article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
this interdisciplinary work tailors and whets some state - of - the - art operations research methods to find solutions in a fundamental class of problems in design of experiments .
our exact algorithms call the margot ( 2007 ) integer linear programming ( ilp ) solver for enumerating orthogonal arrays ( oas ) or finding optimum oas .
we make comparisons to the efficient constraint programming ( cp ) with isomorphism rejection oa enumeration algorithm of schoen , eendebak and nguyen ( 2010 ) and exploit equivalence of orthogonal designs ( ods ) to speed up our exact algorithms .
a factorial design @xmath4 with @xmath5 runs and @xmath6 factors each having @xmath7-levels is an orthogonal array of strength @xmath8 , @xmath9 , denoted by @xmath10 , if each of the @xmath11 level combinations appears exactly @xmath12 times when @xmath4 is projected onto any @xmath8 factors . clearly , an @xmath10 is also an oa@xmath13 for @xmath14 .
the index @xmath15 of an @xmath10 is defined as @xmath12 .
an @xmath10 is universally optimal for estimating the model containing all main effects and all interactions having @xmath16 factors or less ; see cheng ( 1980 ) and mukerjee ( 1982 ) .
when @xmath7 is a prime power , regular fractional factorial designs ( also known as linear codes ) , constructed by solving a linear system of equations over the field gf(@xmath7 ) , are examples of oas .
since a regular fractional factorial design is the solution set of a linear system of equations over gf@xmath17 , the number of runs it contains must be a power of @xmath7 .
such a restriction does not apply to all oas .
there is a vast literature on the construction and existence of regular fractional factorial designs ( see hedayat , sloane and stufken ( 1999 ) ) , but not as much is known about all oas .
the design obtained by permuting factors or runs as well as levels in a subset of factors in an @xmath10 is also an @xmath10 .
let such operations be called _
isomorphism operations_. two @xmath10 are called _ isomorphic _ if one can be obtained from the other by applying a sequence of isomorphism operations . assuming the hierarchical ordering principle ( see section 3.5 of wu and hamada , 2000 ) , two @xmath10 are compared under model uncertainty using the _ generalized minimum aberration _ ( gma ) criterion developed in xu and wu ( 2001 ) .
let @xmath18 $ ] be a @xmath19-level design with entries @xmath20 having @xmath5 runs and @xmath6 factors , and @xmath21 be a nonempty subset of @xmath22 factors .
the gma criterion is based on the concept of the @xmath23-characteristics @xmath24 introduced by tang and deng ( 1999 ) .
note that @xmath25 and larger values @xmath26 imply a stronger degree of aliasing among the factors in @xmath27 .
an average aliasing among all subsets of @xmath22 factors is @xmath28 and the sequence @xmath29 is called the _ generalized word length pattern _ ( gwp ) of @xmath4 .
the gma criterion selects designs that sequentially minimize the gwp .
a design with the same first non - zero gwp entry as a gma design is called a _ weak _ gma design . in general
the gwp for @xmath7-level designs is computed as follows .
let @xmath30 be the number of columns at which the @xmath31th and @xmath32th rows of @xmath4 differ , and define @xmath33 for @xmath34 .
the _ distance distribution _ ( @xmath35 ) of @xmath4 determines the components of the gwp and vice versa ; the direct relationships provided in xu and wu ( 2001 ) are : @xmath36 for @xmath37 , where @xmath38 and @xmath39 are the _ krawtchouk polynomials_. when computing the krawtchouk polynomials , the recursion @xmath40 with initial values @xmath41 and @xmath42 is useful .
it is well known that a factorial design @xmath4 with @xmath5 runs , @xmath6 factors , and @xmath7 levels is an @xmath10 if and only if @xmath43 .
( for @xmath44 this is equivalent to @xmath45 for all @xmath46 and @xmath47 ) .
hence , for a fixed @xmath48 combination the gma criterion seeks an @xmath10 with the largest possible @xmath8 . in general , finding gma designs is a very difficult problem . butler ( 2003 , 2004 )
theoretically constructed 2-level gma designs .
proofs for establishing that the constructed designs are gma involved finding lower bounds for the gwp of 2-level designs for certain number of runs and factors .
xu ( 2005 ) derived lower bounds for the gwp using linear programming in infinite precision and also found 2-level factorial designs based on the nordstorm - robinson code that achieve the bounds .
bulutoglu and kaziska ( 2010 ) improved the lower bounds of xu ( 2005 ) using ilp in infinite precision developed by espinoza ( 2006 ) and applegate _
et al . _ ( 2007 ) .
another way of finding gma designs is by classifying all non - isomorphic oas .
if two @xmath10 are isomorphic , they are indistinguishable under the gma criterion .
on the other hand , there are non - isomorphic @xmath10 with the same gwp .
classifying all non - isomorphic @xmath10 allows the best to be found with respect to the gma or any other ordering criterion that remains invariant between isomorphic designs .
many have studied the problem of classifying all non - isomorphic @xmath10 ( e.g. , stufken and tang 2007 ; bulutoglu and margot 2008 ; and schoen , eendebak and nguyen ( 2010 ) .
bulutoglu and margot ( 2008 ) showed that finding all @xmath10 is equivalent to finding all nonnegative integer solutions to a symmetric ilp with binary coefficients .
two solutions are defined to be _ isomorphic _ if they correspond to isomorphic oas , so finding all non - isomorphic solutions is equivalent to finding all non - isomorphic @xmath10 .
branch - and - cut algorithms ( see padberg and rinaldi , 1991 ) are a standard technique for solving ilps .
however , a presence of symmetry in an ilp requires extending the basic branch - and - cut algorithm to avoid resolving isomorphic subproblems .
such an extension developed by margot ( 2002 , 2003ab , 2007 ) was used by bulutoglu and margot ( 2008 ) to find all isomorphism classes of @xmath10 for many @xmath49 combinations .
schoen , eendebak and nguyen ( 2010 ) provided a thorough review of theoretical and computational oa enumeration literature and developed the minimum complete set ( mcs ) algorithm for enumerating oas up to isomorphism .
mcs is a cp algorithm with isomorphism rejection .
schoen , eendebak and nguyen ( 2010 ) used the mcs algorithm to find all non - isomorphic @xmath10 for many @xmath49 combinations including all those in bulutoglu and margot ( 2008)and mixed level oas where all factors do not have the same number of levels .
the mcs algorithm is up to 3 orders of magnitude faster than the other algorithms in the literature . in section [ sec : enum ] , we introduce four algorithms based on ilp and ilp with isomorphism pruning developed by margot ( 2007 ) .
these algorithms sequentially enumerate non - isomorphic @xmath10 for @xmath50 , where @xmath51 is the largest @xmath6 for which an @xmath10 exists .
our methods require finding all non - isomorphic solutions to sequences of symmetric ilps .
the fastest of our algorithms is at least three orders of magnitude faster than the mcs algorithm for enumerating oa@xmath52 when @xmath53 .
this enabled us to enumerate all non - isomorphic oa@xmath0 for @xmath1 and oa@xmath2 for @xmath3 for the first time .
a catalog of the new gma oas based on our enumerations is documented in section [ sec : results ] .
the new gma designs are available upon request .
for @xmath54 , let @xmath55 be the number of non - isomorphic @xmath10 and @xmath56 be a set of non - isomorphic ( non - od - equivalent when @xmath44 ) @xmath10 .
extension algorithms obtain @xmath57 from @xmath58 .
[ alg : basic ] input : @xmath5 , @xmath6 , @xmath7 , @xmath8 , and @xmath58 . set @xmath59 . 1 .
obtain a set @xmath60 of @xmath10 such that the first @xmath61 columns are the same as @xmath62 and exactly 1 representative from each isomorphism class ( od equivalence class when @xmath44 ) of such @xmath10 is included.[item : extend ] 2 .
increment @xmath63 and then repeat step [ item : extend ] if @xmath64 .
[ item : m ] set @xmath65 .
form @xmath57 from @xmath66 by picking exactly one representative from each isomorphism class ( od equivalence class when @xmath44 ) of @xmath10 in @xmath66 .
output : @xmath57.[item : nauty ] the @xmath11 full factorial design replicated @xmath67 times is the singleton @xmath68 .
given @xmath58 , @xmath69 is obtained after applying the basic algorithm @xmath70 times .
alternatively , if @xmath71 , bulutoglu and margot ( 2008 ) ( with od - equivalence reduction when @xmath44 ) can be used to directly obtain input @xmath58 of the basic algorithm . for step [ item : extend ] and step [ item : nauty ] , we used a new extension of the graph based method based on od - equivalence if @xmath44 and the method itself otherwise with the program ` nauty ` ( see mckay ( 2013 ) , mckay and piperno ( 2013 ) ) as described in ryan and bulutoglu ( 2010 ) .
the contributions here are the efficient versions of step [ item : extend ] and the new extension of the graph based method in step [ item : extend ] and step [ item : nauty ] .
od - equivalence reduction , od - equivalence as well as the new extension of the graph based method are discussed in section [ sec : fold ] .
let the factor levels of an @xmath10 be coded @xmath72 .
lemma [ lem : y ] ( which follows directly from the definition of an @xmath10 ) and definition [ def : cor ] are used to formulate the ilp feasibility problem resulting from extending an @xmath73 to an @xmath10 .
[ lem : y ] let @xmath4 be an @xmath5 run , @xmath6 factor , @xmath7-level factorial design with columns @xmath74 @xmath75
. let @xmath76 be the @xmath77 matrix with columns @xmath78 where for @xmath79 and @xmath80 the @xmath31th entry of @xmath81 is @xmath82 if the @xmath31th entry of @xmath83 is @xmath84 and is @xmath85 otherwise .
then @xmath4 is an @xmath10 if and only if for @xmath86 @xmath87 for any @xmath88 columns @xmath89 of @xmath76 such that @xmath90 for all @xmath91 .
[ def : cor ] let @xmath76 as in lemma [ lem : y ] be a solution to the system of @xmath92 equations ( [ eqn : jy01 ] ) . also , let @xmath4 be the oa@xmath49 obtained from @xmath76 with columns @xmath93 where @xmath94 is the @xmath95 vector of @xmath82s
. then @xmath4 is called _ the @xmath10 corresponding to @xmath76_. [ alg : br3idd ] use algorithm [ alg : basic ] with the following ilp extension step . 1 .
construct @xmath96 from @xmath97 using lemma [ lem : y ] , extend @xmath96 with @xmath98 columns of binary variables @xmath99 for @xmath100 , and obtain all solutions to the ilp : @xmath101 for each pair of replicated run indices @xmath102 in the input design and for any @xmath103 columns @xmath104 as in lemma [ lem : y ] with the last @xmath105 columns of @xmath76 deleted .
each solution matrix @xmath106 $ ] to ilp ( [ ilp : oa ] ) has a corresponding @xmath10 .
take @xmath60 to be the set of all such @xmath10 .
the @xmath107 constraints in ilp ( [ ilp : oa ] ) with right hand sides @xmath108 for @xmath86 imply that any projection of @xmath8 columns which includes the new column will be a @xmath11 full factorial design replicated @xmath15 times . fixing the first run of the new factor at the low level by setting @xmath109 and ordering factor levels in the new column over
replicates within the input design by requiring @xmath110 for @xmath100 boosts speed by reducing @xmath111 .
the objective function @xmath85 was a convenience , so we could apply the ilp solvers provided by professor francois margot .
also , using @xmath85 instead of other objective functions resulted in a faster algorithm . for an @xmath112 ,
let @xmath113 be the @xmath114 matrix obtained from @xmath76 in lemma [ lem : y ] by concatenating columns @xmath115 for @xmath116 .
let @xmath117 , where @xmath118 is the group of all permutations of @xmath119 objects and @xmath120 is the resulting matrix when the rows and columns of @xmath113 are permuted according to @xmath121 and @xmath122 , respectively .
let @xmath123 be the maximum subgroup of @xmath124 that does not have an element that sends a row s index of @xmath4 to the index of a replicate row .
algorithm [ alg : br3idd ] can be run using the group @xmath125 with the isomorphism pruning of margot ( 2007 ) in a branch - and - cut algorithm to implement isomorphism rejection while solving the needed ilps ; see section [ sec : proofbr3idd ] for a proof .
algorithm [ alg : br3idd ] is an example of basic extension algorithm [ alg : basic ] ; it solves an ilp in step [ item : extend ] with @xmath126 binary variables . unlike algorithm [ alg : br3idd ]
, the bulutoglu and margot ( 2008 ) algorithm solves one ilp in @xmath127 nonnegative integer variables to directly find all @xmath10 .
the bulutoglu and margot ( 2008 ) ilp formulation is defined by @xmath128 equality constraints and an unknown solution vector @xmath129 whose @xmath130th entry is the number of times the factor level combination @xmath131 appears in the sought after @xmath10 .
next , we provide an improved version of the bulutoglu and margot ( 2008 ) formulation whose lp relaxation does not have redundant constraints .
the improved formulation has @xmath132 equality constraints and yet it is equivalent to the bulutoglu and margot ( 2008 ) formulation .
[ alg : bf ] input : @xmath5 , @xmath6 , @xmath7 , @xmath8 .
find all solutions to ilp : @xmath133}}= \frac{n}{s^q},\label{ilp : bf}\\ & 1 \leq x_1 , \ \
{ \boldsymbol{x}}\in \{0,\cdots , p_{\max}\}^k \nonumber\end{aligned}\ ] ] for @xmath134 and for each subset of @xmath88 indices @xmath135 using the group @xmath136 from bulutoglu and margot ( 2008 ) to prune the enumeration tree .
compute @xmath137 by solving the ilp in lemma 5 in bulutoglu and margot ( 2008 ) .
output : @xmath57 .
the group @xmath136 in ilp ( [ ilp : bf ] ) maps a solution @xmath138 corresponding to an @xmath10 to that of an isomorphic oa ; this implies that the branch - and - cut algorithm with isomorphism pruning returns the unique @xmath10 that is lexicographically minimum in rows in the orbit of @xmath136 per isomorphism class .
the constraint @xmath139 is valid for the sought after isomorphism class representatives , and its inclusion boosts speed .
next , an extension version of algorithm [ alg : bf ] is defined .
it reduces the number of variables from @xmath127 to @xmath140 , where @xmath141 is the number of distinct runs in the input @xmath73 being extended .
let the rows of input design @xmath4 be ordered lexicographically so that repeated runs appear next to each other .
let @xmath142 for @xmath143 be the number of times the @xmath31th row of @xmath144 appears in the input design , where @xmath144 is all the @xmath145 level combinations with @xmath61 factors ordered lexicographically .
let @xmath146 be such that @xmath147 for @xmath148 .
let @xmath149 be the maximal subgroup of @xmath136 that maps the input @xmath73 to itself .
[ alg : hybrid4 ] use algorithm [ alg : basic ] with the following ilp extension step . 1 .
construct @xmath96 from @xmath150 as in lemma [ lem : y ] .
let @xmath151 be such that the @xmath152th row in the input design is the first row different from the @xmath153th row .
let @xmath138 be the indicator vector of the sought after @xmath10 as in ilp ( [ alg : bf ] ) and @xmath154 be the resulting vector with @xmath155 entries after all the entries of @xmath138 that do not appear in the constraints of the ilp below are deleted .
find all solutions or all non - isomorphic solutions using the group @xmath149 for isomorphism pruning to ilp : @xmath156 and for any @xmath103 columns @xmath157 as in lemma [ lem : y ] with the last @xmath105 columns of @xmath76 deleted .
each solution @xmath154 to ilp ( [ ilp : oahybrid ] ) has a corresponding @xmath10 .
take @xmath60 to be the set of all such @xmath10 .
the validity of algorithm [ alg : hybrid4 ] is theoretically justified in section [ sec : proofhybrid4 ] .
empirically , algorithm [ alg : hybrid4 ] was the most efficient out of those tested in section [ sec : results ] when extending oa@xmath158 to oa@xmath159 and oa@xmath160 to oa@xmath161 . a major improvement in the speed of algorithm [ alg : hybrid4 ] is achieved if the larger group @xmath136 is used for isomorphism pruning .
however , using @xmath136 would only be valid if all @xmath10 that were lexicographically minimum in rows in the orbit of @xmath136 were extensions of oa@xmath162 that were lexicographically minimum in rows in the orbit of @xmath136 , but this is not the case .
theorem [ thmod ] is the basis for a data compression that makes more enumerations possible .
its proof is in section [ sec : proofthmod ] .
[ hadamardeq ] let @xmath164 and @xmath165 be @xmath19-level , @xmath6-factor factorial designs whose levels are @xmath20 .
then , @xmath164 and @xmath165 are called _ hadamard equivalent _ if @xmath165 can be obtained from @xmath164 by applying signed permutations to the columns and/or rows of @xmath164 , see mckay ( 1979 ) .
[ odeq ] let @xmath166 and @xmath167 be @xmath19-level , @xmath6-factor factorial designs whose levels are @xmath20 . then
, @xmath166 and @xmath167 are called _ od - equivalent _ if @xmath168 $ ] is hadamard equivalent to @xmath169 $ ]
. clearly , isomorphic designs are od - equivalent . however , the converse is not true .
[ thmod ] let @xmath8 be even , @xmath170 , and @xmath4 be an oa(@xmath171 ) that extends to an oa(@xmath172 ) . if @xmath173 is an oa(@xmath171
) that is od - equivalent to @xmath4 , then @xmath174 also extends to an oa(@xmath172 ) . for
a 2-level , @xmath6-factor , @xmath5-run design @xmath164 , mckay ( 1979 ) defined a graph @xmath175 with @xmath176 vertices and showed that @xmath164 is hadamard equivalent to @xmath165 if and only if @xmath177 and @xmath178 are isomorphic graphs .
for @xmath44 , step [ item : extend ] of algorithm [ alg : basic ] and optionally algorithm [ alg : bf ] finds a set of all non - od - equivalent oa(@xmath179 ) .
this is done as follows .
first , a column of 1s is added to each oa(@xmath179 ) in @xmath60 or @xmath57 .
second , each resulting design is converted to the graph defined by mckay ( 1979 ) .
third , the set of resulting graphs is reduced to a subset of non - isomorphic graphs by using the ` shortg ` utility in ` nauty ` , see mckay ( 2013 ) .
( at this point algorithm [ alg : bf ] converts the resulting graphs to designs and quits ) . in step [ item : nauty ] ,
isomorphism reduction is applied to the graphs in @xmath66 by using ` shortg ` again .
finally the resulting non - isomorphic graphs are converted to a set of all non - od - equivalent oas . for @xmath44 ,
the major improvement here is implementing od - equivalence reductions in step [ item : extend ] and step [ item : nauty ] instead of oa - isomorphism reductions .
this greatly reduces the number of ilps that need to be solved at each extension step .
also , a set of all non - isomorphic oa(@xmath179 ) can be extracted from a set @xmath66 of non - od - equivalent oa(@xmath179 ) as follows . 1 . for each design @xmath180 ,
compute @xmath181\ ] ] for @xmath182 , where @xmath183 is the @xmath31th column of @xmath4 and @xmath184 .
2 . convert each resulting design to the graph defined in ryan and bulutoglu ( 2010 ) after deleting its column of all 1s .
3 . reduce the resulting set of graphs to a subset of non - isomorphic graphs by using the ` shortg ` utility of ` nauty ` mckay ( 2013 ) .
4 . convert each graph back to its corresponding design as defined in ryan and bulutoglu ( 2010 ) .
we do not report the cpu time needed to extract all non - isomorphic @xmath163 from a set of non - od - equivalent @xmath163 as this is insignificant compared to the time needed to obtain a set of all non - od - equivalent @xmath163 .
we used the algorithms in sections [ sec : enum ] and [ sec : fold ] to extend the known @xmath163 catalog ; summaries are in tables [ tab : enum ] and [ tab : bj ] .
the ` isop1.1 ` version of the margot ( 2007 ) solver that calls the ` cplex12.5.1 ` libraries was used . in table
[ tab : enum ] , we report the number of od - classes only in cases where we used od - equivalence reductions . in such cases steps 1 - 4 of section [ sec : fold ] were necessary to obtain all non - isomorphic @xmath163s from the enumerated ods .
in all the algorithms isomorphism pruning was turned off if the size of the symmetry group was strictly less than 4 . whenever the isomorphism pruning was turned off ilps were solved by interactive ` cplex12.5.1 ` .
( settings in ` cplex ` were optimized by @xmath185 and @xmath186 on a small subset of the corresponding ilps . ) in table [ tab : enum ] enumerations , isomorphism pruning in algorithm [ alg : br3idd ] was always turned off .
jobs were run on ` intel(r ) xeon(r ) ` @xmath187ghz processors , except the cp - based times in table [ tab : schoen ] are those reported by schoen , eendebak and nguyen ( 2010 ) .
an issue was determining the most efficient algorithm to produce a given row in table [ tab : enum ] and [ tab : schoen ] , or else the job might not finish . for small @xmath6 , algorithm [ alg : bf ] was fast .
for the more time consuming problems with moderate and large @xmath6 , we typically tested each extension algorithm on the same small subset of @xmath73 and then used the fastest to finish the full job . with large @xmath6 ,
algorithm [ alg : br3idd ] was the most efficient because many of the ilps had @xmath188 , so use of isomorphism pruning was an unnecessary computational overhead .
this , however , was not true for small or moderate @xmath6 .
as @xmath5 gets larger compared to @xmath6 , the cp - based algorithm suffers from the exponential growth of the search space and can not compete with the best ilp - based algorithm .
however , when @xmath5 is small and the number of non - isomorphic oa@xmath162 is large , the cp - based algorithm is the most efficient algorithm ; see table [ tab : schoen ] .
this happens for two reasons .
1 ) as the number of variables increases solving lp relaxations can detect infeasibilities higher in the enumeration tree compared to the consistency checks of the cp algorithm .
this makes the computational burden of solving lp relaxations worthwhile .
in fact , this phenomenon becomes more pronounced as the number of variables increases .
2 ) the cp - based algorithm does not find all non - isomorphic extensions of each non - isomorphic oa@xmath162 in @xmath60 to @xmath10 , but only those that are lexicographically minimum in columns .
this eliminates many designs that are otherwise produced by the ilp - based algorithms . using the group @xmath136 in algorithm [ alg : hybrid4 ]
would get rid of this inefficiency , but , as stated in section [ sec : enum ] , this does not result in a valid algorithm .
table [ tab : schoen ] times for ilp formulations are based on an earlier version of the margot ( 2007 ) solver .
this solver is built on ` bcp / stable/1.1 ` and uses the lp solver ` clp1.7 ` ; see ` http://www.coin-or.org/projects/bcp.xml ` .
we observed this solver to be more than @xmath189 times slower than ` cplex12.5.1 ` when its isomorphism pruning utility is turned off , so the ilp - based times in table [ tab : schoen ] are conservatively high . in spite of this , the superiority of ilp over cp at large @xmath5 is the trend in table [ tab : schoen ] .
we remind the reader that the caveat with table [ tab : schoen ] comparisons is that they are indirect to begin with because the cp times were taken from schoen , eendebak and nguyen ( 2010 ) , so we did not see the need to rerun these long ilp jobs on a faster solver .
instead , we compiled a copy of the cp algorithm of schoen , eendebak and nguyen ( 2010 ) on our system and used the resulting executable to directly address some fundamental research questions .
one question is whether oa@xmath0 enumerations could have been completed by using the cp - based mcs algorithm together with od - equivalence reductions . to answer this question we extended @xmath190 randomly chosen oa@xmath159 to oa@xmath191 using the mcs algorithm and algorithm [ alg : br3idd ] .
algorithm [ alg : br3idd ] was more than @xmath192 orders of magnitude faster .
so , algorithm [ alg : br3idd ] was essential in this enumeration .
another question is whether isomorphism pruning was essential in our ilp - based enumerations . to answer this question we turned off isomorphism pruning in algorithm [ alg : hybrid4 ] .
this resulted in a one order of magnitude slow down for the extensions to oa@xmath159 , and the ` nauty ` step of the extensions to oa@xmath161 failed because the @xmath193 gb of disk space mounted on the /tmp directory was filled .
one last question is whether we could have enumerated the new cases if we had not had the isomorphism pruning solver of margot ( 2007 ) . to answer this question we tested the mcs algorithm for cases in which we had to use isomorphism pruning .
the extension of all non - od - equivalent oa@xmath158 to all non - od - equivalent oa@xmath159 was @xmath194 times slower with the mcs algorithm .
we estimated the extension from oa@xmath160 to oa@xmath161 to be about @xmath195 times slower .
this estimate was based on @xmath196 randomly selected extensions .
hence , our computational experiments suggest that the use of margot ( 2007 ) solver becomes more crucial in enumerating oa@xmath52 as @xmath5 increases .
our results suggest algorithm [ alg : bf ] when @xmath197 , algorithm [ alg : hybrid4 ] when @xmath198 , and algorithm [ alg : br3idd ] when @xmath199 should be used with od - equivalence reductions to enumerate oa@xmath200 .
ilp - based algorithms with isomorphism pruning enumerated oa@xmath52 with @xmath185 and @xmath186 and established gma designs with the properties in table [ tab : enum ] .
a comparison of these algorithms to the cp algorithm with isomorphism rejection of schoen , eendebak and nguyen ( 2010 ) was made .
we could not solve the oa@xmath200 cases as the number of oa@xmath201 solutions that had to be reduced based on od - equivalence was not manageable .
future research will focus on decreasing the number of oa@xmath200 for @xmath202 that need to be enumerated for finding all gma oa@xmath200 .
the authors are indebted to professor francois margot for providing the ilp solvers and answering questions .
the authors thank professor brendan mckay for answering programming questions regarding ` nauty ` .
this research was supported by the afosr grants f1ata06334j001 , f1ata03039j001 and by an allocation of computing time from the ohio supercomputer center .
the views expressed in this article are those of the authors and do not reflect the official policy or position of the united states air force , department of defense , or the u.s . government .
each element of @xmath66 from algorithm [ alg : br3idd ] is an @xmath10 by lemma [ lem : y ] , and the set @xmath66 is reduced with ` nauty ` to contain no isomorphic designs
. let algorithm [ alg : br3idd ] be implemented with an oa(@xmath203 ) , @xmath164 .
first we show that the set @xmath204 contains at least one copy of each non - isomorphic extension of @xmath164 to an @xmath10 . for each @xmath205
, @xmath206 corresponds to an @xmath10 which is isomorphic to @xmath164 .
next , let @xmath207 denote this @xmath10 .
then each solution @xmath138 of the ilp solved in step [ item : extend ] corresponds to an extension of @xmath164 that is an @xmath10 .
let @xmath208 $ ] be this extension .
then @xmath209 also solves this ilp and corresponds to @xmath210 $ ] , so @xmath208 $ ] and @xmath210 $ ] are isomorphic .
the indicator variables for the ilp with @xmath127 variables for finding all @xmath10 are indexed with the @xmath127 full factorial design .
the action of @xmath136 on the full factorial design defines the action of @xmath136 on the indicator variables .
all elements in @xmath136 are viable to use in isomorphism pruning , except those that produce a different @xmath73 input design when @xmath136 s action is restricted to the first @xmath61 factors .
this is true since the elements that are not viable send a feasible solution ( i.e. , a feasible extension of the input design ) to an infeasible solution ( i.e. , a design whose first @xmath61 factors are not fixed to the input design ) .
the elements that are viable to use in isomorphism pruning send a run that appears @xmath211 times in the input design to a run that appears the same number of times .
these elements form the group @xmath149 .
let @xmath212 $ ] and @xmath213 $ ] .
let @xmath214 $ ] be an oa@xmath215 extension of @xmath4 .
first , by od - equivalence @xmath4 is either isomorphic to @xmath174 or to @xmath216\ ] ] for some @xmath217 . if @xmath4 is isomorphic to @xmath174 there is nothing to prove .
( isomorphic @xmath10 extend to the same maximum number of factors @xmath218 as oa@xmath219s . )
hence , assume @xmath4 is isomorphic to @xmath220 .
since @xmath220 and @xmath4 are isomorphic @xmath220 extends to an oa@xmath215 @xmath221.\ ] ] let @xmath222.\ ] ] then , each @xmath223 value of @xmath224 is equal to a @xmath223 or @xmath225 value of @xmath226 if @xmath22 is even and @xmath223 or @xmath227 value of @xmath226 if @xmath22 is odd .
this implies that @xmath223 values of @xmath224 are zero if @xmath228 .
hence , @xmath224 is the desired extension .
+ @xmath10 & oa classes & od classes & time & algorithm & @xmath232 & @xmath233 & @xmath234 & @xmath235 & @xmath236 + continued from previous page + @xmath10 & oa classes & od classes & time & algorithm & @xmath232 & @xmath233 & @xmath234 & @xmath235 & @xmath236 + + oa(160,5,2,4 ) & 6 & & 0.033 & [ alg : bf ] & 0.00 + oa(160,6,2,4 ) & 29 & & 0.5 & [ alg : bf ] & 0.00 & 0.04 + oa(160,7,2,4 ) & 450 & 106 & 13.5 & [ alg : bf ] & 0.00 & 0.28 & 0.00 + oa(160,8,2,4 ) & 99,618 & 11,712 & 1,522 & [ alg : hybrid4 ] & 0.20 & 0.52 & 0.04 & 0.00 + oa(160,9,2,4 ) & 15,083 & 1,608&6,309.7 & [ alg : br3idd ] & 0.56 & 1.36 & 0.24 & 0.04 & 0.00 + oa(160,10,2,4 ) & 0 & 0 & 4.13 & [ alg : br3idd ] + oa(176,5,2,4 ) & 6 & & 0.01 & [ alg : bf ] & 0.01 + oa(176,6,2,4 ) & 14 & & 0.09 & [ alg : bf ] & 0.05 & 0.01 + oa(176,7,2,4 ) & 945 & 179 & 15.3 & [ alg : bf ] & 0.17 & 0.06 & 0.01 + oa(176,8,2,4 ) & 1,157,443 & 129,138 & 31,113.3 & [ alg : hybrid4 ] & 0.46 & 0.23 & 0.07 & 0.01 + oa(176,9,2,4 ) & 26 & 4 & 378,241.4 & [ alg : br3idd ] & 1.04 & 0.69 & 0.30 & 0.07 & 0.00 + oa(176,10,2,4 ) & 0 & 0 & 0.01 & [ alg : br3idd ] + + @xmath10 & @xmath237 + continued from previous page + @xmath10 & @xmath238
+ + oa(160,5,2,4 ) & 5 , 25 , 50 , 50 , 25 , 5 + oa(160,6,2,4 ) & 2.600 , 14.400 , 39 , 48 , 39 , 14.400 , 2.600 + oa(160,7,2,4 ) & 1.600 , 7 , 29.400 , 42 , 42 , 29.400 , 7 , 1.600 + oa(160,8,2,4 ) & 1.100 , 3.300 , 18.900 , 36.700 , 40.500 , 35.900 , 18.700 , 4.100 , 0.800 + oa(160,9,2,4 ) & 1 , 0.900 , 11.400 , 30 , 37 , 37 , 28.600 , 11.600 , 2 , 0.500 + oa(176,5,2,4 ) & 5.545 , 27.273 , 55.455 , 54.545 , 27.727 , 5.455 + oa(176,6,2,4 ) & 2.909 , 15.818 , 42.273 , 54.545 , 40.909 , 16.909 , 2.636 + oa(176,7,2,4 ) & 1.705 , 8.432 , 30.068 , 48.523 , 46.932 , 29.114 , 10.023 , 1.205 + oa(176,8,2,4 ) & 1.216 , 3.909 , 20.045 , 40.091 , 46.932 , 37.545 , 20.045 , 5.727 , 0.489 + oa(176,9,2,4 ) & 1.068 , 1.330 , 12.272 , 31.500 , 42.955 , 41.523 , 28.636 , 13.500 , 3.068 , 0.148 +
|
statistical design of experiments is widely used in scientific and industrial investigations .
an optimum orthogonal array can extract as much information as possible at a fixed cost .
finding ( optimum ) orthogonal arrays is an open ( yet fundamental ) problem in design of experiments because constructing ( optimum ) orthogonal arrays or proving non - existence becomes intractable as the number of runs and factors increase .
enumerating orthogonal arrays is equivalent to finding all feasible solutions to a class of inherently symmetric constraint satisfaction problems .
the inherent symmetry implies that the solutions remain feasible under the action of a group permuting the variables .
this causes the number of feasible solutions to be huge .
we develop algorithms for enumerating orthogonal arrays that call the margot ( 2007 ) isomorphism pruning algorithm and bring the enumeration of oa@xmath0 for @xmath1 and oa@xmath2 for @xmath3 within computational reach . a catalog of the newly found optimum orthogonal arrays obtained using our algorithms is also presented .
constraint programming ; extension algorithm ; integer linear programming ; isomorphism rejection ; ` nauty ` ; orthogonal designs .
|
since the pioneering papers by watts and strogatz on small - world networks @xcite and barabsi and albert on scale - free networks @xcite , complex networks have received considerable attention as an interdisciplinary subject @xcite .
complex networks describe many systems in nature and society , such as internet @xcite , world wide web @xcite , metabolic networks @xcite , protein networks in the cell @xcite , co - author networks @xcite and sexual networks @xcite , most of which share three apparent features : power - law degree distribution , small average path length ( apl ) and high clustering coefficient . in recent years , many evolving models
@xcite have been proposed to describe real - life networks .
the original ba model @xcite captures the two main mechanisms responsible for the power - law degree distribution of degrees , namely growth and preferential attachment .
dorogovtsev , mendes , and samukhin @xcite gave an exact solution for a class of growing network models thanks to the use of a `` master - equation '' .
krapivsky , redner , and leyvraz @xcite examined the effect of a nonlinear preferential attachment on network dynamics and topology .
amaral et al .
@xcite studied models that incorporate aging and cost and capacity constraints in order to explain the deviations from the power - law behavior in several real - life networks .
dorogovtsev and mendes @xcite also addressed the evolution of networks with aging of sites .
bianconi and barabsi @xcite offered a model addressing the competitive aspect in many real networks such as world wide web .
additionally , in real systems a series of microscopic events shape the network evolution , including the addition or rewiring of new edges or the removal of vertices or edges .
albert and barabsi @xcite discussed a model that incorporates new edges between existing vertices and the rewiring of edges .
dorogovtsev and mendes @xcite considered a class of undirected models in which new edges are added between old vertices and existing edges can be removed .
although it is now established that preferential attachment can explain the power - law characteristic of networks , there is a wide range of microscopic alternative mechanisms that could affect the evolution of growing networks and still lead to the observed scale - free topologies .
kleinberg et al .
@xcite and kumar et al .
@xcite proposed copying mechanisms motivated by the desire to explain the power - law degree distribution of the world wide web .
chung et al .
@xcite introduced also a duplication model for biological networks .
krapivsky and render @xcite presented edge redirection mechanisms which are mathematically equivalent to the model of kumar et al @xcite . inspired by citation networks vzquez proposed in @xcite the walking mechanism . actually scale - free networks can be created by various methods .
comellas , fertin and raspaud @xcite introduced a category of graphs via a recursive construction .
specific recursive and general scale - free constructions are given also in @xcite and scale - free trees ( without clustering ) in @xcite .
zhou et al . construct in @xcite an integer network using a pure mathematical method . in relation to the problem of apollonian packing , a two - dimensional example of which
is shown in fig .
1 , andrade et al . introduced apollonian networks @xcite which were also proposed by doye and massen in @xcite .
it should be pointed out that high - dimensional apollonian networks were already introduced in @xcite , but in this work the emphasis is placed on two - dimensional apollonian networks and to provide a model to help understanding the energy landscape networks .
zhang et al . in @xcite
offer a simple general algorithm producing high - dimensional apollonian networks and derive in detail analytical expressions for their order , size , degree distribution , clustering coefficient and diameter .
inspired also by the apollonian packing , zhou et al .
@xcite proposed a simple rule that generates random two - dimensional apollonian networks with very large clustering coefficient and very small apl .
in this paper , motivated by high - dimensional apollonian packings , we present a simple iterative algorithm for high - dimensional random apollonian networks ( hdran ) which extends the idea introduced in @xcite for dimension two to any dimension .
the algorithm can concretize the problems of abstract high - dimensional random apollonian packings . using the algorithm
we determine relevant characteristics of hdran : scale - free degree distribution and large clustering coefficient which depend on the dimension of the apollonian packing .
based on the algorithm , we estimate also that the average path length of the networks increases logarithmically with the number of vertices .
therefore , these networks ( hdran ) have a small - world scale - free topology .
the name of apollonian packings goes back to apollonius of perga ( c. 262 bc c. 190 bc ) who studied the problem of finding tangent circles to any three given circles .
when the initial three circles are mutually tangent , there exist only two of such tangent circles , known as apollonian circles or soddy inner and outer circles .
frederick soddy @xcite , nobel prize for chemistry in 1921 , published in nature a poem describing the equation which relates the radii of the circles and is now known as the `` kissing circles theorem '' .
this result , however , had been previously given by ren descartes in 1643 and rediscovered in 1842 by the mathematician phillip beecroftan .
a generalization to @xmath0-dimensional hyperspheres , also as a poem , was published as well in nature by thorold gosset @xcite and reads : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ and let us not confine our cares + to simple circles , planes and spheres , + but rise to hyper flats and bends + where kissing multiple appears , + in @xmath0-ic space the kissing pairs + are hyperspheres , and truth declares - + as @xmath1 such osculate + each with an @xmath2 fold mate + the square of the sum of all the bends + is @xmath0 times the sum of their squares . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ in the procedure known as _ apollonian packing _ or _
apollonian gasket _ , we start with three mutually tangent circles , and draw their inner soddy circles .
then we draw the inner soddy circles of this circle with each pair of the original three , and the process is iterated , see fig.1 .
this apollonian packing can be used to design a network , where each circle is associated to a vertex of the network and vertices are connected if the corresponding circles are tangent .
we call this network a two - dimensional apollonian network @xcite . in a similar way , it is possible to construct higher dimensional ( @xmath3-dimensional @xmath4 ) apollonian networks associated with other self - similar packings .
we start with @xmath5 mutually tangent @xmath3-dimensional hyperspheres ( @xmath3-hyperspheres ) in a @xmath3-dimensional simplex which is enclosed and tangent to a larger hypersphere , the interstices of which are curvilinear @xmath3-dimensional simplices ( @xmath3-simplices ) . in the first iteration , and inside each of the interstices , we add @xmath3-hyperspheres tangent to each of the @xmath5 @xmath3-hyperspheres bounding the curvilinear @xmath3-simplex .
the added hyperspheres will not fill the interstice and each produces @xmath5 smaller interstices .
then , in the second iteration , @xmath5 @xmath3-hyperspheres are added inside all of the @xmath5 new interstices , being again tangent to the enclosing @xmath3-hyperspheres . the process is repeated iteratively obtaining the high - dimensional apollonian packing .
if each @xmath3-hypersphere corresponds to a vertex and vertices are connected by an edge if the corresponding @xmath3-hyperspheres are tangent , then we obtain a @xmath3-dimensional apollonian network . in the above @xmath3-dimensional apollonian packings , if we add only one @xmath3-hypersphere inside a randomly selected interstice in every iteration , after a certain number of iterations we have a high - dimensional random apollonian packing .
analogously , if each @xmath3-hypersphere corresponds to a vertex and vertices are connected by an edge if the corresponding @xmath3-hyperspheres are tangent , then one obtains @xmath3-dimensional random apollonian networks . in @xcite
it is discussed the special case @xmath6 . here
we will consider the general case .
it should be noted that we consider the initial bounding @xmath3-hypersphere at the same level as the other @xmath5 initial @xmath3-hyperspheres .
in the iterative process for the construction of hdran , at each step and for each new hypersphere added , @xmath5 new interstices are created in the packing , that will be filled in one of the following iterations .
equivalently , for each new vertex added , @xmath5 new @xmath3-simplices are created in the network , into which vertices will be inserted in one of the following iterations . according to this process
, we introduce a general iterative algorithm for hdran .
we denote the @xmath3-dimensional random apollonian network after @xmath7 iterations by @xmath8 , @xmath4 . before introducing the algorithm
we give the following definition . a complete graph @xmath9 ( also referred in the literature as @xmath3-clique ; see @xcite ) is a graph without loops whose @xmath3 vertices are pairwise adjacent .
then the @xmath3-dimensional random apollonian network @xmath8 is constructed as follows .
initially ( @xmath10 ) , @xmath11 is the complete graph @xmath12 or @xmath13-clique . at each step , we choose an existing subgraph isomorphic to a @xmath14-clique _ that has never been selected before _ , then we add a new vertex and join it to all the vertices of the selected subgraph .
the growing process is repeated until the network reaches the desired size .
note that when @xmath6 , our model gives the maximal planar network of @xcite .
furthermore , since the network size is incremented by one with each step , in this paper , we use the step value @xmath7 to represent a vertex created at this step .
we can see easily that at step @xmath7 , the network consists of @xmath15 vertices .
the total degree equals @xmath16 .
so , when @xmath7 is large the average vertex degree at step @xmath7 is equal approximately to a constant value @xmath17 , which shows that the network @xmath8 is sparse like many real - life networks @xcite .
in this section we show that the dimension @xmath3 is a tunable parameter which controls the relevant characteristics of a high dimensional random apollonian network @xmath8 .
the degree distribution is one of the most important statistical characteristics of a network .
we use the mean - field method @xcite to predict the growth dynamics of the individual vertices , and calculate analytically the connectivity distribution @xmath18 and the scaling exponents . note that , after a new vertex is added , the number of @xmath14-cliques that can be chosen in the following step increases by @xmath3 .
therefore , we can immediately see that after @xmath7 steps , the number of @xmath14-cliques ready for selection are @xmath19 . given a vertex , when its degree increases by one , the number of @xmath14-cliques that contain the vertex increases by @xmath20 , so the number of @xmath14-cliques for selection containing the vertex @xmath21 with degree @xmath22 is @xmath23 .
if we consider @xmath24 continuous , we can write for a vertex @xmath21 @xmath25 the solution of this equation , with the initial condition that vertex @xmath21 was added to the network at @xmath26 with connectivity @xmath27 , is @xmath28 the probability that a vertex has a connectivity @xmath29 smaller than @xmath24 , @xmath30 , is @xmath31 assuming that we add the vertices to the network at equal intervals , the probability density of @xmath26 is @xmath32 substituting this into eq .
