunc ] to determine the true terminal velocity as @xmath59 .
the first quadrant terminal velocity curve for 960 longitude bins of width 3 between @xmath11 and @xmath12 is shown in figure [ fig : vt_vs_l ] .
we also show the offset velocity , @xmath51 , which indicates regions where the terminal velocities are likely effected by significant local non - circular velocities .
the spike in velocity offset at @xmath60 corresponds to small , anomalous velocity clouds found by @xcite . in total
we found spurious fits for much less than @xmath61% of the fitted spectra .
for the analysis that follows we run a median filter of spectral width 3 samples across the @xmath51 values before correcting the data from @xmath56 to @xmath8 .
the final terminal velocity values are provided in tabulated form in table [ tab : termvels ] .
the pre - existing most densely sampled terminal velocity curve of the first quadrant comes from @xmath62co j=1 - 0 data published by @xcite .
this curve , and its associated polynomial fit , has been used for many analyses and kinematic distances over the past decades .
as a tracer of the terminal velocity co has different strengths and weaknesses from h i. while h i is a more ubiquitous tracer , it has large ( @xmath63 @xmath58 ) velocity linewidths , whereas co typically has linewidths of less than @xmath64 @xmath58 @xcite .
it has therefore been argued that co may be a better tracer of the terminal velocity than h i. however , in paper i we compared h i and co for the fourth quadrant and showed that they were remarkably consistent but with larger scatter in the co measurements , which we posited was because the low filling factor of co led the co terminal velocity curve to reflect the individual velocity dispersion of clouds , rather than an ensemble average as traced by h i. the analysis there indicated that neither tracer was significantly better than the other .
figure [ fig : vt_hi+co ] shows the @xcite curve plotted with the h i curve determined for the first quadrant . at all longitudes @xmath65 there is a systematic offset between the two curves , with the @xcite co data approximately 7 @xmath58 higher than the h i. in deriving the co terminal velocity curve , @xcite effectively used a threshold detection method ( three consecutive channels with emission greater than @xmath66 ) for the most extreme velocity line component , corrected by a uniform offset towards permitted velocities ( smaller @xmath67 ) equivalent to the half - width of an average co cloud or @xmath68 @xmath58 . in general
, threshold - crossing methods will result in higher terminal velocities than our spectral fitting method , which locates the inflection point of the spectral drop - off .
the difference between @xcite and our h iresults suggests that the average co cloud 3 @xmath58 correction is insufficient to recover the velocity at the tangent point . in spite of the differences in the magnitude of the velocity at a given longitude , the shape of the two curves are reassuringly similar .
most of the features on scales as small as a few degrees and up to scales of tens of degrees are apparent in both curves .
this lends confidence to the veracity of the small - scale features as true dynamical effects rather than measurement artefacts . for a galaxy in purely circular rotation the run of terminal velocities
will be completely linear with @xmath69 : @xmath70 .
of course , non - circular motions induced by spiral arms , the galactic bar and other discrete dynamic effects produce deviations from linearity , but a linear fit with @xmath69 provides a useful mechanism for comparing data . independently fitting the co and h i terminal velocity curves and then comparing the residuals we find that the scatter in the two curves is nearly identical at @xmath71 @xmath58 .
we compare h i terminal velocity curves for the first and fourth quadrants , each plotted versus @xmath72 , in figure [ fig : vt_both ] .
the absolute values of the velocities are plotted for direct comparison .
the two curves are strikingly similar , despite probing different regions of galactic structure .
most notably , both curves have an abrupt flattening of the slope at @xmath73 ( @xmath74 ) with almost no gradient in velocity until @xmath73 ( @xmath75 ) . in addition , both curves show an inflection at @xmath76 ( @xmath77 ) .
these features will be discussed further in the context of the rotation curve below .
the dashed lines in figure [ fig : vt_both ] show independent linear fits to the run of terminal velocities with @xmath72 in qi and qiv of the form described above ( @xmath78 ) , where the fitted coefficients are @xmath79 @xmath58 , @xmath80 @xmath58 and @xmath81 @xmath58 , @xmath82 @xmath58 for qi and qiv , respectively .
it is important to note that because the rotation curve is not necessarily flat over this range of @xmath69 , the coefficient @xmath83 does not equate to @xmath25 .
however , just as with the comparison of the h i and co curves , the fit is a useful indicator of the consistency of the two quadrants .
we find that the fits to qi and qiv are consistent to within 2% indicating the uniformity of the complete terminal velocity curve presented here .
there are , however , some large - scale offsets between the two curves .
the asymmetry between the the terminal velocity curves in the first and fourth quadrants has been discussed extensively ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in figure
[ fig : vdiff_l ] we plot a smoothed version of the difference between our qi and qiv terminal velocities as a function of galactic longitude .
the difference ranges smoothly from @xmath84 @xmath58 at @xmath85 to @xmath86 @xmath58 at @xmath87 and back down to @xmath88 @xmath58 at @xmath89 .
as discussed in @xcite , the difference between the terminal velocity curves depends only on the difference in azimuthal velocity of the gas at the tangent points in the two quadrants and any potential outward motion of the lsr .
an outward motion of the sun or the lsr , as posited by @xcite and @xcite , will give rise to a velocity difference between qi and qiv at all longitudes of the form @xmath90 , where @xmath91 is the radial velocity of the lsr . @xcite found @xmath92 @xmath58 , which was consistent with stellar data at the time .
@xcite revisited the asymmetry , recommending a value of @xmath93 @xmath58for the magnitude of the radial motion of the lsr . in figure
[ fig : vdiff_l ] we over - plot the functional form , @xmath90 for radial velocities of @xmath94 @xmath58 , @xmath86 @xmath58 and @xmath95 @xmath58 .
it is clear that over the range of longitudes measured here an outward motion of the lsr alone can not account for velocity difference . to fit the terminal velocity difference curve @xcite included a non - axisymmetric velocity field induced by a triaxial spheroidal potential .
alternative fits to the galactic potential , including those associated with the long bar ( e.g. * ? ? ? * ) or spiral arms ( e.g. * ? ? ?
* ) would also produce a noticeable difference curve .
fitting the velocity curves to multiple galactic potentials is clearly an important next step , but beyond the scope of this paper .
an alternative suggestion to explain only the asymmetry between the two curves near the solar circle is that of streaming motions associated with the carina spiral arm in qiv at @xmath96 @xcite . in mcg07
we compared sgps qiv values with qi data from @xcite , which showed the quadrants deviating by about 10 @xmath58 from @xmath97 kpc ( @xmath98 ) . using the consistently measured data here , and shown in figure [ fig : vdiff_l ] , we see that over the longitude range @xmath98 the terminal velocities in qi are @xmath99 @xmath58 ( mean @xmath100 @xmath58 ) lower than qiv , with no indication that the difference is converging towards zero at the solar circle .
from the terminal velocities measured here we can calculate the rotation curve for @xmath101 kpc from the circular velocity relation @xmath102 and assuming @xmath103 , where we use iau standard values of @xmath104 kpc and @xmath105 .
the rotation curves for qi , as calculated in this paper , and qiv , as calculated in paper i , are shown in figure [ fig : rot_fit ] .
in paper i we performed a joint linear fit to the qiv data along with h i rotation curve data of the first quadrant from @xcite , giving a fit of : @xmath106 .
we repeat that exercise here substituting the new uniformly measured , high resolution h i data for first quadrant .
the joint fit is very similar , as can be seen in figure [ fig : rot_fit ] , @xmath107 . the slightly smaller slope in the new fit is not within the formal fitting errors of either fit , but is less than 10% different and therefore on the order of the error in any given velocity measurement .
also shown in this figure is the often used @xcite rotation curve , derived from co terminal velocities for qi .
while this curve fits well the velocity structure of qi , it incorporates streaming motions that are not general to the inner galaxy rotation curve .