( 3 ) we obtain that @xmath33 thus the degree distribution is @xmath34 for large @xmath7 @xmath35 and if @xmath36 then @xmath37 giving a degree exponent @xmath38 .
this value is different from the degree exponent for a high dimensional apollonian network ( non random hdan ) which is @xmath39 @xcite .
this discrepancy would need further study , but a possible explanation is that for two networks with the same number of vertices , hdran and hdan , the distribution of degrees differs because of the different growing process .
the method used to obtain @xmath18 might have also some influence in the value of @xmath40 as it is explained at the end of the next subsection .
when @xmath6 , one has @xmath41
@xcite , while as @xmath3 goes to infinity @xmath42 .
thus by tuning the parameter @xmath3 , it is possible to obtain a variety of scale - free networks with different exponents in the range , @xmath43 . in fig .
2 , the degree distributions @xmath18 at various values of the dimension @xmath3 are displayed and we find that at any value of @xmath3 , the simulated degree distribution is in good agreement with the analytic value . at various values of the dimension @xmath3 .
the squares ( a ) , stars ( b ) , circles ( c ) , and triangles ( d ) denote the simulation results for @xmath6 , @xmath44 , @xmath45 , and @xmath46 , respectively .
the reference dotted lines come from the evaluation of @xmath47 as given by equation ( 7).,width=453 ] also , from eqs .
( 2 ) and ( 7 ) , we can notice that when @xmath7 becomes large , the maximal degree of a vertex is roughly @xmath48 .
the clustering coefficient of a vertex is the ratio of the total number of existing connections between all its @xmath24 nearest neighbors and @xmath49 , the number of all possible connections between them .
the clustering of the graph is obtained averaging over all its vertices .
we can derive analytical expressions for the clustering @xmath50 for any vertex with degree @xmath24 .
when a vertex is created it is connected to all the vertices of a @xmath14-clique whose vertices are completely interconnected .
it follows that a vertex with degree @xmath51 has a clustering coefficient of one because all the @xmath52 possible links between its neighbors actually exist . in the following steps , a newly - created vertex links the considered vertex whose @xmath3 nearest neighbors
are also connected by the new one at the same iteration step .
thus there is a one - to - one correspondence between the degree of a vertex and its clustering . for a vertex @xmath53 of degree @xmath24 ,
the exact expression for its clustering coefficient is @xmath54 which depends on the degree @xmath24 and the dimension @xmath3 .
this expression indicates that the local clustering scales as @xmath55 for large @xmath24 .
note that , as the clustering only depends on @xmath24 for a given dimension , it matches the results found in @xcite for hdan .
it is interesting to notice that a similar scaling has been observed in several real - life networks @xcite .
the clustering coefficient @xmath56 of hdran can be obtained as the mean value of @xmath50 with respect to the degree distribution @xmath18 expressed by eq .
the result is @xmath57 for @xmath6 , equation ( 9 ) gives @xmath58 , which is the same result obtained in @xcite .
for @xmath44 we obtain @xmath59 . for @xmath45
the analytical value is @xmath60 and for @xmath46 it is @xmath61 .
we note that they do not depend on the order of the network .
note also that for @xmath6 the value obtained is not far from the value of the clustering in the coauthorship network in neuroscience ( @xmath62 ) @xcite .
thus the clustering @xmath56 of hdran is high and increases with the dimension @xmath3 and approaches a limit of @xmath63 when @xmath3 is large . in fig .
3 , we present simulation results for @xmath56 vs the dimension @xmath3 for various values of the @xmath64 , the network order . for @xmath65 , the simulation results provide values of @xmath56 of @xmath66 , @xmath67 , @xmath68 and @xmath69 , respectively . we see that the small difference between analytical and simulation values decreases with @xmath3 . this difference might be explained because the continuum approach does not properly account all the vertices , affecting @xmath18 which is used in eq .
( 9 ) to obtain @xmath56 , see @xcite . of hdran at various values of dimension @xmath3 .
the squares , stars , circles and triangles denote @xmath6 , @xmath44 , @xmath45 and @xmath46 , respectively .
the dashed lines are fits to the simulated data.,width=302 ] the average path length of a network ( apl ) is defined as the number of edges in the shortest path between two vertices averaged over all pairs of vertices .
below , using the same approach than in @xcite we will discuss the apl of a high dimensional random apollonian network .
first we note that it is not very difficult to prove that for a hdran and for any two arbitrary vertices @xmath21 and @xmath70 each shortest path from @xmath21 to @xmath70 does not pass through any vertex @xmath24 satisfying that @xmath71 .
if @xmath72 denotes the distance between @xmath21 and @xmath70 , we introduce the total distance of a hdran with order @xmath64 as @xmath73 @xmath74 and we denote the apl by @xmath75 , defined as : @xmath76 according to the former remark , the addition of new vertices will not affect the distance between those already existing and we have : @xmath77 assume that the vertex @xmath78 is added joining the @xmath14-clique @xmath79 composed of vertices @xmath80 then eq . ( 12 ) can be rewritten as : @xmath81 where @xmath82 . constricting the clique @xmath79 continuously into a single vertex @xmath83 ( here we assume that @xmath84 )
, we have @xmath85 . since @xmath86 , eq .
( 13 ) can be rewritten as : @xmath87 where @xmath88 is a vertex set with cardinality @xmath89 .
the sum @xmath90 can be considered as the total distance from one vertex @xmath83 to all the other vertices in our model with order @xmath91 and it can be roughly approximated in terms of @xmath92 : @xmath93 note that , as @xmath75 increases monotonously with @xmath64 , it is clear that : @xmath94 from equations ( 14 ) , ( 15 ) and ( 16 ) , we can obtain the expression : @xmath95 considering eq .
( 17 ) as an equation and not an inequality , we can provide an upper bound for the variation of @xmath73 as @xmath96 which gives @xmath97 where @xmath98 is a constant . as @xmath99
, we have @xmath100 . note that as we have deduced equation ( 19 ) from an inequality , then @xmath75 increases at most as @xmath101 with @xmath64 . in fig .
4 , we report the simulation results on the average path length @xmath102 vs the dimension @xmath3 for various values of the network order @xmath64 , which agree well with the analytic results .
note that the apl decreases with the dimension @xmath3 as the networks become more dense .
the dependence of the apl on @xmath3 , comes from the method used to obtain eq .
( 14 ) as the sum there is on a set @xmath103 with cardinality @xmath89 . of a hdran vs its network size @xmath64 for different values of the dimension @xmath3 .
the squares , stars , circles and triangles denote the simulation results for @xmath6 , @xmath44 , @xmath45 and @xmath46 , respectively .
the lines are fits to the simulated data .
one can see that @xmath102 grows at most as @xmath101.,width=302 ]
in summary , we have introduced an iterative algorithm for the construction of high - dimensional random apollonian networks associated with high dimensional random packings .
the networks have the typical characteristics of many technological and social networks and are small - world with power - law degree distributions , thus they can model a variety of scale - free networks of the real - life world .
we have computed the analytical expressions for the degree distribution and clustering coefficient and we show that they depend on the dimension . also , the average path length of the networks grows logarithmically with the order of the hdran . moreover , hdran are a generalization of the maximal planar network introduced in @xcite
. future work should include studying in detail processes such as percolation , spreading , searching and diffusion taking place on hdrans .
this research was supported by the national natural science foundation of china under grant no .
support for f.c . was provided by the secretaria de estado de universidades e investigacin ( ministerio de educacin y ciencia ) , spain , and the european regional development fund ( erdf ) under project tic2002 - 00155 .
the authors are grateful to the anonymous referees for their valuable comments and suggestions .
j. m.kleinberg , r. kumar , p. raghavan , s. rajagopalan , a. tomkins , the web as a graph : measurements , models and methods , proceedings of the 5th annual international conference , cocoon99 , tokyo , july 1999 ( springer - verlag , berlin ) , ( 1999 ) 1 .
r. kumar , p. raghavan , s. rajalopagan , d. sivakumar , a. s. tomkins , e. upfal , stochastic models for the web graph , proceedings of the 41st ieee symposium on foundations of computer science ( ieee computing society , los alamitos , calif . )
( 2000 ) 5765 .
j. s. andrade jr . , h. j. herrmann , r. f. s. andrade l. r. da silva , apollonian networks : simultaneously scale - free , small world , euclidean , space filling , and with matching graphs , phys .
94 ( 2005 ) 018702 .
west , _ introduction to graph theory_. ( prentice - hall , upper saddle river , nj , 2001 ) .
barabsi , r. albert , h. jeong , mean - field theory for scale - free random networks , physica a 272 ( 199 ) 173187 .
|
we propose a simple algorithm which produces a new category of networks , high dimensional random apollonian networks , with small - world and scale - free characteristics .
we derive analytical expressions for their degree distributions and clustering coefficients which are determined by the dimension of the network .
the values obtained for these parameters are in good agreement with simulation results and comparable to those coming from real networks .
we estimate also analytically that the average path length of the networks increases at most logarithmically with the number of vertices .
complex networks , scale - free networks , small - world networks , disordered systems , networks 89.75.da,89.75.fb,89.75.hc
|
in @xcite we introduced the cosmicfish code as a powerful tool to perform forecast on many different models with future cosmological experiments . + in this set of notes
we show the validation pipeline that was used for the code .
we compared the results obtained with the cosmicfish code to other results in literature .
we find an overall good level of agreement . + together with these notes we release a cosmicfish package that contains the relevant code to produce all the results presented here .
this package is going to be updated as new validation results become available .
the cosmicfish code and its validation package are publicly available at http://cosmicfish.github.io .
the validation package contains forecasted results for the _ planck _ mission , obtained using _ planck _
blue book specifications @xcite , and for the proposed cmbpol satellite @xcite .
+ cosmicfish is used varying the standard 6 parameters ( @xmath0 ) , from now on dubbed as s6 , and the number of relativistic species @xmath1 .
+ the results are compared with what is obtained in section iii.c of @xcite ; even though these results were obtained with mcmc methods , cosmicfish bounds and degeneracies show a very good agreement ( see figure [ fig : cmbfor ] and table [ tab : cmbfor ] ) .
+ we are grateful to ana achcarro , carlo baccigalupi , erminia calabrese , stefano camera , luigi danese , giulio fabbian , noemi frusciante , bin hu , valeria pettorino , levon pogosian , giuseppe puglisi and alessandra silvestri for useful and helpful discussions on the subject .
we are indebted to luca heltai for help with numerical algorithms .
mm is supported by the foundation for fundamental research on matter ( fom ) and the netherlands organization for scientific research / ministry of science and education ( nwo / ocw ) .
mm was also supported by the dfg transregio trr33 grant on the dark universe during the preparation of this work .
mr acknowledges partial support by the italian space agency through the asi contracts euclid - ic ( i/031/10/0 ) and the infn - indark initiative .
mr acknowledges the joint sissa / ictp master in high performance computing for support during the development of this work .
mr thanks the national astronomical observatories , chinese academy of science for the hospitality during the initial phases of development of this work .
mr and mm thank the galileo galilei institute for theoretical physics for the hospitality and the infn for partial support during the completion of this work .
gbz and yw are supported by the strategic priority research program `` the emergence of cosmological structures '' of the chinese academy of sciences grant no .
xdb09000000 , and by university of portsmouth .
yw is supported by the nsfc grant no .
11403034 .
m. raveri , m. martinelli , g. zhao and y. wang , arxiv:1606.06273 [ astro-ph.co ] . m. raveri , m. martinelli , g. zhao and y. wang , arxiv:1606.06268 [ astro-ph.co ] . planck - the scientific programme http://sci.esa.int/jump.cfm?oid=47334 s. galli , m. martinelli , a. melchiorri , l. pagano , b. d. sherwin and d. n. spergel , phys . rev .
d * 82 * ( 2010 ) 123504 doi:10.1103/physrevd.82.123504 [ arxiv:1005.3808 [ astro-ph.co ] ] .
r. adam _ et al . _
[ planck collaboration ] , arxiv:1502.01582 [ astro-ph.co ] .
p. a. r. ade _ et al .
_ [ planck collaboration ] , arxiv:1502.01589 [ astro-ph.co ] .
m. martinelli , s. pandolfi , c. j. a. p. martins and p. e. vielzeuf , phys .
d * 86 * ( 2012 ) 123001 doi:10.1103/physrevd.86.123001 [ arxiv:1210.7166 [ astro-ph.co ] ] .
a. conley _ et al .
_ [ snls collaboration ] , astrophys . j. suppl .
* 192 * ( 2011 ) 1 doi:10.1088/0067 - 0049/192/1/1 [ arxiv:1104.1443 [ astro-ph.co ] ] .
a. albrecht _ et al .
_ , astro - ph/0609591 .
a. zablocki , arxiv:1411.7387 [ astro-ph.co ] .
|
these notes show and comment the examples that have been used to validate the cosmicfish code .
we compare the results obtained with the code to several other results available in literature finding an overall good level of agreement .
we will update this set of notes when relevant modifications to the cosmicfish code will be released or other validation examples are worked out . +
the cosmicfish code and the package to produce all the validation results presented here are publicly available at http://cosmicfish.github.io .
+ the present version is based on cosmicfish jun16 .
|
in many applications , initial value problems of ordinary differential equations are given as _
additively _ partitioned systems @xmath0 where the right - hand side @xmath1 is split into @xmath2 different parts with respect to , for example , stiffness , nonlinearity , dynamical behavior , and evaluation cost .
additive partitioning also includes the special case of _ component _ partitioning where the solution vector is split into @xmath2 disjoint sets , @xmath3 , with the @xmath4-th set containing the components @xmath5 with indices @xmath6 .
one defines a corresponding partitioning of the right hand side @xmath7 where @xmath8 is the @xmath9-th column of the identity matrix , and superscripts ( without parentheses ) represent vector components .
a particular case is _ coordinate _ partitioning where @xmath10 and @xmath11 , @xmath12 : @xmath13 the development of runge - kutta ( rk ) methods that are tailored to the partitioned system started with the early work of rice @xcite .
hairer @xcite developed the concept of p - trees and laid the foundation for the modern order conditions theory for partitioned runge - kutta methods .
the investigation of practical partitioned runge - kutta methods with implicit and explicit components @xcite was revitalized by the work of ascher , ruuth , and spiteri @xcite .
additive runge - kutta methods have been investigated by cooper and sayfy @xcite and kennedy and carpenter @xcite , and partitioning strategies have been discussed by weiner @xcite .
this work focuses on specialized schemes which exploit the different dynamics in the right hand sides @xmath14 ( e.g. , stiff and non - stiff ) , allow for arbitrarily high orders of accuracy , and posses good stability properties with respect to dissipative systems .
the approach taken herein generalizes the additively partitioned runge - kutta family of methods @xcite by allowing for different stage values as arguments of different components of the right hand side .
the paper is organized as follows .
section [ sec : gark ] introduces the new family of generalized additively partitioned runge - kutta schemes .
generalized implicit - explicit runge - kutta schemes are discussed in section [ sec : imex ] .
a stability and monotonicity analysis is performed in section [ sec : stability ] .
section [ sec : imim ] builds implicit - implicit generalized additively partitioned runge - kutta methods .
conclusions are drawn in section [ sec : conclusions ] .
in this section we extend additive runge - kutta to generalized additively partitioned runge - kutta schemes . the order conditions for this new class
are derived using the n - tree theory of sanz - serna @xcite .
kennedy and carpenter @xcite developed additive runge - kutta ( ark ) methods for systems with partitioned right hand sides .
one step of an ark scheme reads [ eqn : ark ] @xmath15 the corresponding butcher tableau is @xmath16 the _ generalized additively partitioned runge - kutta _ ( gark )
family of methods extends the traditional approach by allowing for different stage values with different components of the right hand side .
one step of a gark scheme applied to solve reads : [ eqn : gark ] @xmath17 the corresponding generalized butcher tableau is @xmath18 in contrast to traditional additive methods @xcite different stage values are used with different components of the right hand side .
the methods @xmath19 can be regarded as stand - alone integration schemes applied to each individual component @xmath20 .
the off - diagonal matrices @xmath21 , @xmath22 , can be viewed as a coupling mechanism among components .
for the special case of component partitioning a gark step reads [ eqn : gark - componentwise ] @xmath23 \begin{bmatrix } y_{n+1}^{\{1\ } } \\ \vdots \\ y_{n+1}^{\{n\ } } \end{bmatrix } & = & \begin{bmatrix } y_{n}^{\{1\ } } \\ \vdots \\ y_{n}^{\{n\ } } \end{bmatrix } + h \begin{bmatrix } \sum_{i=1}^{s^{\{1\ } } } b_{i}^{\{1\ } } f^{\{1\}}\left(y_i^{\{1\}}\right ) \\ \vdots \\
\sum_{i=1}^{s^{\{n\ } } } b_{i}^{\{n\ } } f^{\{n\}}\left(y_i^{\{n\}}\right ) \end{bmatrix}\,.\end{aligned}\ ] ] a gark scheme is called internally consistent if @xmath24 in order to understand internal consistency consider the system @xmath25 with the exact solution @xmath26^t$ ] .
the gark stages are @xmath27 the internal consistency condition ensures that all components of the stage vectors are calculated at the same internal approximation times .
we derive the gark order conditions by applying araujo , murua , and sanz - serna s n - tree theory @xcite to , while taking into account the fact that the internal stages @xmath28 and the stage numbers @xmath29 depend on the partition @xmath14 .
n - trees @xcite are a generalization of p - trees from the case of component partitioning to the general case of right - hand side partitioning .
the set nt of n - trees consists of all butcher trees with colored vertices ; each vertex is assigned one of @xmath2 different colors corresponding to the @xmath2 components of the partition .
similar to regular butcher trees each vertex is also assigned a label .
the order @xmath30 is the number of nodes of @xmath31 .
the empty n - tree is denoted by @xmath32 .
the n - tree with a single vertex of color @xmath33 is denoted by @xmath34}$ ] .
the n - tree @xmath31 with @xmath35 and a root of color @xmath33 can be represented as @xmath36^{[\nu]}$ ] , where @xmath37 are the non - empty subtrees ( n - trees ) arising from removing the root of @xmath38 .
similar to regular butcher trees one denotes by @xmath39 the number of symmetries of @xmath31 , and by @xmath40 the density defined recursively by @xmath41}\right ) & = & 1 , \quad \nu=1,\ldots , n , \\
\gamma(u ) & = & \rho(u ) \gamma(u_1 ) \cdots
\gamma(u_m ) \text { for } u=[u_1,\ldots , u_m]^{[\nu]}.\end{aligned}\ ] ] an elementary differential @xmath42 is associated to each n - tree @xmath31 .
the elementary differentials are defined recursively for each component @xmath43 as follows : @xmath44}\right)(y ) & = & f^{[\nu]i}(y ) , \quad \nu=1,\dots , n\ , , \\
f^i(u)(y ) & = & \sum_{i_1,\ldots , i_m=1}^{d } \frac{\partial^m\ , f^{[\nu]i}}{\partial y^{i_1 } \cdots \partial y^{i_m}}(y ) f^{i_1}(u_1)(y ) \cdots f^{i_m}(u_m)(y ) \\ & & \quad \textnormal{for } ~~u=[u_1,\ldots , u_m]^{[\nu]}\,.\end{aligned}\ ] ] an nb - series is a formal power expansion @xmath45 where @xmath46 is a mapping that assigns a real number to each n - tree . for example
, the exact solution of can be written as the following nb - series @xcite @xmath47 the order conditions for a gark method are obtained from the order conditions of ordinary runge - kutta methods .
the usual labeling of the runge - kutta coefficients ( subscripts @xmath48 ) is accompanied by a corresponding labeling of the different partitions for the n - tree ( superscripts @xmath49 ) .
let @xmath50 be a vector of ones of dimension @xmath51 , and @xmath52 .
the specific conditions for orders one to four are as follows .
[ eqn : gark - order - conditions-1-to-4 ] @xmath53 here , and throughout this paper , the matrix and vector multiplication is denoted by dot ( e.g. , @xmath54 is a dot product ) , while two adjacent vectors denote component - wise multiplication ( e.g. , @xmath55 is a vector of element - wise products ) .
these order conditions are given in element - wise form in appendix [ sec : oder - conditions - detailed ] .
we derive the nb - series of the solution of the gark scheme following the approach in @xcite for the nb - series of ark schemes .
the gark solution , the stage vectors , and the stage function values are expanded in nb - series @xmath56 from we have that the method gark is of order @xmath57 iff @xmath58 the coefficients @xmath59 , @xmath60 and @xmath61 are related through the numerical equations . for @xmath62 relation yields [ eqn : nbtrees ] @xmath63 and
relation gives @xmath64 from the properties of a derivative of a b - series on has @xmath65^{\{q\}}\ , , \end{array } \right .
\\ \nonumber & & i=1,\ldots , s^{\{m\ } } , \quad m , q=1,\ldots , n\,.\end{aligned}\ ] ] using equations
recursively , we have that for @xmath66^{\{q\}}$ ] @xmath67 equation becomes @xmath68^{\{q\}}~~ \textnormal{and}~~q \ne m\ , , \\ 1 \quad \textnormal{for}~~ u = \tau^{\{m\}}\ , , \\
\sum_{j_1,\ldots , j_\ell=1}^{s^{\{m\ } } } a_{i , j_1}^{\{m,\nu_1\ } } \cdots a_{i , j_\ell}^{\{m,\nu_\ell\ } } \mathbf{g}_{j_1}^{\{\nu_1\}}(u_1 ) \cdots \mathbf{g}_{j_\ell}^{\{\nu_\ell\}}(u_\ell ) \\
\qquad \textnormal{for}~~ u = [ u_1,\ldots , u_\ell]^{\{m\}}~~ \textnormal{and}~~\textnormal{color(root}(u_j ) ) = \nu_j\ , , \end{array } \right . \\ \nonumber & & i=1,\ldots , s^{\{m\ } } , \quad m=1,\ldots , n\,.\end{aligned}\ ] ] from we have @xmath69^{\{m\}}\ , .
\ ] ] for a regular runge - kutta method we ignore the coloring of the tree nodes and consider @xmath70 , where @xmath71 are the regular butcher trees .
the nb - series expansions are regular b - series expansions @xmath72 whose coefficients are given by @xmath73^{\{\bullet\}}\ , , \end{array } \right . \\
\nonumber & & \quad i=1,\ldots , s\ , , \\
\label{eqn : btrees - solution - expanded } \mathbf{c}(u ) & = & \sum_{i=1}^{s } b_i \mathbf{f}_i(u).\end{aligned}\ ] ] the rk method has order @xmath57 iff @xmath74 for any n - tree @xmath31 the recurrences and mimic the b - series relations and , respectively . each coefficient subscript in , is paired with a unique color superscript in and , for example @xmath75 , @xmath76 , and @xmath77 .
the nb - series coefficient @xmath78 written in terms of the gark scheme coefficients @xmath79 , @xmath80 has precisely the same form as the b - series coefficient @xmath81 written in terms of the rk scheme coefficients @xmath82 , @xmath83 , with color superscripts added to match the corresponding subscripts .
( the calculation of @xmath81 ignores the vertex coloring of @xmath38 . )
since has to hold for any coloring of the vertices of @xmath31 , the corresponding rk condition has to hold for arbitrary sets of superscripts , provided that they are paired correctly with the coefficient subscripts . for n - trees @xmath31 with all vertices of the same color @xmath20 one obtains the traditional rk order conditions for the individual method @xmath84 .
as an example consider the order conditions for equal superscripts @xmath85 .
therefore a necessary condition for a gark method to have order @xmath57 is that each individual component method has order @xmath57 . in addition , the gark method needs to satisfy the coupling conditions resulting from trees with vertices of different colors . for internally consistent gark methods
the order conditions simplify considerably .
conditions , , and become @xmath86 and correspond to order two conditions , and to the first order three condition of each individual component method ( @xmath87 , @xmath88 ) , @xmath89 . condition becomes @xmath90 and gives raise to @xmath2 order three conditions for each component method , and @xmath91 coupling conditions . compare this with the @xmath92 coupling conditions in the absence of internal consistency .
the @xmath93 order four conditions reduce to the following @xmath94 equations @xmath95
we now focus on systems with a two - way partitioned right hand side @xmath96 where @xmath97 is non - stiff and @xmath98 is stiff .
an implicit - explicit ( imex ) gark scheme is a two way partitioned method where one component method @xmath99 is explicit , and the other one @xmath100 implicit : [ eqn : imexrk ] @xmath101 the corresponding generalized butcher tableau is @xmath102 with @xmath103 for all @xmath104 , where @xmath105^t \in { { \mathbbm r}}^{s^{\{\nu\ } } } $ ] . in
we use @xmath106 with @xmath107 lower triangular and @xmath108 strictly lower triangular .
we obtain @xmath109 , @xmath110 , and recover the classical imex rk of ascher , ruuth , and spiteri @xcite : [ eqn : classical - imexrk ] @xmath111 in we use @xmath112 with @xmath107 lower triangular and @xmath108 strictly lower triangular , to obtain @xmath109 , and the interesting family of schemes : [ eqn : transposed - imexrk ] @xmath113 the low storage imex methods recently presented by higueras and roldan @xcite can also be formulated as imex - gark schemes .
the gark order conditions in matrix form are given in .
each of the implicit and explicit methods @xmath114 and @xmath115 has to satisfy the corresponding order conditions for @xmath116 . in addition , two coupling conditions are required for second order , and 12 coupling conditions are required for third order accuracy .
these conditions are listed in appendix [ sec : oder - conditions - imex ] .
the gark internal consistency condition reads @xmath117 these conditions are automatically satisfied in the case of classical - transposed imex rk methods . in case of classical imex
rk equations are equivalent to @xmath118 . assuming that both the implicit and the explicit methods have order at least three , implies that the the second order coupling conditions are automatically satisfied .
the third order coupling conditions reduce to : @xmath119 assume that the implicit and explicit methods have order at least four . with the order conditions
reduce to the following order four coupling relations : @xmath120 these order conditions are listed explicitly in appendix [ sec : oder - conditions - imex-4 ] .
additional simplifying assumptions @xmath121 further reduce the number of order conditions ; these are are listed in appendix [ sec : oder - conditions - imex-4 ] .
consider now the classical - transposed methods and assume that both the explicit and the implicit method have order three .
since holds , the remaining third order coupling conditions read : @xmath122 assuming that the explicit and implicit methods have each order at least four , the order four coupling conditions are : @xmath123 if , in addition , @xmath124 and @xmath125 , then the third order coupling conditions are automatically satisfied .
the remaining order four coupling conditions read : @xmath126 [ [ example - an - order - three - imex - method ] ] example : an order three imex method + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the implicit part is kvaerno s four stages , order three method ( * ? ? ?
* esdirk 3/2 ) : @xmath127 @xmath128 @xmath129\ , , \quad \mathbf{c}^ { { \ { \textsc{i}\ } } } = \begin{bmatrix } 0\\ 0.871733043016918\\ 1\\ 1 \end{bmatrix}\,.\ ] ] the explicit method has @xmath130 , @xmath131 , and @xmath132 [ [ example - an - order - four - imex - method ] ] example : an order four imex method + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the implicit part is kvaerno s five stages , order four method ( * ? ? ?
* esdirk 4/3 ) : @xmath133 with @xmath134 \ , , \quad \mathbf{c } = \begin{bmatrix } 0 \\ 1.145632124964268 \\ 0.597104987652310 \\ 1 \\ 1 \end{bmatrix } \,,\ ] ] together with the explicit method @xmath135 all other coefficients are zero . the above form a transposed - classical imex rk method of order four .
we consider the prothero - robinson ( pr ) @xcite test problem written as a split system @xmath136 where the exact solution is @xmath137 .
an imex - gark method is pr - convergent with order @xmath57 if its application to ( [ prothero - robinson ] ) gives a solution whose global error decreases as @xmath138 for @xmath139 and @xmath140 .
[ thm : pr - convergence ] consider the imex - gark method of order @xmath57 .
assume that the implicit component has a nonsingular coefficient matrix @xmath141 , and an implicit transfer function stable at infinity , @xmath142 assume also that the additional order conditions hold : @xmath143 for @xmath144 , @xmath145 .
then the imex - gark method is pr - convergent with order : * @xmath146 if @xmath147 , and * @xmath20 if @xmath148 .
the method applied to the scalar equation ( [ prothero - robinson ] ) reads [ eqn : imex - on - pr ] @xmath149 here @xmath150^t\ , , \\
\phi^ { { \ { \textsc{i}\ } } } & = & \phi\left(t_{n-1 } + \mathbf{d}^ { { \ { \textsc{i}\ } } } \ , h \right ) = \left [ \phi(t_{n-1 } + d_1^ { { \ { \textsc{i}\ } } } \ , h ) , \ldots,\phi(t_{n-1 } + d_{s^ { { \ { \textsc{i}\ } } } } ^ { { \ { \textsc{i}\ } } } \ , h ) \right]^t\ , , \\\end{aligned}\ ] ] where @xmath151 , @xmath152 are the stage approximation times . due to the structure of the test problem
, the method uses an explicit approach for the time variable , therefore @xmath153 the exact solution is expanded in taylor series about @xmath154 : @xmath155 where the vector power @xmath156 is taken componentwise .
consider the global errors @xmath157 write the stage equation in terms of the exact solution and global errors , and use the taylor expansions to obtain @xmath158 similarly , write the solution equation ( [ eqn : imex - sol2 ] ) in terms of the exact solution and global errors : @xmath159 where the stability function of the implicit component method is @xmath160 since the explicit component method ( by itself ) has at least order @xmath57 , it follows from and the explicit order conditions that @xmath161 consequently , the global error recurrence reads @xmath162 for @xmath163 @xmath164 if conditions hold we infer from that @xmath165 for @xmath144 .
the global error recurrence reads @xmath166 this gives the desired result .
if one defines the test problem such that the time is an implicit variable then becomes @xmath167 equation is satisfied automatically by the imex coupling conditions of order @xmath57 .
the additional order conditions become @xmath168 consider now the semi - linear ( sl ) problem @xmath169 where @xmath170 is smooth and non - stiff .
an imex - gark method is sl - convergent with order @xmath57 if its application to ( [ semi - linear ] ) gives a solution whose global error decreases as @xmath138 for @xmath139 and @xmath140 .
[ thm : sl - convergence ] consider an internally consistent imex - gark method of order @xmath57 .
assume that the implicit component has a nonsingular coefficient matrix @xmath141 , and an implicit transfer function strictly stable at infinity , @xmath171 assume also that the additional order conditions hold : [ eqn : sl - order - conditions ] @xmath172 for @xmath145 .
then the imex - gark method is sl - convergent with order @xmath146 .
the exact solution is expanded in taylor series about @xmath154 : @xmath173 application of the method to problem gives @xmath174 insert the exact solutions in the numerical scheme to obtain @xmath175 where @xmath176 and @xmath177 .
the exact solutions satisfy the numerical scheme only approximately .
the residuals are as follows : @xmath178 for the solution equation we have that @xmath179 where the last equality follows from the order @xmath57 conditions for the implicit and explicit components .
consider the global errors @xmath180 relations for these errors are obtained by subtracting from : @xmath181 using the mean function theorem , @xmath182 the error equations become @xmath183 @xmath184 inserting into and rescaling gives @xmath185 which for @xmath163 gives the following recurrence for the global error : @xmath186 we have that @xmath187 where the last equality follows from the additional order conditions .
since @xmath188 the global error recurrence becomes @xmath189 which gives the desired result in the limit @xmath190 .
the following extension of the stiff accuracy concept @xcite offers a convenient way to satisfy the additional order conditions and .
a gark method is _ stiffly accurate _ if @xmath191 note that stiff stability can be formulated with respect to any component method @xmath20 by replacing @xmath2 with @xmath20 in . a stiffly accurate imex - gark satisfies @xmath192 and consequently @xmath193 the prothero - robinson order conditions are equivalent to @xmath194 and therefore to @xmath195 the conditions are automatically satisfied for @xmath196 as they are part of the explicit component order conditions . for a stiffly accurate imex method applied to prothero - robinson with implicit time
the order conditions are equivalent to @xmath197 and is satisfied automatically for @xmath198 due to the imex coupling conditions of order @xmath57 .
thus a stiffly accurate method is pr - convergent with order @xmath199 regardless of the form of the test problem . for a stiffly accurate imex - gark
the semi - linear oder conditions read @xmath200 and are satisfied automatically through the explicit order conditions .
in this section a stability and monotonicity analysis is performed .
we derive a linear stability theory , as well as nonlinear stability theories for both dispersive and coercive problems .
we apply the gark scheme to the linear scalar test problem @xmath201 with @xmath202 one obtains @xmath203 denote @xmath204 and @xmath205 we have that @xmath206 where the stability function can be written compactly as @xmath207 consider a stiffly accurate gark method and assume that @xmath208 for @xmath209 .
we have @xmath210 assume that the system integrated with the stiffly accurate component is very stiff , @xmath211 .
then the _ entire _ gark stability function becomes zero , @xmath212 .
we now study the nonlinear stability of gark methods applied to partitioned systems where each of the component functions is dispersive with respect to the same scalar product @xmath213 : @xmath214 consider two solutions @xmath215 and @xmath216 of , each starting from a different initial condition .
equation implies that @xmath217 and consequently the norm of the solution difference @xmath218 is non - increasing , @xmath219 .
it is desirable that the difference of the corresponding numerical solutions is also non - increasing , @xmath220 .
the analysis is carried out in the norm associated with the scalar product in .
several matrices are defined from the coefficients of for @xmath221 : @xmath222 with @xmath223 , @xmath224 , and @xmath225 the following definition and analysis generalize the ones in @xcite .
a generalized additive runge - kutta method is algebraically stable if the weight vectors are non - negative @xmath226 and the following matrix is non - negative definite : @xmath227 we have the following result .
an algebraically stable gark method applied to a partitioned system with dispersive component functions is unconditionally nonlinearly stable , in the sense that the difference of any two numerical solutions in non - increasing @xmath228 for any step size @xmath229 .
the difference between solutions @xmath230 advances in time as follows : @xmath231 where @xmath232 from we get @xmath233 from it follows that @xmath234 substituting into leads to @xmath235 equation can be written in the equivalent form @xmath236 where @xmath237^t.\ ] ] from and the positive definiteness of @xmath238 we have that @xmath239 the positivity of the weights and dispersion condition give the desired result : @xmath240 a gark method is _ stability - decoupled _ if @xmath241 for stability decoupled gark methods the interaction of different components does not influence the overall nonlinear stability . if each of the component methods is nonlinearly stable , perhaps under a suitable step size restriction , then the overall method is nonlinearly stable ( under a step size restriction that satisfies each of the components ) . in particular , if each of the component runge - kutta scheme is algebraically stable @xmath242 then equation shows that is a sufficient condition for the algebraic stability of the gark scheme .
equation shows that the nonlinear stability of each of the component methods is sufficient to obtain a nonlinear stable gark scheme in the case of component partitioning .
next , consider partitioned systems where each of the component functions is coercive @xcite : @xmath243 consider a partitioned system with coercive component functions solved by a gark method .
assume that there exist @xmath244 such that the following matrix is positive definite @xmath245 where @xmath238 was defined in .
then the solution is conditionally nonlinearly stable , in the sense that @xmath246 , under the step size restriction @xmath247 equation and condition yield @xmath248 from the positive definiteness assumption @xmath249 we see that if @xmath250 for all @xmath251 then @xmath252 .
this proves the desired result , which extends the one given in @xcite for classical additive runge - kutta methods .
if the gark method is stability decoupled then the weights @xmath253 in are chosen independently for each component . in this case
each component method , applied to the corresponding subsystem , is conditionally stable under a step restriction @xmath254 .
the gark method s step size restriction is given by the bounds for individual components , @xmath255 , i.e. , no additional stability restrictions are imposed on the step size .
we construct a second order imex - gark method where the implicit and explicit parts have different numbers of stages .
the method has a free parameter denoted @xmath256 .
the implicit method @xmath257 is second order accurate and algebraically stable since @xmath258 the explicit method is : @xmath259\ , , \quad \mathbf{b}^ { { \ { \textsc{e}\ } } } = \left[\begin{array}{c } \frac{1}{4}\\ \frac{1}{2}\\ \frac{1}{4 } \end{array}\right ] \ , , \quad \mathbf{c}^ { { \ { \textsc{e } , \textsc{e}\ } } } = \left[\begin{array}{c } 0\\ \frac{1}{2}\\ 1 \end{array}\right ] \,.\ ] ] the explicit method is conditionally stable for coercive problems .
a good value of the free parameter for stability is @xmath260 for which holds with @xmath261 .
the coupling coefficients are @xmath262\ , , \quad \mathbf{c}^ { { \ { \textsc{e } , \textsc{i}\ } } } = \mathbf{c}^ { { \ { \textsc{e } , \textsc{e}\ } } } = \left[\begin{array}{c } 0\\ \frac{1}{2}\\ 1 \end{array}\right]\,,\ ] ] and @xmath263\ , , \quad \mathbf{c}^ { { \ { \textsc{i } , \textsc{e}\ } } } = \mathbf{c}^ { { \ { \textsc{i } , \textsc{i}\ } } } = \left[\begin{array}{c } \frac{1}{4}\\ \frac{3}{4 } \end{array}\right]\,.\ ] ] the imex - gark method is stability - decoupled @xmath264^t \mathbf{b}^ { { \ { \textsc{i}\ } } } + \mathbf{b}^ { { \ { \textsc{e}\ } } } \mathbf{a}^ { { \ { \textsc{e } , \textsc{i}\ } } } - \mathbf{b}^ { { \ { \textsc{e}\ } } } \left[\mathbf{b}^ { { \ { \textsc{i}\ } } } \right]^t = \mathbf{0}\,.\ ] ] this property , and the algebraic stability of the implicit part , imply that the gark method is nonlinearly stable under the same step size restriction for which the explicit component is nonlinearly stable ( e.g. , for @xmath260 we have @xmath265 ) .
this imex - gark scheme is represented compactly by its generalized butcher tableau as : @xmath266 this section studies the contractivity and monotonicity of the generalized additively partitioned runge - kutta methods .
we are concerned with partitioned systems where there exist @xmath267 such that @xmath268 this implies that condition holds for any @xmath269 , i.e. , for each individual subsystem @xmath4 the solution of one forward euler step is monotone under this step size restriction .
the condition also implies that the system has a solution of non increasing norm .
to see this write an euler step with the full system as a convex combination @xmath270 if @xmath271 , and consequently @xmath272 @xcite .
we seek to construct gark schemes which guarantee a monotone numerical solution @xmath273 for under suitable step size restrictions . a comprehensive study of contractivity of runge - kutta methods is given in @xcite .
step size conditions for monotonicity are discussed in @xcite .
strong stability preserving methods suitable for hyperbolic pdes are reviewed in @xcite .
monotonicity for runge - kutta methods in inner product norms is discussed in @xcite .
this study follows the approach of higueras and co - workers , who have extended the monotonicity theory to additive runge - kutta methods @xcite .
the scheme can be represented in matrix form as [ eqn : genaddrk - matrix ] @xmath274 using notation and @xmath275 equations become @xmath276 the following definitions extend the corresponding ones from @xcite .
let @xmath277 and @xmath278 a gark scheme defined by @xmath279 is called _ absolutely monotonic _
( a.m. ) at @xmath280 if [ eqn : am - conditions ] @xmath281 where @xmath282 . here
all the inequalities are taken component - wise .
the region of absolute monotonicity of the gark scheme is @xmath283 \times \dots \times \left[0,r^{\{n\}}\right ] \right\}\,.\ ] ] consider the gark scheme defined by @xmath284 and a point in the interior of its absolute monotonicity region @xmath285 for any step size obeying the restriction @xmath286 the stage values and the solution of are monotonic [ eqn : monotonicity - conclusions ] @xmath287 in practice we are interested in the largest upper bound for the time step that ensures monotonicity .
the proof is a direct extension of the corresponding one for classical additively partitioned runge - kutta methods given in @xcite .
construct the matrix @xmath288 as in .
add the same quantity @xmath289 to both sides of to obtain @xmath290 using the notation of this relation can be written in the equivalent form @xmath291 denote a vector of norms by @xmath292 since @xmath293 we have that @xmath294 and @xmath295 .
taking norms in leads to @xmath296 under the step size restriction we have @xmath297 for any @xmath298 , and from @xmath299 it follows that @xmath300 and @xmath301 multiplication by the matrix @xmath302 , whose entries are all non - negative , implies that @xmath303 and the monotonicity relation follows .
for @xmath304 we have @xmath305 which we write in the equivalent form @xmath306 where the extra stage does not contribute to the final solution .
in particular , for classical imex rk we have @xmath307 consequently @xmath308 with @xmath309 we have @xmath310 and @xmath311 gark absolute monotonicity conditions are equivalent to the absolute traditional additive rk monotonicity conditions obtained by higueras @xcite
@xmath312 here we have @xmath313 and @xmath314 @xmath315 the gark absolute monotonicity conditions lead to @xmath316 consider the following second order imex - gark scheme
. the explicit method has order two , is strong stability preserving , and has an absolute stability radius @xmath317 [ eqn : imex2-monotone ] @xmath318 the implicit method has order two , is stiffly accurate , and has an absolute stability radius @xmath319 .
the coefficients are @xmath320 and @xmath321 the two methods can not be paired as a classical imex runge - kutta method .
the following coupling terms @xmath322 ensure that the gark scheme is second order .
this can be seen from the fact that the internal consistency conditions are satisfied .
the coupling has one free parameter @xmath323 . for @xmath324
the imex scheme is of transposed - classical type .
different values of @xmath323 lead to different regions of monotonicity , as illustrated in figure [ fig : imex2-monotone ] .
a numerical search has revealed that the largest region is obtained for @xmath325 .
the method has the following generalized butcher tableau : @xmath326 monotonicity conditions for several multirate and partitioned explicit runge - kutta schemes are also discussed by hundsdorfer , mozartova , and savcenco in a recent report @xcite .
we now consider systems with two way partitioned right hand sides where both components @xmath97 and @xmath98 are stiff .
we apply a two way partitioned gark method [ eqn : imimrk ] @xmath327 the scheme has the following characteristics : * the discretization is implicit - implicit ( imim ) ; stages are implicit in @xmath328 , while stages are implicit in @xmath329 .
* one solves in succession nonlinear subsystems corresponding to each individual component . *
if each of the implicit schemes is algebraically stable , and the gark scheme is stability decoupled , then the separation of subsystem solutions does not affect the algebraic stability of the overall method .
we consider a pair of dirk schemes and compute the corresponding coupling conditions .
the first method is second order accurate and algebraically stable with @xmath330 @xmath331\,,\quad \mathbf{b}^{\{1\ } } = \left[\begin{array}{c } \frac{1}{4}\\ \frac{3}{4 } \end{array}\right]\ , , \quad \mathbf{c}^{\{1,1\ } } = \left[\begin{array}{c } \frac{1}{8}\\ \frac{5}{8 } \end{array}\right]\,.\ ] ] the second method is second order accurate and algebraically stable with @xmath332 @xmath333\ , , \quad \mathbf{b}^{\{2\ } } = \left[\begin{array}{c } \frac{2}{3}\\ \frac{1}{3 } \end{array}\right]\ , , \quad \mathbf{c}^{\{2,2\ } } = \left[\begin{array}{c } \frac{1}{3}\\ \frac{5}{6 } \end{array}\right]\,.\ ] ] the coupling coefficients @xmath334\ , , \quad \mathbf{c}^{\{1,2\ } } = \left[\begin{array}{c } 0\\ \frac{2}{3 } \end{array}\right]\,,\ ] ] and @xmath335\ , , \quad \mathbf{c}^{\{2,1\ } } = \left[\begin{array}{c } \frac{1}{4}\\ 1 \end{array}\right]\,,\ ] ] ensure that the gark method is second order accurate and is stability - decoupled , @xmath336 .
the generalized butcher tableau of the scheme reads : @xmath337 the method proceeds as follows : @xmath338 where in each stage one nonlinear system is solved for the underlined variable .
this work develops a generalized additive runge - kutta family of methods .
the new gark schemes extend the class of additively partitioned runge - kutta methods by allowing for different stage values as arguments of different components of the right hand side .
the theoretical investigations develop order conditions for the gark family using the nb - series theory .
we carry out linear and nonlinear stability analyses , extend the definition of algebraic stability to the new generalized family of schemes , and show that it is possible to construct stability - decoupled methods .
we also perform a monotonicity analysis , extend the concept of absolute monotonicity to our new family , and prove monotonic behavior under step size restrictions .
we develop implicit - explicit gark schemes in the new framework .
we show that classical implicit - explicit runge - kutta methods are a particular subset , and develop a new set of transposed - classical schemes . a theoretical investigation of the stiff convergence motivates an extension of the stiff accuracy concept .
we construct implicit - implicit gark methods where the nonlinear system at each stage involves only one component of the system .
future work will search for practical methods of high order , and will test their performance on relevant problems . partitioned implicit - implicit methods with the same coefficients on the diagonal of each method ( sdirk type )
are desirable , as are orders higher than two and a partitioning of the right hand side into three or more components .
also stiffly accurate methods with @xmath339 and @xmath340 when @xmath341 are desirable for systems driven by multiple stiff physical processes .
we are developing multirate schemes @xcite as well as symplectic schemes @xcite based on the generalized additive runge - kutta framework presented here .
10 , _ symplectic methods based on decompositions _ , siam journal on numerical analysis , 34 ( 1997 ) , pp .
19261947 . , _ additive runge - kutta methods for stiff ordinary differential equations _ , mathematics of computation , 40 ( 1983 ) , pp .
207218 . , _
contractivity / monotonicity for additive runge - kutta methods : inner product norms _ , applied numerical mathematics , 56 ( 2006 ) , pp .
862878 . , _ strong stability - preserving high - order time discretization methods _ , siam review , 43 ( 2001 ) , pp . 89112 . , _ gark methods for hamiltonian systems_. in preparation , 2013 .
height 2pt depth -1.6pt width 23pt , _ multirate gark methods_. in preparation , 2013 . , _ order conditions for numerical methods for partitioned ordinary differential equations _
, numerische mathematik , 36 ( 1981 ) , pp .
431445 . , _ solving ordinary differential equations ii : stiff and differential - algebraic problems _ ,
springer , 1993 .
, _ on strong stability preserving time discretization methods _ , journal of scientific computing , 21 ( 2004 ) , pp
. 193223 .
height 2pt depth -1.6pt width 23pt , _
monotonicity for runge - kutta methods : inner product norms _ , journal of scientific computing , 24 ( 2005 ) , pp .
97117 . height 2pt depth -1.6pt width 23pt , _ representations of runge - kutta methods and strong stability preserving methods _ , siam journal on numerical analysis , 43 ( 2005 ) , pp .
924948 . height 2pt depth -1.6pt width 23pt , _ strong stability for additive runge - kutta methods _ , siam journal on numerical analysis , 44 ( 2006 ) , pp .
17351758 . height 2pt depth -1.6pt width 23pt , _ characterizing strong stability preserving additive runge - kutta methods _ , journal of scientific computing , 39 ( 2009 ) , pp .
115128 . , _ efficient implicit - explicit runge - kutta methods with low storage requirements_. presentation at scicade 2013 , valladolid , spain , september 2013 . ,
_ monotonicity conditions for multirate and partitioned explicit runge - kutta schemes_. cwi report , unpublished , 2013 . ,
_ additive runge - kutta schemes for convection - diffusion - reaction equations _ , appl
, 44 ( 2003 ) , pp .
139181 . , _ contractivity of runge - kutta methods _ , bit numerical mathematics , 31 ( 1991 ) , pp .
482528 . , _ singly diagonally implicit runge - kutta methods with an explicit first stage _ ,
bit numerical mathematics , 44 ( 2004 ) , pp .
489502 . , _ on the stability and accuracy of one - step methods for solving stiff systems of ordinary differential equations _ , mathematics of computation , 28 ( 1974 ) , pp .
145162 . ,
_ partitioned runge - kutta methods with stepsize control and stiffness detection _ , numerische mathematik , 47 ( 1985 ) , pp .
545564 . , _ split runge - kutta methods for simultaneous equations _
, journal of research of the national institute of standards and technology , 64 ( 1960 ) . ,
_ stepsize conditions for general monotonicity in numerical initial value problems _ , siam journal on numerical analysis , 45 ( 2007 ) , pp .
12261245 . , _ implicit - explicit runge - kutta methods for time - dependent partial differential equations _ , applied numerical mathematics , 25 ( 1997 ) , pp .
151167 . ,
_ partitioning strategies in runge - kutta type methods _ , i m a journal on numerical analysis , 13 ( 1993 ) , pp .
the specific conditions for orders one to four are as follows .
[ eqn : gark - order - conditions-1-to-4-detail ] order 1 : @xmath342 order 2 : @xmath343 order 3 : @xmath344 order 4 : @xmath345
each of the implicit and explicit methods @xmath114 and @xmath346 has to satisfy the corresponding order conditions for @xmath116 . in addition , the following coupling conditions are required for third order accuracy .
the imex order two coupling conditions are : [ eqn : coupling-2 ] @xmath347 these are equivalent to the requirement that the coupling methods @xmath348 for @xmath349 are second order .
the imex order three coupling conditions read : [ eqn : coupling-3 ] @xmath350
the order four imex order conditions under the simplifying assumption are as follows .
we have two compatibility relation between the implicit and the explicit methods : @xmath351 and 16 conditions involving the coupling terms : @xmath352 @xmath353 @xmath354 @xmath355 when both simplifying assumptions and are satisfied , each of the coupling methods @xmath356 and @xmath357 needs to be fourth order accurate in its own right . in addition the following 12 coupling conditions are required : @xmath358 @xmath359 @xmath360 @xmath361 moreover , if the two coupling terms are equal , @xmath362 , then @xmath363 needs to be fourth order accurate .
the remaining six coupling conditions are @xmath364
we consider the modified prothero - robinson ( pr ) @xcite test problem written as a split system @xmath365 the method applied to the scalar equation reads here @xmath150^t\ , , \\
\phi^ { { \ { \textsc{i}\ } } } & = & \phi\left(t_{n-1 } + \mathbf{d}^ { { \ { \textsc{i}\ } } } \ , h \right ) = \left [ \phi(t_{n-1 } + d_1^ { { \ { \textsc{i}\ } } } \ , h ) , \ldots,\phi(t_{n-1 } + d_{s^ { { \ { \textsc{i}\ } } } } ^ { { \ { \textsc{i}\ } } } \ , h ) \right]^t\ , , \\\end{aligned}\ ] ] where @xmath151 , @xmath152 are the stage approximation times . due to the structure of the test problem
, the method uses an explicit approach for the time variable , therefore @xmath367 the exact solution is expanded in taylor series about @xmath154 : @xmath368 where the vector power @xmath156 is taken componentwise . write the stage equation @xmath372 in terms of the exact solution and global errors , and use the taylor expansions to obtain @xmath373 similarly , write the solution equation ( [ eqn : imex - sol ] ) in terms of the exact solution and global errors : @xmath374 the stability function of the implicit component method is @xmath160 since the explicit component method ( by itself ) has at least order @xmath57 , it follows from and the explicit order conditions that @xmath375 consequently , the global error recurrence reads @xmath376
|
this work generalizes the additively partitioned runge - kutta methods by allowing for different stage values as arguments of different components of the right hand side .
an order conditions theory is developed for the new family of generalized additive methods , and stability and monotonicity investigations are carried out .
the paper discusses the construction and properties of implicit - explicit and implicit - implicit , methods in the new framework .
the new family , named gark , introduces additional flexibility when compared to traditional partitioned runge - kutta methods , and therefore offers additional opportunities for the development of flexible solvers for systems with multiple scales , or driven by multiple physical processes .
computational science laboratory technical report csl - tr-5/2013 + adrian sandu and michael gnther `` a class of generalized + additive runge - kutta methods '' computational science laboratory + computer science department + virginia polytechnic institute and state university + blacksburg , va 24060 + phone : ( 540)-231 - 2193 + fax : ( 540)-231 - 6075 + email : [email protected] + web : http://csl.cs.vt.edu [ cols="^,^,^ " , ] partitioned runge - kutta methods , nb - series , algebraic stability , absolute monotonicity , implicit - explicit , implicit - implicit methods 65l05 , 65l06 , 65l07 , 65l020 .
|
the @xmath0 x ray observatory ( cxo ) is providing us with unprecedented high spatial ( and spectral ) resolution data on extended structures . the very first observation of a radio loud quasar , pks 0637752 , revealed the presence of an x - ray jet extending for @xmath2 10 arcsec ( chartas et al . 2000 ; schwartz et al . 2000 ) , and since then other x ray jets have been detected with high spatial resolution ( cen a , pictor a , see cxo www page http://chandra.harvard.edu/photo/cycle1.html ) . while the detection of an x ray jet is not unprecedented ( e.g. m87 , biretta , stern & harris , 1991 ; 3c273 , harris & stern 1987 ; rser et al .
2000 ) , it is interesting to notice that both radio galaxies and blazar like sources ( i.e. with a jet oriented close to the line of sight ) have been observed .
models involving inverse compton scattering of the cosmic microwave background ( cmb ) and nuclear hidden quasar radiation to produce large scale x
rays have been proposed ( e.g. brunetti , setti & comastri 1997 in the case of emission from the lobes ) .
however it has been found that the local broad band spectral energy distributions from the knots in both pks 0637752 and m87 , are not straightforward to interpret , since the level of the x ray emission is higher than what simple models predicts ( e.g. chartas et al .
2000 ; schwartz et al .
2000 , rser et al .
2000 ) . here
we propose a scenario involving both highly and mildly relativistically moving plasma in the jet and consider the dominant radiation fields on which relativistic particles compton scatter .
this can satisfactorily account for enhanced large scale x
ray emission in all jetted sources , both blazars and radio galaxies .
we also consider the spatial distribution of emission at different frequencies ( section 3 ) .
the predictions are compared to the results on pks 0637752 in section 4 and our conclusions drawn in section 5 .
comparatively little is observationally known on the structure and dynamics of extragalactic jets .
theoretically one would expect the existence of a velocity structure along the jet , from a faster inner core to slower outer regions of the flow dissipation can be produced by shocks in the relativistic plasma flow , or deceleration of the jet material can be caused by obstacles in the jet or by the interaction of the relativistic flow with the ( steady ) walls .
however only recently observational evidence has been accumulating in support of such possibility and on the relevant role of the outer layers in contributing to the observed emission .
this evidence includes the morphology and polarization structure of large scale low power jets ( komissarov 1990 , laing 1993 , 1999 ; giovannini et al .
1999 ) as well as apparent inconsistencies on the value of the relativistical beaming parameters as inferred from the flux observed at large angles with the jet direction and the statistic properties of beamed and parent populations according to the unification schemes ( chiaberge et al . 2000 ) .
these facts have been satisfactorily accounted for by schematically consider the jet as constituted by two main components : a slow , but still mildly relativistic , outer `` layer '' and a high relativistically moving `` spine '' .
strong support for such a fast component even at large ( arcsec ) distances from the active nucleus comes from the detected superluminal motion of m87 on kpc scales ( biretta , sparks & macchetto 1999 ) . within this assumption on the jet structure ,
let us consider the electromagnetic fields ( magnetic and radiative ) which can contribute to the emission of relativistic electrons at different distances along the jet , and thus determine the corresponding dominant cooling process for relativistic particles .
indeed , if dissipation occurs in knots at large distances , the accelerated particles can cool by the synchrotron and synchrotron self
compton ( ssc ) processes , but also by scattering any radiation produced externally to the knot itself .
the relative importance of the fields clearly depends on the bulk lorentz factor of the emitting regions in the jet , i.e. the spine and the layer bulk lorentz factors .
theoretical advances in recent years ( especially due to high energy @xmath3ray observations ) have convinced us that the bulk of the emission seen in blazars is produced in a rather well defined region of the jet , which on one hand can not be too compact ( hence , close to the jet apex ) in order not to absorb @xmath3ray photons through @xmath3@xmath3 interactions , and on the other hand can not be too large in order to vary on short timescales ( ghisellini & madau 1996 ) .
this region should be located at some hundreds of schwarzschild radii from the black hole , and produce radiation collimated in the beaming cone of semi aperture angle @xmath4 , where @xmath5 is its bulk lorentz factor . at larger distances ,
the jet is therefore illuminated by this radiation : it turns out that it can be observationally relevant as scattered radiation if the large scale plasma has only mildly relativistic speeds , as supposed for the jet layer .
in fact , let us assume that on the parsec and sub parsec scale the jet contains a blazar like emitting region , moving with a bulk lorentz factor @xmath5 at an angle @xmath6 with the line of sight .
this region emits an intrinsic ( comoving ) synchrotron luminosity @xmath7 .
further out , at a distance @xmath8 from the jet apex , there is a knot where dissipation occurs and electrons are accelerated and therefore radiate .
if the bulk lorentz factor @xmath9 of this region is significantly smaller than @xmath5 this region will see the nuclear ( inner ) flux enhanced by beaming , with a corresponding radiation energy density @xmath10 where @xmath11^{-1}$ ] is the beaming or doppler factor of the inner jet , and @xmath12 is the corresponding velocity .
this external energy density must be compared with the local magnetic ( @xmath13 ) and synchrotron radiation ( @xmath14 ) energy densities .
if the magnetic field is in equipartition with the locally produced synchrotron radiation , we have @xmath15 , and the self
compton emission is then of the same order .
alternatively , the field can be estimated by conservation of magnetic flux @xmath16 , which at large distances gives @xmath17 where we assume that the emitting region located at @xmath8 has a transverse dimension @xmath18 . in general @xmath19
can be a function of @xmath8 if the jet is not conical , especially at @xmath20kpc scale , where some recollimation may occur ( see e.g. kaiser & alexander 1997 ) .
the synchrotron energy density can be written as @xmath21 where @xmath22 is the beaming factor of the knot radiation .
further fields which can provide seed photons for compton scattering are those reprocessed in the narrow line region ( nlr ) and by any large scale dust structure surrounding the nucleus .
we consider these two components as produced within @xmath2kpc scale and therefore they would be seen de amplified by plasma moving along the jet on larger scales : @xmath23 where @xmath24 is the fraction of the disk emission reprocessed by the nlr .
dust is assumed to re - emit as a quasi - blackbody at temperature @xmath25 at a distance @xmath26 .
finally , the cmb radiation provides a uniformly distributed contribution to the radiation fields : @xmath27 the inverse compton process on the cmb photons is particularly relevant if the knot is moving relativistically , since in this case the cmb energy density as seen in the comoving frame is amplified by @xmath28 .
a possible contribution to the seed photons comes from the interplay between the spine and the layer of the jet , if both are active and relatively close .
radiation produced in one zone can in fact be seen amplified ( by beaming ) by the other zone .
we have however verified that these contributions are less important than the other ones , at least for the range of parameters considered in this work .
we have also neglected any stellar contribution .
the relative importance of the radiation produced by the blazar component and the local ssc emission is given by @xmath29 for illustration , consider the two cases of @xmath30 ( corresponding to a fast moving spine ) and @xmath31 ( corresponding to a layer at rest relative to the center ) : \i ) @xmath30 : in this case @xmath32 dominates if @xmath33 , i.e. if the observed power coming from the nucleus is two orders of magnitude ( for @xmath34 ) greater than the power of the knot ; \ii ) @xmath31 : in this case , for @xmath35 of the order of unity , @xmath32 dominates if the power dissipated in the inner jet is greater than the power dissipated at large scales , i.e. if @xmath36 .
we conclude that it is likely that some part of the outer jet can produce depending on the relative velocities of the inner and outer regions an inverse compton spectrum dominated by seed photons coming from the inner regions , for @xmath37@xmath38 @xmath39 . in order to fully compare all the contributions discussed above to the compton emission of the knot , in fig .
1 we report the relevant energy densities as a function of the distance along the jet , for two values of the bulk lorentz factors ( schematically corresponding to a `` spine '' and `` layer '' ) . for semplicity , in fig .
1 we also assumed that @xmath19 is constant ( and equal to 0.1 ) , i.e. a conical jet . clearly ,
if the jet progressively recollimates at large @xmath8 , the scaling of @xmath40 and @xmath41 changes and their values increases relatively to the other energy densities .
-0.5 true cm fig .
1 shows that along most of the jet the cmb is the dominant field for the spine : the energy densities of the magnetic field and of the nuclear blazar emission becomes relatively important only in the inner parts ( @xmath42 few kpc ) .
on the contrary for a slowly moving layer the beamed inner radiation can dominate the radiation energy density up to large scales . reprocessed radiation from the nlr and dust structures do not provide a relevant contribution at any distance .
for the spine emission , a systematic increase of the compton dominance ( ratio of the luminosity in the high energy and low energy spectral components ) is expected with increasing distance .
clearly in terms of the observed luminosity the relative contributions of the spine and layer have to be weighted by the corresponding beaming factors .
finally , let us consider the relative importance of the energy density in relativistic electrons and fields ( see fig .
the electron density can be estimated from the intrinsic synchrotron luminosity @xmath43 of the knot : @xmath44 then the electron energy density @xmath45 is @xmath46 where @xmath47 .
we have assumed that the particle distribution is a power law , @xmath48 , between @xmath49 and @xmath50 , with @xmath51 . for illustration ,
assume @xmath52 , @xmath53 , @xmath54 , and equal synchrotron and magnetic energy densities .
in this case @xmath55 erg @xmath56 . it is then clear that if @xmath57 holds , then the energy density of relativistic electrons is bound to dominate by a large factor , especially at large distances , as long as @xmath58 increases with @xmath8 ( since @xmath59 scales as @xmath60 while the other energy densities scale as @xmath61 ) . the only way to decrease the particle dominance is to increase the relative importance of @xmath13 . but
this immediately implies that the ssc component of the spectrum becomes weaker than the synchrotron one .
if a large compton x ray flux is instead observed , then a more moderate particle dominance can be achieved only if this emission is not due to ssc .
equipartition between @xmath13 and @xmath59 is however possible if the external radiation fields are orders of magnitude greater that the local synchrotron energy density ( if the x ray flux is comparable to the low energy synchrotron one ) . such large ratios of external to local radiation fields are possible for the spine component ( see fig .
1 ) at large distances .
the spine component of jets can then have a compton dominated spectrum even in the case of equipartition between magnetic field and particles , providing that the magnetic field in the emitting region is greater than the values shown in fig .
1 ( possibly due to shock compression and field amplification ) . in summary , suppose to observe a significant inverse compton x ray flux in an extended jet structure . if the source is a radio galaxy , we have a strongly particle dominated emitting region , since we are likely observing the emission from the layer , for which the external photon fields can only slightly reduce the particle to magnetic energy density ratio .
if the source is an extended jet of a blazar , we are observing the emission from the spine , with a strongly dominating cmb field . in this case
it is possible ( even if not guaranteed ) , that @xmath62 .
in aligned sources ( i.e. superluminal sources and blazars ) the more visible parts of the jet will be those still moving at relativistic speeds , producing radiation highly beamed towards the observer ( i.e. the spine emission will dominate ) . on the contrary , in misaligned radio galaxies the slower moving parts ( layer )
dominate with quasi
isotropic radiation providing most of the observed flux .
a further crucial point to be considered in such extended structures is the relative role of radiative versus adiabatic cooling .
the bottom panel of fig .
1 shows the radiative cooling timescales of @xmath63 electrons of both the spine and the layer . due to the different radiation fields dominating in the spine and in the layer , the spine is radiatively efficient in the outer parts and the layer in the inner parts .
therefore at large distances the adiabatic losses are negligible for the spine , and are instead important for the layer . in fig .
2 we show examples of the sed for the spine and layer , where the corresponding jets are assumed to be observed at @xmath64 and @xmath65 from the jet axis , respectively . the particle distribution is normalized to give a fixed amount of _ total _ ( synchrotron plus compton ) luminosity .
the addition of the external radiation field then changes also the synchrotron flux level , and greatly changes the compton to synchrotron flux ratio .
this can be seen in the upper panel of fig .
2 , comparing the sed corresponding to ssc only with the one including the contribution of the external nuclear photons . note that the pure ssc case can be appropriate for a curved or bent jet , in which the beamed radiation from the center does not illuminate the large scale jet .
a further prediction of the model due to the energy dependent cooling timescales is that the relative extension from the dissipation region where particles are accelerated would be larger for slower cooling particles , i.e. will depend on the mapping frequency , as @xmath66 .
-0.5 true cm -0.5 true cm
let us consider the specific case of the jet associated with pks 0637752 , and in particular of its knot w7.8 , for which detailed information are reported ( chartas et al . 2000 ) .
the spatially resolved x ray images allowed to identify knots of emission similar in structure and intensity to those seen in the radio ( tingay et al .
1998 ) and optical ( hst ; schwartz et al .
2000 ) bands . according to the cxo and radio results ,
the source can be resolved at 5 ghz with a dimension of 0.3 `` , at a distance from the central source of about 7.8 '' , yielding @xmath67 .
vsop observations detected superluminal motion in the ( pc scale ) jet ( lovell 2000 ) , setting a lower limit on the ( inner ) bulk lorentz factor @xmath68 and an upper limit on the viewing angle @xmath69 degrees ( assuming @xmath70 km @xmath71 mpc@xmath72 and @xmath73 ) .
the small scale jet appears to be well aligned with the kpc scale one . as discussed by chartas et al . and schwartz et al .
, the radio spectrum can be explained as synchrotron emission , as also supported by the detection of strong linear radio polarization , while the x ray emission largely ( @xmath2 2 orders of magnitude ) exceeds the extrapolation from the radio optical spectrum . on the other hand ,
if the x ray flux is due to ssc emission , large deviation from equipartition between magnetic field and particle energy density and/or strong inhomogeneities have to be invoked in the emitting region . thermal emission by e.g.
shocked plasma also appears to be unlikely , as it would imply high particle density and thus unobserved large rotation measures . on the contrary in the proposed scenario the x
ray emission can be satisfactorily interpreted as scattered cmb radiation .
3 shows the sed of both the nuclear blazar component and of the knot w7.8 . to fit the core emission ,
we have assumed an emitting region of size @xmath74 cm , magnetic field @xmath75 gauss , bulk lorentz factor @xmath76 and viewing angle @xmath77 ( yielding @xmath78 ) .
the emitted intrinsic luminosity is @xmath79 erg s@xmath72 .
for this nuclear jet component we could estimate the kinetic power carried by the jet : in the form of poynting flux and emitting electrons we have respectively @xmath80 erg s@xmath72 @xmath81 .
assuming one ( cold ) proton for each electron , the estimated kinetic power is of order @xmath82 erg s@xmath72 .
-0.5 true cm -0.5 true cm for the knot , we have assumed that the emission region has a size of @xmath23 kpc , at a distance of 640 kpc from the center ( corresponding to the deprojected observed distance , with a viewing angle of @xmath83 ) .
the magnetic field is @xmath84 g , the intrinsic luminosity is @xmath85 erg s@xmath72 and @xmath86 .
the viewing angle is the same as the core . with these parameters , the cmb contribution to the radiation fields as seen by the knot is largely dominating . at the redshift of pks
0637752 this corresponds to @xmath87 erg @xmath56 , to be compared with @xmath88 and @xmath89 erg @xmath56 .
the jet bulk kinetic powers are @xmath90 and @xmath91 erg s@xmath72 .
if there is one proton for emitting electron @xmath92 erg s@xmath72 .
these numbers , typical for powerful radio sources , are in rough agreement with the corresponding values for the core , suggesting that the kinetic power is conserved along the jet ( see e.g. celotti & fabian 1993 ) .
note that both the large bulk lorentz factor and the small size ( i.e. a much smaller value of @xmath19 than what adopted in the previous section ) refer to the _ spine _ component , while the jet could appear wider and slower if the layer dominates ( i.e. at large viewing angles ) .
however , so far direct evidence for layers exists only for @xmath2kpc scale jets ( swain , bridle & baum 1998 ; laing et al . 1999 )
. the optical emission can be self consistently interpreted as the higher energy part of the synchrotron component .
this in turn implies that the emission is due to high energy electrons , whose cooling length corresponds to distances @xmath93 kpc .
radio and x ray emission are instead accounted for by particles of similar ( lower ) energies and therefore images in these two bands are expected to resemble each other ( with cooling lengths @xmath94 kpc ) , as indeed observed ( schwartz et al . 2000 ) . in our model
, @xmath49 is required to be small ( 1020 ) in order to fit the x ray flux and spectrum .
the presence of such low energy electrons can cause some depolarization of the radio emission of the knot ( wardle et al .
1998 ) possibly indicating the contribution of electron positron pairs ( which do not affect polarization ) .
we have discussed the role of different electromagnetic fields for the radiative dissipation of relativistically accelerated electrons in large scale jets . the dissipation regions ( knots ) act as scattering mirrors for any externally produced photon field as well as the nuclear beamed radiation .
a jet structure comprising components with different speeds can account for large scale emission in jets associated with both blazar and radio galaxies . in the former case
, we expect to detect more intense radiation from a fast moving component ( spine ) , spatially concentrated in the outer jet regions and dominated by the cmb scattered field at distances greater than a few kpc . on the contrary , weaker and more
isotropic emission would dominate in radio galaxies , produced by a mildly relativistic portion of the jet , mainly compton scattering the beamed radiation coming from the nucleus .
deviation from this behavior in radio galaxies is expected for jets which are subject to strong bending : in this case the inner beamed field would not reach any misdirected component .
while the above estimates imply values of the kinetic power associated with the emitting knots typical of high power radio sources , a quasi stationary hydrodynamical confinement of such structures appears to be a problem as extreme high gas densities for the external gas would be required .
we expect that optical and x ray emission to be common in large scale jets of both blazar and radio galaxies .
the promising results by @xmath0 and hst will thus play a key role in the understanding of the jet structure and dynamics as well as the dissipation processes .
we thank the anonimous referee for useful suggestions .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , caltech , under contract with the nasa .
|
x ray emission from large scale extragalactic jets is likely to be due to inverse compton scattering of relativistic particles off seed photons of both the cosmic microwave background field and the blazar nucleus .
the first process dominates the observed high energy emission of large scale jets if the plasma is moving at highly relativistic speeds and if the jet is aligned with the line of sight , i.e. in powerful flat radio spectrum quasars .
the second process is relevant when the plasma is moving at mildly bulk relativistic speeds , and can dominate the high energy emission in misaligned sources , i.e. in radio galaxies . we show that this scenario satisfactorily accounts for the spectral energy distribution detected by @xmath0 from the jet and core of pks 0637752 . 0_h_@xmath1 0_q_@xmath1 galaxies : active - galaxies : jets - galaxies : quasars : individual : pks 0637752 - radiation mechanisms : non thermal
|
the acceleration of the fast solar wind is a long - standing problem in solar physics .
it is likely that alfvn waves play a significant role in this process ( e.g. leer et al .
. they can propagate over long distance , which allows them to reach the outer corona , because to lowest order they are incompressible and do not dissipate .
however dissipative damping is required at some point to avoid too high wind velocities ( e.g. holzer et al .
a closer examination shows though that a linearly polarized alfvn wave is compressible to second order , since the magnetic pressure , @xmath0 , varies with half the wave length of the alfvn wave itself ( alfvn & flthammar 1963 ) . in a circularly polarized alfvn wave , on the other hand , the magnetic pressure is constant along the wave , which is the physical reason why the circularly polarized alfvn wave in a homogeneous medium is an exact solution to the nonlinear mhd equations .
several groups have studied different mechanism through which alfvn waves can dissipate . in two dimensions phase mixing due to a transverse gradient in the phase velocity ( e.g. heyvaerts & prist 1983 ) can lead to a strong damping of the wave .
this mechanism is not available in one dimension though , and one has then rather considered the nonlinear coupling of the alfvn wave with other modes ( e.g. wentzel ( 1973 ) , for circularly polarized alfvn waves in particular see also del zanna et al .
( 2001 ) ) .
cohen & kulsrud ( 1974 ) found in an approximate solution of the nonlinear mhd equations that the linearly polarized alfvn wave steepens and can form current sheets .
this result has later been confirmed by numerical simulations by boynton & torkelsson ( 1996 ) and ofman & davila ( 1997 ) .
in this paper we use the numerical code of boynton & torkelsson ( 1996 ) ( see also torkelsson & boynton ( 1998 ) ) to simulate propagating circularly polarized alfvn waves . in the following
we will consider both models with a constant background density , and stratified models .
for all the simulations the temperature is @xmath1 k and the period of the waves is 300 s. the isothermal speed of sound is @xmath2 ms@xmath3 .
= 13.0 cm for the unstratified models , we differentiate between two cases , when the magnetic field is weak , that is @xmath4 and @xmath5 , and when it is strong that is @xmath6 and @xmath7 .
for both the runs @xmath8kgm@xmath9 .
we compare the runs in fig .
1 . if we compare our fig 1a to fig .
1 of boynton & torkelsson ( 1996 ) we notice that in both cases there is an acoustic precursor moving at a speed close to @xmath10 , but in their case the precursor is a sinusoidal oscillation in @xmath11 and @xmath12 .
apparently the gradual increase of the magnetic pressure at the front of the linearly polarized alfvn wave drives an oscillatory wave , while the sharp discontinuity of the magnetic pressure at the front of the circularly polarized wave works as a piston .
the amplitude of the acoustic precursor in fig . 1a scales with the square of the amplitude of the alfvn wave . when @xmath13 ( fig .
1c and 1d ) a density enhancement propagates with the wave . @xmath12 and @xmath11 are then related by @xmath14 . while the right edge of the enhancement coincides with the front of the alfvn wave , the left edge propagates with the velocity @xmath15 . in the leftmost part of fig .
1c we see the effect of the non - linear interaction between the acoustic wave and the alfvn wave .
boynton & torkelsson ( 1996 ) found that when @xmath16 approaches @xmath17 in strength , the alfvn wave starts to steepen into current sheets ( fig .
2a ) which enhance the dissipation of the wave
. the circularly polarized alfvn wave does not develop such current sheets ( fig .
= 6.0 cm we discus in this paper two models : one with @xmath18 and one with @xmath19 , both have a background magnetic field @xmath20 t. at the bottom of the grid @xmath21 is smaller than @xmath15 and equal to @xmath22 ms@xmath3 , but it increases with @xmath23 and eventually becomes higher than @xmath15 . density is taken to be equal to @xmath24kgm@xmath9 at the bottom of the corona and varies with @xmath23 as @xmath25 where @xmath26 m is the scale height at @xmath27 , and @xmath28 is the solar radius .
the wave front of the alfvn wave pushes matter upwards forming a density perturbation which behaves as an acoustic precursor as long as @xmath21 is smaller than @xmath15 .
when @xmath21 becomes larger than @xmath15 the alfvn wave begins to overtake the precursor , which however persists and propagates forward with a speed close to the speed of sound ( fig .
3a ) . for a low amplitude alfvn wave
the density enhancement almost keeps its shape as it propagates while for a high amplitude wave the density enhancement changes gradually .
= 8.8 cm in fig .
3c we see a decrease in the poynting flux in front the acoustic precursor .
4 shows the time average over one alfvn wave period of the poynting flux and the integrated work it does via the lorentz force . in the linearly polarized case
the wave loses @xmath29 of its poynting flux below @xmath30 r@xmath31 and very little of it is dissipated after that ( see boynton & torkelsson ( 1996 ) fig .