@xcite have recently used parallax measurements of masers to carefully measure the milky way rotation curve and derive galactic constants , @xmath25 and @xmath24 . in @xcite
they explored several rotation curve fits to their data .
their simplest fit , which is an excellent match to their data at @xmath108 kpc , is given by @xmath109 , where @xmath110 and @xmath111 @xmath58 .
this produces an almost flat rotation curve for large radii , as shown in figures [ fig : rot_fit ] & [ fig : rot_fit_reid ] . including data interior to @xmath112 kpc
, @xcite found that the so - called `` universal '' rotation curve for spiral galaxies from @xcite captured the decrease in velocities at @xmath113 kpc , while still maintaining the nearly flat , though slightly declining , rotation curve for larger radii .
their fit to the `` universal '' curve is also shown in figure [ fig : rot_fit ] using their value of @xmath114 @xmath58 .
both the @xcite rotation curves are above our h i velocities calculated from iau value of @xmath115 @xmath58 . for comparison , in figure [ fig : rot_fit_reid ]
we show our h i rotation curves calculated assuming @xmath116 @xmath58 and @xmath117 kpc , as in @xcite . for these values ,
the ` universal' rotation curve gives a reasonable match to the data .
because of the small range of galactic radii available in our data we do not perform a new fit with the ` universal' rotation curve .
future work should include a careful compilation of the extremely accurate maser parallax measurements with these new h i curves for a complete ` universal' milky way rotation curve .
for now we recommend the joint linear fit as a good estimate of the h i rotation curve at @xmath118 kpc .
an advantage of our simple linear fit to the rotation curve over the range @xmath119 to @xmath120 kpc range , i.e. @xmath121 , where @xmath122 @xmath123 , is that kinematic distances are given by the quadratic equation , based on equation 1 and the law of cosines . scaling the observed velocity of an object of interest , @xmath67 , by @xmath124 gives dimensionless parameter @xmath125 .
we can therefore define the kinematic distance , @xmath126 , in terms of the parameter @xmath127 as : @xmath128 where the constants are @xmath129 and @xmath130 .
the square root term gives the line of sight distance measured on either side of the subcentral point , whose distance is defined as @xmath131 . using the joint polynomial fit described above
we can subtract the overall rotation to examine the structure of the residuals in the rotation curve .
these residuals should reflect bulk non - circular motions in the galaxy .
the residual curves for both quadrants are shown in figure [ fig : resids ] . once again , qi is shown in red and qiv in blue .
the values for the two quadrants are extremely close at @xmath132 kpc and for @xmath133 kpc .
the agreement at @xmath133 kpc is expected as we trace gas towards the cardinal points @xmath134 and @xmath135 , where circular rotation projects towards @xmath136 @xmath58 .
the agreement at @xmath137 kpc is more surprising . while there are clearly small scale variations with @xmath138 , i.e. @xmath139 pc , the magnitude of those features are generally @xmath140 @xmath58 .
the largest deviations occur in the range @xmath141 kpc . in this region
the two curves are offset from each other by @xmath142 @xmath58 .
this region also encompasses some of the salient features noted in the terminal velocity curve discussed in [ subsec : vt_comp ] . notably the inflection seen in both quadrants terminal velocity curves at @xmath143
is observed as the rise in both quadrants rotation curves at @xmath144 kpc and the flattening of the slope of the terminal velocity curves between @xmath74 and @xmath75 are observed as the smooth rise in the range @xmath145 kpc .
the most striking aspect of the residuals shown on figure [ fig : resids ] is the similarity in the saw - tooth patterns seen in both quadrants between @xmath146 kpc .
in addition , by eye it appears that there is a great deal of structure in both quadrants that has a characteristic width of a few hundred pc , even though the resolution of the survey is much finer than this , typically @xmath14710 pc x @xmath148 .
although the effects of h i absorption toward compact continuum sources have been removed , there are a few very narrow features on figure [ fig : resids ] that correspond to local velocity perturbations , most of the structure in both residual curves is much broader in @xmath138 . to robustly estimate the characteristic width of the structure in the residuals on figure [ fig : resids ] , we compute the autocorrelation functions of both quadrant s rotation curves .
these are shown on figure [ fig : correlation ] and labelled `` acf '' . to construct the autocorrelations , we first re - sample the residuals to a constant step size of 5 pc in galactic radius .
all correlation functions are computed using the routine _ scipy - signal - correlate _ from the python numerical package _
numpy _ , which handles edge effects in a conservative way .
but given that our data covers a range of only @xmath1475 kpc , edge effects still lead to increasing uncertainty for lags of more than about @xmath1492 kpc .
we have compared the scipy correlation program with other algorithms , and they give similar results for lags less than half the total interval of @xmath138 used for the correlation , which is 3.5 to 7.5 kpc . the half - width to half maximum of the autocorrelation functions for the first and fourth quadrants are 290 and 210 pc , respectively .
their half widths to first zero crossing are 690 pc ( qi ) and 730 pc ( qiv ) .
this shows that the residuals have generally quite broad structure , in spite of a few sharp variations that can be seen in figure [ fig : resids ] . that we observe a consistent scale size for velocity fluctuations in the rotation curve that is much greater than the resolution of the data suggests that we are probing a characteristic size scale for dynamical effects
this size scale is of the same order as that often suggested for the energy injection scale of turbulence in the milky way , @xmath150 pc ( e.g. * ? ? ? * ) , and similar to the scale of the largest dynamical objects in the galaxy , specifically potential energy injection sites of clustered supernovae @xcite .
in addition to auto - correlation analysis , we can more robustly estimate the degree of similarity between the rotation residuals of the first and fourth quadrants with a cross - correlation analysis .
the cross - correlation function between the residuals two quadrants shows a distinct peak near zero lag on figure [ fig : correlation ] ( `` ccf '' ) .
the peak has normalised correlation of 0.27 at @xmath151 pc shift ( positive @xmath152 meaning that the qiv residuals are shifted to higher radius relative to the qi residuals at the peak ) .
the non - zero normalised correlation coefficient confirms the visual impression that the two curves are very similar in their structure .
furthermore , a radial shift between the two curves is what we expect if the large - scale structure is induced by spiral arms .
the magnitude of this shift , however , is surprisingly small .
the large sidelobes at @xmath1492 kpc lag for each of the cross - correlation function and auto - correlation functions show that the residuals have structure that is roughly sinusoidal with wavelength about 2 kpc .
this corresponds to the sawtooth pattern that is apparent by eye between @xmath2 kpc .
it has long been assumed that these large departures from a smooth circular rotation are induced by the spiral pattern of the inner galaxy ( e.g. * ? ? ?
* ; * ? ? ?
other grand design spirals like m81 the perturbations give sinusoidal departures from circular rotation ( e.g. * ? ? ?
* ) with amplitude ( peak - to - peak ) about 20 km s@xmath153 . unfortunately ,
interpretation of the structure in the residuals ( figure [ fig : resids ] ) is complicated by the changing projection of the line of sight as we move along the locus of sub - central points and cross spiral arms at various angles . furthermore , the velocity field predictions are different for linear and non - linear density wave theories @xcite . the cross - correlation function and
autocorrelation functions presented in figure [ fig : correlation ] will be useful constraints for hydrodynamical models of galactic structure and dynamics ( e.g. * ? ? ?
underlying most of our analysis is the assumption that the measurement of the terminal velocity represents gas at the tangent point and probes the true rotation curve . of course
, that premise is flawed slightly because of non - circular rotation due to streaming motions near spiral arms or the elliptical orbits around the central bar . given the observational indications that the milky way has a long bar , extending over a half - length of @xmath154 kpc and an angle of @xmath155 from the sun - galactic center line ( e.g. * ? ? ?
* ) , we expect non - axisymmetric effects in the inner galaxy rotation curve .
recently @xcite explored this topic thoroughly using a hydrodynamical simulation of a barred milky way - like galaxy . from the simulation they measured the velocity field and therefore both the true rotation curve and the rotation curve assuming the maximum velocity is found at the tangent point .
their analysis found that for @xmath156 kpc kpc . ]
the average tangent point derived curve is very close to the true galactic rotation curve , however , in the range @xmath157 kpc , the tangent point analysis deviates by as much as @xmath158% for bar angles of @xmath159 .