9 ) while in the circularly polarized case the wave loses very little of its poynting flux below @xmath32 , but it has lost almost half of it at @xmath33 r@xmath31 by doing mechanical work on the plasma .
in constructing models of alfvn wave driven stellar winds it is important to understand the mechanism by which the alfvn waves are damped .
a linearly polarized alfvn wave of high amplitude can steepen and form current sheets even in a homogeneous medium , which leads to a quick damping of the wave .
while a circularly polarized alfvn wave does not dissipate its energy in a homogeneous medium , we have found that it may do so in a stratified medium .
the dissipation sets in later than for a linearly polarized alfvn wave though , but a significant fraction of its poynting flux may still be spent on doing work on the medium through which the alfvn wave propagates .
|
we study propagating alfvn waves by solving the time - dependent equations of magnetohydrodynamics ( mhd ) in one dimension numerically . in a homogeneous medium
the circularly polarized alfvn wave is an exact solution of the ideal mhd equations , and therefore it does not suffer from any dissipation .
a high - amplitude linearly polarized alfvn wave , on the other hand , steepens and form current sheets , in which the poynting flux is lost . in a stratified medium , however , a high - amplitude circularly polarized alfvn wave can also lose a significant fraction of its poynting flux .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
we consider a homogenous and isotropic conducting body , assumed to be contained in @xmath0 , a bounded , lipschitz domain of @xmath1 , @xmath2 .
we assume that there exist @xmath3 , a lipschitz domain contained in , and different from , @xmath0 , and a closed set @xmath4 such that the interior of @xmath5 is not empty and @xmath5 has a positive distance from @xmath6 .
we assume that @xmath5 is known and accessible to measurements
. in the body there might be present some defects , which we assume to be perfectly insulating and outside @xmath3 .
namely , we model these defects by a closed set @xmath7 such that @xmath8 is empty .
we notice that @xmath9 represents the union of the boundaries of these defects and that we denote with @xmath10 the connected component of @xmath11 containing @xmath3 , that is the region of @xmath0 reachable from @xmath5 without crossing @xmath9 .
the defects may have different geometrical properties .
for instance , we may have , even at the same time , _ cracks _ ( either interior or surface - breaking ) , or _ material losses _ ( either interior , that is cavities , or at the boundary ) .
we recall that a defect @xmath9 is a _ material loss _ if @xmath10 coincides with the interior of its closure .
let us consider the following experiment . if a current density @xmath12 is applied on @xmath5 , then the electrostatic potential in @xmath0 , @xmath13 , is the solution to the following ( normalized ) neumann boundary value problem @xmath14 the current density is modeled by a function @xmath15 , for some constant @xmath16 , such that @xmath17 .
the electrostatic potential @xmath18 may then be measured on @xmath5 .
we call such a measurement @xmath19 and we observe that @xmath20 and @xmath21 . in this way we obtain an electrostatic boundary measurement of voltage , @xmath22 , and current , @xmath12 , type on @xmath5 . in mathematical words
, we measure the cauchy data @xmath23 of the harmonic function @xmath18 on @xmath5 .
clearly , prescribed the current @xmath12 , the voltage @xmath22 depends on @xmath9 . if @xmath9 is unknown , then the measured voltage @xmath22 may provide information about the unknown defect .
in fact , the aim of the inverse problem is to reconstruct an unknown defect @xmath9 by prescribing one or more current densities @xmath12 and measuring the corresponding value of the potentials on @xmath5 .
such a problem arises , for instance , in non - destructive evaluation , for the determination of flaws like cracks or cavities in conducting bodies by non - invasive methods .
we refer to this problem as the _ inverse crack problem _ , in the general case . instead , when we a priori know that the defect is a material loss , we denote it as the _ inverse cavity problem_. for results on the inverse crack problem and related problems , we refer to the review article @xcite .
here we simply wish to note that a single measurement ( that is performing the experiment previously described only once ) is enough to determine uniquely a material loss . in the general crack case , instead
, one measurement may not be enough , however two suitably chosen measurements ( corresponding to two suitable prescribed current densities ) are enough for unique identification of any kind of defects at least in the planar case .
let us remark here that if the unknown defect is a priori assumed to be interior ( that is @xmath24 and if the whole boundary of @xmath0 is accessible , then we may simply take @xmath25 .
our approach to this inverse problem is the following .
we observe that @xmath18 is smooth outside @xmath9 , whereas it may , and generally does , jump across @xmath9 .
therefore , starting from the cauchy data , we wish to reconstruct the function @xmath18 in @xmath0 , and in particular its discontinuity set @xmath26 .
we notice that this is not a classical cauchy problem for @xmath18 , since @xmath18 is harmonic in @xmath11 with @xmath9 unknown !
rather , it looks more like a free - discontinuity problem for @xmath18 , since its discontinuity set @xmath26 is unknown and it is actually the aim of our reconstruction .
if we are able to reconstruct @xmath18 and @xmath26 , then we obtain valuable information on @xmath9 , given the fact the @xmath27 .
actually , for the inverse cavity problem , @xmath26 determines the whole @xmath28 . on the contrary , in the inverse crack problem
, it may happen that a crack is not visible for a particular measurement , that is @xmath26 does not detect the whole @xmath28 . in this case
, we may change the prescribed current density , reconstruct again the electrostatic potential from its values on @xmath5 , and recover another portion of @xmath28 .
the uniqueness results tell us how many times and with which kind of prescribed current densities we need to repeat this procedure to fully reconstruct the unknown defect .
the main difficulties in the reconstruction of @xmath18 from its cauchy data are the following .
first of all , the problem is severely ill - posed , as cauchy problems for elliptic equations are .
second , since the potential @xmath18 to be reconstructed is a discontinuous function whose discontinuities are unknown ( actually they are the aim of our reconstruction ) , the problem is not even linear .
thus all the main difficulties of the original inverse problem are still present in the reconstruction of @xmath18 .
the way to tackle ill - posedness is crucial .
in fact , since the boundary data are measured , the data which are really available are not the exact cauchy data @xmath23 but some noisy perturbation of them .
namely , the available data we assume to know are @xmath29 . here
@xmath30 belongs to @xmath31 and satisfies @xmath32 and @xmath33 , whereas @xmath34 belongs to @xmath35 and satisfies @xmath36 .
we assume that @xmath37 where @xmath38 , @xmath39 , denotes the noise level .
as mentioned , rather than a classical cauchy problem , we consider such a reconstruction as a free - discontinuity problem for the unknown potential @xmath18 .
we follow the variational approach developed in @xcite for cracks and material losses , respectively .
such a method is based on the following two features .
the first one is the choice of the regularization . in order to regularize the problem a perimeter - like penalization
namely , we penalize the @xmath40-dimensional measure of the unknown defect @xmath9 ( actually of the discontinuity set of the unknown potential ) .
second , we model discontinuity sets through phase - field functions , thus obtaining a formulation in which a discontinuous function @xmath41 and its discontinuity set @xmath42 are replaced , respectively , by a smooth function @xmath41 and by a smooth phase - field function @xmath43 .
such a formulation is amenable to numerical implementation . in particular , for the crack case , in @xcite
it has been used a regularization based on the so - called mumford - shah functional , @xcite , and its approximation , in the sense of @xmath44-convergence , with phase - field functionals due to ambrosio and tortorelli , @xcite . for material losses , @xcite , it has been used a more classical perimeter penalization and its approximation , again in the sense of @xmath44-convergence , with phase - field functionals due to modica and mortola , @xcite . for further details on free - discontinuity problems and their approximations
we refer for instance to @xcite .
in this paper we develop the numerics of the approach in @xcite .
the convergence analysis done in these papers provides a justification of the numerical method , in particular for the material loss case , see ( * ? ? ?
* theorem 4.2 ) . for what concerns the crack case , we do not have a precise convergence result for the method implemented here .
however this is quite simpler from a numerical point of view than the one developed in @xcite and for which we have convergence results .
moreover we believe that this simplification might still lead to good reconstructions , see also the discussion in section 5 of @xcite .
we shall use the following notation .
we fix a constant @xmath45 , depending on some regularity properties of the defect @xmath9 to be reconstructed .
namely , we assume that there exist a constant @xmath45 and a constant @xmath46 , independent of @xmath12 , such that @xmath47 where @xmath13 solves ( [ dirpbm ] ) .
we also set @xmath48 the function @xmath49 is continuous and non - decreasing and such that @xmath50 , @xmath51 if @xmath52 , and @xmath53 .
then , for @xmath39 , we define @xmath54 we introduce a single - well potential @xmath55 centered at @xmath56 , that is a non - negative continuous function such that @xmath57 if and only if @xmath58 .
we shall also need a double - well potential @xmath59 centered at @xmath60 and @xmath56 , that is a non - negative continuous function such that @xmath61 if and only if @xmath62 .
we assume that the functions @xmath59 , @xmath55 and @xmath63 are @xmath64 and are bounded all over @xmath65 and also their derivatives are bounded and uniformly continuous all over @xmath65 .
we shall also assume that @xmath66 , @xmath67 and @xmath68 are hlder continuous for some exponent @xmath69 , @xmath70 , all over @xmath65 .
furthermore , the following assumption will be made .
about @xmath63 we require that for any @xmath71 we have @xmath72 and @xmath73 for any @xmath74 .
in particular , we have that @xmath75 .
we also require that for any @xmath71 we have @xmath76 .
obviously , we have that , for any @xmath71 , @xmath77 , and , for any @xmath74 , @xmath78 and @xmath79 . for example
, the following choices may be made .
for any @xmath80 $ ] @xmath81 with straightforward extensions beyond @xmath82 $ ] .
we define the space @xmath83 . to any @xmath84 we associate the function @xmath85 .
we remark that @xmath86 and @xmath87 almost everywhere in @xmath3 .
we finally fix positive tuning parameters @xmath88 , @xmath89 and @xmath90 , and a noise level @xmath38 , @xmath39 .
all these constants and the notation will be kept fixed throughout the paper .
basically , the method is the following . beginning from the crack case
, we wish to minimize , with respect to the phase - field variable @xmath84 , with the constraint @xmath91 , the functional @xmath92 , which is defined as follows . for any @xmath84 , recalling that @xmath85 , we set @xmath93 here @xmath94 solves @xmath95 we notice that the first term is the fidelity term with respect to the measured boundary datum , the other three terms are the ambrosio - tortorelli functional .
the link with the prescribed boundary datum and with the presence of cracks is through @xmath96 , the solution of the weighted elliptic equation .
we observe that the single - well potential @xmath55 forces the phase - field function @xmath85 to be equal to @xmath56 except in a small region , which is where the crack should be located .
the tuning parameters @xmath88 , @xmath89 and @xmath90 allow to put more emphasis on one or the other of the features of the functional .
namely , @xmath88 controls the match with the dirichlet datum , @xmath89 the smoothness of the reconstructed potential away from its discontinuities and @xmath90 the penalization on the @xmath40-dimensional measure of the discontinuities .
therefore @xmath90 may be seen as a regularization parameter .
for the material loss case , we simply replace the single - well potential @xmath55 with the double - well potential @xmath59 .
namely , we define @xmath97 in an analogous way by simply replacing @xmath55 with @xmath59 , that is , for any @xmath84 , we set @xmath98 we then minimize , with respect to the phase - field variable @xmath84 , with the constraint @xmath91 , the functional @xmath99 .
we notice that in this case the last two terms are the modica - mortola functional , which penalizes the perimeter of @xmath10 in @xmath0 . here
the double - well potential @xmath59 forces the phase - field function @xmath43 to be either @xmath60 ( inside the material loss ) or @xmath56 ( outside the material loss ) , with a quick transition between these two regions . summarizing , we shall minimize the functional @xmath100 , when we aim to reconstruct defects such as cracks , and the functional @xmath99 , when we aim to reconstruct material losses .
namely , we wish to solve numerically the following minimization problems ( depending on the properties of the unknown defect @xmath9 ) a. @xmath101 on @xmath102 , with the constraint @xmath91 , if @xmath9 contains portions of cracks .
b. @xmath103 on @xmath102 , with the constraint @xmath91 , if @xmath9 is a material loss defect .
let us notice that , by the direct method , these minimum problems admit a solution . about the numerical method , in order to find the minimizers , we formulate the corresponding optimality system and we use a gradient method , see section [ optsec ] for details . in section [ numsec ] numerical simulations
are presented for both the single- and double - well approximations .
numerical experiments are performed for various types of defects with noise - free and noisy data - sets .
the second author is partially supported by gnampa under 2008 and 2009 projects and by the italian ministry of university and research under prin 2008 project .
part of this work was done during a visit of the first author to the university of trieste , supported by gnampa under 2008 project , and during a visit of the second author at the university of graz , supported by the special research center `` mathematical optimization and applications in biomedical sciences '' .
the authors wish also to thank alfio borz for useful discussions .
we now look towards the numerical implementation of the method .
we begin by recalling the differentiability properties of the functionals @xmath100 and @xmath99 , which have been investigated in ( * ? ? ?
* section 6 ) .
we define the following spaces . for
any @xmath104 , @xmath105 , let us call @xmath106 and @xmath107 , with norm @xmath108 and @xmath109 . to any @xmath110 we as usual associate the function @xmath85 .
if @xmath111 belongs either to @xmath112 or to @xmath113 , then @xmath114 , @xmath87 almost everywhere in @xmath3 , and , provided @xmath91 almost everywhere in @xmath0 , we also have @xmath115 almost everywhere in @xmath0 .
we observe that @xmath116 as previously defined . for any @xmath117 , @xmath118
, we define @xmath119 we observe that , by a generalized poincar inequality , on @xmath120 the usual @xmath121 norm and the norm @xmath122 are equivalent .
therefore , we shall set this second one as the natural norm of @xmath120 .
we define @xmath123 as follows @xmath124 there exist constants @xmath125 and @xmath126 , depending on @xmath38 and @xmath69 , such that all the following results hold . first , @xmath127 , with bounded image in @xmath128 , and , for any @xmath129 , such an operator @xmath130 is differentiable in @xmath131 with respect to the @xmath132 , with @xmath133 , and @xmath128 norms . let @xmath134 be the differential in @xmath131 . then for any @xmath111 in @xmath112 we have @xmath135=u_{\varepsilon}(\tilde{v}_0,\tilde{v})\ ] ] where @xmath136 solves the following problem @xmath137 obviously , @xmath138 .
we recall that for any vector valued function @xmath139 , @xmath140 defines a functional on @xmath141 in the following way @xmath142=-\int_{\omega}g\cdot \nabla \phi\quad\text{for any } \phi\in w^{1,2}(\omega).\ ] ] therefore , the weak formulation of ( [ der ] ) is looking for a function @xmath143 such that @xmath144 here , and analogously in the sequel , the differentiability has to be understood in the following sense .
for any @xmath111 in @xmath112 @xmath145+r(\tilde{v})\ ] ] where @xmath146 we conclude that , for any @xmath147 , @xmath100 and @xmath99 are differentiable in @xmath131 with respect to the @xmath113 norm , with @xmath133 .
let @xmath148 be the differentials in @xmath131 of @xmath100 and @xmath99 , respectively .
then , for any @xmath149 we have @xmath150= \frac{2a}{\varepsilon^{q_1}}\int_{\gamma}(\mathcal{h}_{\varepsilon}(\tilde{v}_0)-g_{\varepsilon})u_{\varepsilon}(\tilde{v}_0,\tilde{v})+\\ b \int_{\omega}\left(2\psi_{\varepsilon}(v_0)\nabla \mathcal{h}_{\varepsilon}(\tilde{v}_0)\cdot\nabla u_{\varepsilon}(\tilde{v}_0,\tilde{v})- \psi'_{\varepsilon}(v_0)|\nabla \mathcal{h}_{\varepsilon}(\tilde{v}_0)|^{2}\tilde{v}\right)+\\ \frac{c^2}{\varepsilon}\int_{\omega}(-v'(v_0)\tilde{v})+2\varepsilon \int_{\omega}\nabla\tilde{v}_0\cdot\nabla \tilde{v}\end{gathered}\ ] ] and @xmath151= \frac{2a}{\varepsilon^{q_1}}\int_{\gamma}(\mathcal{h}_{\varepsilon}(\tilde{v}_0)-g_{\varepsilon})u_{\varepsilon}(\tilde{v}_0,\tilde{v})+\\ b \int_{\omega}\left(2\psi_{\varepsilon}(v_0)\nabla \mathcal{h}_{\varepsilon}(\tilde{v}_0)\cdot\nabla u_{\varepsilon}(\tilde{v}_0,\tilde{v})- \psi'_{\varepsilon}(v_0)|\nabla \mathcal{h}_{\varepsilon}(\tilde{v}_0)|^{2}\tilde{v}\right)+\\ \frac{c^2}{\varepsilon}\int_{\omega}(-w'(v_0)\tilde{v})+2\varepsilon \int_{\omega}\nabla\tilde{v}_0\cdot\nabla \tilde{v}.\end{gathered}\ ] ] an important remark is the following . if @xmath152 , then we may actually choose @xmath153 , and we observe that @xmath154 is a hilbert space , with the scalar product @xmath155 for any @xmath156 , @xmath157 . if @xmath158 , then it might happen that @xmath159 and that @xmath160 has not a hilbert space structure anymore
. however , since @xmath161 is finite , @xmath160 is still a strictly convex real reflexive banach space . in the sequel we shall fix @xmath162 , ( with @xmath153 if @xmath152 ) and we call @xmath163 the following functional , which is defined on @xmath164 , @xmath165 such
a functional is finite for any @xmath166 . by similar reasonings , for any @xmath167
, we have that @xmath163 is differentiable in @xmath168 and for any @xmath166 we have @xmath169= \frac{2a}{\varepsilon^{q_1}}\int_{\gamma}(u_0-g_{\varepsilon})u+\\ b \int_{\omega}\left(2\psi_{\varepsilon}(v_0)\nabla u_0\cdot\nabla u- \psi'_{\varepsilon}(v_0)|\nabla u_0|^{2}\tilde{v}\right)+ \frac{c^2}{\varepsilon}\int_{\omega}(-v'(v_0)\tilde{v})+2\varepsilon \int_{\omega}\nabla\tilde{v}_0\cdot\nabla \tilde{v}.\end{gathered}\ ] ] we observe that @xmath170 .
analogously , we define @xmath171 simply by replacing @xmath55 with @xmath59
. analogous properties of differentiability hold for @xmath171 as well .
let us finally define @xmath172 such that for any @xmath173 we have @xmath174 in an analogous way we define @xmath175 replacing @xmath163 with @xmath171 .
we observe that @xmath176 ( and @xmath175 as well ) is differentiable in any @xmath177 . for any @xmath178 we have @xmath179=\\ \frac{2a}{\varepsilon^{q_1 } } \int_{\gamma}(u_0-g_{\varepsilon})u+ 2b\int_{\omega}\psi_{\varepsilon}(v_0)\nabla u_0\cdot\nabla u+ \int_{\omega}\psi_{\varepsilon}(v_0)\nabla \phi_0\cdot \nabla u,\end{gathered}\ ] ] and @xmath180= -b\int_{\omega}\psi'_{\varepsilon}(v_0)|\nabla u_0|^2\tilde{v}+\frac{c^2}{\varepsilon}\int_{\omega } ( -v'(v_0)\tilde{v } ) + \\2\varepsilon\int_{\omega}\nabla\tilde{v}_0\cdot\nabla \tilde{v } -\int_{\omega}\psi'_{\varepsilon}(v_0)\tilde{v}\nabla u_0\cdot\nabla \phi_0,\end{gathered}\ ] ] and , finally , @xmath181=
\int_{\omega}\psi_{\varepsilon}(v_0)\nabla u_0\cdot\nabla \phi-\int_{\partial\omega}f_{\varepsilon}\phi.\ ] ] then the resulting optimality system is the following .
we look for critical points , or better minimizers , of @xmath100 , or , equivalently , of @xmath182 subject to the constraint @xmath183 .
we use a gradient method , whose algorithm is divided into steps .
a completely analogous method may be used for finding minimizers of @xmath99 .
we initialize the algorithm by putting @xmath184 and choosing an initial guess @xmath147 such that @xmath185 almost everywhere .
we observe that taking @xmath186 ( that is @xmath187 ) is not a good choice because this is a critical point of the functional @xmath100 , thus the gradient method fails in this case .
we solve @xmath189 that is we look for @xmath190 such that @xmath191 we notice that @xmath192 and @xmath193 actually belongs to @xmath194 . by ( [ lagrangian ] ) and by ( [ dpweak ] )
, we have that for any @xmath195 @xmath196 we solve the following boundary value problem @xmath198 here @xmath199 denotes the characteristic function of @xmath5 , that is @xmath200 the weak formulation of ( [ dustrong ] ) is looking for @xmath201 such that @xmath202 such a solution @xmath203 exists and is unique . then @xmath204 and , by ( [ duweak ] ) , @xmath205 we compute the differential of @xmath100 at the point @xmath207 .
we observe that if @xmath183 , then for any @xmath195 we have @xmath208 therefore , since @xmath192 , and if we pick @xmath209 , then @xmath210 we conclude that , by ( [ dvweak ] ) , we have for any @xmath211 @xmath212= -b \int_{\omega}\psi'_{\varepsilon}(v_k)|\nabla u_k|^2\tilde{v}+\frac{c^2}{\varepsilon}\int_{\omega } ( -v'(v_k)\tilde{v } ) + \\2\varepsilon\int_{\omega}\nabla\tilde{v}_k\cdot\nabla \tilde{v } -\int_{\omega}\psi'_{\varepsilon}(v_k)\tilde{v}\nabla u_k\cdot\nabla \phi_k.\end{gathered}\ ] ] let us now consider the space @xmath160 .
we recall that either @xmath213 ( if @xmath152 ) , that is @xmath160 is a hilbert space , or @xmath160 is a strictly convex real reflexive banach space ( if @xmath158 ) . in either cases ,
if @xmath214 , we fix an operator @xmath215 such that for any @xmath216 , we have @xmath217 where @xmath218 is the usual duality between @xmath219 and @xmath59 .
we may choose @xmath220 as the duality mapping from @xmath219 into @xmath221 .
if @xmath59 is a hilbert space and we also identify @xmath219 with @xmath59 , then @xmath220 is actually the identity .
see , for instance , ( * ? ? ?
* section 42.6 ) .
let us call @xmath222 the corresponding operator for @xmath160 .
for a positive constant @xmath223 , we then update @xmath207 by setting @xmath224 we observe the following .
if @xmath225 , then @xmath226 is a critical point of @xmath176 and @xmath207 is a critical point of @xmath100 and the algorithm comes to a stop .
otherwise , provided @xmath223 is small enough , an easy computation shows that @xmath227 .
we normalize @xmath229 by truncation as follows .
we set @xmath230 . in such a way we obtain that @xmath231 and @xmath232 almost everywhere in @xmath0 .
let us note that , by our hypotheses , such a truncation does not increase the value of the functional , in fact for any @xmath233 , if @xmath234 , then @xmath235 therefore , we have found that either @xmath225 , and the algorithm stops , or , otherwise , provided @xmath223 is small enough , @xmath236 .
once we have computed @xmath237 , we iterate the algorithm by going back to step 1 .
the data for the numerical experiments are generated by solving laplace equation numerically on an domain with certain prescribed defects ( cracks or cavities ) .
we solve the neumann problem with given flux on the boundary of the computational domain , and read off the corresponding dirichlet data to get a feasible pair of neumann and dirichlet boundary data on a discrete set of measurement points on the boundary from which the defect has to be reconstructed . as input fluxes we choose pairs of plus - shaped current profiles with opposite sign located at two different sides of the rectangular computational domain .
the laplace equation is solved on a very fine irregular grid using linear finite elements .
the boundary data are genuinely defined on the unevenly distributed nodal points of elements on the boundary and are interpolated onto a much courser regular grid of measurement points .
when experimenting with noisy input data , both boundary values are contaminated by adding gaussian distributed artificial noise to the data , usually with different noise levels for @xmath238 and @xmath239 . for the numerical implementation of step 1 in the algorithm described in the previous section ( that is the numerical solution of equation ( [ dpweak2 ] ) for @xmath193 with given @xmath240 and prescribed @xmath241 )
, we also use linear finite elements for the discretization of @xmath193 .
in contrast to the data generation routine , we discretize the potential on a
_ regular , structured grid _ which is usually much coarser than the grid used for the data generation .
later on , we shall assume that the phase - field @xmath240 is also an element in the space of piecewise linear functions on the same underlying regular grid as for @xmath193 .
for the assembling of the stiffness matrix for ( [ dpweak2 ] ) , however , we replace the phase - field @xmath240 by its @xmath242-projection onto the space of functions which are piecewise constant on the triangles of the finite element space . a completely analogous procedure is applied for the solution of the adjoint equation ( [ duweak2 ] ) described in step 2 for the adjoint variable @xmath203 .
note that both systems share the same stiffness matrix and that the right - hand side of ( [ duweak2 ] ) can be easily assembled using a slightly modified stiffness matrix .
we shall use up to six different cauchy data - sets for the reconstruction of the defect .
the data - sets correspond to all possible combinations of pairs of electrodes where each electrode is located on a different side of the computational rectangle .
we can use the same factorization of the stiffness matrix for all different right - hand sides of ( [ dpweak2 ] ) and ( [ duweak2 ] ) .
the calculation of the descent direction for the cost functional as described in step 3 requires another solution of an elliptic boundary value problem for the variable @xmath243 .
as mentioned above , the update @xmath244 is discretized using linear triangular elements on a regular grid . to find @xmath244 we have to solve an elliptic equations with system matrix defined by a discretization of the operator @xmath245 . in our 2-dimensional test examples
, we always set @xmath246 and for any @xmath216 we set @xmath247 where @xmath43 solves in a weak sense @xmath248 with some parameter @xmath249 and homogeneous dirichlet boundary conditions .
the choice of dirichlet boundary conditions is motivated by the desire to keep the phase - field constantly at the value 1 on the boundary .
the assembling of the right - hand side of the equation for @xmath244 is done by evaluating ( [ dvweak0 ] ) for piecewise linear in all bases functions @xmath111 . the projection required in step 4
is easily implemented for piecewise linear functions by thresholding the nodal values . moreover , a suitable step - length for the update of the phase - field is found using an armijo - type line search .
we use a maximum number of five reduction steps for the correction of the step - length .
since each evaluation of the cost functional requires one solution of the state equation , we try to steer the step - size modification in a rather conservative way . within this setup , the following numerical experiments have been performed .
for all experiments , the phase - field parameter @xmath38 was decreased in several steps from an initial value of @xmath250 down to @xmath251 for the single - well potential and to @xmath252 for the double - well case .
we run 2500 iterations of our algorithm in the single - well case and 1000 in the double - well case .
figure [ fig1 ] shows the final phase - field together with the linear crack ( as a white line ) which was used for the data generation .
we use all six available data - sets with electrode positions on ( up / down ) , ( left / right ) , ( down / left ) , ( up / left ) , ( down / right ) , and ( up / right ) sides of the rectangle for the reconstruction and set the noise - level to zero . in this simple situation where the crack is located rather close to the boundary
we obtain very good reconstruction of the crack location with the single - well approximation . in figure [ fig2 ]
it is shown a comparison between reconstructions using 3 measurements ( left image ) with electrode positions on ( left / right ) , ( left / up ) , and ( right / up ) edges and 6 measurements ( right figure ) , again in the single - well case .
it is notable that in the reconstruction with 3 data - sets the crack tips are accurately identified but the reconstructed crack is strongly curved which is probably due to the fact that we have no electrode located on the lower edge of the computational domain .
in contrast the overall geometrical shape of the crack is reconstructed much better with 6 data - sets but the position of the crack tips is less accurate . in these two simulations
we added one percent of normally distributed noise to neumann and dirichlet data .
figure [ fig3 ] shows results for a situation with two cracks and different noise levels . here
we fixed the noise - level for the neumann data to 1% for both experiments whereas the dirichlet data were contaminated with 1% ( left image ) and 5% ( right image ) of noise .
we used three measurements ( left / right ) , ( left / up ) , ( right / up ) and the single - well potential . there is no big difference in the quality of the reconstructions . in both cases
the placement of the smaller crack in the upper right corner is inaccurate and the larger crack in the lower left corner is curved .
nonetheless the convergence of the algorithm is not heavily effected by the presence of ( moderately strong ) noise and the reconstructions are stable .
the next series of experiments presented in figure [ fig4 ] shows the tendency of the single - well based algorithm to produce dendrite - like structures .
in fact , the dendrite - shaped crack in the leftmost image is reconstructed quite well .
the polygonal crack in the middle image is approximated by a cloth - hanger like structure which has a satisfactory data fit with a shorter overall length than the polygonal curve .
finally the cavity in the rightmost image is approximated by a one - dimensional structure which looks roughly like the skeleton of the cavity . in all these three experiments
noise level is 1% for neumann data and 5% for dirichlet data and the three measurements ( left / right ) , ( left / up ) , ( right / up ) are used .
figure [ fig5 ] shows reconstructions obtained by using the double - well approximation . as expected , the phase - field approximates the characteristic functions of one cavity ( left image ) and two cavities ( right image ) . in these two tests noise level
is 1% for neumann data . in the left image noise level for dirichlet data is 5% and the three measurements ( left / right ) , ( left / up ) , ( right / up ) are used . in the right image noise level for dirichlet data is 1% but only one measurement , namely ( left / right ) , is used .
the overall location of the cavities is satisfactory , but the lower left quadrilateral is approximated by a non - convex shape . in this respect
the experiment with the double - well potential for two cavities resembles the results shown in figure [ fig2 ] where the lower left crack also has a strong tendency to bend inward . for our final numerical experiment , documented in figure [ fig5 ] ,
the double - well approach was used for the reconstruction of one - dimensional defects like the polygonal crack shown in the left image and the star - shaped crack shown on the right - hand side of the figure . in both cases
the defect is approximated by a two dimensional structure .
an interesting feature is the occurrence of a self - intersection of the boundary curve of the reconstructed defect in the case of the star - shaped crack .
also in these two final tests , noise level is 1% for neumann data and 5% for dirichlet data and the three measurements ( left / right ) , ( left / up ) , ( right / up ) are used . as a conclusion we can state that both algorithms give reconstructions of the defects with a satisfactory accuracy for an exponentially ill - posed problem .
the algorithms show a quite stable behaviour in the presence of data noise .
the single - well and double - well models develop the types of structures for which they are designed ( one - dimensional for the single - well and two dimensional for the double - well potential ) , so the single - well approach approximates cavities by dendrites and the double - well approach approximates cracks by cavities .
the double - well approach looks more stable with respect to noise , is slightly less sensitive with respect to the adjustment of the phase - field parameter @xmath38 and usually needs less iterations for convergence .
this may be in accordance with the theory , in fact for the double - well case a convergence analysis is proved , whereas the single - well model we use is a modification of the one for which we have convergence results . finally , it turned out to be important to update the phase - field parameter @xmath38 adaptively during the algorithm . if the parameter @xmath38 is chosen too small initially or decreased too fast , sharp interfaces develop too early , sometimes at incorrect locations , and the algorithm is not able to move well established interfaces to other locations .
on the other hand , if the parameter @xmath38 is decreased too much , the term containing the potential might prevail and not well established defects , usually the smaller ones , may disappear .
k. bryan and m. s. vogelius , _ a review of selected works on crack identification _ , in c. b. croke , i. lasiecka , g. uhlmann and m. s. vogelius eds .
, _ geometric methods in inverse problems and pde control _ , springer - verlag , new york , 2004 , pp .
|
we numerically implement the variational approach for reconstruction in the inverse crack and cavity problems developed by one of the authors .
the method is based on a suitably adapted free - discontinuity problem .
its main features are the use of phase - field functions to describe the defects to be reconstructed and the use of perimeter - like penalizations to regularize the ill - posed problem .
the numerical implementation is based on the solution of the corresponding optimality system by a gradient method .
numerical simulations are presented to show the validity of the method .
* ams 2000 mathematics subject classification * primary 35r30 .
secondary 65n21 , 65k10 .
* keywords * inverse problems , cracks , cavities , phase - field , perimeter penalization , optimality system .
|
neutrinos are considered possible new messengers of the universe since their properties differ from those of photons , that currently provide most information .
they could improve our understanding of the sources and mechanisms capable of accelerating cosmic rays up to energies larger than @xmath0 ev , that are observed by extensive air shower arrays .
photons of energies larger than 10 tev can not bring information from distances @xmath1 mpc since they interact by pair production with background photons . a deep connection between high energy @xmath2 emissions and @xmath3 production exists according to the model of the `` beam dump '' .
protons or nuclei accelerated by an engine , interact on a gas of matter or photons .
these interactions result mainly in neutral and charged pions which decay into photons and neutrinos .
this explains the expectation that , if @xmath2 absorption could be ignored , observed @xmath2 fluxes and @xmath3 ones should have almost the same normalization and spectral shape .
this should be a power law with spectral index around @xmath4 , as expected from fermi acceleration processes .
gamma - ray experiments are currently providing interesting results and a new era for the observations above 1 tev has been recently inaugurated by the hess experiment @xcite .
the experiment observed tev photons from the galactic centre with 9 standard deviations in only 12 h of observation , reconstructing a spectral index of the differential energy spectrum above 165 gev of @xmath5 , in disagreement with cangaroo @xcite .
this region is extremely complicated due to the presence of sgr a@xmath6 , most probably a super - massive black hole , and of the supernova remnant ( snr ) sgr a east .
the measured spectrum could be interpreted as due to purely electromagnetic processes , but also to proton acceleration and photo - meson interactions with ambient radiation or interactions with the plasma close to sgr a@xmath6 @xcite .
both these last two scenarios imply high energy @xmath3 production . from about 18 h of data ,
hess estimated a similar spectral dependence than for the galactic centre for the snr rx j1713 - 3946 .
the observed photon spectrum by cangaroo was pointed out to be explained by @xmath7 decay and not by electromagnetic processes @xcite , though the result has been considered controversial since the predicted spectrum seems to exceed the egret observed flux from the nearby region .
the spectrum measured by hess does not show the cut - off in the cangaroo spectrum and it is compatible with egret observations .
if the @xmath3 flux were equal to the one measured by hess , event rates expected in antares would be @xmath8 event / yr .
nevertheless , it is clear that the galactic centre region is getting more and more interesting for what concerns the possibility of proton acceleration .
the observation of this region can be done only from the upper hemisphere using @xmath3-induced muons , that is the topology with the best angular resolution in @xmath3 telescopes .
preliminary studies show that antares will perform this observation with unprecedented angular resolution , comparable to recent and forthcoming @xmath2 experiments , such as hess , magic , integral , swift and glast .
these experiments will provide alerts in case of bursting sources , such as gamma - ray bursters ( grbs ) , and clues to select interesting candidates in order to enhance antares sensitivity .
other galactic sources , such as fast rotating neutron stars in snrs , plerions , associations of stars , and magnetars , are foreseen to predict up to hundreds of muon neutrino events in a cubic - kilometer detector and a few events in antares ( for a review see @xcite and references therein ) .
antares effective area for muons reaches 0.05 km@xmath9 for @xmath10 gev .
the most promising models are already constrained by the amanda - ii limits @xcite .
this is the case of the micro - quasar model in @xcite for what concerns the persistent source ss433 , though this peculiar source is surrounded by a nebula and hence the model could be particularly uncertain .
extragalactic sources , such as active galactic nuclei and grbs , if transparent to nucleons , can produce event rates up to from a few to a few hundreds of events in a km@xmath11 telescope . as a matter of fact ,
if the neutrons can escape sources and produce the uhe protons observed by extensive air shower arrays above @xmath12 ev , an upper limit can be derived as firstly shown in @xcite ( hereafter w&b limit ) .
nevertheless scenarios exist that evade the limit by considering other proton spectral dependences than @xmath13 and sources that are not transparent to nucleons ( see ref .
@xcite , hereafter mpr limit ) .
amanda - ii @xcite is already at the level of testing the limit for completely opaque sources in @xcite , while the w&b flux limit will be tested after 1 yr of operation of the km@xmath11 detector icecube @xcite .
it is probable that neutrino detection will allow also to address the problem of the observation of the ultra - high energy cosmic rays ( crs ) , since these could be produced by the same processes in sources .
other granted high energy @xmath3 sources should be cr interactions with the cosmic microwave background and with the galactic interstellar matter ( for the former event rates in antares are of the order of a few events per year ) .
moreover , speculative top - down production processes can be envisaged , where supermassive particles or topological defects decay into neutrinos .
neutrino telescopes detect the cherenkov light emitted by charged particles produced in @xmath3 interactions by 3-d arrays of optical modules ( oms ) .
oms are pressure resistant glass spheres containing phototubes ( pmts ) , located in polar ice or sea / lake water depths in order to reduce the surface @xmath14 flux by orders of magnitude . from the times of pmts hit by the cherenkov light ,
tracks can be reconstructed , while the amplitudes allow an energy reconstruction .
neutrino telescopes were originally optimized to detect upward - going muons from @xmath15 charged current ( cc ) interactions .
their direction of flight allows to discriminate atmospheric muons , since @xmath3s are the only atmospheric shower secondaries capable of crossing the earth . since the @xmath3 cross - sections and the @xmath14 range increase with energy , the effective target mass for @xmath3 interactions becomes larger with energy . as a matter of fact
, the larger the energy the better the performances of this technique , up to a saturation that depends on the detector dimensions .
moreover , for @xmath16 tev the muon has the same direction of the parent neutrino allowing to point sources . in view of recent results on neutrino oscillations ,
it is now considered important to detect other flavor neutrinos .
flavor ratios at sources from meson decays , in the hypothesis that the environment density is such that all muons decay , are @xmath17 @xcite . in light of solar and atmospheric @xmath3 results and of constraints from reactor experiments ,
oscillations through baselines larger than tens of kpc would lead to @xmath18 .
since the @xmath3 cross section increases with energy , cc interactions during propagation through the earth prevent @xmath15 and @xmath19 from reaching the detector ( unless for @xmath15 the muon range is larger than the distance of the @xmath3 vertex to the detector ) .
the earth shadowing becomes important above 1 pev and neutrinos can reach the detector only from the horizon or from the upper hemisphere .
hence for uhe @xmath3 detection it is necessary to have good efficiencies in reconstructing horizontal events , to use energy estimators to suppress the atmospheric @xmath14 background , and discriminate showering events inside the instrumented region from crossing atmospheric @xmath14 tracks .
the case of tau @xmath3s is peculiar : they are never absorbed during propagation through the earth , though they loose energy , since they regenerate in @xmath20 decays @xcite .
they can produce background free topologies , such as the double bang events , where the shower from the @xmath3 vertex and the shower from the electronic and hadronic @xmath20 decay channels are connected by a @xmath20 track , long enough to separate the showers .
these interesting events , that would prove @xmath3 oscillations in an astrophysical beam , are expected to be very rare , depending on the assumed spectrum , due to the fact that a @xmath20 track is @xmath21 m for @xmath22 pev ) .
the antares collaboration was formed in 1996 and comprises physicists , astronomers , engineers and sea experts from france , germany , italy , the netherlands , russia , spain and the united kingdom .
the installation of the detector has started in a 2.5 km deep site about 40 km off - shore toulon ( @xmath2350n,@xmath2410e ) .
the set - up of the telescope is shown in fig .
twelve lines form an octagon with average distances of about 65 m. being 450 m high , they are kept taut by buoys at the top end and by anchors at the bottom . [
cols="^,^ " , ]
antares will observe the southern hemisphere sky using neutrinos .
it will complement the detectors at the south pole and represents an important step to acquire the proper know - how to deploy a km@xmath11 kilometer detector in the mediterranean .
the first line of the detector is planned to be deployed and connected to the junction box in the summer of 2005 . in the meanwhile , two more lines will be deployed .
the milom combines a storey with optical modules , calibration devices ( laser , led beacons and acoustic positioning modules ) , and instruments for the monitoring of optical and mechanical environment of the site .
the so - called line zero , with 25 storeys but no pmts inside glass spheres , will allow to realistically test the electro - mechanical cables and all the containers .
the assembly and test of both lines will validate the final line integration procedure .
|
antares is a project aiming at the operation of an underwater detector at a depth of 2.5 km close to toulon in the south of france .
the detector is expected to be completed at the beginning of 2007 .
the main purpose of the experiment is the detection of high energy neutrinos produced in astrophysical sources .
being weakly interacting , neutrinos could potentially be more powerful messengers of the universe compared to photons , but their detection is challenging .
the technique employs phototubes to detect the arrival time and the amplitude of photons emitted by neutrino charged secondaries due to the cherenkov effect .
antares will contribute significantly in the field of neutrino astronomy , observing the galactic centre with unprecedented pointing capabilities .
|
to collect data needed to provide coastal communities with timely tsunami warnings , the national oceanic and atmospheric administration ( noaa ) has deployed an array of deep - ocean assessment and reporting of tsunamis ( dart^^ ) buoys at strategic locations in the pacific and atlantic oceans ( gonzlez _ et al . _ , 2005 ; titov _ et al .