@xcite found for their fiducial bar of orientation @xmath160 , qi tangent point velocities over - estimated the true azimuthal velocity and qiv tangent point values under - estimated the azimuthal velocities .
similar results appear in the velocity field modelling by @xcite and seem to agree well with the residual rotation curves derived here . because of the complexities in attributing measured terminal velocities to the tangent point the next steps for understanding the milky way rotation curve must be either comparison with simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) or surface density mass modelling ( e.g. * ? ? ?
we have used the vla galactic plane survey of h i and 21-cm continuum emission @xcite to derive the terminal velocity curve of the first galactic quadrant with a sampling of 4 .
we find excellent consistency between the structure of the first quadrant h i terminal velocity curve and the co terminal velocity curve derived by @xcite , but with a systematic offset in the h i suggesting that the co curve slightly over - estimates the terminal velocities .
combining our first quadrant h i data with the terminal velocity curve of the fourth quadrant from the southern galactic plane survey @xcite published in our companion paper @xcite , we have produced the most uniform , high resolution , terminal velocity curve for the milky way interior to the solar circle .
as seen previously @xcite there is a clear asymmetry between the terminal velocity curves of the two quadrants .
we confirm that an outward motion of the lsr alone can not explain the asymmetry @xcite , highlighting the need to include a non - axisymmetric azimuthal velocity component for the inner milky way .
we find that the measured terminal velocities in two quadrants are almost identical over the galactic radius ranges , @xmath161 kpc and from @xmath133 kpc to the solar circle . over the intermediate radius range ( @xmath146 kpc ) ,
the amplitudes of the two quadrants differ by @xmath162 to @xmath163 , but they show remarkably similar structure . using the rotation curves for first and fourth quadrants together we have performed a new joint fit to the milky way rotation curve over the range @xmath164 kpc .
this curve can be used for estimating kinematic distances .
this joint fit smooths over the streaming motions contained in the previous first quadrant fit by @xcite and should be used in preference to that curve for inner galaxy kinematic distances .
we also compared our data with the flat and `` universal '' spiral galaxy rotation curves from @xcite , finding both to be reasonable fits to the h i rotation curve when calculated assuming @xmath114 @xmath58 and @xmath117 kpc , instead of the iau recommended values for the galactic constants .
finally , we compared the velocity residuals for the two quadrants after subtraction of the fitted rotation curve .
the residuals show remarkably similar structure that is roughly sinusoidal with a wavelength of about 2 kpc . with this high resolution , uniform terminal velocity curve
we have overcome the remaining observational problems limiting our knowledge of the milky way terminal velocities .
there is no indication from our measurements that either higher resolution or higher sensitivity observations would change the h i rotation curve derived here over the range @xmath165 kpc .
further progress on the milky way rotation curve will need to come from the compilation of these h i values with other measurements such as on - going maser parallax ( e.g. * ? ? ? * ) or upcoming stellar parallax from gaia .
those measurements , together with our h i terminal velocity curve can be used to improve the estimates of the surface density of the milky way disk and spiral structure .
we are grateful to s. mcgaugh whose encouragement lead us to complete this long overdue companion paper and who provided us with an electronic version of the clemens data and to m. reid for comments on an earlier version of this manuscript . n. m. m .-
g . acknowledges the support of the australian research council through future fellowship ft150100024 . | using atomic hydrogen ( h i ) data from the vla galactic plane survey we measure the h i terminal velocity as a function of longitude for the first quadrant of the milky way .
we use these data , together with our previous work on the fourth galactic quadrant , to produce a densely sampled , uniformly measured , rotation curve of the northern and southern milky way between @xmath0 .
we determine a new joint rotation curve fit for the first and fourth quadrants , which is consistent with the fit we published in mcclure - griffiths & dickey ( 2007 ) and can be used for estimating kinematic distances interior to the solar circle .
structure in the rotation curves is now exquisitely well defined , showing significant velocity structure on lengths of @xmath1 pc , which is much greater than the spatial resolution of the rotation curve .
furthermore , the shape of the rotation curves for the first and fourth quadrants , even after subtraction of a circular rotation fit shows a surprising degree of correlation with a roughly sinusoidal pattern between @xmath2 kpc . |
the general relativity theory defines the physical spacetime as a differentiable pseudo - riemmanian 4-dimensional manifold ( hawking & ellis , 1973 ) .
the spacetime is seen through the _ intrinsic _ geometry .
however , many interesting results have shown that the _ extrinsic _ geometry gives sometimes a better understanding of the physical structure of the spacetime .
as an example of its growing interest we can cite the recent studies of gravity theories in more than four dimensions and the study of the geometry of extended objects as in string and membrane theories .
in contrast to general relativity , where the metric uniquely specifies the geometry , to describe a spacetime locally and isometrically embedded in a pseudo - euclidean manifold with dimension @xmath2 , two new quantities have to be considered : the _ second fundamental form _ and the _ torsion _ vector ( or the third fundamental form ) .
these two quantities are well known in the study of differentiable manifolds ( eisenhart , 1946 ) and come out from the gauss - codazzi - ricci equations , which are the integrability conditions for the existence of the embedding ( maia , 1986 ) .
rigorously it is not necessary to know the embedding to obtain the second fundamental form and the torsion vector . both can be obtained directly through the field equations ( maia & roque , 1989 ) which are highly non - linear parcial differential equations in the 4 spacetime coordinates . however , if we know the embedding _ a priori _ , to determine these two quantities we need to get first the set of vectors normal to the spacetime v@xmath0 . in this paper
we will be concerned with the discussion of an algorithm which helps to determine these vectors from the embedding coordinates and then to find out the second fundamental form and the torsion vector .
the following section sets up the main equations that rule the embedding theory . in section 3 the algorithm for determining the vectors normal to the spacetime v@xmath0 is described .
this algorithm has been implemented in the algebraic computing system reduce and its application is done in section 4 for the case of schwarzchild embedding . some comments and remarks on an extension of this algorithm to select and determine the rank of any @xmath3 matrix
are left to section 5 of the paper .
a local embedding of a spacetime v@xmath0 in a pseudo - euclidean m@xmath1 manifold is done when a set of cartesian coordinates @xmath4 is specified as functions of the spacetime coordinates @xmath5 ( greek indices run from 1 to @xmath6 , lowercase latin letters run from 1 to 4 and capital latin letters run from 5 to @xmath6 ) . at any point of the manifold we can find a set of n vector fields @xmath7 orthogonal to v@xmath0 and to themselves .
thus if @xmath8 denotes the cartesian components of the metric of m@xmath1 , then the following set of equations are valid to to to where @xmath9 is the spacetime metric , @xmath10 , @xmath11 , depending on the signature of @xmath12 , @xmath13 are the components of the tangent vectors to v@xmath0 ( partial derivatives with respect to spacetime coordinates are indicated by a comma , as usual ) and @xmath14 is a constant .
the second fundamental form and the torsion vector are given , respectively , by to to in a matricial form the set of equations ( 1.b ) can be written as @xmath15 where * s * is the @xmath16 matrix formed by the components of the tangent vectors to v@xmath0 multiplied by the metric components of m@xmath1 and * y*@xmath17 is the column matrix @xmath18 formed by the components of the vectors @xmath7 .
the homogeneous system described by equation ( 3 ) can be solved ( for the non - trivial solution ) by taking into account pure algebraic considerations : we need to find a square submatrix of * s * of order @xmath19 that is invertible .
that is always possible as the rows of the matrix * s * are exactly the components of the vectors that generates the tangent space of the spacetime .
thus , they are linearly independent . therefore from linear algebra we know that there exist a submatrix @xmath20 of * s * that is non - singular .
let * p * be a @xmath19 submatrix of * s * that is invertible and * q * the matrix formed from * s * taking out the elements of * p*. * q * is a @xmath21 matrix .
the system ( 3 ) can be written in the equivalent form @xmath22 where @xmath23 are the components of @xmath24 associated to the invertible submatrix and @xmath25 the components of @xmath26 associated to the remaining columns .
thus , from ( 4 ) we have that @xmath27 which allows us to write , @xmath28 taking into account the above definitions we write in the following section an algorithm to determine these quantities explicitly .
algorithm a given the set of @xmath6 cartesian coordinates @xmath29 as a @xmath30 column matrix and the metric tensor @xmath31 as a @xmath32 square matrix , compute the * s * matrix as @xmath33 . using the algorithm b , decompose * s * and @xmath26 matrices in submatrices * p * , * q * , @xmath23 and @xmath34 such that @xmath35 and such that @xmath23 contains the components of @xmath7 corresponding to * p * and @xmath36 those corresponding to * q*. substitute * p * , * q * , @xmath23 and @xmath36 in @xmath37 , obtaining a system of four linear equations in those components of @xmath7 corresponding to * p*. solve that system of equations for the four components of each @xmath7 in @xmath23 in terms of the @xmath38 others . * for * @xmath39 * do * * for * @xmath40 * do * substitute the expressions for @xmath7 in @xmath41 obtaining a non - linear equation in the components of @xmath7 corresponding to * q*. solve this equation for one of the remaining components of @xmath7 in terms of the others .
return this solution to the next equation generated in step a5a . at the end of loop
@xmath42 components of @xmath7 will remaing arbitrary . *
endfor * compute the second fundamental form and the torsion vector from eqns .
( 2.a ) and ( 2.b ) .