_ , 2005 ; spillane _ et al .
_ , 2008 ; mofjeld , 2009 ) .
when a tsunami event occurs , data from these buoys are analyzed at u.s .
tsunami warning centers ( twcs ) using the short - term inundation forecast for tsunamis ( sift ) application ( gica _ et al .
_ , 2008 ; titov , 2009 ) .
the sift application was developed by the noaa center for tsunami research to rapidly and efficiently forecast tsunami heights at specific coastal communities .
sift compares dart^^ buoy data with precomputed models as one step in creating the forecast . matching precomputed models with data
requires that any tidal components in the data be either removed or compensated for in some manner , an operation that we refer to as detiding . to facilitate the operational needs of sift , detiding of data from dart^^ buoys
must be done as soon as possible after the data become available .
in the course of developing the sift application , we have entertained multiple methods for detiding dart^^ buoy bottom pressure ( bp ) data nearly in real time . in this paper
we compare five such methods using archived data collected by eleven dart^^ buoys .
these buoys are deployed in both the pacific and atlantic oceans in places with different tidal regimes . besides the dominant tides ,
the archived data contain other bp fluctuations of the kind that would be present during an actual tsunami event .
we take data from periods when no known significant tsunamis occurred and introduce an artificial tsunami signal .
each signal is patterned after a precomputed computer model for an actual tsunami event .
the magnitude of the artificial event is controlled by a source coefficient @xmath0 .
we consider five different methods for extracting the artificial tsunami signal , each of which handles the tidal component in a different manner .
we assess how well each of the five detiding methods allows us to extract the known @xmath0 . by repeating this scheme for many different combinations of dart^^ data and artificial tsunami signals
, we can evaluate how well the five detiding methods work under idealized conditions ( thus , while we make use of observed tidally - dominated data , we do not take into consideration confounding factors , an important one being a mismatch between the actual tsunami event and our model for it ) .
the remainder of the article is organized as follows . in sect .
[ sec : bpmeasurements ] we review the format of bp measurements from dart^^ buoys as received by u.s .
we next describe construction of simulated tsunami events , which are formed by adding archived data from eleven representative buoys to associated models for tsunami signals ( sect .
[ sec : scenarios ] ) .
we give details about the five methods for extracting tsunami signals in sect .
[ sec : detidingpressuremeasurements ] .
the first two are well - known methods based on harmonic models .
the next two are non - standard linear filters that utilize either empirical orthogonal functions or kalman smoothing in conjunction with a local - level state space model . the final method , which is based on a regression model that includes terms for both the tsunami signal and tidal components , proves to be the method of choice in study described in sect . [ sec : results ] .
we compare the five methods using dart^^ buoy data collected during the devastating 2011 japan tsunami in sect .
[ sec : tohoku ] .
we state our conclusions and discuss our results in sect .
[ sec : conclusions ] .
a dart^^ buoy actually consists of two units : a surface buoy and a unit located at the bottom of the ocean with a pressure recorder ( gonzlez _ et al .
_ , 2005 ; noaa data management committee , 2008 ; mofjeld , 2009 ) . the bottom unit stores measurements of water pressure integrated over non - overlapping 15-sec time windows , for a total of @xmath1 measurements every hour .
these internally recorded measurements only become fully available when the bottom unit returns to the surface and is recovered for servicing ( time between servicings can be as long as two or three years ) .
we refer to the internally recorded data as the 15-sec stream .
normally the buoy operates in standard reporting mode in which the bottom unit packages together one measurement every 15 minutes ( a 60 fold reduction in data ) over a 6-h block for transmission via acoustic telemetry up to the surface buoy once every 6 h. the surface buoy then relays the data up to the iridium satellite system for dissemination to the outside world .
we refer to data collected in standard reporting mode as the 15-min stream .
when a typical tsunami event occurs , the bottom unit detects a seismic event , which causes the dart^^ buoy to go into event reporting mode .
the transmission of the 15-min stream is suspended while the dart^^ buoy is in event reporting mode . as part of this mode , the bottom unit averages together four consecutive 15-sec measurements and transmits these averages up to the surface buoy .
we refer to these data as the 1-min stream .
an additional 2 hours of 1-min data are transmitted on the hour during the event reporting mode .
when the outside world first gets access to the 1-min stream , there can be a gap between it and the most recent value of the 15-min stream ranging up to almost 6 h. after the on - hour data transmission an additional 1 to 2 hours of 1-min data before the event will be available .
if we assume that tsunami events commence at random within the 6-h reporting cycle for the 15-min stream , then the average size of this gap will be 3 h. the data that are thus typically available during a tsunami event are a portion of the 15-min stream prior to the event and a 1-min stream that becomes available piece - by - piece in real time as the event evolves and that contains at least 1 h of data occurring before the event .
the goal of this paper is to objectively compare different detiding methods for extracting tsunami signals in near real - time .
a model for what is recorded by a dart^^ buoy during a tsunami event has three components : a tsunami signal , tidal fluctuations and background noise .
the latter is due to seismic , meteorological , measurement and other nontidal effects ( cartwright _ et al .
_ , 1987 ; niiler _ et al .
_ , 1993 ; webb , 1998 ; cummins _ et al .
_ , 2001 ; mofjeld _ et al . _ , 2001 ; zhao and alford , 2009 ) .
the purpose of detiding is to compensate for tidal fluctuations in the recorded data ( consoli _ et al .
_ , 2014 ) .
although dart^^ buoys have recorded a number of tsunami events , use of these data to evaluate different detiding methods is problematic : we ideally need to know the true tidal fluctuations , and these are not known to sufficient accuracy for actual data due to the presence of background noise .
one solution is to simulate each of the three components . in combining these components to simulate tsunami events
, we would then know the true tidal fluctuations and thus be in a position to quantify how well different detiding methods did ; however , simulating tidal fluctuations and background noise in a manner that does not give an unfair advantage to certain detiding methods is tricky .
we can bypass simulation of both tidal fluctuations and background noise by making use of actual bp measurements from dart^^ buoys under the realistic assumption that significant tsunami signals are rare events .
in this paper we use archived 15-sec streams retrieved from eleven representative dart^^ buoys .
the buoys are in the atlantic and pacific oceans and are listed in table [ tab : buoys ] .
their locations are displayed in fig .
[ fig : detidingpaperchart ] .
these particular buoys were chosen because the data they collect represent four well - known types of tides , hence offering the study of detiding over a broad range of ocean tidal conditions .
all data were obtained from noaa s national geophysical data center ( ngdc ; the reader is referred to mungov , _ et al . _ , 2012 , for details about data collection and processing ) .
the units for the data archived by ngdc are in pounds per square inch absolute ( psia ) , which we convert to water depth in meters by multiplication by @xmath2 m / psia ( this conversion factor is based on a standard ocean , but its value does not impact any of the results we present ) . as documented in table [ tab : buoys ] , we have from 321 to 998 days of 15-sec streams for each of the eleven buoys .
we use these streams to construct ` scenarios ' mimicking the tidal fluctuations and background noise that might have been present in 15-min and 1-min streams available during an actual tsunami event . to do so for a particular buoy
, we start by selecting a random starting time @xmath3 . the first part of the scenario associated with @xmath3 consists of a 29-day segment of a 15-min stream extracted by subsampling from the 15-sec stream immediately prior to @xmath3 . to mimic typical operational conditions , we create a 3-h gap prior to @xmath3 by eliminating part of the constructed 15-sec stream . using data from the 15-sec stream occurring immediately after @xmath3 , we form a 1-day segment of a 1-min stream by averaging four adjacent values .
the constructed 15-min and 1-min streams constitute one scenario . for each buoy
we form a total of 1000 scenarios with starting times @xmath3 chosen at random without replacement from the set of all possible starting times ( for example , as indicated in table [ tab : buoys ] for buoy 21416 , this set ranges from 29 days after 07/25/2007 up to 1 day prior to 07/01/2009 ) .
figure [ fig : scenario ] shows an example of one scenario .
( there are occasional missing values in the 15-sec streams archived by ngdc .
these can lead to gaps in a constructed 15-min stream in additional to the usual 3-h gap prior to @xmath3 .
all of the detiding methods in our study can handle these gaps . if , however , the missing values lead to gaps in the constructed 1-min stream , we have elected to disregard the chosen @xmath3 and randomly pick a new one merely to simplify evaluation of the performance of the various detiding methods .
thus , a given scenario can have gaps at arbitrary times in its 15-min stream , but none in its 1-min stream . ) with tidal fluctuations and background noise being handled using actual data from dart^^ buoys , we now turn our attention to simulating tsunami signals and combining these with the other two components to simulate the 1-min streams observed during a tsunami event .
let @xmath4 be a vector containing a simulated 1-min stream observed during a tsunami event starting at time @xmath3 and lasting for one day ( the bar over @xmath5 is a reminder that the 1-min stream is produced from four - point averages of the 15-s stream ) .
we model this vector as @xmath6 where @xmath7 , @xmath8 , , @xmath9 and @xmath10 are vectors representing , respectively , tidal fluctuations , components derived from @xmath11 unit sources ( these collectively serve to model the tsunami signal ) and background noise , while @xmath12 , , @xmath13 are @xmath14 nonnegative scalars known as source coefficients ( percival _ et al . _ ,
2011 ) . forming the @xmath15 s
is discussed in detail in titov _
et al . _ ( 1999 ) and gica _ et al . _ ( 2008 ) , and the following overview is taken from percival _ et al . _ ( 2011 ) .
tsunami source regions are defined along subduction zones and other portions of oceans from which earthquake - generated tsunamis are likely to occur ( see the solid curves on fig .
[ fig : detidingpaperchart ] ) .
each source region is divided up into a number of unit sources , each of which has a fault area of @xmath16 @xmath17 . for each pairing of a given dart^^ buoy with a particular unit source , a model
@xmath15 has been calculated predicting what would be observed at the buoy from time @xmath3 and onwards based upon the assumption that the tsunami event was caused by a reverse - thrust earthquake with standardized moment magnitude @xmath18 starting at @xmath3 and located within the unit source .
the source coefficient @xmath19 is used to adjust the standardized magnitude in the model to reflect the magnitude in an actual tsunami event .
the sum of the @xmath19 s is a measure of the overall strength of the tsunami event and provides initial conditions for models that predict coastal inundation . as discussed in percival _
et al . _ ( 2011 ) , given @xmath14 candidate unit sources and based on varying amounts of data , a least squares procedure can be used to obtain statistically tractable estimates of the @xmath19 s under the assumption that the tidal fluctuations @xmath7 are known to reasonable accuracy . to focus on detiding , we assume a simplified version of equation [ eq : modelk ] , namely , @xmath20 i.e. , the simulated 1-min stream involves just a single unit source and a single source coefficient .
the scenarios for a given buoy are stand - ins for @xmath21 .
we take the unit source - based @xmath22 and multiply it by @xmath0 to form an artificial tsunami signal @xmath23 . from an examination of actual tsunami events
, we set @xmath24 as a representative source coefficient .
figure [ fig : oneevent ] shows an example of constructing a simulated tsunami event based upon the 1-min stream shown in fig . [ fig : scenario ] and a unit source chosen for dart^^ buoy 52402 . for each of the eleven buoys in our study
, we picked from three to seven unit sources with different orientations with respect to the buoy
these choices are listed in table [ tab : unitsources ] and depicted as solid circles in fig .
[ fig : detidingpaperchart ] . for dart^^ buoys near a subduction zone , we selected the closest unit source so that the buoy would be in the main beam of the radiated tsunami energy .
we then added two additional unit sources , bearing 30 to 45 degrees to either side of the first , to represent waves traveling obliquely toward the buoy . in the case of dart^^
buoy 21416 , which is representative of the northwest pacific and proximate to both the aleutian and kamchatka source regions , we selected unit sources from both regions .
buoy 32411 ( located close to both central and south america ) has a similar geometry , so we chose three unit sources from central america and three from south america . for mid - ocean buoys ( 51406 , 51407 and 44401 ) , we selected unit sources from several subduction zones to which the buoy might respond .
there are 42 unit sources in all , 5 of which are used by two buoys , for a total of 47 pairings of buoys and unit sources .
each pairing leads to a different artificial tsunami signal @xmath25 .
twelve representative signals are shown in fig .
[ fig : unitsources ] ( here the coefficient @xmath0 is adjusted separately for each signal merely for plotting purposes the actual range for all 47 @xmath26 s is listed in table [ tab : unitsources ] and varies from 0.2 to 13.1 cm ) .
each plot shows a 120-min segment of a given artificial tsunami signal @xmath25 , one point for each minute , but a different segment for each signal .
for use later on , five hand - picked points are colored red . from left to right , the first four of these points mark the approximate occurrence of a quarter , a half , three - quarters and all of the first full wave comprising the signal ( table [ tab : unitsources ] lists the actual times associated with these points ) . the final ( right - most ) point marks one hour past the end of the first full wave .
each of the eleven buoys is represented once in fig .
[ fig : unitsources ] , except for buoy 32411 , which has two signals associated with it that are visually quite different ( left - hand and middle plots on second row ) .
one of these two depicted signals ( cs027b , left - hand plot , second row ) is generated from the same @xmath16 @xmath17 region as the signal depicted for 51406 ( left - hand plot , bottom row ) .
again these two signals are visually quite different .
these duplicate - buoy / duplicate - unit - source pairings illustrate the fact that each signal depends upon both the location of the buoy and the location of the unit source .
figure [ fig : tsunamievents ] shows 120-min segments of twelve simulated tsunami events , each of which make use of the artificial tsunami signals shown in corresponding plots of figure [ fig : unitsources ] .
these events were formed by adding @xmath27 ( with @xmath0 now set to 6 ) to the 1-min stream from a randomly chosen scenario for each buoy ( the scenario chosen for buoy 52402 is the same one used in fig . [
fig : oneevent ] thus the left - most plot in the third row of fig .
[ fig : tsunamievents ] is a zoomed - in version of bottom plot of fig .
[ fig : oneevent ] ) .
the vertical axis for each plot in fig . [ fig : tsunamievents ] spans 1.2 m .
it is easy to visually pick out the tsunami signal in some plots , but harder in others , which illustrates the fact that , even though @xmath28 in all cases , the signal - to - noise ratio varies substantially .
here we describe five methods that take bp measurements @xmath4 and use them to estimate the source coefficient @xmath0 in equation [ eq : model1 ] .
each method deals with the tidal fluctuations @xmath29 in a different manner .
the first four methods do so by predicting or estimating the fluctuations using , say , @xmath30 .
this prediction is then subtracted from @xmath4 to yield detided bp measurements : @xmath31 where @xmath32 is an error term encompassing both background noise and inaccuracies in predicting the tidal fluctuations .
given the detided measurements , we then use the ordinary least squares ( ols ) method to estimate @xmath0 , yielding the estimator @xmath33 where @xmath34 denotes the transpose of the vector @xmath22 .
the fifth method is different from the other four in that it uses ols to create @xmath35 and to estimate @xmath0 jointly . the inversion algorithms for estimating source coefficients from multiple buoys currently in sift ( percival , _ et al .
_ , 2011 ) and under development ( percival , _ et al .
_ , 2014 ) are based on least squares methods , but go beyond ols by including nonnegativity constraints and penalties to induce automatic unit source selection .
these algorithms assume the model of equation [ eq : modelk ] for the multiple buoys , but proceed under the assumption that the bp measurements have been adjusted so that tidal fluctuations have been removed as much as possible .
for studying how best to deal with tides , it suffices to use a single buoy and the simplified model of equation [ eq : model1 ] , and there is no real gain in using anything other than the ols estimator of equation [ eq : alphahat ] .
the first two of the five methods are based on harmonic modeling , which we describe in sect .
[ sec : harmonic ] .
the third method is based on empirical orthogonal functions ( sect .
[ sec : eof ] ) , while the fourth employs kalman smoothing , but makes use of harmonic modeling for initialization purposes ( sect .
[ sec : ks ] ) .
[ sec : joint ] describes the fifth method , which in part involves a simplified version of harmonic modeling .
the harmonic prediction method is the standard one used by noaa to predict the tides at coastal stations . for detiding dart^^ data ,
it can be thought of as the classic method of tidal prediction .
it assumes that the tides are sums of sinusoidal constituents , each with its own frequency . to make tidal predictions at a dart^^ station , the amplitude and phase lag of each constituent
are determined via a tidal analysis of observed bp data .
the first detiding method is a harmonic prediction method that carries out the tidal analysis in the following manner .
consider one of the scenarios described in section [ sec : scenarios ] .
each scenario consists of a one - day segment of a 1-min stream starting at time @xmath3 , which is preceded by 29 days ( less 3 hours ) of a 15-min stream . without loss of generality , set @xmath36 to simplify the discussion , and let @xmath37 denote an observation from the 15-min stream , but with @xmath38 indexing the underlying 15-sec stream from which the 15-min stream is subsampled ; i.e.
, the actual time associated with @xmath37 is @xmath39 , where @xmath40s .
assuming @xmath37 to be tidally dominated , we entertain a harmonic model of the form @xmath41 + e_n \label{eq : harmmodelbc}\end{aligned}\ ] ] for @xmath42 ( note that index @xmath43 corresponds to the last value in the observed 15-min stream , which occurs prior to the 3-h gap , while @xmath44 indexes the first value , which occurs 29 days prior to @xmath36 ) . in the above
@xmath45 is an unknown overall mean level ; @xmath46 is one of @xmath47 known tidal frequencies ; @xmath48 and @xmath49 are unknown coefficients that can be used to deduce the amplitudes @xmath50 and phase lags @xmath51 ; and @xmath52 is a residual term ( hopefully small ) .
we use an ols fitting procedure to estimate @xmath45 , @xmath48 and @xmath49 via , say , @xmath53 , @xmath54 and @xmath55 .
if we replace @xmath45 , @xmath48 and @xmath49 by their estimates , we can then use the right - hand side of equation [ eq : harmmodelbc ] with the residual term set to zero to predict what the tidal fluctuation should be at any desired time index @xmath38 .
after an artificial tsunami signal is added to the 1-min stream of the scenario to form a tsunami event , we detide this stream by forming @xmath56\,\delta ) + c_m \sin(\omega_m [ n+k]\,\delta ) \right]\ ] ] for @xmath57 , where @xmath58 is an element from the 1-min stream @xmath4 . for 29 days of prior data ,
we make use of @xmath59 tidal frequencies @xmath60 , , @xmath61 commonly referred to as n2 , m2 , s2 , q1 , o1 and k1 ( see , e.g. , table 1 in ray and luthcke , 2006 ) .
the @xmath62 s given above form the elements of the vector @xmath35 used to form the estimator @xmath63 of equation [ eq : alphahat ] . the first row of fig .
[ fig : detidingfive ] shows detided bp measurements @xmath62 corresponding to the simulated tsunami event shown in the bottom plot of fig .
[ fig : oneevent ] .
visually there is evidence that this detiding is not entirely satisfactory : there is a systematic drift downwards over the first hour that arguably is due to tidal fluctuations and thus should not be present in @xmath62 .
as described in the caption to the figure , the five black circles mark various time points associated with the artificial tsunami signal .
if we place the data from @xmath3 up to one of these five time points into the vector @xmath35 and create the corresponding vector @xmath26 , we can obtain an estimate of the source coefficient @xmath0 using equation [ eq : alphahat ] .
the resulting estimates @xmath63 are listed in the first row of table [ tab : alphaestimates ] .
recalling that the true value of @xmath0 is 6 , we see that , not unexpectedly , the better estimates are associated with larger amounts of data . looking at estimates based on varying amounts of data is of considerable operational interest .
as more bp measurements become available as a tsunami event evolves , we can expect in general to get better estimates of @xmath0 , but at the expense of a delay in issuing timely warnings .
determining how much an estimate of @xmath0 is likely to improve by waiting for more data is vital for managing the trade - off between accuracy and timeliness .
for this example , there is improvement in waiting until the first full wave occurs , but none in waiting an hour past that time .
the second detiding method is based on a harmonic analysis that , for a given buoy , is based on _ all _ the 15-s data listed for it in table [ tab : buoys ] .
this ` blanket ' harmonic model has the same form as equation [ eq : harmmodelampphase ] , but now the time index @xmath38 increases in steps of one rather than 60 , and we use @xmath64 sinusoidal constituents . for optimal accuracy of this type of an analysis , the measurements should span at least one year , which table [ tab : buoys ] indicates is true for all buoys with the exception of 41420 ( this buoy has 321 days of data , slightly less than a year ) .
the tidal predictions are made by adjusting the lunar harmonic constants for perigean ( 8.85-year ) and nodal ( 18.6-year ) variations in the moon s orbit , computing the height associated with each constituent at the times of interest , and then summing these heights to yield the prediction .
detiding of the 1-min series in a scenario is accomplished using an equation similar to equation [ eq : detideoneminuteusingha ] , the only difference being that the overall mean level @xmath45 is estimated using the scenario s 15-min stream rather than being pre - specified .
a particular strength of the second detiding method is that it requires minimal use of the 15-min stream ( it is only used to estimate @xmath45 ) .
in one extreme case , this stream was entirely missing for buoy 21416 over a six - week period prior to 1/15/2009 , when a kuril islands event triggered reporting of the 1-min stream .
if the 15-min stream is not available , it is possible to estimate @xmath45 using just the 1-min stream , but care would be needed to ensure that the tsunami signal does not unduly distort the estimate .
a number of software packages are available to do tidal analyses of observations and to make the tidal predictions , a standard one being the foreman fortran 77 package ( foreman , 1977 , revised 2004 ) . for our study , we used tidal predictions generated by ngdc ( see mungov _
et al . _ , 2012 , for details ) . the second row of fig .
[ fig : detidingfive ] shows detided bp measurements @xmath62 produced by this second harmonic method .
in contrast to the first method , we no longer see a systematic drift downwards over the first hour ; however , the @xmath62 s during that hour are systematically elevated above zero , which is questionable .
the five estimates @xmath63 corresponding to different amounts of data are listed in the second row of table [ tab : alphaestimates ] .
these estimates are worse than the ones we obtained from the first harmonic method except when using the largest amount of data ; however , we should nt rely on this single example to draw conclusions about the relative merits of the two harmonic methods see the discussion in section [ sec : results ] .
when tidal predictions are subtracted from bp measurements , fluctuations always remain in the tidal bands .
they are due to non - stationary fluctuations , non - linear tides , and tidal constituents not accounted for in the tidal analysis . of these ,
internal tides are certainly significant .
they are generated around the ocean margins and shallow ridges and then propagate elsewhere in the ocean ( e.g. , cummins _
et al . _ , 2001 ; zhou and alford , 2009 ) .
the residual tides limit the degree to which the total tide can be removed from bp data through simple subtraction of a predicted tide , even when the tidal analysis is performed on the same time series ( as is the case here ) .
the two best known approaches for detiding data are to subtract off a prediction from a harmonic model ( as described in the previous section ) and to subject the data to a linear time - invariant ( lti ) high - pass filter . to isolate a tsunami signal without distortion , the high - pass filter should retain components with periods as long as 2 hrs .
the so - called ` edge effects ' of such a filter distort at least 1-hr sections at the beginning and at the end of the tsunami signal .
thus an lti filter can not reliably isolate a tsunami signal immediately after it is registered by a dart^^ buoy , whereas it is desirable to make use of this data as soon as possible following the start of a tsunami event .
moreover , most digital filters are designed to work with regularly sampled data and are not easy to adapt if there are gaps in the data . while the bp unit in a dart^^ buoy internally records a measurement once every 15 s , only gappy segments of these data ( or 1-min averages thereof ) are typically available externally following an earthquake event . in this section and
the next , we explore two approaches to detiding involving linear ( but not time - invariant ) filters .
these approaches are tolerant of data gaps and are less prone to edge effects , thus overcoming the two problems we noted about standard lti high - pass filters .
the first approach is based on extracting the tidal component in a segment of data by back and forth projection onto a specific sub - space in an @xmath65-dimensional space of vectors .
the sub - space is spanned by the empirical orthogonal functions ( eofs ) of segments of length @xmath65 of archived 15-min streams from dart^^ buoys ( tolkova , 2009 ; tolkova , 2010 ) .
the following description of this approach is based on tolkova ( 2010 ) , to which we refer the reader for more details .
the eof method relies on the premise that , due to the structure of tides in the deep ocean , the sub - space spanned by the leading eofs of tidally dominated data segments of up to 3-days in length is fairly universal across various dart^^ buoys .
oceanic tidal energy is concentrated in the long - period , diurnal , and semidiurnal frequency bands centered around 0 , 1 , and 2 cycles per day ( cpd ) .
the effective diurnal band is from 0.8 to 1.1 cpd , and the semidiurnal , from 1.75 to 2.05 cpd , so the bandwidth for both bands is 0.3 cpd ( munk and cartwright , 1966 ) .
tidal motion thus has two inherent time scales : one day ( the apparent tidal quasi - period ) and 3.3 days ( the shortest tidal segment from which we can in theory resolve individual constituents within either of the two major bands ) .
since tides at different locations differ only in the fine structure of the tidal bands , the premise behind the eof method says that the sub - space spanned by the leading eofs of tidal segments is essentially the same at all dart^^ locations as long as the segment length is so short as to not allow resolution of the fine structure in the tidal bands .
tolkova ( 2010 ) computed the basis of the eof sub - space of tidally dominated data encompassing one lunar - day ( 24 h 50 min ) using an ensemble of 250 segments taken from dart^^ buoy 46412 in 2007 .
each segment consists of 99 readings sampled at 15 min intervals .
the eofs associated with the seven largest eigenvalues , complemented with a constant vector with elements all equal to @xmath66 , provide the basis for a sub - space spanned by @xmath67 vectors of dimension @xmath68 .
this sub - space is sufficient to capture the bulk of tidal variations in segments of length 99 from a 15-min stream recorded by any dart^^ buoy . to process 1-min streams within the sift system , the @xmath47 vectors were re - sampled to a 1-min interval , yielding vectors of length @xmath69 .
re - orthogonalization and re - normalization produces vectors @xmath70 , @xmath71 , satisfying @xmath72 for @xmath73 and @xmath74 for @xmath75 .
figure [ fig : eighteofs ] shows the resulting eight vectors .
three steps are needed to accomplish detiding using the @xmath70 vectors . for the sake of argument , suppose the vector @xmath76 contains a segment from a 1-min stream of length @xmath77 ( i.e. , approximately one lunar day ) .
first , we project this segment onto the vectors @xmath70 to obtain @xmath47 coefficients @xmath78 .
second , we estimate the tidal component by taking the @xmath47 vectors , multiplying them by their corresponding coefficients and then adding together the resulting scaled vectors . finally , this estimate of the tidal component is subtracted from @xmath76 , yielding the detided data @xmath35 .
mathematically , we can write @xmath79 where @xmath80 is a @xmath81 matrix whose columns are @xmath82 , and @xmath83 is a vector containing the @xmath47 coefficients . letting @xmath58 and @xmath84 denote , respectively , the @xmath38th element of @xmath76 and the @xmath85th element of @xmath80 and momentarily regarding @xmath86 as a function of @xmath87 , we note that setting @xmath88 equal to @xmath89 results in minimizing the above sum of squares . we can generalize this technique to handle the case of irregular sampling by the following simple procedure , which ignores the distinction between the 1-min and 15-min streams . for a span of @xmath77 minutes of interest , construct a vector @xmath90 of length @xmath65 that contains all available values from either one of the streams .
let @xmath91 if the @xmath38th element @xmath92 of @xmath90 is actually available from one of the streams , and let @xmath93 if it is not available .
we set @xmath94 such that @xmath95 is minimized with respect to @xmath96 .
minimization of ( [ eq : aminim ] ) is a least squares problem whose associated normal equations are @xmath97 where @xmath98 is an @xmath81 matrix whose @xmath99th element is @xmath100 .
the system ( [ eq : aequ ] ) has a numerically viable solution as long as the symmetric @xmath101 matrix @xmath102 is not poorly conditioned . in this case the detided data are contained in @xmath103 the third row of fig .
[ fig : detidingfive ] shows detided bp measurements @xmath62 produced by the eof method .
in contrast to the two methods based on harmonic analysis , the values that are subtracted from the 1-min stream to accomplish eof detiding depend upon the stream itself and hence can change as different amounts of this stream are utilized , as this example illustrates . in particular
we see a systematic upward drift when utilizing data less than or equal to the first full wave , but this drift disappears when we use data one hour past the end of the first full wave .
the five estimates @xmath63 corresponding to different amounts of data are listed in the third row of table [ tab : alphaestimates ] .
for the three largest amounts of data , these estimates are an improvement in this example over the ones we obtained from the two harmonic methods .
kalman smoothing ( ks ) has been used in numerous applications as a method for optimally smoothing time series as new values of the series become available over time .
the optimality of this procedure is contingent upon our ability to adequately describe the underlying dynamics of the time series in terms of a so - called state space model .
ks - based detiding has been advocated before ( see consoli _
et al . _ , 2014 , and references therein ) ,
but the approach we describe here for detiding data from dart^^ buoys differs from previous approaches in important aspects ( for detailed expositions on ks , see brockwell and davis , 2002 ; durbin and koopman , 2012 ; shumway and stoffer , 2011 ) .
our ks approach is a two - stage procedure .
as before , let @xmath58 represent the 1-min stream from a given scenario to which we have added an artificial tsunami signal .
the first stage is to detide @xmath58 using the first harmonic modeling method , yielding the detided series @xmath62 , @xmath104 , via equation [ eq : detideoneminuteusingha ] . merely to simplify equations that follow ,
we reindex this first - stage detided series by defining @xmath105 , @xmath106 .
the second stage applies ks to @xmath107 , for which we assume a local level model ( also called a random walk plus noise model ; see brockwell and davis , 2002 ; durbin and koopman , 2012 ) .
this model consists of two equations , the first of which is known as the state equation , and the second , the observation equation .
the state equation takes the form @xmath108 where @xmath109 is the initial state variable , and @xmath110 is a white noise sequence with mean zero and variance @xmath111 .
the observation equation takes the form @xmath112 where @xmath113 is another white noise sequence with mean zero , but now with time - varying variance @xmath114 ( the sequences @xmath113 and @xmath110 are uncorrelated ) .
the intent with this model is to use @xmath115 to track any tidal component left over in the first - stage detiding @xmath107 and to compensate for the presence of the tsunami signal by adjusting @xmath114 appropriately .
based upon the initial state variable @xmath109 , the first - stage detided data @xmath116 and the parameters @xmath111 and @xmath114 , there is an elegant set of equations known as the kalman recursions that give the best ( in the sense of minimum mean square error ) linear estimates of the unknown state variables @xmath117 .
we denote these estimates as @xmath118 . for a fixed index @xmath119 , the estimate @xmath120 of @xmath121 changes as @xmath38 increases , i.e. , as more and more of the first - stage detided series becomes available . given @xmath116
, we define the ks - based detided series to be @xmath122 the corresponding estimate of the source coefficient @xmath0 is given by equation [ eq : alphahat ] , where now the vector @xmath35 contains the @xmath123 variables . the unknown parameters that we must set to implement ks - based detiding are @xmath109 , @xmath114 and @xmath111 . for purposes of this paper
, we just use @xmath124 to estimate @xmath109 ; however , to offer some protection against rogue values , the operational version of sift has an option for using the median of @xmath125 .
our estimate @xmath126 of @xmath114 is the sample variance of the seven variables @xmath127 , with estimates for the first three variances @xmath128 , @xmath129 and @xmath130 being set to @xmath131 ( estimates for the last three variances are handled in an analogous manner ) .
we determined a setting for the final parameter @xmath111 in the following manner .
for all 1000 scenarios for a given pairing of a buoy and an artificial tsunami signal with the source coefficient @xmath28 , we used the ks detiding method to compute source coefficient estimates @xmath132 , @xmath133 , over a grid of values for @xmath111 and for various amounts of data .
we then determined which value minimized @xmath134 .
different pairings of buoys and signals and different data lengths led to different minimizing values .
we set @xmath135 after considering a large collection of representative pairings and data lengths while this value was not optimal for all such pairings and lengths , it performed well overall . the fourth row of fig .
[ fig : detidingfive ] shows detided bp measurements @xmath136 produced by the ks method . the starting point for this method
is the detided series shown in the first row , which was produced by the first harmonic modeling method .
a comparison of these two detided series shows that the ks method has eliminated the downward trend that is evident in the first seventy minutes of the detrended series produced by harmonic modeling .
in contrast to the first three methods , there is no evidence in the ks detided series of a drift or offset that might be attributable to lingering tidal fluctuations .
similar to eof detiding , ks detided values can change as different amounts of the 1-min stream are utilized , but the changes in the ks method are smaller than those for the eof method in this example .
the five ks - based estimates of @xmath63 corresponding to different amounts of data are listed in the fourth row of table [ tab : alphaestimates ] .
for this example , these estimates are closer to the true value @xmath28 than the ones corresponding to the first three methods .
the four detiding methods we have considered so far are similar in that they all produce a detided series @xmath35 .
we then use @xmath35 to produce an estimate @xmath63 of the source coefficient via the ols estimator of equation [ eq : alphahat ] .
the fifth and final method estimates the tidal component jointly with the source coefficient based on just the 1-min stream ( merely to simplify the description below , we assume this stream to be gap - free , but it is easy to reformulate this method to handle gaps ) .
this joint estimation method is based on the model @xmath137 where @xmath76 is an @xmath65 dimensional vector containing a portion of the 1-min stream from a scenario to which the artificial tsunami signal @xmath138 has been added ; @xmath139 is a vector of ones ; @xmath140 is a vector whose elements are @xmath141 , @xmath142 , with @xmath143 and @xmath60 being , respectively , the tidal frequency m2 and half that frequency and with @xmath144 min ; @xmath145 is analogous to @xmath140 , but with sines replacing cosines ; @xmath146 is a vector of errors ( presumed to have mean zero and a common variance ) ; and @xmath45 , @xmath147 , @xmath148 , @xmath149 , @xmath150 and @xmath0 are unknown parameters . in essence
, this method estimates the tidal component via a two - constituent harmonic model ( as alternatives to this model , we also considered harmonic models with other than two constituents and polynomial models of various orders , but the two - constituent harmonic model worked best overall for joint estimation of @xmath0 ) .
equation [ eq : joint ] can be rewritten as @xmath151 where @xmath152 is an @xmath153 design matrix whose columns are @xmath139 , @xmath154 , @xmath155 , @xmath156 , @xmath157 and @xmath22 , while @xmath158 is a vector of coefficients .
the ols estimate @xmath159 of @xmath160 satisfies the normal equations @xmath161 subject to the invertibility of @xmath162 ( in the study discussed in sect .
[ sec : results ] , no instances of non - invertibility were encountered ) .
we can take the detided series for this method to be @xmath163 a strength of this detiding method is that , in contrast to the other four methods , it does not make any use of the 15-min stream , which , as we ve noted before , was entirely missing in one actual tsunami event . the fifth row of fig .
[ fig : detidingfive ] shows detided bp measurements @xmath164 produced by the joint estimation method , with the five estimates of @xmath63 corresponding to different amounts of data being listed in the fifth row of table [ tab : alphaestimates ] .
although in this example the detided series for this and the ks method are visually similar to each other , the joint estimation - based @xmath63 estimates are always closer to the true value of @xmath28 than the ks - based estimates .
here we compare the five detiding methods described in the previous section by considering how well each method estimates the source coefficient @xmath0 from simulated tsunami events constructed as per equation [ eq : model1 ] ( an example of one such event is shown in fig .
[ fig : oneevent ] ) . for each of the 47 buoy / unit source pairings listed in table [ tab : unitsources ]
, we constructed 1000 simulated events based upon the 1000 scenarios created for each buoy .
for each such event and for five different amounts of data from the 1-min stream ranging from a quarter of the first full wave ( 1/4 ffw ) of the tsunami signal up to an hour past the end of the ffw , we estimated @xmath0 using the five detiding methods , thus yielding @xmath165 estimated coefficients for each method .
the task at hand is to summarize how well each method did .
we start by considering results for the pairing of buoy 52402 with unit source ki060b ( this same combination is used in all or part of figs .
[ fig : scenario ] to [ fig : detidingfive ] and in table [ tab : alphaestimates ] ) . figure [ fig : scatterfive ] has five rows , one for each of the five methods .
the dots in a given row show the 1000 estimates @xmath63 derived from data just up to 3/4 ffw .
ideally we would like to see estimates that cluster tightly around the true value @xmath24 ( indicated by a blue dashed line ) .
the scatter in the estimates for the two methods based on harmonic analysis ( top two rows ) is much larger than that for the three remaining methods . with only a few exceptions ,
this pattern persists for all 47 buoy / unit source pairings and for all five amounts of data and tells us that these two methods are not competitive with the other methods . the distribution of the estimates in each row of fig .
[ fig : scatterfive ] is summarized on the right - hand side by a boxplot ( chambers _ et al .
_ , 1983 ) .
the central box in each boxplot has three horizontal lines , which indicate , from bottom to top , the lower quartile , the median and the upper quartile of the data .
the short horizontal line below the central box is the lower hinge , which indicates the estimate @xmath63 that is closest to but not less than the value of the lower quartile minus 1.5 times the interquartile distance ( the upper quartile minus the lower quartile ) . the upper hinge has a similar interpretation , with ` not less than ' replaced by ` not greater than ' and with ` lower quartile minus ' replaced by ` upper quartile plus ' .
any estimates @xmath63 that happen to be either smaller than the lower hinge or greater than the upper hinge are indicated by circles . because the variability in the estimates for the first two methods is so much greater than those for the three remaining methods , only the boxplots for the former are clearly visible in fig .
[ fig : scatterfive ] .
figure [ fig : boxplots ] shows boxplots for the just latter three methods on a common scale , but now for all five amounts of data rather than just data up to 3/4 ffw .
these boxplots show that the eof estimates tend to be biased low when using data less than or equal to 3/4 ffw , whereas the ks estimates are biased high , but to a lesser degree ( the boxplots show medians , but there would be no noticeable differences had we shown sample means rather medians in fig . [
fig : boxplots ] ) .
by contrast there is little evidence of bias in the joint - estimation estimates . for estimates based on data up to the ffw or smaller amounts ,
the spreads of the distributions are generally greatest for the eof estimates and smallest for the joint - estimation estimates .
figure [ fig : onermse ] summarizes the spreads in the distributions involved in figs .
[ fig : scatterfive ] and [ fig : boxplots ] via root - mean - square errors ( rmses ) , i.e. , @xmath166 ) . by this measure
the two methods based on harmonic analysis are about an order of magnitude worse than the best method ( joint estimation ) . with almost no exceptions
, this poor rmse performance persists through all 47 buoy / unit source pairings and all five amounts of data under study .
the joint estimation method outperforms the eof and ks methods , but the former becomes competitive when the largest amount of data is used , and the latter , for the smallest and two largest amounts . increasing
the amount of data from 1/4 ffw to ffw results in approximately half an order of magnitude drop in rmse for all five methods ( table [ tab : unitsources ] indicates that the time it takes to collect the extra data from 1/4 ffw to ffw is 14 min for this particular buoy / unit source pairing ) .
figure [ fig : stackedrmses ] is similar to fig .
[ fig : onermse ] , but now shows rmse plots for the twelve representative buoy / unit source pairings shown in figs . [ fig : unitsources ] and [ fig : tsunamievents ] . to simplify this figure , we do not show results for the two non - competitive methods based upon harmonic analysis .
the joint estimation method tends to outperform the eof and ks methods when more than 1/4 ffw of data is involved , but not uniformly so ( the pairing of buoy 51407 with unit source cs100b in the middle plot on the bottom row offers a counterexample ) .
the ks method generally outperforms the eof method .
there are a number of instances in which rmse increases for the eof and ks methods when the amount of data increases from 1/4 ffw to 1/2 ffw .
this pattern is counterintuitive since more data should imply a more stable estimate of @xmath0 ; however , this behavior is confirmed by an analytic theory in which the tsunami s partial waves are regarded as a filter on the tides . figure [ fig : bestrmseqw ] has five plots , one for each amount of data under study , with points indicating ratios of rmses involving all 47 twelve buoy / unit source pairings .
the ratios are formed by taking the rmse for either the eof , ks or joint estimation method and dividing it by the best rmse amongst all five detiding methods .
a ratio of one for a particular method indicates that it is the best method .
the top plot is for 1/4 ffw and shows the ks and joint estimation methods about evenly divided for best - method honors .
the buoys are ordered such that those that are separated most in distance from their unit sources ( 51406 , 51407 and 44401 ) are on the right - hand side of the plots .
the ks method generally outperforms the joint estimation method for these buoys , which tend to have lower signal - to - noise ratios than the buoy / unit source pairings on the left - hand side of the plot . as the amount of data increases beyond 1/4 ffw ( four bottom plots ) ,
there are increasingly fewer pairings where the joint estimation method fails to be the method of choice .
we also note that the disparity amongst the three methods tends to decrease as the amount of data increases .
finally , we note that , as alternatives to rmse as a summary measure , we also considered the mean absolute error @xmath167 , the maximum absolute error @xmath168 and the maximum absolute negative error , i.e. , @xmath168 with @xmath169 ranging over values such that @xmath170 .
these last two measures are based on worst - case scenarios , which are of considerable operational interest ; in particular , the last measure focuses on worst - case underestimation of @xmath0 , which can lead to forecasting coastal inundations that are too small .
all three additional measures lead to the same conclusions we drew using the rmse : the joint estimation method is generally to be preferred over the eof and ks methods , and the two methods involving only harmonic analyses are not competitive .
here we apply the five detiding methods to data collected during the devastating 2011 japan tsunami , which was generated by the great mw 9.0 earthquake that occurred on march 3rd at 05:46:23 ut ( @xmath3 ) ( tang _ et al .
_ , 2012 ; wei _ et al .
_ , 2012 ; wei _ et al .
_ , 2014 ) .
several dart^^ buoys recorded this event , including buoy 52402 .
figure [ fig : tohokudata ] shows the data from 52402 we make use of here .
the data consist of a 15-min stream starting 29 days prior to @xmath3 and ending 2.0 hrs before @xmath3 , and a 1-min stream starting 1.8 hrs prior to @xmath3 and ending a day after @xmath3 ( there is a small gap in the 15-min stream around 14 days prior to @xmath3 ) .
the first method uses a 6 constituent harmonic analysis based on the 29 days of data prior to the event time @xmath3 ( these are mostly from the 15-min stream , but there are 1.8 hours from the 1-min stream ) . the top plot in fig .
[ fig : detidingtohoku ] shows the corresponding detided series , which is the difference between data from the 1-min stream recorded after @xmath3 and predicted values based on the fitted harmonic model .
this series shows rapid fluctuations starting about 10 min after @xmath3 and dissipating after about 90 min .
these are evidently due to seismic noise from the earthquake . ignoring this noise ,
the detided series starts at a positive intercept and then rises almost linearly until about 210 minutes , at which point the tsunami signal becomes evident .
the five black circles are subjectively chosen markers of a quarter of the first full wave ( ffw ) , a half , three quarters , the end of the ffw and one hour past the end of the ffw ( these occur at , respectively , 225 , 232 , 236 , 245 and 305 min after @xmath3 ) .
the true tidal fluctuations are of course unknown , but a reasonable conjecture is that the linear increase in evidence here is actually a tidal component that the first method failed to properly extract .
the second method uses a 68 constituent harmonic analysis based on 465 days of data collected by buoy 52402 between 12/13/2006 and 3/21/2008 , i.e. , well before the 2011 japan tsunami . the detided series , shown below the top plot in fig .
[ fig : detidingtohoku ] , is again the difference between data from the 1-min stream and predicted values based on the fitted harmonic model , but with the mean level of the predictions adjusted using the 29 days of data prior to @xmath3 . in comparison to the first method , there is now only a slight linear increase in the first three hours , but the positive intercept is larger .
the third method is eof - based and yields detided series that depend on the amount of the 1-min stream after @xmath3 we wish to detide . the middle plot in fig .
[ fig : detidingtohoku ] shows five detided series color coded to indicate the amount of data used ( red , green , cyan , magenta and black for data ending at , respectively from left to right , the locations of the five solid circles , and starting 1470 minutes earlier ) .
differences between the displayed detided series at the same location in time are typically small , but do get as large as 5.7 cm . ignoring seismic noise , all five detided series are relatively flat out to 150 min , after which they exhibit a noticeable dip prior to the arrival of the tsunami signal . while we might be tempted to regard this dip as a remanent tidal component ,
caution is in order since true tidal fluctuations are unknown .
the fourth method
kalman smoothing is similar to the eof method in that the detided series depends on the amount of the 1-min stream to be detided , but is dissimilar in that detiding involves data prior to @xmath3 only indirectly because the starting point for the ks method is the detided series provided by the first method . the next - to - bottom plot in fig .
[ fig : detidingtohoku ] shows five detided series color coded in the same manner as before .
the differences between the detided series at the same point in time are again typically small , with the largest difference now being 0.7 cm .
there is only a slight hint here of the dip readily apparent in the eof detided series .
similar to the eof and ks methods , the final method ( a 2 constituent local harmonic analysis with joint source coefficient estimation ) yields detided series that depend upon the amount of the 1-min stream to be detided , but , in contrast , it makes no direct or indirect use of any data prior to @xmath3 .
this method , however , does depend upon a suitable model for the tsunami signal .
_ ( 2014 ) discuss selection of unit sources for the japan tsunami using an objective automatic method .
the selection is based on data from three dart^^ buoys ( 21401 , 21413 and 21418 ) located much closer to the epicenter of the earthquake than buoy 52402 is .
these buoys registered the tsunami signal within 5 min after @xmath3 , well before it arrived at buoy 52402 more than 3 h later .
the automatic method selected seven unit sources @xmath8 ,
, @xmath171 to model the tsunami signal .
accordingly , we need to adjust the joint estimation method to make use of seven unit sources rather than just one .
we do so by suitably redefining the design matrix @xmath152 and the vector of coefficients @xmath172 in equation [ eq : jointrewritten ] .
thus @xmath152 is now of dimension @xmath173 , with its first five columns being as before and with its next seven columns now being @xmath8 , , @xmath171 ; correspondingly , @xmath174 is augmented to dimension 12 , with its last seven elements being the source coefficients for the unit sources .
after these adjustments to @xmath152 and @xmath174 , we produce the detided series using the same equations as before ( [ eq : normalequations ] and [ eq : jointdetided ] ) . the bottom plot in fig .
[ fig : detidingtohoku ] shows detided series for the joint estimation method ( five in all corresponding to different amounts of data ) .
these series are visually quite similar to the ks detided series , but the largest difference between the joint detided series is larger ( 3.6 cm ) than for the ks series ( 0.7 cm ) .
it is important to note that , in contrast to the other four methods , the joint estimation method is dependent upon a suitable model for the tsunami signal . to demonstrate this fact , the two plots in fig . [ fig : detidingtohokuoneus ] show detided series using the joint estimation method , but based upon different models .
the top plot is the same as the bottom plot in fig .
[ fig : detidingtohoku ] , for which the detided series utilize a model involving seven unit sources . in the bottom plot
the detided series use a model based on only one of these seven sources ( ki026b ) .
the magnitude of the earthquake that generated the 2011 japan tsunami was so large that it is physically unrealistic for the signal to be well - modeled by a single unit source ( papazachos _ et al . _ , 2004 ) .
the detided series from this presumably inadequate model have low - frequency fluctuations in the first three hours that are not evident in the top plot ( or in the ks - based detided series , which are shown in the next - to - bottom plot of fig .
[ fig : detidingtohoku ] ) .
these fluctuations are best attributed to a failure on the part of the joint estimation method due to an inappropriate model .
we have undertaken a comprehensive comparison of five methods for estimating the source coefficient @xmath0 based upon dart^^ buoy bottom pressure ( bp ) data collected during a tsunami event ( this coefficient reflects the strength of the event and is used to provide initial conditions for predicting coastal inundation ) .
any method for estimating @xmath0 must deal with the fact that the variability in bp data is typically dominated by tidal fluctuations , and hence all viable methods must detide the data in some explicit or implicit manner .
the five methods under study have been entertained as part of the on - going development of the sift application , a tool developed at noaa for use by u.s .
tsunami warning centers for real - time assessment of tsunami events .
the clear method of choice is a scheme by which @xmath0 is estimated jointly in a regression model that accounts for the tidal components using sinusoidal constituents involving the tidal frequency m2 and half that frequency .
this method is particularly convenient from an operational point of view in that it does not make direct use of data occurring prior to a tsunami event , as is true in varying degrees for the four other methods under study . amongst the four remaining methods , the kalman smoothing ( ks ) method performed best overall in our study . in some cases ,
the eof method is competitive with the joint estimation and ks methods , but the two methods based on only harmonic analyses proved to be markedly inferior to the other three methods .
note that we evaluated the performance of the eof method using a specific set of eight basis vectors adapted for use within the sift system .
we limited this set to eight vectors due to the fact that only this number of vectors defines a location - independent tidal sub - space , should the set be derived using data from a single buoy , as was the case here ( tolkova , 2010 ) . expanding
the set of vectors by deriving them from multiple buoys might allow for more accurate detiding . despite the fact that our study points to the joint estimation method as the method of choice , some caution is in order .
a presumption behind our study is that the model for the tsunami signal is known perfectly , which is obviously unrealistic in practice .
the issue of model mismatch must therefore temper our conclusions , a point that is reinforced by the discussion surrounding fig .
[ fig : detidingtohokuoneus ] . for graphical presentation of detided series within sift
, the fact that the ks method does not depend on an assumed model for the tsunami signal but often compares favorably with the joint estimation method suggests its use . as currently implemented , the ks method uses the output from the 29-day harmonic analysis method as its starting point . in cases where the 15-min stream is not available over much of the preceding 29 days , the blanket harmonic method could provide a model - free detiding for display purposes , as could the eof method if the 15-min stream were available going back about a lunar day from the start of the tsunami signal .
there is thus potential use within sift for all five methods we have studied . as of this writing ,
the current version of sift supports the 29-day harmonic analysis , eof and ks methods , and there are plans to implement the joint estimation method within an overall scheme for automatically selecting unit sources to serve as models for the tsunami signal .
this work was funded by the joint institute for the study of the atmosphere and ocean ( jisao ) under noaa cooperative agreement no .
na17rj1232 and is jisao contribution no .
this work is also contribution no .
4089 from noaa / pacific marine environmental laboratory .
the authors thank george mungov of noaa s national geophysical data center for supplying dart^^ buoy data and predictions based upon harmonic analyses with 68 sinusoidal constituents .
brockwell , p.j . , and davis , r.a .
( 2002 ) , _ introduction to time series and forecasting _ , 2nd edition .
springer , 448 p. cartwright , d.e . , spencer , r. , and vassie , j.m .
( 1987 ) , _ pressure variations on the atlantic equator _ , j. geophys .
92_(c1 ) , 725741 , doi:10.1029.jc092ic01p00725 . chambers , j.m . ,
cleveland , w.s . ,
kleiner , b. , and tukey , p.a .
( 1983 ) , _ graphical methods for data analysis_. wadsworth & brooks / cole , 409 p. consoli , s. , recupero , d.r . , and zavarella , v. ( 2014 ) , _ a survey of tidal analysis and forecasting methods for tsunami detection _ , j. tsunami soc . int . _
33_(1 ) , 156 .
cummins , p.f . ,
cherniawsky , j.y . , and foreman , m.g.g .
( 2001 ) , _ north pacific internal tides from the aleutian ridge : altimeter observations and modeling _ ,
j. marine res . _
59_(2 ) , 167191 . durbin , j. , and koopman , s.j .
( 2012 ) , _ time series analysis by state space methods _ , 2nd edition .
oxford university press , 368 p. foreman , m.g.g .
( 1977 , revised 2004 ) , _
manual for tidal heights analysis and prediction .
pacific marine science report . _ 7710 .
institute of ocean sciences , patricia bay , 58 pp .
gica , e. , spillane , m.c . ,
titov , v.v . , chamberlin , c.d . , and newman , j.c .
( 2008 ) , _ development of the forecast propagation database for noaa s short - term inundation forecast for tsunamis ( sift)_. noaa tech . memo .
oar pmel139 , 89 pp .
gonzlez , f.i . ,
bernard , e.n . ,
meinig , c. , ebl , m.c . ,
mofjeld , h.o . , and
stalin , s. ( 2005 ) , _ the nthmp tsunameter network _ , natural hazards _
35_(1 ) , special issue , u.s .
national tsunami hazard mitigation program , 2539 .
mofjeld , h.o .
( 2009 ) , _ tsunami measurements_. in the sea , volume 15 : tsunamis ( eds . e.n .
bernard and a.r .
robinson ) , harvard university press , cambridge , ma , pp .
mofjeld , h.o . , whitmore , p.m. , ebl , m.c .
, gonzlez , f.i . , and newman , j.c .
( 2001 ) , _ seismic - wave contributions to bottom pressure fluctuations in the north pacific implications for the dart tsunami array_. proceedings of the international tsunami symposium 2001 ( its 2001 ) ,
session 5 - 10 , seattle , wa , 710 august 2001 , 633641 .
mungov , g. , elb , m. , and bouchard , r ( 2012 ) , _
dart^^ tsunameter retrospective and real - time data : a reflection on 10 years of processing in support of tsunami research and operations _ , pure appl .
geophys . _
170_(910 ) , 13691384 , doi:10.1007/s00024 - 012 - 0477 - 5 munk , w.h . , and cartwright , d.e .
( 1966 ) , _ tidal spectroscopy and prediction _ , phil .
london , series a , mathematical and physical sciences _ 259 _ , 533581 .
niiler , p.p . ,
filloux , j. , liu , w.t . ,
samelson , r.m . ,
paduan , j.d . , and paulson , c.a .
( 1993 ) , _ wind - forced variability of the deep eastern north pacific : observations of seafloor pressure and abyssal currents _ , j. geophys .
98_(c12 ) , 2258922602 , doi:10.1029/93jc01288 noaa data management committee ( 2008 ) , _ tsunami data management : an initial report on the management of environmental data required to minimize the impact of tsunamis in the united states _ , version 1.0 , 87 pp .
http://www.ngdc.noaa.gov/noaa_pubs/index.shtml papazachos , b.c . ,
scordilis , e.m . ,
panagiotopoulos , d.g . ,
papazachos , c.b . , and karakaisis , g.f .
( 2004 ) , _ global relations between seismic fault parameters and earthquake moment _ , horizontal parker , b.p .
( 2007 ) , _ tidal analysis and prediction_. noaa special publication nos co - ops 3 ; 378 p. percival , d.b . , denbo , d.w . ,
elb , m.c . ,
gica , e. , mofjeld , h.o . , spillane , m.c .
, tang , l. , and titov , v.v .
( 2011 ) , _ extraction of tsunami source parameters via inversion of dart^^ buoy data _ , natural hazards _
58_(1 ) , 567590 , doi:10.1007/s11069 - 010 - 9688 - 1 percival , d.m . ,
percival , d.b . ,
denbo , d.w . ,
gica , e. , huang , p.y . , mofjeld , h.o . , and spillane , m.c .
( 2014 ) , _ automated tsunami source modeling using the sweeping window positive elastic net _ , j. amer . stat .
assoc . under review
. ray , r.d . , and luthcke , s.b .
( 2006 ) , _ tide model errors and grace gravimetry : towards a more realistic assessment _ , geophys .
j. int . _
167_(3 ) , 10551059 , doi:10.1111/j.1365 - 246x.2006.03229.x shumway , r.h . , and stoffer , d.s . ( 2011 ) , _ time series analysis and its applications with r examples _
, 3rd edition .
springer , 608 p. spillane , m.c .
, gica , e. , titov , v.v . , and mofjeld , h.o .
( 2008 ) , _ tsunameter network design for the u.s .
dart^^ arrays in the pacific and atlantic oceans_. noaa tech
. memo .
oar pmel143 , 165 pp .
http://nctr.pmel.noaa.gov/pubs.html tang , l. , titov , v.v . ,
bernard , e.n .
, wei , y. , chamberlin , c.d . , newman , j.c .
, mofjeld , h.o . , arcas , d. , eble , m.c . ,
moore , c. , uslu , b. , pells , c. , spillane , m. , wright , l. , and gica , e. ( 2012 ) , _ direct energy estimation of the 2011 japan tsunami using deep - ocean pressure measurements _ , j. geophys .
117_(c8 ) , c08008 , doi:10.1029/2011jc007635 .
titov , v.v .
( 2009 ) , _ tsunami forecasting_. in the sea , volume 15 : tsunamis ( eds . e.n .
bernard and a.r .
robinson ) , harvard university press , cambridge , ma , pp .
titov , v.v . ,
gonzlez , f.i . ,
bernard , e.n . ,
ebl , m.c .
, mofjeld , h.o . , newman , j.c . , and venturato , a.j .
( 2005 ) , _ real - time tsunami forecasting : challenges and solutions _ , natural hazards _
35_(1 ) , special issue , u.s . national tsunami hazard mitigation program , 4158 .
titov , v.v . ,
mofjeld , h.o .
, gonzlez , f.i . , and newman , j.c .
( 1999 ) , _ offshore forecasting of hawaiian tsunamis generated in alaskan aleutian subduction zone_. noaa tech
. memo .
oar pmel114 , 22 pp . http://nctr.pmel.noaa.gov/pubs.html tolkova , e. ( 2009 ) , _ principal component analysis of tsunami buoy record : tide prediction and removal _ , dyn .
46_(14 ) , 6282 , doi:10.1016/j.dynatmoce.2008.03.001 tolkova , e. ( 2010 ) , _ eof analysis of a time series with application to tsunami detection _ , dyn .
50_(1 ) , 3554 , doi:10.1016/j.dynatmoce.2009.09.001 webb , s.c .
( 1998 ) , _ broadband seismology and noise under the ocean _ , rev .
geophys . _
36_(1 ) , 105142 .
wei , y. , chamberlin , c. , titov , v.v . ,
tang , l. , and bernard , e.n .
( 2013 ) , _ modeling of the 2011 japan tsunami : lessons for near - field forecast _ , pure appl .
geophys . _
170_(68 ) , 13091331 , doi : 10.1007/s00024 - 012 - 0519-z wei , y. , newman , a.v . ,
hayes , g.p . ,
titov , v.v . , and tang , l. ( 2014 ) , _ improving tsunami forecast by joint inversion of real time tsunami waveforms and seismic or gps data : application to the tohoku 2011 tsunami _ , pure appl .
geophys . in press .
zhao , z. , and alford , m.h .
( 2009 ) , _ new altimetry estimates of mode-1 m2 internal tides in the central north pacific ocean _ , j. phys .
ocean . _
39_(7 ) , 16691684 , doi:10.1175/2009jpo3922.1 .eleven representative dart^^ buoys used in detiding study .
the buoy locations are shown by triangles on fig .
[ fig : detidingpaperchart ] .
the tidal range is a mean daily range as estimated from the harmonic constituents described in sect .
[ sec : harmonic ] for the second harmonic method ( mofjeld _ et al .
_ , 2004 ) .
the type of tide is also based upon these constituents ( parker , 2007 ) .
the start / stop dates are those associated with the bottom pressure data retrieved from the particular buoy , and the column marked ` days ' gives the number of days spanned by the data . [ cols="^,^,^,^,^,^ " , ] set to 9:21:00 ut on 27 june 2007 .
the points show the 15-min stream , which starts 29 days prior to @xmath3 , but has a 3-h gap just prior to it .
the dark curve on the right - hand side is the 1-min stream , which starts at @xmath3 and ends one day later .
, title="fig : " ] + -6pt as the dark curve on the right - hand side ) .
this stream is assumed to consist of tidal fluctuations and background noise summed together , i.e. , @xmath21 from equation [ eq : model1 ] .
the middle plot shows @xmath175 with @xmath0 set to 6 , where @xmath26 is based on ki060b , one of the unit sources associated with 52402 ( see table [ tab : unitsources ] ) . the bottom plot is a simulated tsunami event @xmath4 formed by adding @xmath23 to @xmath21 .
, title="fig : " ] + -25pt .
each signal is based upon a model for what would be observed eventually over a 120-min segment of time at a particular buoy due to an earthquake originating within a particular unit source ( the starting time of each segment is different for each signal ) .
the source coefficient @xmath0 is adjusted separately for each signal merely for plotting purposes ( the actual ranges for the unscaled @xmath26 s are listed in table [ tab : unitsources ] ) .
there are five hand - picked red points in each plot .
these are associated with varying amounts of data related to the first full wave of the signal , namely , the first quarter of the full wave , half , three - quarters , all the full wave and one - hour past the end of the first full wave .
, title="fig : " ] + -25pt .
each event is formed by adding the artificial tsunami signal @xmath27 shown in fig .
[ fig : unitsources ] ( but with @xmath0 now set to 6 ) to the 1-min stream of a scenario randomly chosen from amongst the 1000 scenarios for a given buoy .
the vertical height of each plot is 1.2 m .
, title="fig : " ] + -6pt . each plot corresponds to one of the five methods under study . from top to bottom , these are ( 1 ) a 6 constituent harmonic analysis ( ha ) based on 29 days of the 15-min stream before @xmath3 ( the start of the 1-min stream ) , ( 2 ) a 68 constituent ha based on all available 15-sec data , ( 3 ) an empirical orthogonal function approach , ( 4 ) a kalman smoother approach and ( 5 ) a 2 constituent local ha with joint estimation of the source coefficient @xmath0 . in plots ( 3 ) to ( 5 ) , there are five color - coded curves , each corresponding to detiding based upon one of five different amounts of data ( some curves in ( 4 ) and ( 5 ) are partially obscured because they are virtually identical to other curves ) .
red indicates use of data from @xmath3 up to a quarter of the first full wave ( ffw ) of the tsunami signal ; green , to half of the ffw ; cyan , to three quarters ; magenta , to the end of the ffw ; and black , to one hour past the end of the ffw .
the five ending times are shown by black circles ( the first four are listed in table [ tab : unitsources ] on the row pertaining to buoy 52402 and unit source ki060b , and all five are shown as red circles in the left - hand plots on the third rows of figs .
[ fig : unitsources ] and [ fig : tsunamievents ] ) .
, title="fig : " ] + , @xmath176 , used in eof - based detiding .
the vectors consist of a constant vector @xmath177 and vectors related to eofs having the seven largest eigenvalues .
the vectors are arranged in the plot according to their number @xmath178 of zero crossings .
, title="fig : " ] + -6pt ( dots ) given by , from top to bottom , ( 1 ) a 6 constituent harmonic analysis ( ha ) based on 29 days of the 15-min stream before @xmath3 ( the start of the event ) , ( 2 ) a 68 constituent ha based on all available 15-sec data , ( 3 ) an empirical orthogonal function approach , ( 4 ) a kalman smoother approach and ( 5 ) a 2 constituent local ha with joint estimation of the source coefficient @xmath0 .
each estimate is based on data up to three quarters of the first full wave from one of the 1000 scenarios for buoy 52402 , to which has been added an artificial tsunami signal based on unit source ki060b with source coefficient @xmath28 .
the dots for the estimates for scenario 943 are surrounded by a red triangle ( this scenario is used as an example in figs . [
fig : scenario ] , [ fig : oneevent ] and [ fig : detidingfive ] and in table [ tab : alphaestimates ] ) .
the boxplot on each row summarizes the distribution of the corresponding @xmath63 estimates ( see text for details ) .
, title="fig : " ] + based on simulated tsunami events constructed from buoy 52402 and unit source ki060b .
the estimates are based on , respectively , the eof , ks and joint estimation methods ( top to bottom ) and data up to ( from left to right ) a quarter of the first full wave ( ` qw ' ) , half of the full wave ( ` hw ' ) , three - quarters ( ` tqw ' ) , the full wave ( ` fw ' ) and one hour past the end of the first full wave ( ` hr ' ) . , title="fig : " ] + based on simulated tsunami events constructed from buoy 52402 and unit source ki060b . for each of five methods ,
rmses are shown for estimates based on data up to ( from left to right ) a quarter of the first full wave ( ` qw ' ) , half of the full wave ( ` hw ' ) , three - quarters ( ` tqw ' ) , the full wave ( ` fw ' ) and one hour past the end of the first full wave ( ` hr ' ) . , title="fig : " ] + set to 05:46:23 ut on 3/11/2011 .
the points show the 15-min stream , which goes from 29 days up to 2.0 hours prior to @xmath3 .
the dark curve on the right - hand side is the 1-min stream , which starts 1.8 hours prior to @xmath3 and ends one day after @xmath3 .
, title="fig : " ] + -6pt .
each plot corresponds to a different detiding method . from top to bottom , these are ( 1 ) a 6 constituent harmonic analysis ( ha ) based on 29 days of data before @xmath3 ( the start of the event ) , ( 2 ) a 68 constituent ha based on 15-sec data recorded from 12/13/2006 to 3/21/2008 ; ( 3 ) an empirical orthogonal function approach , ( 4 ) a kalman smoother approach and ( 5 ) a 2 constituent local ha with joint estimation of source coefficients .
see the caption to fig .
[ fig : detidingfive ] for an explanation of the black circles and the color - coded curves in the bottom three plots .
, title="fig : " ] + -6pt using the method based on a two constituent local harmonic analysis with joint estimation of source coefficients .
the top plot here is identical to the bottom plot of fig .
[ fig : detidingtohoku ] , for which the detided series are based on a model for the tsunami signal that is the linear combination of six unit sources selected by an objective automatic procedure ( percival _ et al . _ , 2014 ) . in the bottom plot ,
the detided series are based on a single unit source that is presumably an inadequate description of the underlying tsunami signal .
see the caption to fig .
[ fig : detidingfive ] for an explanation of the black circles and the color - coded curves in these plots .
, title="fig : " ] +
|
u.s .
tsunami warning centers use real - time bottom pressure ( bp ) data transmitted from a network of buoys deployed in the pacific and atlantic oceans to tune source coefficients of tsunami forecast models . for accurate coefficients and therefore forecasts , tides at the buoys
must be accounted for . in this study ,
five methods for coefficient estimation are compared , each of which accounts for tides differently .
the first three subtract off a tidal prediction based on ( 1 ) a localized harmonic analysis involving 29 days of data immediately preceding the tsunami event , ( 2 ) 68 pre - existing harmonic constituents specific to each buoy , and ( 3 ) an empirical orthogonal function fit to the previous 25 hrs of data .
method ( 4 ) is a kalman smoother that uses method ( 1 ) as its input .
these four methods estimate source coefficients after detiding .
method ( 5 ) estimates the coefficients simultaneously with a two - component harmonic model that accounts for the tides .
the five methods are evaluated using archived data from eleven dart^^ buoys , to which selected artificial tsunami signals are superimposed .
these buoys represent a full range of observed tidal conditions and background bp noise in the pacific and atlantic , and the artificial signals have a variety of patterns and induce varying signal - to - noise ratios . the root - mean - square errors ( rmses ) of least squares estimates of sources coefficients using varying amounts of data are used to compare the five detiding methods .
the rmse varies over two orders of magnitude between detiding methods , generally decreasing in the order listed , with method ( 5 ) yielding the most accurate estimate of source coefficient .
the rmse is substantially reduced by waiting for the first full wave of the tsunami signal to arrive . as a case study , the five method
are compared using data recorded from the devastating 2011 japan tsunami .
|
the seyfert 1 galaxy ngc5548 was the target of intensive space- and ground - based monitoring designed to measure the light travel time across its broad optical - line - emitting region ( @xcite , and references therein ) .
these reverberation studies imply that the broad line region ( blr ) has a characteristic diameter @xmath7 40 light days , or @xmath8as for h@xmath9 km s@xmath10 mpc@xmath10 and the galaxy s recession velocity relative to the 3 k background ( 5354 km s@xmath10 ; @xcite ) .
given this small size , is it possible to constrain models for the ultimate energy source for the blr in ngc5548 ?
the common view is that the energy source is a supermassive black hole fuelled by an accretion disk ( e.g. , @xcite ) , and three recent findings support this scenario for ngc5548 .
first , analysis of variations in the emission line profiles of ngc5548 suggests that the blr material moves along randomly inclined keplerian orbits defined by a central mass of order @xmath11 ( @xcite ) .
second , the central continuum source must irradiate the blr anisotroptically ( @xcite ) , perhaps because the continuum source is surrounded by an optically - thick accretion disk ( cf .
figure 1 of @xcite ) .
third , ngc5548 exhibits very broad and asymmetric x - ray emission lines whose profiles are well matched by gravitationally - redshifted and fluorescing lines from accretion disk material ( @xcite ) . still , the starburst hypothesis of terlevich et al .
( 1992 ) potentially offers an alternative energy source : a compact star cluster hosts a supernova and the blr is identified with the compact supernova remnant formed as the supernova ejecta interacts with dense circumstellar material .
although the star cluster hypothesis has yet to be tested against the three findings listed above for ngc5548 , it appears consistent with analysis of the optical continuum variability and the reverberation results ( @xcite ) .
astrometric monitoring of seyfert nuclei offers another strong test of of the starburst hypothesis ( @xcite ) .
the low supernova rates and their short radiative lifetimes mean that as the compact star cluster evolves and new supernovae occur , the blr should dance within the star cluster . in constrast , black hole models for low - power objects like
seyfert nuclei predict no motion of the nucleus over time , or simple linear motion if the emission arises from propogating shocks in a jet and counterjet ( @xcite ) .
this paper imposes observational constraints on the astrometric stability of the radio continuum nucleus of ngc5548 , to help distinguish between star cluster and black hole models for the energy source of the blr .
although the star cluster models do not predict light curves for the radio continuum emission from a compact supernova remnant , terlevich et al .
( 1995 ) do remark on the relevance of sn1988z to their models , as being an example of a ( radio ) supernova expanding into dense circumstellar material ( @xcite ) .
therefore this paper also examines the photometric variability of the radio continuum nucleus of ngc5548 , for comparison with the photometric evolution of sn1988z at radio wavelengths .
table 1 gives a log of the astrometric imaging observations of ngc5548 with the very large array ( vla , @xcite ) and the very long baseline array ( vlba , @xcite ) .
both instruments acquired data in dual circular polarizations at the center frequencies tabulated . for each polarization ,
the vla bandwidth was 100 mhz with 3-level sampling and the vlba bandwidth was 32 mhz with 4-level sampling .
all vla and vlba observations after 1990 assumed a coordinate equinox of 2000 .
earlier vla observations assumed an equinox of 1950 and were transformed to 2000 using a reference date of 1979.9 for the phase calibrator grid used at the vla prior to 1990 ( @xcite ) . on the vla ,
an 8-minute observation of ngc5548 was preceded and followed by an observation of the phase calibrator oq208 ( b1404 + 286 ) about @xmath12 from ngc5548 ( figure 1 ) .
vla observations of 3c286 were used to set the amplitude scale to an accuracy of about 3% . on the vlba ,
a 3-minute observation of ngc5548 was preceded and followed by an observation of the phase , rate , and delay calibrator j1419 + 27 about @xmath13 from ngc5548 ( figure 1 ) .
the position assumed for j1419 + 27 was @xmath14 and @xmath15 , with an error in each coordinate of less than 1 mas ( m. eubanks , private communication ) .
sources oq208 and j1310 + 32 were also observed with the vlba to check the astrometric accuracy and to align the phases of the baseband channels , respectively .
vlba system temperatures and gains were used to set the amplitude scale to an accuracy of about 5% , after first correcting for sampler errors .
for both the vla and vlba , the position assumed for oq208 was @xmath16 and @xmath17 ( m. eubanks , private communication ) .
these assumed positions for both j1419 + 27 ( b1417 + 273 ) and oq208 agree with those in ma et al .
the vla data were calibrated using the 1994 january 15 release of the nrao aips software , while the vlba data were calibrated with the 1999 april 15 release .
the aips task imagr was used to form and deconvolve images of the stokes @xmath18 emission from ngc5548 .
neither the vla nor vlba data on ngc5548 were self - calibrated .
the vla images were made with a weighting scheme that achieved high angular resolution without seriously degrading sensitivity .
furthermore , vla imaging tests showed that the peak source strength increased as angular resolution decreased , suggesting that the source was slightly resolved .
thus only vla images with matched resolutions were analysed .
the vlba images were made with natural weighting to optimize sensitivity and were restored with a matched elliptical gaussian beam .
photometric and astrometric comparisons will be limited to observations at matched angular resolutions of 500 mas @xmath0 260 pc ( figure 2 , 9 vla observations ) , 1210 mas @xmath0 640 pc ( figure 3 , 9 vla observations ) , and 2.3 mas @xmath0 1.2 pc ( figure 4 , 3 vlba observations ) .
image rms values ( @xmath19 ) appear in table 1 in units of microjanskys per beam area ( @xmath20jy ba@xmath10 ) ; all rms values match those expected given the array and observation parameters ( @xcite ) . the image resolutions , expressed either as the fwhm dimension of a circular gaussian beam or the fwhm dimensions and elongation orientation of an elliptical gaussian beam , also appear in table 1 .
the tabulated position peaks were derived from quadratic fits in the image plane . for the vla images ,
these quadratic fits also yielded the peak flux densities listed in the table , along with errors that are quadratic sums of a 3% scale error and @xmath19 .
for the vlba images , the source appeared to be resolved , so total flux densities were derived by integration over @xmath21 beam areas and tabulated along with errors that are quadratic sums of a 5% scale error and @xmath22 .
this apparent resolution could be due either to unmodelled atmospheric and/or station motion effects ( @xcite ) , or due to real structure . that the structure is similar in the vlba images
suggests the latter but does not prove it , because some phase - referencing errors can be systematic from observation to observation .
the table also notes the 1@xmath23 error in each coordinate for the astrometry of the peak position .
@xmath23 is the quadratic sum of a term measuring the peak signal - to - noise ratio ( @xcite ) and a term measuring the accuracy of the absolute astrometry .
this latter term was _ estimated _ in the vla case as follows .
the imaging at 500-mas resolution was done in the same configuration and season as patnaik et al .