* stop * algorithm b ( inicialization ) @xmath43 ( @xmath44 points to a candidate to be the first column of * p * ( or @xmath45 ) ) . *
while * @xmath46 * and * @xmath47 can result an expression that can be zero or not depending of physical informations unavaliable to reduce . in this case , if it is not immediatly zero , the actual program could ask the user if it should be taken as zero by use of the internal ( symbolic ) procedure yesp . ] * do * * begin * @xmath48 ( @xmath49 points to a candidate to be the second column of * p * ( or @xmath45 ) . * while * @xmath50 * and * @xmath47 * do * * begin * @xmath51 ( @xmath52 points to a candidate to be the third column of * p * ( or @xmath45 ) . * while * @xmath53 * and * @xmath47 * do * * begin * @xmath54 ( @xmath55 points to a candidate to be the fourth column of * p * ( or @xmath45 ) . * while * @xmath56 * and * @xmath47 * do * * begin * @xmath57 , @xmath58 ( @xmath59 points to a column of * s * ( or * y * ) , @xmath14 to a column of * q * ( or @xmath60 ) ) . * repeat * * begin * * if * @xmath61 * then * store the column of * s * pointed by @xmath59 as the first column of @xmath62 , the one of * y * as the first of @xmath45 * else * * if * @xmath63 * then * store the column of * s * pointed by @xmath59 as the second column of @xmath62 , the one of * y * as the second of @xmath45 * else * * if * @xmath64 * then * store the column of * s * pointed by @xmath59 as the third column of @xmath62 , the one of * y * as the third of @xmath45 * else * * if * @xmath65 * then * store the column of * s * pointed by @xmath59 as the fourth column of @xmath62 , the one of * y * as the fourth of @xmath45 * else * store the column of * s * pointed by @xmath59 as the @xmath14-th column of @xmath66 , the one of * y * as the @xmath14-th column of @xmath60 and @xmath67 * endif*. @xmath68 * end * * until * @xmath69 * endrepeat*. @xmath70 * endwhile * @xmath71 * endwhile * @xmath72 * endwhile * @xmath73 * endwhile*. * return * @xmath62 , @xmath66 , @xmath45 , and @xmath60
. the termination of the algorithm at step b3 is guaranteed by the existence of the non - singular submatrix *
p*. the algorithm above has been implemented in the algebraic computing system reduce ( hearn , 1986 ; rayna , 1987 ; stauffer et .
al . , 1988 ) making use of its matrix facilities ( see davenport et .
al . , ( 1988 ) , for a good introduction to matrix representation in computer algebra ) .
however , for shortage of space , we left the program out of the paper .
in the specific case of schwarzchild spacetime the embedding ( rosen , 1965 ) is done in a 6-dimensional pseudo - euclidean manifold ( @xmath74 ) with metric @xmath75 .
the schwarzchild embedding is given by the coordinates , @xmath76 where @xmath77 , @xmath78 is a well defined function of @xmath79 , and @xmath79 , @xmath80 , @xmath81 , and @xmath82 denote the spacetime coordinates . the four vectors tangent to v@xmath0
are determined taking the derivative of the coordinates @xmath4 with respect to each one of the spacetime coordinates .
we denote by @xmath83 , @xmath84 the components of the normal vectors with @xmath85 , respectively . to determine these vectors the following conditions have to be considered : i ) the orthogonality of the normal vectors with respect to the tangent vectors ( eq .
1.b ) ( from this we obtain a set of 8 equations ) and ii ) orthonormality of the normal vectors ( eq .
1.c)(from this we get 3 equations ) . out of a total of 11 equations we have now to determine the 6 components of the two vectors @xmath86 e @xmath87 .
we have a set of 11 equations for 12 unknowns .
notice that our unknowns are functions of the spacetime coordinates . according to the algorithm ( and program ) developed in the previous section
, we just need to set @xmath74 and ask reduce to calculate the matrix * s * ( step a1 ) . after some algebraic manipulation we obtain for the schwarzchild spacetime embedding the normal vectors , @xmath88 where @xmath89 , @xmath90 and @xmath91 .
it is easy now to calculate with reduce ( but tedious by hand ) the _ second fundamental form _ and the _ torsion _ vector , from the eqs.(2.a and 2.b - step a6 ) : @xmath92,@xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 .
the geometrical and physical analysis of these quantities are not the main concern of this paper .
however it is important to point out that geometrically the second fundamental form and the torsion vector are fundamental quantities as they determine , together with the metric , the structure of the embedding manifold and physically if general relativity has to be considered as part of a more general theory of embedded manifolds then , besides the metric which represents the classical gravitation , these two quantities have also to be considered : the second fundamental form may be interpreted as the source of the matter fields and the torsion vector may represent a yang - mills gauge field ( see maia , 1986 , for details ) .
the calculations were initially done in interactive form with the version 3.2 of reduce running in an ibm pc - xt and later on ( by demand ) in a microvax running vms .
finally we coded a fairly general program for the 3.3 version of reduce requiring only as input the number of extra dimensions @xmath38 , the set of cartesian coordinates @xmath4 , and the metric @xmath99 of the embedding manifold @xmath12 . the available physical memory of the pc ( 640 kb , but less
when the system is loaded ) is a great limiting factor for the execution of calculus with more general functions and/or higher dimensions ( this would involve matrices with order greater then @xmath100 ) . to circumvent the very often problem of free storage cell explosion in the pc , we had to make the trick of using the output of the results as input for the following steps .
though this initial limitation at the pc , the problem above would have been far more difficult to solve with paper and pencil than with the interactive use of reduce .
the algorithm developed here can be extended to determine all non - singular submatrices of a given matrix determining , in addition , its rank matrix , the number of submatrices of order @xmath101 of * a * is given by @xmath102 , where @xmath103 and @xmath104 . ] .
thus it can also be used to establish the existence and type of solution of a system of linear equations by the simple analysis of its coefficients matrix and extended matrix ranks , for either symbolic ( functions ) or numeric matrix entries , as the manipulation is purely algebraic .
w.l.roque is grateful to the cnpq for financial support through a research grant .
davenport , j. h. , siret , y. , tournier , e. ( 1988 ) .
_ computer algebra : systems and algorithms for algebraic computation_. academic press . | in this paper we describe an algorithm to determine the vectors normal to a space - time v@xmath0 embedded in a pseudo - euclidean manifold m@xmath1 .
an application of this algorithm is given considering the schwarzchild spacetime geometry embedded in a 6 dimensional pseudo - euclidean manifold , using the algebraic computing system reduce . |
over the last two decades the macdonald polynomials @xcite have beeen the subject of much attention in combinatorics and representation theory .
these polynomials are symmetric in @xmath4 variables @xmath5 and also depend on two parameters denoted by @xmath6 and @xmath7 .
they are labelled by partitions of @xmath8 with no more than @xmath4 parts . up to normalization
, they can be defined as eigenvectors of certain family of commuting linear operators acting on the space @xmath9 of all symmetric polynomials in the variables @xmath5 with coefficients from the field @xmath10 .
these operators were introduced by macdonald @xcite as the coefficients of a certain operator valued polynomial of degree @xmath4 in a variable @xmath11 with the constant term @xmath12 . in particular
, macdonald observed that all eigenvalues of the operator coefficient at @xmath11 are already free from multiplicities .
hence this coefficient alone can be used to define the macdonald polynomials .
it is quite common in combinatorics to extend various symmetric polynomials to an infinite set of variables .
these extensions are called _
symmetric functions_. the ring @xmath13 of symmetric functions is defined as the inverse limit of the sequence @xmath14 in the category of graded algebras .
the defining homomorphism @xmath15 here is just the substitution @xmath16 . in particular ,
the macdonald polynomials are extended to infinitely many variables @xmath17 by using their _ stability property _ @xcite and by passing to their limits as @xmath18 .
the limits are the macdonald symmetric functions .
they belong to the ring @xmath13 and are labelled by partitions of @xmath19 .
they have been also studied very well . in particular , the limit at @xmath20 of the renormalized macdonald operator coefficient at @xmath11 was considered in @xcite .
other expressions for the same limit were given in @xcite .
since the higher operator coefficients are not required to define the macdonald polynomials , the limits of these coefficients at @xmath20 received due attention only later . from the geometric point of view
they were studied in @xcite , see also the works @xcite .
explicit expressions for the limits were given in @xcite and independently in @xcite .
all these expressions involved hall - littlewood symmetric functions @xcite in the variables @xmath21 .
these symmetric functions are labelled by partitions of @xmath8 but depend on the parameter @xmath3 only .