( 1992 ) or in the same configuration and contemporaneous with browne et al .
those authors used phase calibrators @xmath24 or less from their targets , cycled back to a calibrator every 6 minutes , and measured a 1-d positional accuracy of about 10 mas rms .
the present work involves a calibrator @xmath12 from the target ( figure 1 ) but observed with a 10-minute cycle time , so 20 mas was adoted as an estimate for the 1-d accuracy of the absolute vla astrometry . in the vlba case ,
the accuracy of the absolute astrometry was _ derived _ from analysis of the astrometric check source oq208 , as referenced to j1419 + 27 . as figure 1 shows , oq208 and j1419 + 27
are separated by about @xmath25 , while the separation between j1419 + 27 and ngc5548 is about @xmath13 .
thus the position error for oq208 as referenced to j1419 + 27 , plotted in figure 5 , serves as a cautious upper limit to the position error for ngc5548 as referenced to j1419 + 27 . indeed ,
residual position errors of order 1 mas and 0.7 mas are expected for separations of @xmath25 and @xmath13 , respectively , due to known deficiencies in the 1980 iau nutation model adopted at the correlator ( @xcite ) .
photometric monitoring results from table 1 are plotted in figure 6 .
each figure panel shows results for a given angular resolution within 100 days centered on the mean observation epoch indicated .
data at 8.4 ghz are plotted as dark symbols , while data at 4.9 ghz appear as light symbols .
the figure legend indicates which symbols encode which matched angular resolutions .
photometric comparisons should only be made among data at matched resolutions .
to aid such comparisons the pairs of panels at the top , middle , and bottom of figure 6 show data at matched resolution , with the left columns having the earlier mean epoch of observation .
figure 6 demonstrates that the radio nucleus of ngc5548 is photometrically variable , on timescales that are both short ( comparing within a panel ) and long ( comparing across paired panels ) . specifically , the maximum flux densities are higher than minimum flux densities , by @xmath1% between vla observations separated by 41 days ( panel [ d ] ) and by @xmath2% between vla observations separated by 4.1 yr ( panel [ a ] _ versus _ [ b ] ) .
panel ( d ) also establishes that the 41-day `` flare '' is milder ( @xmath3% ) at 4.9 ghz than at 8.4 ghz .
furthermore , the spectral index @xmath26 ( @xmath27 ) is @xmath28 prior to the 41-day `` flare '' but @xmath4 during the `` flare '' .
thus the `` flare '' is optically thick and milder at the lower frequency , traits typical for variable active nuclei .
the `` flare '' is too poorly sampled to derive a meaningful variability timescale .
the flux densities plotted in figure 6 correspond to 8.4-ghz powers @xmath29 w hz@xmath10 , while the associated luminosities are @xmath30 erg s@xmath10 .
the radio sampling in figure 6 is too coarse for useful comparisons with concurrent continuum monitoring from the ground or space .
however , some relevant comparison can be made in connection with the 41-day radio variability between 1993 march 15 and april 25 .
first , that radio variability is much slower then the 1-day and 3-day variability reported for the hard and soft x - rays , respectively , during the period 1992 december to 1993 january ( @xcite ) .
second , the 33% change during the 41-day radio `` flare '' is comparable to the 40% changes documented for the continuum at @xmath31 during the period 1993 march 19 to may 27 ( @xcite ) . radio continuum emission can arise from ( 1 ) an advection - dominated accretion flow ( adaf ) onto a black hole ( @xcite ) , ( 2 ) jets emerging from an adaf ( @xcite ) , or ( 3 ) jets emerging from a standard accretion disk which itself emits no radio photons ( @xcite ) .
comparisions with these models will assume a black hole mass @xmath11 ( @xcite ) and will try to match the spectral indices @xmath26 , luminosities @xmath32 , and photometric variability observed for the radio nucleus of ngc5548 , plus be consistent with vlba emission being predominantly unresolved .
the radio continuum from an adaf is synchrotron emission fron an x - ray emitting and optically - thin plasma ( @xcite ) , and is characterized by three key traits .
first , equation 1 of yi & boughn ( 1999 ) predicts that the radio power should have an inverted spectral index @xmath33 , consistent with the index measured in the `` flaring '' state ( @xmath4 ) but not in the the quiescent state ( @xmath28 ) .
second , equation 2.7 of yi & boughn ( 1998 ) implies that the 8.4-ghz photons should originate about 6 schwarzschild radii from the black hole , placing them at radii less than @xmath34as , which is consistent with the vlba emission being predominantly unresolved .
finally , equations 1 and 3 of yi & boughn ( 1999 ) predict an 8.4-ghz luminosity @xmath35 erg s@xmath10 .
this is about @xmath36 times weaker than observed and argues for an alternative source of radio photons .
radio emission from jets is a plausible alternative , but before turning to that topic the existing evidence for outflow from the nucleus of ngc5548 will be described .
figure 7 displays a vla image at 8.4 ghz with resolution 660 mas @xmath0 350 pc and rms sensitivity @xmath37jy ba@xmath10 .
this deep image was obtained by adding the 1988 - 1989 vla data lacking evidence for photometric variability ( see panels [ a ] and [ c ] in figure 6 ) and two further 1989 observations at lower resolution .
areal integration over the figure yields a total flux density of @xmath38 mjy , corresponding to a total power at 8.4 ghz of @xmath39 w hz@xmath10 .
figure 7 shows that bipolar lobes straddle the radio nucleus and span 12 arcsec @xmath0 6.4 kpc .
the lobes are very faint at 8.4 ghz , with peak intensities @xmath40jy ba@xmath10 .
the lobe morphologies confirm previous hints from images at lower resolution ( figure 3 , @xcite ) that the southern lobe is amorphous and the northern lobe is edge - brightened , possibly due to a bubble - like structure in 3-d .
these synchrotron - emitting lobes could be driven by a bipolar wind emanating from the nucleus of ngc5548 , as has been suggested for ngc3079 ( @xcite ) .
models for the absorption of the x - ray and ultraviolet continuum from ngc5548 do imply outflowing ionized gas on subparsec scales ( @xcite ) . the shape of the subparsec outflow ( radial _ versus _ bipolar _ versus _ cylindrical ) is not constrained , but its kinetic luminosity of @xmath41 erg s@xmath10 is four orders of magnitude higher than the photon luminosities of the radio lobes ( @xcite ) . if the subparsec outflow contains relativistic plasma , then figure 4 could be used to trace the shape of the flow on parsec scales provided the reality of the apparent vlba structure was verified ( see section 2 ) .
it is worth noting that weak synchrotron emission from relativistic plasma on scales 100 - 500 pc could be responsible for the source resolution effects , discussed in section 2 , that prompted the use of matched angular resolutions .
suppose the synchrotron luminosity of the jets on scales @xmath42 pc is powered by a rotating black hole accreting from a magnetized plasma .
then the jet luminosity can be estimated from the blandford - znajek process given values for the spin paramter @xmath43 of the black hole , the efficiency @xmath44 of the jet radio emission , and the mass accretion rate @xmath45 , normalized to the eddington rate ( @xcite ) .
adopting @xmath45 from an adaf analysis of ngc5548 leads to radio jet luminosities of @xmath46 erg s@xmath10 .
this process could easily match the required radio jet luminosites on scales @xmath47 pc , but the predictions are woefully imprecise compared to the predictions from the adaf models .
this process could also in principal match the kinetic luminosity of @xmath41 erg s@xmath10 for the subparsec outflow ( @xcite ) .
alternatively , in the jet - disk symbiosis model of falcke & biermann ( 1996 ) , the radio jet luminosity and the ( standard ) accretion disk luminosity are correlated .
gauging the latter for ngc5548 by the nuclear bolometric luminosity of @xmath48 erg s@xmath10 ( @xcite ) , then the radio jet luminosity is as expected from that jet / disk correlation .
moreover , the spectral index of the quiescent radio emission is as expected for an inhomogeneous jet model , while the inverted spectrum during the 41-day `` flare '' could arise from a brightening in the optically - thick base of the jet .
terlevich et al . (
1995 ) note the relevance of sn1988z to their models , as being an example of a ( radio ) supernova expanding into dense circumstellar material ( @xcite ) .
it is thus of interest to compare the radio properties of this most distant known radio supernova ( @xcite ) , with those of the radio nucleus in ngc5548 .
van dyk et al . ( 1993 ) conclude that sn1988z is a rare , peculiar type ii supernova , which represents the endpoint of stellar evolution for a massive progenitor ( @xmath49 ) that underwent a period of high mass loss ( @xmath50 yr@xmath10 ) before explosion .
sn1988z achieved a maximum power at 8.4 ghz of about @xmath51 w hz@xmath10 , similar to the minimum power recorded in figure 6 for the radio nucleus of ngc5548 .
van dyk et al .
( 1993 ) present models quantifying the photometric evoluton of sn1988z at 8 and 5 ghz .
( while the evolution appears to be smooth , the defining light curves are not well sampled in time . )
their models are plotted in figure 8 with an arbitrary epoch shift for comparison with the vla data from panel ( d ) of figure 6 .
the radio supernova models appear able to account for the photometric variability of ngc5548 observed on time scales of years .
those models can not , however , account for the variability observed on time scales of tens of days .
all astrometric monitoring results from table 1 are plotted in figure 9 .
each panel shows data at the matched angular resolution indicated , with the panels ordered as in figure 6 to aid astrometric comparisons among data at matched resolutions . the origin for each panel is taken to be the position of the nucleus measured from the vlba image at epoch 1995.337 .
data at 8.4 ghz are plotted as dark symbols , while data at 4.9 ghz appear as light symbols .
the symbol shapes encode the matched angular resolution and have the same meaning as in figure 6 .
figure 9 implies that the radio nucleus of ngc5548 is astrometrically stable , on timescales that are both short ( comparing within a panel ) and long ( comparing across paired panels ) . panels
( a ) and ( b ) demonstrate that the nucleus is astrometrically stable at 8.4 ghz , to an accuracy of @xmath52 mas @xmath0 28 mas @xmath0 15 pc between vla observations spanning both tens of days and 4.1 yr . panels ( c ) and ( d ) verify this stability but with a poorer accuracy of @xmath53 mas @xmath0 35 mas @xmath0 19 pc .
the nucleus is as astrometrically stable at 4.9 ghz as it is at 8.4 ghz , based on comparisons within panels ( b ) and ( d ) .
finally , panels ( e ) and ( f ) imply that the nucleus is astrometrically stable at 8.4 ghz , to an accuracy of @xmath54 mas @xmath0 1.8 mas @xmath0 0.95 pc between vlba observations spanning 13 days and 3.1 yr . the observed astrometric stability is consistent with any models based on black holes , including adaf and jet models , but an adaf model was ruled out photometrically in section 3.1 . if the reality of the apparent vlba structure in figure 4 could be verified , then those images could be used to constrain the evolution of the lengths of , or features within , any jets or counterjets .
furthermore , if a vlba image was acquired during a radio flare , then the peak in that image would arise from the optically - thick base of the jet and would be expected to be located closer to the black hole .
suppose that the blue luminosity from the seyfert `` nucleus '' of the ngc5548 arises from a compact star cluster .
then the absolute magnitude of the cluster can be equated to the historical minimum of @xmath55 mag recorded for the seyfert nucleus ( @xcite ) .
a star cluster with that magnitude will be characterized by an optical supernova rate of 0.6 yr@xmath10 ( @xcite ) and an effective radius @xmath56 pc @xmath0 42 mas ( @xcite ) .
each new optical supernova event should yield a radio counterpart if sn1988z is indeed as relevant to the starburst models as claimed by terlevich et al .
( 1995 ) .
a circle with the effective radius @xmath57 of the star cluster is shown in panels ( a)-(d ) in figure 9 , with the circle arbitrarily centered at the origin .
figure 9 implies that radio nucleus of ngc5548 does not move with the characteristic step size @xmath57 between supernova events , as predicted by melnick et al .
the strongest constraints follow from the vlba - vlba astrometric stability at the 1-pc level for observations separated by 3.1 yr
. however , in the star cluster model it is the massive stars which are the supernova progenitors , so if the initial mass function is biased toward massive stars and/or mass segregation is occuring , then the radio supernova would occur within a region of radius much smaller than @xmath56 pc @xmath0 42 mas ( @xcite ) .
the optical supernova rate of 0.6 yr@xmath10 for ngc5548 is moderately high , leading to a new supernova event about every 1.7 yr .
this motivates a deep search for secondary radio sources at the matched vlba resolution .
data from vlba epochs 1998.445 and 1998.481 were combined to yield an image with rms @xmath58jy ba@xmath10 , corresponding to about 100 times the strength of casseopiaa were it placed at the distance of ngc5548 ( @xcite ) .
no secondary sources stronger than @xmath59 , or about 200 times casseopiaa , were found in about 7600 vlba beam areas within an annular region , centered as for figure 9 and with inner radius 10 mas @xmath0 5.3 pc and outer radius 84 mas @xmath0 45 pc @xmath60 .
this outer radius is well within the limitations to the field of view given the vlba observing strategies ( @xcite ) .
the absence of secondary vlba sources , in conjuction with the vlba constraints on the astrometric stability of the radio source , requires the presence of no more than one radio supernova during a 3.1-yr interval .
this condition is improbable if the detected vlba source really is a radio supernova whose slow decline , based on the vlba photometry , is shared by other radio supernovae .
thus either the radio supernova rate is lower than 0.6 yr@xmath10 and/or , as mentioned above , the radio supernovae occur within a region of radius much smaller than @xmath61 pc @xmath0 42 mas .
the vla and vlba were used to monitor the radio continuum counterpart to the optical broad line region ( blr ) in the seyfert galaxy ngc5548 .
the nucleus is photometrically variable at 8.4 ghz by @xmath1% and @xmath2% between vla observations separated by 41 days and 4.1 yr , respectively .
the 41-day changes are less pronounced ( @xmath3% ) at 4.9 ghz and exhibit an inverted spectrum ( @xmath4 , @xmath5 ) from 4.9 to 8.4 ghz .
the nucleus is astrometrically stable at 8.4 ghz , to an accuracy of 15 pc between vla observations separated by 4.1 yr and to an accuracy of 0.95 pc between vlba observations separated by 3.1 yr .
the photometric variability and astrometric stability of the radio nucleus is not consistent with star cluster models that treat the blr as a compact supernova remnant and , for ngc5548 , required a new supernova event like sn1988z every 1.7 yr within an effective radius @xmath56 pc .
the radio supernova models for sn1988z can not account for the vla variability observed on time scales of tens of days .
the combined vlba evidence for astrometric stability , no secondary sources , and slow photometric evolution of the detected source mean that the radio supernova rate is lower than predicted and/or the radio supernovae occur within a region much smaller than predicted .
additional vlba monitoring would thus be benefical , to provide further astrometric and photometric constraints on the detected radio source and to continue searches for secondary sources .
the photometric variability and astrometric stability of the radio nucleus in ngc5548 is consistent with models of jets on scales @xmath47 pc emerging from a black hole .
if the subparsec outflow inferred from x - ray / uv absorber models contains relativistic plasma , then the vlba images could be used to trace the shape of the flow on parsec scales .
however , the reality of the apparent vlba structure must first be verified .
tracing flow kinematics to 100 pc should also be feasible through an hst study of the narrow optical emission lines , because those lines are so dominated by gas within 70 pc @xmath0 130 mas from the nucleus ( @xcite and references therein ) .
a deep image at 8.4 ghz with a resolution 350 pc was obtained by adding data from quiescent vla observations .
this image shows faint bipolar lobes straddling the radio nucleus and spanning 6.4 kpc .
the northern lobe is edge - brightened , suggesting a bubble - like structure in 3-d .
these lobes are plausibly driven by twin jets or a bipolar wind from the seyfert 1 nucleus .
the kinematics of the flow could be followed to kiloparsec scales if neutral or molecular gas was detected in absorption against the radio continuum lobes .
sensitive polarimetry of the radio lobes could also constrain the properties of an enveloping magnetoionic medium and the topology of the magnetic field in a synchrotron - emitting bubble .
the author thanks dr .
m. eubanks for providing vlbi source positions ; thanks drs .
j. conway , r. terlevich , j. ulvestad and c. walker for discussions ; and acknowledges support from a serc visiting fellowship held at nuffield radio astronomy laboratories , england .
nrao is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
aretxaga , i. , & terlevich , r. 1993 , , 205 , 69 aretxaga , i. , & terlevich , r. 1994 , , 269 , 462 ball , j. a. 1975 , in methods in computational physics , volume 14 , eds .
b. alder , s. fernbach , & m. rotenberg ( new york : academic press ) , 177 baan , w. a. , & irwin , j. a. 1995 , , 446 , 602 baars , j. w. m. , genzel , r. , pauliny - toth , i. i. k. , & witzel , a. 1977 , , 61 , 99 baum , s. a. , odea , c. p. , dallacassa , d. , de bruyn , a. g. , & pedlar , a. 1993 , , 419 , 553 beasley , a. j. , & conway , j. e. 1995 , in asp conf .
82 , very long baseline interferometry and the vlba , eds .
j. a. zensus , p. j. diamond , & p. j. napier ( san francisco : asp ) , 327 browne , i. w. a. , patnaik , a. r. , wilkinson , p. n. , & wrobel , j. m. 1998 , , 293 , 257 de vaucouleurs , g. , de vaucouleurs , a. , corwin , h. g. , jr . , buta , r. j. , paturel , g. , & fouque , p. 1991
, third reference catalog of bright galaxies ( new york : springer ) done , c. , pounds , k. a. , nandra , k. , & fabian , a. c. 1995 , , 275 , 417 falcke , h. , & biermann , p. 1996
, , 308 , 321 goad , m. , & wanders , i. 1996 , , 469 , 113 korista , k. t. et al .
1995 , , 97 , 285 kraemer , s. b. , crenshaw , d. m. , filippenko , a. v. , & peterson , b. m. 1998 , , 499 , 719 kukula , m. j. , pedlar , a. , baum , s. a. , & odea , c. p. 1995 , , 276 ,
1262 ma , c. , et al .
1998 , , 116 , 516 mathur , s. , elvis , m. , & wilkes , b. 1995 , , 452 , 230 melnick , j. , aretxaga , i. , & terlevich , r. 1997 , in science with the vlt interferometer , ed .
f. paresce ( berlin : springer ) , 112 nagar , n. m. , wilson , a. s. , mulchaey , j. s. , & gallimore , j. s. 1999 , , 120 , 209 nandra , k. , george , i. m. , mushotzky , r. f. , turner , t. j , & yaqoob , t. 1997 , , 477 , 602 napier , p. j. , bagri , d. s. , clark , b. g. , rogers , a. e. e. , romney , j. d. , thompson , a. r. , & walker , r. c. 1994 , proc .
ieee , 82 , 658 patnaik , a. r. , browne , i. w. a. , wilkinson , p. n. , & wrobel , j. m. 1992 , , 254 , 655 peterson , b. m. 1993 , , 105 , 247 peterson , b. m. 1997 , an introduction to active galactic nuclei ( cambridge : cup ) peterson , b. m. , & wandel , a. 1999 , , 521 , l95 stathakis , r. a. , & sadler , e.m .
1991 , , 250 , 786 terlevich , r. , tenorio - tagle , g. , franco , j. , & melnick , j. 1992 , , 255 , 713 terlevich , r. , tenorio - tagle , g. , rozyczka , m. , franco , j. , & melnick , j. 1992 , , 272 , 198 thompson , a. r. , clark , b. g. , wade , c. m. , napier , p. j. , 1980 , , 44 , 151 trotter , a. s. , greenhill , l. j. , moran , j. m. , reid , m. j. , irwin , j. a. , & lo , k. y. 1998 , , 495 , 740 van dyk , s. d. , weiler , k. w. , sramek , r. a. , & panagia , n. 1993 , , 419 , l69 walker , r. c. 1999 , the sched user manual , nrao user document wanders , i. et al . 1995 , , 453 , l87 wanders , i. , & peterson , b. m. 1996 , , 466,174 wilson , a. s. , & ulvestad , j. s. 1982 , , 260 , 56 wrobel , j. m. 1995 , in asp conf . ser .
82 , very long baseline interferometry and the vlba , eds .
j. a. zensus , p. j. diamond , & p. j. napier ( san francisco : asp ) , 413 wrobel , j. m. , & walker , r. c. 1999 , in asp conf .
180 , synthesis imaging in radio astronomy ii , eds .
g. b. taylor , c. l. carilli , & r. a. perley ( san francisco : asp ) , 171 yi , i. , & boughn , s. p. 1998
, , 499 , 198 yi , i. , & boughn , s. p. 1999
, , 515 , 576 young , p. j. 1976 , , 81 , 807 lccccllcc 1992.814 ... & vla & 500 & 8.4 & 75 & @xmath62 & @xmath63 & 20 & @xmath641988.913 ... & vla & 500 & 8.4 & 43 & 14 17 59.5409 & + 25 08 12.603 & 20 & [email protected] ... & vla & 500 & 8.4 & 44 & 14 17 59.5398 & + 25 08 12.607 & 20 & [email protected] ... & vla & 500 & 8.4 & 36 & 14 17 59.5394 & + 25 08 12.608 & 20 & [email protected] ... & vla & 500 & 8.4 & 44 & 14 17 59.5402 & + 25 08 12.604 & 20 & [email protected] ... & vla & 1210 & 8.4 & 37 & 14 17 59.5411 & + 25 08 12.626 & 25 & [email protected] ... & vla & 1210 & 8.4 & 33 & 14 17 59.5401 & + 25 08 12.616 & 25 & [email protected] ... & vla & 1210 & 8.4 & 38 & 14 17 59.5391 & + 25 08 12.596 & 25 & [email protected] ... & vla & 1210 & 8.4 & 32 & 14 17 59.5411 & + 25 08 12.636 & 25 & [email protected] ... & vla & 500 & 8.4 & 67 & 14 17 59.5401 & + 25 08 12.604 & 20 & [email protected] ... & vla & 500 & 4.9 & 76 & 14 17 59.5406 & + 25 08 12.594 & 20 & [email protected] ... & vla & 500 & 8.4 & 66 & 14 17 59.5395 & + 25 08 12.608 & 20 & [email protected] ... & vla & 500 & 4.9 & 96 & 14 17 59.5388 & + 25 08 12.600 & 20 & [email protected] ... & vla & 500 & 8.4 & 56 & 14 17 59.5402 & + 25 08 12.613 & 20 & [email protected] ... & vla & 500 & 4.9 & 105 & 14 17 59.5406 & + 25 08 12.603 & 20 & [email protected] ... &
vla & 500 & 8.4 & 50 & 14 17 59.5396 & + 25 08 12.598 & 20 & [email protected] ... & vla & 500 & 4.9 & 78 & 14 17 59.5413 & + 25 08 12.600 & 20 & [email protected] ... & vla & 500 & 8.4 & 69 & 14 17 59.5397 & + 25 08 12.596 & 20 & [email protected] ... & vla & 500 & 4.9 & 111 & 14 17 59.5396 & + 25 08 12.602 & 20 & [email protected] ... & vla & 1210 & 8.4 & 49 & 14 17 59.5400 & + 25 08 12.603 & 25 & [email protected] ... & vla & 1210 & 4.9 & 64 & 14 17 59.5400 & + 25 08 12.603 & 25 & [email protected] ... & vla & 1210 & 8.4 & 47 & 14 17 59.5380 & + 25 08
12.653 & 25 & [email protected] ... & vla & 1210 & 4.9 & 68 & 14 17 59.5390 & + 25 08 12.593 & 25 & [email protected] ... & vla & 1210 & 8.4 & 47 & 14 17 59.5410 & + 25 08 12.613 & 25 & [email protected] ... & vla & 1210 & 4.9 & 65 & 14 17 59.5390 & + 25 08 12.603 & 25 & [email protected] ... & vla & 1210 & 8.4 & 43 & 14 17 59.5400 & + 25 08
12.603 & 25 & [email protected] ... & vla & 1210 & 4.9 & 64 & 14 17 59.5400 & + 25 08 12.603 & 25 & [email protected] ... & vla & 1210 & 8.4 & 46 & 14 17 59.5390 & + 25
08 12.623 & 25 & [email protected] ... & vla & 1210 & 4.9 & 62 & 14 17 59.5390 & + 25 08 12.623 & 25 & [email protected] ... & vlba & 2.3,1.1,@xmath66 & 8.4 & 92 & 14 17 59.53973 & + 25 08 12.6048 & 1.3 & [email protected] ... & vlba & 2.3,1.1,@xmath66 & 8.4 & 55 & 14 17 59.53972 & + 25 08 12.6049 & 1.3 & [email protected] ... & vlba & 2.3,1.1,@xmath66 & 8.4 & 58 & 14 17 59.53972 & + 25 08 12.6053 & 1.3 & [email protected]
|
the nrao vla and vlba were used from 1988 november to 1998 june to monitor the radio continuum counterpart to the optical broad line region ( blr ) in the seyfert galaxy ngc5548 .
photometric and astrometric observations were obtained at 21 epochs .
the radio nucleus appeared resolved , so comparisons were limited to observations spanning 10 - 60 days and 3 - 4 yr , and acquired at matched resolutions of 1210 mas @xmath0 640 pc ( 9 vla observations ) , 500 mas @xmath0 260 pc ( 9 vla observations ) , or 2.3 mas @xmath0 1.2 pc ( 3 vlba observations ) .
the nucleus is photometrically variable at 8.4 ghz by @xmath1% and @xmath2% between vla observations separated by 41 days and 4.1 yr , respectively .
the 41-day changes are milder ( @xmath3% ) at 4.9 ghz and exhibit an inverted spectrum ( @xmath4 , @xmath5 ) from 4.9 to 8.4 ghz .
the nucleus is astrometrically stable at 8.4 ghz , to an accuracy of 28 mas @xmath0 15 pc between vla observations separated by 4.1 yr and to an accuracy of 1.8 mas @xmath0 0.95 pc between vlba observations separated by 3.1 yr .
such photometric variability and astrometric stability is consistent with a black hole being the ultimate energy source for the blr , but is problematic for star cluster models that treat the blr as a compact supernova remnant and , for ngc5548 , require a new supernova event every 1.7 yr within an effective radius @xmath6 42 mas @xmath0 22 pc . a deep image at 8.4 ghz with resolution 660 mas
@xmath0 350 pc was obtained by adding data from quiescent vla observations .
this image shows faint bipolar lobes straddling the radio nucleus and spanning 12 arcsec @xmath0 6.4 kpc .
these synchrotron - emitting lobes could be driven by twin jets or a bipolar wind from the seyfert 1 nucleus .
|
2rxp j130159.6 - 635806 , first discovered by the _ rosat _
observatory , was later rediscovered in hard x - rays by the _ integral_/ibis telescope and designated with the name igrj13020 - 6359 @xcite .
the first comprehensive analysis of the temporal and spectral x - ray properties of this source was done by @xcite using data from the _ asca _ , _ bepposax _ , _ integral _ and _ xmm - newton _ observatories . in particular , _ xmm - newton _ data showed coherent pulsations with a period of around 700 s. joint spectral analysis of _ xmm - newton _ and _ integral _ data demonstrated that the spectral shape is very typical for accretion - powered x - ray pulsars ( namely , an absorbed power law with a high - energy cut - off ) . based on 2mass archival data , @xcite proposed that the binary companion to 2rxp j130159.6 - 635806 is a be star at a distance of 47 kpc . this suggestion was later confirmed by , who reported the presence of emission lines of hei @xmath42.0594 @xmath5 m and br(74 ) @xmath42.1663 @xmath5 m , which are typical for a be star .
the spectral type of the optical counterpart was determined to be b0.5ve .
the orbital period of the binary remains unknown .
x - ray pulsars in binary systems with be companions ( bexrps ) typically manifest themselves as transient sources through either type i ( periodic flares related to the periastron passage ) or type ii outbursts ( powerful rare transient events ) , or a combination of both .
2rxpj130159.6 - 635806 shows several differences from a standard transient bexrp .
specifically , it has a relatively low persistent flux , a long pulse period , and it does not demonstrate either type i or type ii outbursts .
@xcite , however , did report some variability of its x - ray flux .
therefore , there are substantial reasons to consider 2rxpj130159.6 - 635806 as a member of the subclass of persistent bexrps @xcite .
so far , only a few members of this relatively small category of objects have been studied in detail : 4u 0352 + 309/x persei , rx j0146.9 + 6121/ls i + 61 235 , rx j0440.9 + 4431 , and rx j1037.5 - 564 @xcite . in this paper , we present results of a comprehensive analysis of the temporal and spectral properties of 2rxpj130159.6 - 635806 in a broad energy range , finding some properties that are very unusual for bexrps .
all errors are quoted at the 90% confidence level unless otherwise stated .
rxp was initially serendipitously observed with _ nuclear spectroscopic telescope array ( nustar ) _
@xcite during observations of the gamma - ray binary , with three data sets taken in 2014 may - june ( chernyakova et al .
, in prep . ) . in one observation , rxp appears in the corner of @xmath6 s field of view ( fov ) , and , in two more , the source is at the extreme edge of the fov . despite the large off - axis angles , the data were successfully used to extract coherent pulsations .
this motivated the _
team to trigger an on - axis 30 ks observation of rxpin order to obtain high - quality data for spectral and timing analysis
. table [ tab : log ] lists the _ nustar _
observations used in this work .
cclccccccccc seq . &
obs . & & exp .
& & period + num .
& i d & & [ ks ] & & [ s ] + 1 & 30002017004 & 2014 - 05 - 04 10:01 & 33.3 & @xmath7 & @xmath8 & @xmath9 + 2 & 30002017008 & 2014 - 06 - 02 19:21 & 26.4 & @xmath10 & @xmath11 & @xmath12 + 3 & 30002017010 & 2014 - 06 - 14 17:21 & 29.1 & @xmath13 & @xmath14 & @xmath15 + 4 & 30001032002 & 2014 - 06 - 24 00:06 & 31.6 & @xmath16 & @xmath17 & @xmath18 + _ nustar _ carries two co - aligned identical x - ray telescopes operating in wide energy band from 3 to 79 kev with angular resolution of 18
( fwhm ) and half - power diameter ( hpd ) of 58 .
spectral resolution of @xmath19 ev ( fwhm ) at @xmath20 kev is provided by independent focal planes for each telescope , usualy referred as focal plane module a and b ( fpma and fpmb ) .
_ nustar _ data can have systematic positional offsets as high as @xmath21 .
prior to extraction , we therefore corrected the world coordinate system ( wcs ) of the event files for the four observations to match the and rxp centroid positions based on cataloged coordinates .
since the psf has wide wings @xcite , we investigated how the surface brightness of the source changes with radius in order to determine regions where the source dominates over the background .
we found that rxp is well above background within @xmath22 ( @xmath23 of psf enclosed energy ; see , e.g. , * ? ? ? * ) , and that the background count rate can be robustly measured at radii @xmath24 @xmath25 from the source .
taking this into consideration , we defined the corresponding extraction regions shown in fig .
[ fig : map ] for the 4th ( on - axis ) observation .
the other three observations , for which rxp is far off - axis , have been treated similarly . following _
nustar _ recommended standard practice , we chose the background regions to be on the same detector chip as the source .
rxp is a known source of coherent x - ray pulsations at a period of @xmath26 s with an average spin - up rate of @xmath27 hz s@xmath28 @xcite .
we performed timing analysis of the _ nustar _ data using the xronos ( epoch folding tool efsearch ; * ? ? ?
* ) after barycentering the data with _
barycorr_. for each nustar observation , the pulse period and its uncertainty were calculated following the procedure described in @xcite .
namely , a large number ( @xmath29 ) of source light curves were simulated , the pulse period of each one was determined with efsearch , and the distribution of the corresponding pulse periods was constructed .
the mean value of this distribution and its standard deviation were taken as the pulse period and its @xmath30 uncertainty , correspondingly .
table [ tab : log ] lists period results derived from the fpma and fpmb combined light curves .
the inset of fig .
[ fig : period ] shows the evolution of the spin period as a function of time . as seen from table [ tab : log ] and fig .
[ fig : period ] , all four datasets are suitable for pulsation detection .
it is also quite evident that periods recorded over the time span of 50 days are not consistent with each other , clearly showing a decrease in the period .
we utilized the 1st and 4th observations , which have the most accurate period measurements and also span the full duration of the _ nustar _ coverage , to measure a period derivative of @xmath31 s / day , equivalent to @xmath32 s s@xmath28 , or @xmath33 hz s@xmath28 .
this is in agreement with the @xcite spin - up measurement of the second interval of their data , after the ` break ' at mjd @xmath34 ( @xmath35 hz s@xmath28 ) .
this is quite remarkable since there is almost a decade between the period measurements .
rxp regularly fell into the fov of various x - ray telescopes thanks to extensive observational campaigns dedicated to which is located only @xmath36 away .
this allows us to investigate the long - term evolution of the pulse period .
we analyzed the _ xmm - newton _
@xcite archival data from 2007 - 2011 using the procedures described by @xcite and science analysis software ( sas ) version 14.0.0 .
the list of selected _ xmm - newton _ observations with corresponding period measurements are shown in table [ tab : xmm ] .
cccccccccccc obs .
& start time & exp . &
period + i d & [ utc ] & [ ks ] & [ s ] + 0504550501 & 2007 - 07 - 08 12:01 & 14.1 & @xmath37 + 0504550601 & 2007 - 07 - 16 19:59 & 55.3 & @xmath38 + 0504550701 & 2007 - 08 - 17 08:38 & 11.4 & @xmath39 + 0653640401 & 2011 - 01 - 06 17:37 & 19.9 & @xmath40 + 0653640501 & 2011 - 02 - 02 18:59 & 26.3 & @xmath41 + 0653640601 & 2011 - 03 - 04 05:59 & 12.9 & @xmath42 + fig .
[ fig : period ] presents the long - term evolution of the period , showing that the rxp neutron star has undergone very strong and steady spin - up during the last 20 years .
the first available value of spin period , measured in 1994 , is 735 s @xcite .
the most recent measurements by , from 2014 , show the period to be around 643 s ( table [ tab : log ] ) .
this means that during the last @xmath43 years the spin period decreased by @xmath44 s , corresponding to a mean spin - up rate of @xmath45 s s@xmath28 .
[ fig : period ] shows a change in the average spin - up rate , first reported by @xcite .
we approximated the long - term period evolution with a linear function with one change in slope .
a fit shows that the break occurred at mjd @xmath46 ( mid 1999 ) with a spin - up rate before and after the break of @xmath47 s s@xmath28 and @xmath48 s s@xmath28 , respectively .
as seen from the fit parameters , the spin - up rate becomes significantly higher after the break . as noted above , the inset of fig .
[ fig : period ] shows that the data points fit the long - term spin - up rate with remarkably high precision .
similar behavior was observed for the x - ray pulsar gx 1 + 4 , which showed steady spin up for more than a decade .
however , gx 1 + 4 belongs to the subclass of accreting x - ray pulsars known as symbiotic x - ray binaries ( syxbs ) . for bexrps ,
persistent sources typically demonstrate pulse periods that are relatively stable .
examples include the population of be systems in the small magellanic cloud @xcite and the well - known low luminosity galactic system x persei @xcite .
transient bexrps show strong spin - up during type i and type ii outbursts @xcite with significant spin - down episodes in between ( see , e.g. , * ? ? ?
therefore , 2rxp j130159.6 - 635806 is a unique source among the bexrps because it demonstrates steady and high long - term spin - up with a relatively low and stable luminosity .
pulsar pulse profiles and their evolution with luminosity and energy band depend on the geometrical and physical properties of the emitting regions in the vicinity of the neutron star . in fig .
[ fig : pprof ] , the _ nustar _ pulse profiles of rxp are shown in three different energy bands : 310 , 1020 and 2040 kev .
the lower panel shows `` soft '' ( ( 1020)/(310 ) kev ) and `` hard '' ( ( 2040)/(1020 ) kev ) hardness ratios . at all energies
, the pulse profile can roughly be divided into one main peak at phases 0.00.5 and two smaller peaks at phases 0.50.75 and 0.751.0 .
the main feature that changes with energy is the depth of the minimum at phase 0.0 . as shown in the lower panel of fig .
[ fig : pprof ] , while the `` hard '' hardness ratio is almost constant , the `` soft '' hardness ratio shows a maximum at phase 0.0 due to an increase in the depth of the minimum at 310 kev .
such behavior is caused by differences in the source spectrum with pulse phase ( see section 4 ) . fig . [ fig : ppfrac ] shows the pulsed fraction as a function of energy .
the pulsed fraction is defined as @xmath49 , where @xmath50 and @xmath51 are the maximum and minimum intensities in the pulse profile , respectively .
defined in this way , the pulsed fraction is very high ( about 80% ) . the alternative way to characterize
the pulsed fraction is the relative root mean square ( rms ) , which can be calculated using the following equation : @xmath52 where @xmath53 is the background - corrected count rate in a given bin of the pulse profile , @xmath54 is the count rate averaged over the pulse period , and n is the total number of phase bins in the profile ( @xmath55 in our analysis ) .
the @xmath56 deviation obtained in this way reflects the variability of the source pulse profile in a manner that is not sensitive to outliers like the narrow features seen in the profile of rxp .
therefore , this quantity has a value of around 30% that is much lower than the classically determined pulsed fraction and also independent of the energy band ( see fig .
[ fig : ppfrac ] ) .
it is interesting to note that in contrast to the majority of x - ray pulsars @xcite , rxp does not show an increase in the pulsed fraction at higher energies .
such uncharacteristic behavior was previously observed for another persistent bexrp rxj0440.9 + 4431 @xcite . on the other hand ,
[ fig : pprof ] shows that the pulsed fraction increases somewhat with energy if @xmath51 is defined from the the pulse plateau rather than from the pulse minimum . in other words , the peak - to - plateau difference slightly grows with energy .
the observed 20-year strong and steady spin - up of rxp reveals the existence of a long - term accelerating torque , which indicates that the binary interaction must lead to regular accretion over a decade - long time scale , although this could certainly be episodic ( e.g. , at periastron passages ) .
the torque can be transferred by matter accreted from either the disc around the neutron star or a stellar wind from the optical counterpart .
unfortunately , there is no strong observational evidence allowing us to distinguish between these two different accretion channels . in both scenarios ,
this process is defined mainly by the mass accretion rate and the magnetic field strength ( see , e.g. , * ? ? ?
* ) . due to the unknown distance of rxp ,
the luminosity and mass accretion rate are highly uncertain ; the magnetic field is also unknown since no cyclotron line is found in the energy spectrum ( sect .
[ section : spec ] ) .
however , some qualitative conclusions about the interaction between the accretion disk and the neutron star magnetosphere can be made from the noise power spectrum of the x - ray pulsar . according to the `` perturbation propagation '' model , stochastic variations of viscous stresses in the accretion disc cause variations of the mass accretion rate @xcite .
this , in turn , results in a specific shape of the power density spectrum ( pds ) of the emerging light curve .
namely , it will appear as a power law with slope 1 to 1.5 ( but the exact value is not well established for x - ray pulsars ) up to the break frequency , which is the highest frequency that can be generated in the accretion disk @xcite . in the case of a highly magnetized neutron star , this maximal frequency
is limited by the keplerian frequency at the magnetospheric radius , above which one can expect a cutoff in the source pds .
if the source stays in spin equilibrium ( corotation ) , the cutoff frequency will coincide with the spin frequency of the pulsar , whereas , in the case of spin - up ( increased mass accretion rate ) , the magnetosphere will be squeezed and additional noise will be generated at higher frequencies . if the mass accretion rate is known ( i.e. , the distance to the source
is known ) , this property of the pds can be used to estimate the magnetic field strength of the neutron star .
the appearance of the break in the pds does not necessarily indicate that accretion is from a disk .
there are wind - accreting sources in spin equilibrium also showing a break in their pdss around the pulse frequency @xcite .
however , the evolution of the pds shape aw a function of mass accretion rate in such systems is not well studied . in fig .
[ fig : pds ] , we show the pds of rxp obtained with the _ nustar _ data in the 4th observation after subtracting the pulse profile folded with the measured period from the light curve .
the solid line represents the fitting model in the form of a broken power law .
the measured break frequency is 0.0066 hz ( shown by dotted line ) , and it is clear that it is shifted towards higher frequencies relative to the spin frequency in this observation ( 0.0015 hz ; shown by dashed vertical line ) .
the power - law slope above the break frequency is fixed at 2 .
the best - fit value of the slope below the break is @xmath57 .
given the steady persistent luminosity , we conclude that the spin - up observed during last @xmath58@xmath59 years is caused by a long - term mass accretion rate that is high enough to squeeze the magnetosphere inside the corotational radius .
this finding confirms the uniquness of rxp among the other x - ray pulsars in binary systems with be companions .
we used _ nuproducts _ , a part of the nustardas package , to extract source and background spectra and to generate response matrix ( rmf ) and effective area ( arf ) files for a point source . in our analysis we utilized the most recent calibration database ( caldb ) , version 20150316 .
the extracted spectra were then grouped to have more than 20 counts per bin using the _ grppha _ tool from the heasoft 6.15.1 package .
we fit the spectra using xspec version 12.8.1 @xcite . according to measurements from the _ asca _ and _
xmm - newton _ observatories , the spectrum of rxp is characterized by a moderate absorption value @xmath60 @xmath61 , which is stable over a dozen years @xcite .
this value was obtained by approximating the source spectra as a power law model modified by interstellar absorption ( wabs model in the xspec package ) .
note that it is just slightly higher than the value of the interstellar hydrogen absorption @xmath62 @xmath61 determined by @xcite in the direction of rxp . in the first three observations rxp serendipitously appeared highly offset from the optical axis , and at the large
off - axis angles the absolute flux measurements can be affected by inaccurate psf positioning and subsequent inapropriate weighting in the spectrum extraction procedure .
this leads to large systematics for the extracted spectra .
therefore , we restrict detailed analysis of the source spectrum to the data obtained in the 4th ( on - axis ) observation , which provides high - quality data . in general , the spectrum of rxp has a shape which is typical for accreting pulsars in binary systems , showing a hard spectrum with an exponential cutoff at high energies .
[ fig : spectr_onaxis ] ( a ) presents the phase - averaged spectrum approximated with the most suitable spectral model determined below .
we initially modeled data with a cutoff power - law model ( cutoffpl in the xspec package ) @xmath63 where @xmath64 is the photon index , @xmath65 is the characteristic energy of the cutoff ( e.g. , the folding energy ) , and @xmath66 is a normalization .
this model was modified by interstellar absorption in the form of the wabs model . as the working energy range of the _ nustar _ observatory begins at 3 kev ,
it is not very sensitive to measuring low absorption columns . therefore , in the following analysis , the interstellar absorption was fixed to the value @xmath67 @xmath61 measured by @xcite .
note that we simultaneously fitted spectra obtained by both _
_ modules . to take into account the uncertainty in their relative calibrations , which may be even more of a concern for the observations where the source is highly offset from the optical axis
, we added a cross - calibration constant between the modules .
the fitting parameters for different data sets are shown in table [ tab : spectr ] .
the ` cutoff ' model approximates the source spectrum relatively well with @xmath68 for 921 degrees - of - freedom ( _ dof _ ; see table [ tab : spectr ] ) .
nevertheless , a wave - like structure is clearly seen in the residual panel fig .