they also emerge in the calculus of _ vertex operators _
@xcite which underlies the results of @xcite . in the present article
we construct a different family of commuting operators on @xmath13 such that the macdonald symmetric functions are their eigenvectors . unlike in @xcite our operators
are expressed in terms of the hall - littlewood symmetric functions corresponding to the partitions with one part only .
our construction uses the lax operator formalism , see subsection 2.1 for details .
our theorem 1 gives a relation between the new family of commuting operators and the one we constructed in @xcite .
the proof is based on the notion of baker - akhiezer function corresponding to the lax operator , see subsection 2.2 . in our case
this function is given by theorem 2 .
the proof of the latter theorem is given in subsection 2.3 .
to find the eigenvalues of our new operators we still need the results of @xcite .
it would be interesting to prove directly that the eigenvectors of these operators are the macdonald symmetric functions , see for instance @xcite . also notice that by setting @xmath22 and tending @xmath23 one obtains the _ jack symmetric functions _
@xcite as limits of macdonald symmetric functions .
our new lax operator can be regarded as a discretization of the operator we found in the limiting case @xcite .
the latter operator has been in turn a quantized version of the lax operator for the classical benjamin - ono equation @xcite .
our new lax operator is a quantized version of the one for the classical benjamin - ono equation with discrete laplacian @xcite . in this article
we generally keep to the notation of the book @xcite for symmetric functions .
when using results from @xcite we simply indicate their numbers within the book .
for example , the statement ( 2.15 ) from chapter iii of the book will be referred to as [ iii.2.15 ] assuming it is from @xcite .
fix any field @xmath24 .
denote by @xmath13 the @xmath25-algebra of symmetric functions in infinitely many variables @xmath26 .
let @xmath27 be any partition of @xmath28 .
we will always assume that @xmath29 .
the number of non - zero parts is called the _ length _ of @xmath30 and denoted by @xmath31 .
let @xmath32 be the multiplicities of the parts @xmath33 of @xmath30 respectively
. then @xmath34 .
for @xmath35 let @xmath36 be the _ power sum symmetric function _ of degree @xmath37 : @xmath38 more generally , for any partition @xmath30 put @xmath39 where @xmath40 .
the elements @xmath41 form a basis of @xmath42 . in other words ,
the elements @xmath43 are free generators of the commutative algebra @xmath13 over @xmath24 .
define a bilinear form @xmath44 on @xmath13 by setting for any two partitions @xmath30 and @xmath45 @xmath46 in the above notation .
this form is obviously symmetric and non - degenerate .
we will indicate by the superscript @xmath47 the operator conjugation relative to this form .
in particular , by for the operator conjugate to the multiplication in @xmath13 by @xmath48 with @xmath49 @xmath50 for @xmath35 let @xmath51 be the _ elementary symmetric function _ of degree @xmath37 .
by definition , @xmath52 we will also use a formal power series in the variable @xmath53 @xmath54 by taking logarithms of the left and right hand side of the above display and then exponentiating , @xmath55 the _ complete symmetric functions _ @xmath56 can be determined by the relation @xmath57 where @xmath58 the degree of the element @xmath59 is @xmath37 .
furthermore , by we get an equality @xmath60 the elements @xmath61 as well as the elements @xmath62 are free generators of the commutative algebra @xmath13 over the field @xmath24 .
let @xmath25 be the field @xmath63 with @xmath3 a parameter
. put @xmath64 note that then by and we have @xmath65 in this article we will employ the _ jing vertex operator _
@xmath66 this is a formal series in @xmath53 with coefficients acting on @xmath13 as linear operators .
these operators do not commute , see ( * ? ? ?
* proposition 2.12 ) for commutation relations between them . using another variable @xmath67 instead of @xmath53 in the equalities we get @xmath68 due to and .
the fraction at the right hand side of the equalities should be expanded as a power series in the ratio @xmath69 .
it follows by that @xmath70 following ( * ? ? ?
* proposition 3.9 ) we will use the relation @xmath71 to prove note that due to [ ex.i.5.25 ] the series @xmath72 with operator coefficients showing in the definition is comultiplicative , so that @xmath73 hence the equality is obtained by using with @xmath74 .
now let @xmath30 be any partition with @xmath75 . recall that @xmath76 by our assumption .
introduce a rational function of the variables @xmath77 @xmath78 let us expand every factor with @xmath79 in the product as a power series in @xmath80 respectively . by [ iii.2.15 ]
the _ hall - littlewood symmetric function _
@xmath81 is the coefficient at @xmath82 in the formal series @xmath83
if the partition @xmath30 consists of only one part @xmath84 then @xmath85 is @xmath86 by above definition .
the elements @xmath85 constitute a basis of the vector space @xmath42 .
furthermore , define a bilinear form @xmath87 on the vector space @xmath13 by setting for any partitions @xmath30 and @xmath45 @xmath88 in the notation .
it is obviously symmetric and non - degenerate . by [ iii.4.9 ]
@xmath89 where @xmath90 along with the symmetric function @xmath85 it is convenient to use the symmetric function @xmath91 which is a scalar multiple of @xmath92 . by definition , @xmath93 so that due to @xmath94 now let @xmath25 be the field @xmath95 with @xmath6 and @xmath3 parameters independent of each other . generalizing define a bilinear form @xmath96 on @xmath13 by setting for any partitions @xmath30 and @xmath45 @xmath97 in the notation of .
this form is again symmetric and non - degenerate .
we will indicate by the superscript @xmath98 the operator conjugation relative to the latter form .
in particular , by and for any @xmath49 we have @xmath99 hence by we get @xmath100 using when @xmath30 consists of only one part @xmath84 we get @xmath101 .
now consider the linear operator acting on the vector space @xmath13 as the sum @xmath102 for future discussion note that equals the coefficient at @xmath103 of the series in @xmath53 @xmath104 the operator is clearly self - conjugate relative to the bilinear form . by
* eq.32 ) for any partition @xmath30 the _ macdonald symmetric function _
@xmath105 can be defined up to normalization as an eigenvector of with the eigenvalue @xmath106 for different partitions @xmath30 the eigenvalues are pairwise distinct in @xmath107 .
it follows that the eigenvectors @xmath108 with different @xmath30 are pairwise orthogonal relative to . in the present article we will not be choosing any normalization of @xmath109 .
we will use only the fact [ vi.4.7 ] that the @xmath108 form a basis of the vector space @xmath42 .
in @xcite we introduced the following generalization of the operator .
for any @xmath110 consider the linear operator on @xmath13 @xmath111 then @xmath112 while @xmath113 is the operator . for any @xmath114 the operator
@xmath115 is obviously self - conjugate relative to .
consider a series in a variable @xmath11 @xmath116 where as usual @xmath117 in @xcite we proved @xmath118 if follows from that the operators @xmath119 on @xmath13 pairwise commute .
the eigenvalue of the operator @xmath113 can also be obtained from this equality .
the equality can be derived from the results of ( * ? ? ? * sec.3 ) which in turn are modifications of those of ( * ? ? ?
* sec.9 ) .
our proof ( * ? ? ? * sec.3 ) was independent of all those results .
let us now establish a relation between the works @xcite and @xcite
. for @xmath120 denote by @xmath121 the constant term of the formal series in @xmath77 @xmath122 here the product is regarded as a series in @xmath77 using the expansion rule explained just after displaying it .
this @xmath121 is a certain linear operator on the vector space @xmath42 .
it is convenient to set @xmath123 .
it turns out that @xmath108 for each @xmath30 is an eigevector of the operators @xmath124 like that of @xmath125 .
this fact goes back to @xcite .
it also follows from by the next proposition .
we have the relation @xmath126 where as usual @xmath127 for every partition @xmath30 let us denote by @xmath128 the specialization of the symmetric function @xmath91 to @xmath129 and this is a homogeneous symmetric polynomial in the variables @xmath77 of degree @xmath130 . by using the first equality in and then the expansion [ iii.4.4 ] @xmath131 it follows from the latter equality that @xmath132 hence @xmath133 where @xmath134 denotes the constant term of the formal series in @xmath77 @xmath135 it is known that @xmath136 see ( * ? ? ?
* app.b ) for an elementary proof of this fact .
thus we obtain the relation @xmath137 by substituting the latter expression for @xmath121 in our proposition and by using the definition with the running index @xmath114 replaced by @xmath138 it remains to prove @xmath139 by equating here the coefficients at @xmath140 we have to prove that for every @xmath141 @xmath142 but the last relation follows by setting @xmath143 and @xmath144 in the equality @xmath145 note that the relation established above is equivalent to ( * ? ? ?
* eq.3.3 ) . by using a variation of the mbius inversion (
* lemma 5.1 ) we get from @xmath146
in this section we will construct yet another family of pairwise commuting operators on @xmath13 with the macdonald symmetric functions @xmath108 being their eigenvectors .
let @xmath147\ ] ] be the ring of polynomials in @xmath148 with the coefficients from @xmath13 but without the constant term .
introduce the linear operator @xmath149 on the vector space @xmath150 by setting @xmath151_{{\hspace{0.75pt}}-}\ ] ] where the symbol @xmath152_{{\hspace{0.75pt}}-}$ ] means taking the only negative degree terms of the series .
let us extend the bilinear form from @xmath13 to @xmath150 so that the subspaces @xmath153 are orthogonal to each other , while each one carries the bilinear form determined by identifying that subspace with @xmath154 .
for the operator @xmath155 on @xmath150 conjugate to @xmath156 we then have @xmath157 moreover , using the decomposition of @xmath150 into the direct sum of subspaces the operators @xmath156 and @xmath155 are represented by infinite matrices with operator entries @xmath158 \ 0\,&1&q_1&\cdots \\[2pt ] \,0&0&1 & \cdots \\[-2pt ] \,\vdots&\vdots&\vdots&\ddots\ , \end{pmatrix } \ \quad\text{and}\ \quad \begin{pmatrix } 1&0&\,0&\cdots\ \\[2pt ] q_1^*&1&\,0&\cdots \\[2pt ] q_2^*&q_1^*&\,1&\cdots \\[-2pt ] \vdots&\vdots&\,\vdots&\ddots\ , \end{pmatrix } .\ ] ] further , let @xmath159 be the linear operator on @xmath150 defined by setting @xmath160 accordingly , let @xmath161 the operators @xmath159 and @xmath162 are clearly self - conjugate relative to the bilinear form on @xmath150 defined above .
they are represented by diagonal matrices with the entries @xmath163 and @xmath164 respectively .
our _ lax operator _ on @xmath150 is the composition @xmath165 furthermore introduce the operator @xmath166 by setting @xmath167_{{\hspace{0.75pt}}0}\ ] ] where @xmath168_{{\hspace{0.75pt}}0}$ ] means taking the constant term of a series in @xmath53 .
the conjugate operator @xmath169 is the application of @xmath170 to elements of @xmath171 the operators @xmath172 and @xmath173 are represented by an infinite row and a column with operator entries @xmath174 q_2^ * \\[-2pt ] \vdots \end{pmatrix } .\ ] ] now put @xmath175 it expands as a formal power series in @xmath176 without constant term .
the coefficients of that series are self - conjugate operators on @xmath13 by definition .
these operators will form our new commuting family .
the commutativity follows from the theorem below which relates @xmath177 to @xmath178 .
the proof of the theorem shall be given later .
we have an equality of formal power series in @xmath176 with operator coefficients acting on @xmath13 @xmath179 by using this theorem , the eigenvalues of the operator coefficients of the series @xmath177 on the macdonald symmetric functions @xmath105 can be obtained from
. now introduce the formal power series in @xmath176 @xmath180 the coefficients of this series are operators @xmath181 . by the definitions , @xmath182 we will call the series @xmath183 the _ baker - akhiezer function _ for the lax operator .
our proof of theorem 1 will be based on an expression for @xmath183 given next .
we have an equality of formal power series in @xmath176 with operator coefficients mapping @xmath13 to @xmath150 @xmath184 we will prove theorem 2 in the next subsection .
let us now derive theorem 1 from it .
multiplying both sides of the relation in theorem 1 by @xmath185 on the right and then using we get an equivalent relation to prove : @xmath186 but by using theorem 2 along with definitions , we get the equalities @xmath187_{{\hspace{0.75pt}}0 } \\ & = [ \,j(z)\,a(u\,t^{-1})\,h^\perp(z^{-1}q^{-1}){\hspace{0.75pt}}]_{{\hspace{0.75pt}}0}{\hspace{0.75pt}}.\end{aligned}\ ] ] by our proposition , the right hand of these equalities can be rewritten as the sum @xmath188_{{\hspace{0.75pt}}0 } { \hspace{0.75pt}}/{\hspace{0.75pt}}{({\hspace{0.75pt}}t^{{\hspace{0.75pt}}-1};t^{{\hspace{0.75pt}}-1}{\hspace{0.75pt}})_k}\ ] ] divided by @xmath189 .
note that by using , and then we have @xmath190 therefore by recalling the definition of the operator @xmath121 on @xmath13 and then by using the comultiplicativity [ ex.i.5.25 ] of @xmath191 , the factor @xmath192_{{\hspace{0.75pt}}0}\ ] ] in the summand of is equal to the constant term of the series in @xmath77 @xmath193_{{\hspace{0.75pt}}0}\ , f(z_1{,\ldots,}z_k)\,.\end{gathered}\ ] ] by and this constant term is exactly @xmath194 .
so the sum equals @xmath195 where we first replaced @xmath196 by @xmath114 and then formally included the zero summand corresponding to @xmath197 . using our proposition
, the last displayed sum equals @xmath198 dividing this difference by @xmath199 we get the right hand side of the relation .
this we have derived theorem 1 from theorem 2 .
let us prove the latter .
due to the definition we have to prove the relation @xmath200 let us multiply both sides of this relation by @xmath201 on the left .
in this way we get an equivalent relation to prove : @xmath202 by the definitions of the operators @xmath173 and @xmath203 the latter relation can be written as @xmath204 using the definitions of the operators @xmath205 and @xmath206 it can be rewritten as @xmath207_{{\hspace{0.75pt}}-}\,.\end{gathered}\ ] ] here both sides are series in @xmath11 with operator coefficients that map @xmath13 to @xmath150 .
we can also regard both sides as series in @xmath11 and @xmath53 with operator coefficients mapping @xmath13 to @xmath154 .
this allows us to perform obvious cancellations hence getting to prove @xmath208_{{\hspace{0.75pt}}-}\,.\end{gathered}\ ] ] using the definition the last displayed relation can be also written as @xmath209_{{\hspace{0.75pt}}-}\,.\end{gathered}\ ] ] here @xmath210 is a formal power series in @xmath176 with leading term @xmath103 by .
we also know that @xmath211 see the previous subsection
. hence the coefficents at @xmath11 of both sides of the relation coincide .
let us now multiply both sides of the relation by @xmath212 and take the coefficients at @xmath213 for any @xmath214 .
due to our proposition we obtain @xmath215_{{\hspace{0.75pt}}-}{\hspace{0.75pt}}/{\hspace{0.75pt}}{({\hspace{0.75pt}}t^{{\hspace{0.75pt}}-1};t^{{\hspace{0.75pt}}-1}{\hspace{0.75pt}})_{k-1}}\,.\end{gathered}\ ] ] next multiply both sides by @xmath216 and divide by @xmath217 on the left .
we get @xmath218_{{\hspace{0.75pt}}-}\ , .
\label{divided}\end{gathered}\ ] ] the operator @xmath121 at the left hand side of the relation is the constant term of the series in @xmath219 . using this definition along with , and
the comultiplicativity of @xmath220 , the first summand of the right hand side of is equal to the sum of those terms of the formal series in @xmath221 @xmath222 that are free of @xmath219 . to present in a similar way the expression displayed in the second line of we
will use the following simple lemma .
let @xmath223 be a formal series in @xmath53 with coefficients in any algebra over @xmath63 .
let @xmath224_{{\hspace{0.75pt}}0}$ ] be the constant term of this series .
let @xmath224_{{\hspace{0.75pt}}-}$ ] be the sum of the negative degree terms .
the sum @xmath225_{{\hspace{0.75pt}}0}+(1-t^{-1})\,[\,g({\hspace{0.75pt}}z\,t^{-1}{\hspace{0.75pt}}){\hspace{0.75pt}}]_{{\hspace{0.75pt}}-}\!\ ] ] is equal to the sum of those terms of the series in @xmath226 @xmath227 that are free of @xmath228 . here
the fraction should be expanded as a power series in @xmath69 . verifying this lemma is straighforward .
let us now set @xmath229 by the definition of @xmath230 this @xmath223 is the sum of those terms of the series @xmath231 in @xmath232 that are free of @xmath233 .
the last displayed series equals @xmath234 by and .
note that we have used a similar argument in the previous subsection .
denote @xmath235 applying the lemma , the expression displayed in the second line of is equal the constant term of the series in @xmath232 @xmath236 minus those terms of the series in @xmath237 @xmath238 that are free of @xmath239 .
as we are taking the constant term , the variables @xmath232 in can be replaced by @xmath240 respectively .
the variables @xmath241 in can be replaced by @xmath240 as well .
recall that the coefficients @xmath242 of the series @xmath243 are free generators of the algebra @xmath42 , while the operator product @xmath244 is symmetric in @xmath245 . by the above presentation of the terms of ,
that relation is equivalent to the equality between the symmetrization of @xmath246 and that of @xmath247 here symmetrizing means taking the sum over all @xmath248 permutations of @xmath249 .
let us prove the latter equality . at @xmath250
the sum equals @xmath246 even before symmetrization .
we may assume that @xmath77 are pairwise distinct .
then it suffices to check that that the symmetrization of has no poles at @xmath251 . by the symmetry in @xmath77
taking only @xmath252 will suffice .
the product over @xmath253 showing in the first line of is symmetric in @xmath249 .
at @xmath252 it has a simple pole . by [ iii.1.4 ]
the symmetrization of @xmath246 is @xmath254 hence the residue at @xmath252 of the symmetrization of the whole expression in the first line of is @xmath255 when symmetrizing the negative term of the difference displayed in the second line of , we get a pole at @xmath252 only from the permutations of @xmath77 preserving @xmath256 .
applying [ iii.1.4 ] once again but to the @xmath257 variables @xmath258 instead of @xmath77 the residue at @xmath252 of the symmetrization of the difference displayed in the second line of is @xmath259 to verify that the sum of two products and is zero , we can cancel in both of them the product of the common factors @xmath260 and then use the relation @xmath261 .
this verification completes our proof of theorem 2 .
the first named author has been supported by epsrc grant ep / i014071 , and also acknowledges support of ihs where he stayed in october - november 2014 .
f.bergeron , a.m.garsia , m.haiman and g. tesler , _ identities and positivity conjectures for some remarkable operators in the theory of symmetric functions _ , methods and applications of analysis * 6 * ( 1999 ) , 363420 . | using the lax operator formalism , we construct a family of pairwise commuting operators such that the macdonald symmetric functions of infinitely many variables @xmath0 and of two parameters @xmath1 are their eigenfunctions .
we express our operators in terms of the hall - littlewood symmetric functions of the variables @xmath2 and of the parameter @xmath3 corresponding to the partitions with one part only .
our expression is based on the notion of baker - akhiezer function .
1200 -7 mm |
sustaining large , reliable magnetic fields is a major requirement across a broad range of applications in cold atom experiments , including zeeman slowers@xcite , magnetic traps@xcite , magnetic conveyor belts@xcite and feshbach control@xcite .
all of these applications typically require magnetic fields on the order of 1,000 g or higher .
the need to reach high fields has led to substantial evolution and innovation in the design of electromagnets for both resistive magnets and those based on superconducting elements@xcite .
in contrast to superconducting magnets , resistive magnets allow fast field control for the experiment , but exhibit significant heat dissipation ; heat removal is of critical importance in the optimization of magnet design@xcite .
most electromagnets used in cold atom experiments are constructed out of copper wire or refrigeration tubing@xcite wound into cylinders .
these designs are easy to manufacture and can easily fit around a vacuum component .
there are many drawbacks to these designs , however .
electrical power is dissipated into cooling water that flows in series with the current , leading to a large temperature gradient in the device , a high resistance to cooling water , and significant differences in temperature between the water supply and return .
furthermore , construction of conventional coils requires a significant amount of epoxy , which can degrade over time , and the coil can not be easily modified .
finally , simple wire - wound electromagnets produce weaker magnetic fields by comparison to more advanced designs .
copper arcs ( red ) plus one brass arc ( gold ) as an end cap .
( b ) the current ( red arrows ) flows through the copper arcs ( red ) in a helical pattern through the device .
the cooling water ( blue arrows ) flows parallel to the current through the device in channels cut into the polyester insulators ( grey ) . for simplicity , we only show two copper arcs .
the actual electromagnet contains @xmath9 copper arcs .
] recent investigations by ricci _
et al _ into electromagnet designs for cold atom experiments have focused on improved liquid cooling efficiency by introducing parallel cooling channels@xcite .
the device supports current higher than that allowable in conventional designs ; at @xmath10 a the device runs at @xmath11 @xmath8c .
another modified version of the wire - wound electromagnet design with a parallel cooling water channel within the coil has demonstrated significant improvement to the heat removal ; this device reaches a field of @xmath12 g while dissipating @xmath13 kw . here
we present a modified design for large bore bitter type electromagnets to suit cold atom experiment , made watertight without epoxy .
each electromagnet delivers @xmath1 g while dissipating @xmath14 kw with modest machining and cooling requirements .
the surface area of copper in contact with cooling water compared to the volume of copper is superior to copper wire and refrigeration tubing designs .
a flow rate of @xmath5 @xmath6/min is obtained from a typical laboratory supply ( @xmath15 psi or @xmath16 kpa ) , which keeps the heating of the electromagnet to a mere @xmath7 @xmath8c , and a factor of up to @xmath17 times less heating than traditional and modern electromagnet designs .
the design is suitable for high field tuning of feshbach resonances@xcite , zeeman slowing@xcite and evaporation@xcite in cold atom experiments and related atomic physics applications .
our design follows the classic bitter - type electromagnet @xcite , consisting of @xmath9 annular copper arcs and one annular brass arc , separated by insulating polyester spacers of the same shape . in each successive layer ,
both copper and polyester layers are rotated by one - eighth of a turn to create a helical conduction path .
the assembly is held together by seven steel screws and one brass screw bolted through a series of eight holes bored parallel to the coil axis .
the brass screw serves to deliver electrical current to the annular brass arc and in turn to the first copper arc , furthest from the mount .
the current returns through the last copper arc . in order to ensure a clear optical path ,
the inner bore of the coil is significantly larger than more common bitter coil designs of similar size@xcite .
the steel mount has threaded holes for optical mounts and mounting holes for attachment to our vacuum chamber .
our electromagnets reside inside recesses surrounding the viewports of our vacuum chamber .
two independent electromagnets on opposite sides of the vacuum chamber allow for either a nearly homogenous field in a parallel current configuration or a quadrupole field in an antiparallel current configuration .
we deviate from the conventional bitter - type design by using eight bores to deliver and remove cooling water to a series of ducts with rectangular cross - sections across the copper surface .
an advantage of this design geometry over the original bitter coil is that all electrical and cooling water feed throughs leave the device on the same side of the electromagnet instead of entering one side and exiting the other@xcite , see fig .
[ fig:1 ] .
these water channels between copper layers are defined by removed material from the polyester spacers .
the channels in the polyester spacers are rimmed with nitrile gaskets to ensure a watertight seal .
a complete turn consists of a copper arc , a polyester spacer , a copper insert and five nitrile gaskets , see fig
. [ fig:2 ] .
copper arcs are @xmath18 mm in thickness and the filling fraction of copper in a complete turn is @xmath19 .
coil elements were machined using standard milling machines and machine lathes .
copper arcs were formed from @xmath18 mm sheets of ultra - pure ( 99.99% ) ultra - conductive copper alloy @xmath20 , cut into @xmath21 mm squares , stacked and then clamped between two steel discs .
arcs were machined to the described form as a single stack .
the mount and cooling water distribution elements were constructed from steel alloy @xmath22 and polycarbonate , respectively .
the steel mount provides a rigid seat for the steel clamping screws .
nitrile gaskets and polyester insulators were cut with a co@xmath23 laser cutter ( universal vls4.60 ) .
the entire device is sealed by physical compression to greater than @xmath15 psi above atmosphere without chemical sealants .
the steel screws have acetal washers to prevent shorts to the annular brass arc ; this isolates the whole steel mount from the coil with a measured resistivity over @xmath24 k@xmath25 .
a schematic of the electromagnet is shown in fig .
[ fig:2 ] and a picture of the coil is shown in its test scaffolding in fig .
[ fig:3_1 ] .
it is useful to know how cooling water flows through the channels .
the cross section of the rectangular ducts is @xmath26 mm@xmath27 where @xmath28 mm is the height and @xmath29 mm is the width .
each channel length @xmath30 is about @xmath31 mm .
each insulator has three channels and the insulator between the last copper arc and the polycarbonate cooling block has four channels .
this gives our design a total of @xmath32 channels with a total flow of @xmath5 @xmath6/min , making the average velocity of cooling water in a channel @xmath33 m / s .
the channel height is less than the laminar boundary layer of fluid flow over a flat plate characterized by the displacement thickness@xcite @xmath34 where @xmath35 m@xmath27/s is the kinematic viscosity for water .
the reynolds number , @xmath36 , for the case of a wide rectangular duct is @xmath37 where @xmath38 m@xmath39/s is the volumetric flow rate of @xmath40 @xmath6/min of cooling water and @xmath41 mm is the hydraulic diameter for our rectangular ducts . for a reynolds number of @xmath42
the flow would be turbulent@xcite , implying that the flow along the length of the cooling channels is in the laminar flow regime .
laminar flow induces less vibrational noise . to match
the cooling water flow of our bitter electromagnets , refrigeration tubing and flat wire designs would need to run at very high water pressure on the order of @xmath43 psi or @xmath44 mpa , which is inconvenient and hazardous .
further , our use of parallel cooling channels and the fact that flow conductance for this design increases roughly in proportion to the total number of turns results in a highly scalable design . with copper tubing wound into a cylindrical electromagnet , increasing the number of turns would cause a proportional increase in the hydrodynamic impedance .
we investigate two ways to characterize the thermal characteristics of the electromagnet .
each cooling channel touches two copper arcs which leads to a large surface area of copper in contact with cooling water .
the ratio of the surface area of copper in contact with cooling water to the volume of copper for the electromagnet is @xmath45 which suggests that dissipated heat in the coil can be efficiently removed within an average propagation distance of merely @xmath46 mm .
we can further characterize the efficiency to remove heat by the nusselt number , @xmath47 . for laminar flow in a duct with a rectangular cross - section we obtain@xcite
@xmath48 the nusselt number parameterizes the dominance of convective over conductive flow .
a nusselt number of @xmath49 to @xmath50 corresponds to heat removal by convective flow , implying the heat removal of our device comes mostly from conductive flow .
( a ) + ( b ) in order to obtain an operational profile for the coils , we use the setup shown in fig . [ fig:3_2 ] .
we measure the magnetic field , the resistance , the inductance and thermal characteristics .
magnetic field measurements and thermal characteristics were obtained together during electrical stress tests .
the resistance and inductance were measured separately .
( a ) + ( b ) to experimentally investigate the external magnetic field , we increased the current from @xmath51 a to @xmath3 a in @xmath52 a intervals over @xmath53 minutes .
this allowed us to monitor the temperature in order to determine stability of the operating mode .
the axial magnetic field @xmath2 mm away from the end cap of the coil was found to be @xmath54 g / a , see fig .
[ fig:4_1 ] .
application of @xmath3 amps would produce a peak field of @xmath55 g for a pair of coils separated by @xmath56 mm .
to predict the field , we start with the averaged current density @xmath57 of a bitter electromagnet , given by @xmath58 where @xmath59 is the radial position , @xmath60 mm is the inner radius of the electromagnet , @xmath61 mm is the outer radius , @xmath62 is the copper filling factor and @xmath63 is the applied current .
the @xmath64 current distribution results from the cylindrical geometry of the electromagnets .
the magnetic field generated by a single copper arc along the axial direction can be expressed as@xcite @xmath65,\ ] ] where @xmath66 h / m is the magnetic permeability and @xmath67 is the position from the center of the arc .
this equation predicts that we should output @xmath68 g / a for our electromagnet which agrees well with our measurement of @xmath54 g / a .
the electrical response of the bitter electromagnet is measured by the set up shown in fig .
[ fig:3_2 ] .
the resistance can be estimated by @xmath69 where @xmath70 is the number of turns , @xmath71 mm is the thickness of a copper arc and @xmath72 @xmath25 m is the resistivity of copper .
this equation predicts a resistance of @xmath73 m@xmath25 .
this is in agreement with our measured value of @xmath74 m@xmath25 .
we measure the impedance of the coil over a frequency range of @xmath75 hz to @xmath76 khz subjected to ac voltage at @xmath77 v peak to peak and extract the inductance @xmath78 , see fig .
[ fig:4_2 ] . for the inductance we obtained @xmath79 @xmath80h .
the predicted bitter coil inductance is expressed by @xmath81 where @xmath82 is the half height of the electromagnet and @xmath83 is the self - inductance factor which depends on the geometry .
the dimensionless factor of self - inductance @xmath83 depends only on the current distribution and geometry of the electromagnet .
this factor for our bitter electromagnets was found to be @xmath84 , compared to the typical bitter electromagnet with @xmath85 from literature@xcite .
the calculated inductance of @xmath86 @xmath80h is in good agreement with our measurement of @xmath79 @xmath80h .
our device exhibits an inductance at least two orders of magnitude below conventional designs .
the bitter electromagnets can thus respond faster . as we increase the current through the electromagnet .
we perform this measurement for various flow rates to determine the dependance of coil temperature on the flow rate .
inset : dependence of thermal resistance @xmath87 on the inverse water flow rate @xmath88 . ] to measure how the power dissipation varies with the water flow we took multiple electrical tests at three different flow rates .
we observe a linear dependance of the coil temperature on the dissipated power .
the change in temperature of the coil , @xmath89 , as a function of the flow rate @xmath90 and dissipating power @xmath91 is@xcite @xmath92 where @xmath87 is the absolute thermal resistance , the water heat capacity is @xmath93 j / kg c@xmath8 and density is @xmath94 g / cm@xmath39 .
the thermal resistance @xmath96 of the tests ranges from @xmath77 @xmath8c / kw to @xmath17 @xmath8c / kw and depends inversely upon the flow rate @xmath90 , see fig .
[ fig:5 ] .
we found the slope of the thermal resistance versus the inverse flow rate to be @xmath97 @xmath8c cm@xmath39/w s which is in good agreement with the predicted value , @xmath98 @xmath8c cm@xmath39/w s. the thermal resistance of the device at a given water pressure is of particular merit ; during continuous use , for every kilowatt , the device only gains a few degrees celsius .
conventional devices dissipate more power at comparable amperage and further exacerbate this difference through serial cooling schemes .
both the thermal resistance and water resistance of our bitter electromagnet are very low .
the predicted failure mode for continuous operation of a pair of symmetric coils is one in which the cooling water boils within the devices , which may force cracks to open in the seals .
based on the above measurements , this mode is predicted to occur around a power dissipation of @xmath99 kw and a current of @xmath100 ka per coil with cooling water flowing at @xmath40 @xmath6/min . the peak magnetic field along the axis of symmetry of the two coils , when separated by @xmath101 mm , is @xmath102 g. in summary , we describe a simple , scalable resistive bitter electromagnet design suitable for high - field cold atom experiments .
our design provides stable temperatures close to the ambient during high field operation .
our electromagnet significantly outperforms conventional electromagnets made of refrigeration tubing or flat wire due to high copper filling fraction and efficient heat removal .
further , our modified bitter electromagnet exhibits an inductance at least two orders of magnitude below that of other electromagnet designs .
modified elements from the original bitter electromagnet design include tapped holes for optics , all feedthroughs on one side of the device , and a large central bore for clear optical access , which are essential for cold atom experiments .
this design requires basic machining skills and readily available materials to produce and scale up .
the authors wish to thank l. clark , e. hazlett , k. jimnez - garca , j. johansen , p. scherpelz and s.k .
tung for insightful conversation and h. krebs for machining advice .
the authors acknowledge support from the nsf - mrsec program , nsf award no .
phy-1206095 , afosr - muri , and aro - muri 63834-ph - mur .
t. barrett , s. daporeschwartz , m. ray , and g. lafyatis , phys .
lett . * 67 * , 3483 ( 1991 ) .
e. streed , a. chikkatur , t. gustavson , m. boyd , y. torii , d. schneble , g. k. campbell , d. e. pritchard , w. ketterle , rev .
instrum . *
77 * , 023106 ( 2006 ) . | we present the design , construction and characterization of bitter - type electromagnets which can generate high magnetic fields under continuous operation with efficient heat removal for cold atom experiments .
the electromagnets are constructed from a stack of alternating layers consisting of copper arcs and insulating polyester spacers .
efficient cooling of the copper is achieved via parallel rectangular water cooling channels between copper layers with low resistance to flow ; a high ratio of the water - cooled surface area to the volume of copper ensures a short length scale @xmath0 to extract dissipated heat .
high copper fraction per layer ensures high magnetic field generated per unit energy dissipated .
the ensemble is highly scalable and compressed to create a watertight seal without epoxy . from our measurements , a peak field of @xmath1 g
is generated @xmath2 mm away from a single electromagnet with a current of @xmath3 a and a total power dissipation of @xmath4 kw . with cooling water flowing at @xmath5 @xmath6/min ,
the coil temperature only increases by @xmath7 @xmath8c under continuous operation . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.