[ fig : spectr_onaxis ] ( b ) . to improve the quality of the fit
, we applied another continuum model in the form of a power - law multiplied by a high - energy cutoff ( highecut model in the xspec package ; @xcite ) .
this model can be written as : @xmath69 where @xmath70 is the energy where the cutoff starts .
as in the previous case , we fixed the interstellar absorption value .
residuals of modeling the source spectrum with ` highecut ' are presented in fig .
[ fig : spectr_onaxis ] ( c ) , and best fit parameters are listed in table [ tab : spectr ] .
the ` highecut ' model significantly improves the quality of the fit ( @xmath71 for 920 _ dof _ ) .
the high - energy cutoff value @xmath70 is found to be @xmath72 kev , which is significantly lower than the value of @xmath73 kev reported by @xcite using simultaneous _ xmm - newton _ and _ integral _ data .
the apparent discrepancy is probably due to the lower statistical quality of the _ integral _ data and the gap between the energy bands covered by the _ xmm - newton _ and _ integral _ observatories .
an additional possible problem is that the cutoff energy @xmath74 kev is very close to the energy of the iron fluorescent line at 6.4 kev .
the simplistic ` highecut ' model might hide the presence of the emission line in the source spectrum . observed deviations of the measured spectrum from the model near this energy argue in favor of this possibility . to investigate this issue
, we added a gaussian emission line to the ` highecut ' model , fixing its energy to 6.4 kev and width to @xmath75 kev , allowing its normalization to be a free parameter .
this resulted in an additional improvement of the fit to @xmath76 for 919 _ dof _ for a normalization of @xmath77 ph @xmath61 s@xmath28 .
the corresponding equivalent width of the line is @xmath78 ev ( @xmath79 error ) .
we determined the significance of the line using the xspec script _ simftest _ with @xmath80 trials , and found that presumption against the null hypothesis , or no line required by the data is @xmath81 , which corresponds to @xmath2 line detection , assuming a normal distribution .
the residuals of the ` highecut ' model with the 6.4 kev iron line are shown in fig .
[ fig : spectr_onaxis ] ( d ) . the model itself , together with spectral data points ,
is shown in fig .
[ fig : spectr_onaxis ] ( a ) .
ccllccccccccc & & & & & & & @xmath82 ( dof ) + num . & & & & & & fpma & fpmb & & & + 1 & hi & @xmath83 & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & 1.06 ( 458 ) +
2 & hi & @xmath89 & @xmath90 ( frozen ) & @xmath91 & @xmath92 & @xmath93 & @xmath94 & 0.92 ( 124 ) + 3 & hi & @xmath95 & @xmath90 ( frozen ) & @xmath96 & @xmath97 & @xmath98 & @xmath99 & 0.81 ( 264 ) + 4 & cu & @xmath100 & @xmath101 & & @xmath102 & @xmath103 & @xmath104 & 1.07 ( 921 ) + & hi & @xmath100 & @xmath105 & @xmath72 & @xmath106 & @xmath103 & @xmath104 & 1.01 ( 920 ) + & hi+fe & @xmath100 & @xmath107 & @xmath108 & @xmath109 & @xmath103 & @xmath104 & 1.00 ( 919 ) + considering the moderate spectral resolution of the observatory and the low significance of the detected iron line , we investigated the possibility of the presence of an iron fluorescence line in the _ xmm - newton _ spectrum of rxp .
as mentioned above , the x - ray pulsar rxp was observed with _
xmm - newton _ many times during programs studying . for our purposes , we chose the two observations with the longest exposures
obsid 0092820301 ( @xmath5841.2 ksec ) and obsid 0504550601 ( @xmath5855.3 ksec ) . as for the observations
, we modeled the spectra of rxp including the gaussian line at 6.4 kev . for both of the _ xmm - newton _ observations , we did not find a significant improvement in the fits when the iron line was added and obtained a conservative upper limit ( @xmath79 ) for the equivalent width of the iron line of 110 ev , which is consistent with the results .
despite the fact that the first three observations were made at large offset angles , and the corresponding statistics are significantly lower than for the 4th observation , some useful spectral information can still be extracted . for these observations we used the same ` highecut ' model .
due to low statistics and poor fit we fixed power - law slope in the 2nd and 3rd data set at @xmath110 value measured in the 4th observation . as seen from table [ tab : spectr ] , where the best fit parameters are listed , the principal parameters power - law slope ( @xmath64 ) , cut - off energy ( @xmath70 ) , and folding energy ( @xmath65 ) of the first three high - offset observations are in general agreement with 4th ( on - axis ) observation modeled with ` highecut ' . the estimated @xmath111 kev flux of rxp is about @xmath112 during our observations , which is in agreement with the value of @xmath113 measured by @xcite .
an absence of strong transient activity from rxpon long time scales is confirmed with the rxte / asm and swift / bat instruments .
note that @xcite reported on a flaring episode , but the flux of the flare was relatively low raising by a factor of a few ( up to @xmath114 ) .
finally , in order to check for the possible presence of a cyclotron absorption line in the source spectrum , we modified the best - fit model ( wabs*(powerlaw*highecut+gau ) ) by including an absorption feature in the form of a lorentzian optical depth profile ( cyclabs model in xspec ; * ? ? ?
the search procedure was performed following the prescription from @xcite .
namely , we varied the energy of the line over the range between 5 and 50 kev with 3-kev steps and the line width between 2 and 12 kev with 2-kev steps , leaving the line depth as a free parameter .
we did not find strong evidence for a cyclotron line in the spectrum of rxp ( no trials gave a significance higher than @xmath115 ) . in order to study the evolution of the source spectrum over the pulse period , we performed pulse phase - resolved spectroscopy using the data from the 4th observation .
the period was divided into 10 phase bins with zero phase coinciding with the main minimum in the pulse profile . according to the pulse phase - averaged spectral analysis ,
the fitting model was chosen in the form of ( wabs*(powerlaw*highecut+gau ) ) . however , due to much lower statistics , the photon index was fixed at the value from the averaged spectrum ( @xmath110 ) .
we selected this parameter due to its virtual constancy over the pulse in our preliminary analysis of the same data .
the @xmath116 value is not very well constrained by the data ; however we checked that it is consistent with being constant in phase , and we fixed it as well .
the results are shown in fig .
[ fig : spectr_resol ] .
the average pulse profile of rxp across the entire _ nustar _
energy range is presented in the upper panel .
the lower panels demonstrate the behavior of the two free spectral parameters : the cut - off energy ( @xmath70 ) and the folding energy ( @xmath65 ) .
the cut - off energy is quite stable over the pulse except at the pulse minimum where its value increases approximately by a factor of two .
the folding energy demonstrates an apparent correlation with the pulse intensity , which is probably caused by increasing a spectral hardness around the pulse maximum .
such behaviour of the spectral parameters over the pulse can explain the corresponding behavior of the hardness ratios constructed from the pulse profiles in different energy bands , previously shown in fig .
[ fig : pprof ] . the @xmath117 kev hardness ratio demonstrates correlation with the pulse intensity and the @xmath118 kev ratio peaks at the pulse minimum right in place where @xmath70 shifts from @xmath119 to @xmath120 kev . finally , it is necessary to note that the observed behavior of spectral parameters with the pulse phase can be caused by both physical and artificial reasons ( in particular , due to limitations of available data , the adopted spectral model , and which model parameters were fixed ) .
more data are required to confidently constrain all the spectral parameters and trace their behavior with the pulse phase .
we summarize the results of spectral and timing analysis of serendipitous and dedicated observations of the accreting x - ray pulsar rxp in 2014 may - june . *
the source demonstrates strong pulsations with a period of @xmath121 s. the @xmath3 pulsed fraction is measured to be constant with energy up to 40 kev . *
the pulse profile is virtually independent of energy and can roughly be divided into one main peak at phases 0.00.5 and two smaller peaks at phases 0.50.75 and 0.751.0 .
the only feature that is changing with energy is the depth of the main minimum at phase 0 . *
the measured period shows a significant change over the @xmath122 day time span of the observations , which is consistent with a spin - up rate of @xmath33 hz s@xmath28 .
this rate is in remarkable agreement with measurements made by @xcite almost a decade ago . * together with the results of @xcite , the _ xmm - newton _ data taken in 2007 and 2011 , and the observations , we show a long - term spin - up trend of the source during the last 20 years . during the last 15 years
, the source has undergone a strong and steady spin - up rate at the level of @xmath123 s s@xmath28 . *
the power density spectrum of the source shows a clear break at 0.0066 hz , which is higher than the period frequency of 0.0015 hz .
this fact , together with the steady persistent luminosity of the source , implies that the spin - up observed during the last 20 years is likely caused by a long - term mass accretion rate high enough to squeeze the magnetosphere inside the corotational radius , which makes rxp unique among the other x - ray pulsars in binary systems with be companions .
* the phase - averaged spectrum of the source has a typical shape for accreting neutron stars in binary systems , in particular , for x - ray pulsars , and demonstrates an exponential cutoff at high energies .
our best - fit model contains an absorbed power - law with @xmath124 modified by a high - energy spectral drop with a cut - off energy of @xmath125 kev and a folding energy of @xmath126 kev .
the spectrum also shows @xmath2 evidence for an iron 6.4 kev emission line , with an improvement in the fit when it is included . *
the observed flux corresponds to an _ unabsorbed _ luminosity in the range @xmath127 , assuming a source distance between 4 and 7 kpc @xcite .
these luminosity values imply that the source is a member of the subclass of persistent low luminosity be systems .
* the phase - resolved spectroscopy shows some differences in source spectrum with phase .
the cut - off energy is very stable over the pulse except at zero phase where its value increases by a factor of two .
the apparent correlation of the folding energy with pulse intensity is attributed to the change in hardness of the source spectrum with orbital phase .
this research has made use of data obtained with , a project led by caltech , funded by nasa and managed by nasa / jpl , and has utilized the nustardas software package , jointly developed by the asdc ( italy ) and caltech ( usa ) .
this research has also made use of data obtained with xmm - newton , an esa science mission with instruments and contributions directly funded by esa member states .
al and st acknowledge support from russian science foundation ( grant 14 - 12 - 01287 ) .
an , h. , madsen , k. k. , westergaard , n. j. , et al .
2014 , , 9144 , 91441q arnaud , k. a. 1996 , astronomical data analysis software and systems v , 101 , 17 bildsten , l. , chakrabarty , d. , chiu , j. , et al .
1997 , , 113 , 367 bird , a. j. , barlow , e. j. , bassani , l. , et al .
2006 , , 636 , 765 boldin p. a. , tsygankov s. s. , lutovinov a. a. , 2013 , astl , 39 , 375 chernyakova , m. , lutovinov , a. , rodrguez , j. , & revnivtsev , m. 2005 , , 364 , 455 churazov , e. , gilfanov , m. , & revnivtsev , m. 2001 , , 321 , 759 coleiro a. , chaty s. , zurita heras j. a. , rahoui f. , tomsick j. a. , 2013 , a&a , 560 , aa108 dickey , j. m. , & lockman , f. j. 1990 , , 28 , 215 doroshenko , v. , santangelo , a. , doroshenko , r. , et al . 2014 , , 561 , aa96 ghosh , p. , & lamb , f. k. 1979 , , 234 , 296 gonzlez - galn , a. , kuulkers , e. , kretschmar , p. , et al .
2012 , , 537 , aa66 haberl , f. , angelini , l. , motch , c. , & white , n. e. 1998 , , 330 , 189 harrison , f. a. , craig , w. w. , christensen , f. e. , et al .
2013 , , 770 , 103 hoshino , m. , & takeshima , t. 1993 , , 411 , l79 jansen , f. , lumb , d. , altieri , b. , et al .
2001 , , 365 , l1 klus , h. , ho , w. c. g. , coe , m. j. , corbet , r. h. d. , & townsend , l. j. 2014 , , 437 , 3863 krivonos , r. a. , tomsick , j. a. , bauer , f. e. , et al .
2014 , , 781 , 107 leahy d. a. , darbro w. , elsner r. f. , weisskopf m. c. , kahn s. , sutherland p. g. , grindlay j. e. , 1983 , apj , 266 , 160 lutovinov a.a .
, tsygankov s.s . , 2009 ,
lett . , 35 , 433 lutovinov , a. , tsygankov , s. , & chernyakova , m. 2012 , , 423 , 1978 lyubarskii , y. e. 1997 , , 292 , 679 masetti , n. , pretorius , m. l. , palazzi , e. , et al .
2006 , , 449 , 1139 mihara t. , makishima k. , ohashi t. , sakao t. , tashiro m. , 1990 , natur , 346 , 250 postnov , k. a. , mironov , a. i. , lutovinov , a. a. , et al .
2015 , , 446 , 1013 reig p. , roche p. , 1999 , mnras , 306 , 100 reig , p. 2011
, , 332 , 1 revnivtsev , m. g. , sazonov , s. y. , molkov , s. v. , et al .
2006 , astronomy letters , 32 , 145 revnivtsev , m. , churazov , e. , postnov , k. , & tsygankov , s. 2009 , , 507 , 1211 tsygankov s. s. , lutovinov a. a. , 2005 , astl , 31 , 88 tsygankov s. s. , krivonos r. a. , lutovinov a. a. , 2012 , mnras , 421 , 2407 verner , d. a. , ferland , g. j. , korista , k. t. , & yakovlev , d. g. 1996 , , 465 , 487 white n. , swank j. , holt s. , 1983 , , 270 , 771 wilms , j. , allen , a. , & mccray , r. 2000 , , 542 , 914
|
we present spectral and timing analysis of observations of the accreting x - ray pulsar rxp .
the source was serendipitously observed during a campaign focused on the gamma - ray binary and was later targeted for a dedicated observation .
the spectrum has a typical shape for accreting x - ray pulsars , consisting of a simple power law with an exponential cutoff starting at @xmath0 kev with a folding energy of @xmath1 kev .
there is also an indication of the presence of a 6.4 kev iron line in the spectrum at the @xmath2 significance level .
measurements of the pulsation period reveal that the pulsar has undergone a strong and steady spin - up for the last 20 years .
the pulsed fraction is estimated to be @xmath3 , and is constant with energy up to 40 kev .
the power density spectrum shows a break towards higher frequencies relative to the current spin period .
this , together with steady persistent luminosity , points to a long - term mass accretion rate high enough to bring the pulsar out of spin equilibrium .
|
a double stranded dna ( dsdna ) molecule consists of two polynucleotide strands connected loosely by hydrogen bonds through the base pairs and base - stacking between nearest neighbour pairs of base pairs and wound around each other to make a helix .
the constraints of this helical structure require that the two base sequences on opposite strands must be complementary , with adenine ( a ) always binding to thymine ( t ) and guanine ( g ) binding to cytosine ( c ) @xcite .
the force that holds the complementary strands of dna together is an important regulator of life s processes because the binding of regulatory proteins to dna often involves the procedure of mechanical separation of its strands .
the intermolecular forces of dna have been studied extensively using a variety of techniques , which cover a broad range of forces from a few piconewton(pn ) up to several hundred piconewtons ( pns ) .
these techniques use either atomic force microscopy ( afm ) or laser optical traps and magnetic tweezers @xcite .
the experiments fall into two general categories ; those conducted on short dna and those on long dna .
experiments on long dna focussed on the overall properties of the molecule @xcite and resulted in the discovery of s - dna @xcite ; the occurance of this so called b - s transition has also been found in a short dna chain of 30 base pairs @xcite . those on short duplexes focussed on the reaction pathway of melting , and contributed to the understanding of the local unbinding of dna @xcite . in the single molecule experiments ,
the result may depend on choice of which variables are fixed and which can fluctuate . in some of the experiments ( like afm ) extension is fixed and force is allowed to fluctuate ( constant extension ensemble ) while in others ( like magnetic tweezers ) the force is fixed and extension is allowed to fluctuate ( constant force ensemble ) .
when a force is applied to pull apart the two strands from one end of a dsdna molecule in a direction perpendicular to the helical axis , bases are sequentially stretched as the duplex is unzipped . on the other hand , in shear unzipping
in which the applied force pulls the two strands from opposite ends as shown in fig.1 , the stretching is spread out over many base pairs .
the study of distribution of shearing force along the length of a dna molecule may lead to valuable informations about the force distribution across the phosphate backbone of a single strand of dna versus the force distribution across the paired bases of complementory dna .
such informations are of relevance not only to understanding the biological processes but also in material science application such as determining the strength of dna / gold nanoparticle assemblies @xcite . following the work of lee et al .
@xcite many workers @xcite measured the value of shear force which separates the two strands of a dsdna molecule .
it was found that the force increases linearly with the length .
using a simple ladder model of dna and expressing the backbone bond energy as well as the interaction energy among complementary bases in the form of harmonic springs , de gennes @xcite in 2001 predicted that the critical force for shear unzipping which for short chains shows linear dependence would saturate to a finite value in the limit where the number of base pairs approaches infinity and that the shear stress relaxes over a distance ( number of base pairs ) @xmath2 ( = @xmath3 , where @xmath4 is the spring constant characteristic of stretching the backbone , and r is the spring constant characterstic of stretching the hydrogen bonds between base pairs ) on either side of the chain . the critical force @xmath0 for shear unzipping as a function of number of bound base pairs was given @xcite as f_c=2f_c[^-1tanh()+1 ] .
where @xmath5 is the rupture force of a single bond and ( n+1 ) is the number of base pairs in the molecule .
chakrabarti and nelson @xcite have generalized the simple harmonic model of degennes @xcite by representing the interaction among complimentary bases by the lennard - jones ( 12 - 6 ) potential model and found that the strain is indeed localized over a narrow range of @xmath2 on either side of the chain and the chain unzips when the force exceeds a critical value of @xmath6 for short chains and @xmath7 for long ones as predicted by degennes @xcite .
it may , however , be noted that the calculations of degennes @xcite and also of chakrabarti and nelson @xcite correspond to zero temperature . the langevin dynamics simulation has recently been used to study shear unzipping of dsdna at finite temperature @xcite .
the results found for @xmath0 are in agreement with the results of degennes @xcite and chakrabarti and nelson @xcite .
hatch et al .
@xcite have measured the value of @xmath0 for several dsdna molecules of length ranging from 12 to 50 base pairs at the room temperature and found that @xmath0 is linear function of molecular length only up to 20 base pairs and approaches to an asymptotic value as the number of base pairs increases .
in fact they found that @xmath0 for a molecule of 32 base pairs has already reached within @xmath8 of the asymptotic value which was found to be 61.4 pn .
but when hatch et al .
@xcite tried to fit their data to eq.(1 ) they found that this can be done only by assuming 7 base pairs of all chains to be in the open state , i.e. the effective length of a molecule of ( n+1 ) base pairs is ( n-6 ) irrespective of the value of n. they attributed this to temperature effect as degennes calculation did not include temperature . in this paper
we calculate the value of @xmath0 as a function of the number of base pairs at room temperature and compare our results with the experimental data of hatch et al .
@xcite and show that the need to adjust the length of a molecule is not as much due to temperature as due to use of the values of @xmath5 and @xmath9 in eq.(1 ) and the nature of the curve of @xmath0 @xmath10 n. we derive another form of eq.(1 ) and show that with reasonable choice of values of @xmath5 and @xmath9 one gets values of @xmath0 which are in better agreement with experiment without adjusting the length of molecules than that found from eq.(1 ) . the model which we describe in sec .
ii is similar to the nonlinear scaler displacement model of chakarbarti and nelson @xcite .
we calculate the force in both the constant force and constant extension ensembles . the paper is organized as follows . in sec
ii we describe the model and calculational procedures . in sec iii
we give results found for distribution of effects created by shear force along the length of a molecule and the value of @xmath0 as a function of number of base pairs . in sec
iv we compare our results with those of degennes @xcite and with the experimental values @xcite .
the paper concludes with a brief comment given at the end of sec iv .
we consider a dsdna molecule of length ( n+1 ) base pairs which both 5@xmath11-ends ( or both 3@xmath11-ends ) are pulled along the helical ( molecular ) axis by a force @xmath12 as shown in fig.1 .
the displacements of @xmath13 nucleotide from its equilibrium position are denoted by @xmath14 for one ( the lower one in fig.1 ) single stranded dna ( ssdna ) chain and by @xmath15 for the other ( the upper one ) chain .
the effective hamiltonian of the system can be written as @xcite h=&&_i =- n/2^n/2 - 1+_i =- n/2^n/2v(|*u*_i -*v*_i| ) + & & -*f*.(*u*_n/2-*v*_-n/2 ) .
the first term of this equation represents the stretching energy of nucleotides , excecuting a simple harmonic motion with a spring constant @xmath4 along each chains in dsdna . in the second term
, @xmath16 represents the potential energy of interaction between bases in the @xmath17 pair and the last term of eq.(2 ) represents the energy contribution due to the shear stress . in writing eq.(2 )
we did not include the contribution arising due to the helicity of dna as this effect was found to be negligible by hatch et al . @xcite and also by lavery and lebrun @xcite . for the potential @xmath16 we use a simple model which has a hard - core repulsion and a long range attraction , @xmath18 = - for > 0 .
here @xmath19 is the magnitude of increase in length of hydrogen bonds connecting bases in the @xmath20 pair from its equilibrium value , @xmath21 is the depth of potential at the equilibrium separation and @xmath22 is the diameter of dsdna ( see fig.2 ) .
the repulsion represents the steric hindrance which forbids the molecule from getting compressed along the bond linking the bases with respect to its equilibrium value . from fig.2
one finds , ( 1+)= ( 1+)^1/2 , we take the value of @xmath22 equal to @xmath23 which is the diameter of the _ canonical _ dna ( b - dna ) at room temperature . since in shear unzipping
, a molecule is stretched along its axis we consider the longitudinal displacements of nucleotides and neglect the transverse displacements ; the transverse displacements have been found in ref @xcite an order of magnitude smaller than those in the direction of shear .
we define new variables , x_i= , y_i= , where , @xmath24 and @xmath25 now represent the longitudinal displacements of respective strands . in the notations used here
@xmath24 and @xmath25 are positive when the @xmath26 nucleotide of upper and lower strands move to _ r.h.s . _ and negative when they move to _
l.h.s._. for the experimental situation shown in fig .
1 it is clear that for @xmath27 , @xmath28 and magnitute of @xmath24 is greater than that of @xmath25 and for @xmath29 , @xmath30 and the magnitute of @xmath25 is greater than that of @xmath24 .
this leads to following relation for variables @xmath31 and @xmath32 ; x_-i =- x_i , x_0=0 and y_-i = y_i .
when we substitute these variables in eq.(2 ) it decouples into two independent components ; h = h_x+h_y , where h_x=_i =- n/2^n/2 - 1(x_i+1-x_i)^2 -(x_n/2-x_-n/2 ) , and h_y=_i =- n/2^n/2 - 1(y_i+1-y_i)^2 + _ i =- n/2^n/2v(y_i)-(y_n/2+y_-n/2 ) . here
the potential @xmath16 defined in eq.(3 ) is expressed in terms of variable @xmath32 .
using eqs .
( 4 ) and ( 5 ) , one can rewrite eq.(3 ) as v(y_i)=- .
note that the expression of @xmath33 does not contain the on - site potential @xmath34 and simply corresponds to a harmonic chain which is being pulled at the two ends by a force @xmath35 whereas the expression of @xmath36 contains the on - site potential @xmath34 as well as the force term . in view of the relations given by eq.(6 ) the average value of displacement @xmath37 of @xmath38 base pair can be calculated from the relation , = , where @xmath39 , @xmath40 being the boltzmann constant and @xmath41 is the temperature . since for @xmath33 given by eq.(7 ) the integrals in eq.(11 ) are gaussians , one can solve them analytically to give = n. this result agrees with the one found by degennes @xcite
. the average value of displacement @xmath42 of @xmath43 base pair can be found from the relation , = .
the integrals appearing in this expression can not be evaluated analytically because of the form of on - site potential @xmath34 . however , for the expression of @xmath36 given by eq.(9 ) the integral appearing in eq.(13 ) reduces to multiplication of ( @xmath44 ) matrices .
the discretization of the coordinate variable and introduction of a proper cut - off on the maximum values of @xmath45 determine the size of the matrices .
we have taken -40@xmath46 and 40@xmath46 as the lower and upper limit of integration for each co - ordinate variable and discretized space using the gaussian - legendre method with the number of grid points equal to 900 . by changing the limits of integration as well as the number of grid points we made it sure that the values of @xmath47 are independent of the limit of integration and the number of grid points chosen to discretize coordinate variable .
when a shear force is applied on a dsdna molecule , its two ssdna strands get pulled in opposite directions as shown in fig.1 .
the bonds in the backbone of dna as well as bonds connecting bases in a pair are stretched . in a case of @xmath4
being infinitely large the two strands will move like a rigid body pulling all base pairs in the sequence in parallel .
the effect of the shear force will then be uniformly distributed across all of the base pairs . however , if @xmath4 is finite , then both the backbone and the base pairs will stretch when a shear force is applied .
the effect of the shear force may then be confined to limited lengths on both ends of the molecule . to see how the effect caused by shearing of a dsdna molecule is distributed along the length of a molecule and how this depends on the energies associated with the stretching of backbone and base pairs we calculate the value of @xmath47 from eq.(13 ) for different values of @xmath48 when the two end base pairs are stretched to a given length by the shear force .
in fig.3 we plot our results for four dsdna molecules of length 17 , 25 , 33 and 49 base pairs and for @xmath49 , @xmath50 , @xmath51 and @xmath52 .
the values shown in the figure correspond to @xmath53 and @xmath54 .
when @xmath55 and @xmath56 are allowed to be free ( i.e.@xmath57 ) , then @xmath47 is found to be zero for all @xmath48 . from fig.3
it is clear that for short chains the effect created by shear force is distributed along the entire length of a molecule affecting all base pairs , whereas , for relatively larger molecules the base pairs in the central part of molecules are only marginally affected .
for example , for a molecule of length 17 base pairs the value of @xmath58 is 1.76@xmath59 when @xmath60 , whereas , for the similar situation ( i.e. @xmath61 ) the value of @xmath58 for a molecule of length 49 base pairs is only 0.27@xmath59 .
the qualitative nature of these results are in agreement with the results found in refs @xcite and @xcite .
the other point to be noted from the figure is that the qualitative nature of the distribution of the effect of shear stretching along length of a molecule is same for all values of @xmath62 plotted in the figure .
in fig.4 we plot the value of @xmath58 when @xmath62 = 2.38@xmath59 as a function of length of dsdna molecules .
as discussed below , the value of @xmath62 = 2.38@xmath59 is assumed to be the critical value of stretching in the sense that initiation of separation of two strands starts at this value of @xmath56 and the shear force which creates this value of stretching of the end base pairs is equal to @xmath0 , the minimum ( or critical ) force required to separate the two strands of a dsdna molecule . from figs . 3 and 4 it is clear that as one moves from either ends the differential force across the base pairs decreases and becomes very small at the centre for larger molecules but
has not become zero at the centre even for a molecule of length 49 base pairs .
the extension of the curve of fig.4 shows that it would be zero at the centre only for molecules of length larger than 80 base pairs .
the average value of displacements @xmath63 and @xmath64 of @xmath65 base pair can be found from the known values of @xmath37 and @xmath47 . from eq.(5 ) one gets = and < v_n>=. in fig.5 we plot the values of @xmath63 and @xmath64 as a function of @xmath48 for molecules of length 17 , 21 and 49 base pairs . the values plotted in this figure correspond to @xmath66 = 2.38@xmath59 and f=40.9 , 47.4 and 60.6 pn , respectively for molecules of length 17 , 21 and 49 base pairs .
these values of force , as is shown below , are the critical force of shearing of the respective molecules . from the figure we note that while for a molecule of 17 base pairs the entire ssdna strand moves in the direction of applied force whereas in the case of a molecule of 49 base pairs nearly half of the strand moves in the opposite direction . while for a molecule of 17 base pairs , @xmath67 , @xmath68 and @xmath69 , @xmath70 , for a molecule of 49 base pairs @xmath71 , @xmath72 and @xmath73 , @xmath74 .
as long as the entire ssdna strand moves in the direction of applied force the shear force , as explained in ref @xcite , depends linearly on molecular length .
the departure from the linear dependence of the shear force on molecular length is expected to take place when @xmath75 and @xmath76 become zero i.e. the displacement of nucleotides on the opposite side of a ssdna strand remains unaffected by the applied force .
indeed , we find that @xmath77 and @xmath78=0 for a molecule of length 21 base pairs whcih is in very good agreement with the experimental value @xcite .
since for molecules of base pairs larger than 21 , part of a ssdna moves in the opposite direction , a region develops inbetween on each strand which remains unaffected by the force .
this region moves towards the centre on increasing the molecular length ; the force gets saturated as soon as the region reaches the centre of the chain and stays there on further increasing the molecular length .
the value of shear force needed to separate the two strands of a dsdna molecule can be calculated in two different ways . in one ,
we follow a method proposed by degennes @xcite and which from hereon is referred to as a method of constant force ensemble .
the other method is based on the constant extension ensemble . in the method of constant force ensemble
one first defines a critical distance @xmath79 for the rupture of a base pair ( i.e. when @xmath42 of @xmath65 base pair becomes larger than @xmath79 the bases of the pair become free ) and calculate the value of force which can stretch a base pair to the critical distance @xmath79 .
this force can be found from the on - site potential @xmath80 . thus f_c=
-|_y=|y = ( 1+)^-4 .
if we take the value of @xmath81 , we find @xmath82 for the value of @xmath21 and @xmath4 given above .
this value of @xmath5 is close to the value used by hatch et al @xcite to fit their experimental data and the value estimated by chakrabarti and nelson @xcite . to rupture the end base pairs of a given molecule
they must be stretched by the shear force to distance @xmath79 .
the balance of force at one ends of one of the ssdna gives @xcite ; f_c=(<u_n/2>-<u_n/2 - 1>)+f_c . using the relations of eq.(5 ) and eq.(12 )
we get f_c=(<y_n/2>-<y_n/2 - 1>)+2f_c . taking the value of @xmath83 = 2.38 @xmath59
we calculate the values of @xmath84 from eq.(13 ) for several molecules of length 10 - 60 base pairs . the value of force @xmath0 found from eq.(17 )
is shown by dotted line in fig.6 . in the constant extension ensemble one
first calculate the work done in stretching the two end base pairs of a given molecule to distance @xmath85 .
this work is found from the relation w(y)=[ln z_n+1(y)-lnz_n+1 ] , where z_n+1(y)=_i =- n/2^n/2dy_i(y_-n/2-y)(y_n/2-y ) , is the constrained partition function and z_n+1=_i =- n/2^n/2dy_i . is the partition function .
@xmath86 is the dirac function .
we used the matrix multipliction method to evaluate @xmath87 from the above equations for @xmath88 , @xmath89 and @xmath54 .
the derivative of @xmath87 with respect to @xmath85 gives the average force @xmath90 that is needed to keep the extension of end base pairs of a given molecule equal to @xmath85 .
thus f(y)=. to get the value of critical force we have to chose a value of @xmath85 which corresponds to rupturing of base pairs .
the values shown in fig.6 by dashed line are found when @xmath85 was taken equal to @xmath50 ; this value is slightly lower than the one taken for the constant force ensemble .
the difference in the value of critical stretching in the two ensembles may be due to difference in the path of unzipping .
we note that the values of @xmath0 found by methods of the constant force and the constant extension ensembles are close but not identical .
the difference between the two as expected @xcite is large for molecules of smaller lengths but they becomes close as molecular length increases .
both methods give the same asymptotic value , equal to 61.2pn , which is in very good agreement with the experimental value 61.4pn @xcite .
the experimental values shown in the figure are of hatch et al @xcite . in view of large spread in experimental data
we find good agreement between experimental values of @xmath0 and the theoretical values found using the constant force ensemble as well as the constant extension ensemble .
since the experimental values @xcite of @xmath0 given in fig.6 are found using the fixed force ensemble , we concentrate our discussion with the theoretical values found using the same ensemble and shown in fig.6 by dotted line .
the agreement between theory and experiment is excellent except for a molecule of length 12 base pairs .
we know that at a given temperature shorter dsdna molecules are less stable compared to longer molecules .
it is quite possible that at room temperature due to surface effects few of base pairs at the two ends of a molecule of length 12 base pairs are nearly in open state .
the curve shown in fig.6 by solid line is found when the effective length of a molecule of length 12 base pairs is taken to be equal to 10 base pairs and that of a molecule of length 16 base pairs equal to 15 base pairs .
the values of force @xmath0 shown by solid line is in excellent agreement with the experimental values for the entire range of molecular length investigated by hatch et al . @xcite .
the saturation value of @xmath0 found to be 61.2 pn is also in very good agreement with the experimental value of 61.4 pn @xcite . in order to compare our results with those of degennes @xcite
we first estimate the value of @xmath9 defined as @xmath91 , where r is the spring constant of a simple harmonic potential between the bases of a pair .
expanding @xmath80 in ascending powers of @xmath85 one gets v(y_i)= -+ ( ) y^2_i , r = and^2 = . substituting the values of @xmath21 , @xmath4 and @xmath22 given above we find @xmath92 which corresponds to @xmath93 .
this value of @xmath94 is in good agreement with the predicted value of 77 based on calculation of the spring constants for base pairs and backbones @xcite .
the expression for the displacement @xmath47 found by degennes @xcite can be written as = < y_0>(e^n+e^-n ) .
taking the value of @xmath2=6.44 and the values of @xmath58 determined above ( shown in fig 4 ) we calculate the value of @xmath47 from eq.(24 ) for molecules of length 23 and 49 base pairs and compare them in fig 7 with the values of @xmath47 found from eq.(13 ) .
the two values do not agree ; the difference increases with the length of molecules .
a good agreement is , however , found when @xmath2=9.4 is taken .
this value of @xmath2 is about @xmath95 larger than the value found above .
we calculate @xmath0 from the degennes equation given by eq.(1 ) taking @xmath2=6.44 and 9.4 and @xmath5=4.1 pn . the results are plotted in fig 8 in which we also plot experimental values and the values found from our approach and shown in fig 6 by dotted line .
the values found from eq.(1 ) are very different from both experimental values and values shown by dotted line . in ref
@xcite the value of @xmath96 as a function of n were found to be in good agreement with the values found from eq.(1 ) for @xmath97 .
since eq.(1 ) involves the rupture force @xmath5 for a base pair and @xmath9 , the value of @xmath0 depends on the values of these quantities .
most often quoted value of @xmath5 is in the range of 4 - 5 pn @xcite and that of @xmath98 @xcite .
hatch et al .
@xcite took @xmath5=3.9 pn and @xmath2=6.8 to fit their data to eq.(1 ) and found that they have to adjust the length of molecules which amounted to shifting the curve of @xmath0 to the right direction by 7 unit of base pairs in order to get good agreement with the experimental values .
we now examine whether such an adjustment of length is essential or there is some other reason for not getting good agreement . in the appendix we derive eq.(24 ) from eq.(9 ) and show that @xmath9 which appears in eq.(24 ) should be defined as @xmath9=@xmath99 . substituting the values of @xmath4 and r we find @xmath2=9.2 which is close to 9.4 found by using the values
@xmath47 found from eq.(13 ) . with this definition of @xmath9
we find the following expression for @xmath0 ( see eq.(a8 ) ) .
f_c=2f_c[^-1tanh()+1 ] .
the appearance of @xmath100 instead of @xmath101 ( in eq.(1 ) ) is due to difference in the definition of @xmath9 .
the definition @xmath102 seems more appropriate in the sense that it is ratio of the two harmonic spring constants than the definition @xmath91 of ref @xcite in which one spring constant is multiplied by two .
the value of @xmath0 found from eq.(25 ) when @xmath2=9.4 and @xmath5=4.1 pn are compared in fig 9 with the values using the method of sec .
iii and shown in fig 6 by dotted line and values which are found from eq.(1 ) when @xmath2=6.8 and @xmath5=3.9 pn @xcite . from the figure we first note that eq.(25 ) gives values of @xmath0 as a function of molecular length which are close to the values found in sec .
iii ( dotted line ) ; the difference between the two is due to the combined effect of nonlinearity and the temperature .
thus the effect of temperature is not as large as has been suggested in ref @xcite . the necessity for adjusting the length of molecules is due to the values of @xmath9 and @xmath5 used in eq.(1 ) by hatch et al .
if one takes @xmath2=9.4 and @xmath5=4.1/@xmath95 pn and calculate @xmath0 using eq.(1 ) one gets values shown by full line in fig 9 which are found from eq.(25 ) with @xmath2=9.4 and @xmath5=4.1 pn . in conclusion
, we developed a method to calculate the distribution of shear force along the length of a dsdna molecule at a finite temeprature .
the value of shear force @xmath0 which is required to separate the two strands of a molecule has been calculated in both the constant force and the constant extension ensembles .
the values of @xmath0 found by these methods differ for molecules of shorter length but approach to each other as length increases .
both methods gave the saturation value equal to 61.2 pn which is in very good agreement with the experimental value 61.4pn .
the value of @xmath0 is found to increase linearly with length up to 21 base pairs in agreement with experimental results .
the plots of @xmath63 and @xmath64 given in fig.5 show that as long as the applied force pulls the entire strand in its direction of application , @xmath0 depends linearly on length of molecules .
the departure from linear behaviour takes place when part of a strand moves in opposite direction under the influence of force pulling the other strand .
the saturation value is achieved when half of a strand moves in the direction of force applied on it and the other half in the opposite direction .
it is shown that the value of @xmath0 quickly attains its saturation value on increasing length of dna molecule ; for a molecule of 32 base pairs @xmath0= 57.0pn as compared to saturation value 61.2pn .
the agreement between theorectical and experimental values of @xmath0 shown in fig.6 indicates that the model proposed in this paper is capable of describing the responses of dsdna molecules to shear stress .
the model , however , neglects the effect of helicity of dna and has assumed the molecule to be homogeneous . as far as the effect of helicity is concerned , it has already been shown to be negligible by hatch et al .
@xcite and by lavery and lebrun @xcite .
the effect of heterogenity arising due to random distribution of a - t and c - g base pairs in a sequence is also expected to be small as long as the shear unzipping from the two ends is symmetric .
this is corroborated by the fact that although the molecules investigated by hatch et al @xcite are heterogeneous with half g - c and half a - t base pairs , yet their response to shear stress is described very well by a model which assumes molecules to be homogeneous .
the values of parameters @xmath21 and @xmath4 will , however , depend on the percentage of a - t and c - g base pairs in a given sequence and on the temperature .
but , in the absence of symmetry in the unzipping from the two ends a qualitative new features , as argued in ref @xcite , may arise .
we are grateful to sanjay kumar for drawing our attention to the paper of p. g. degennes and to b. p. mandal for useful discussions .
one of us ( sp ) acknowledges the financial support provided by the university grants commission , india .
y_0= ( a5 ) let the force on the last hydrogen bonds ( @xmath48=@xmath108 ) is @xmath109 when we reach the threshold @xmath5 .
( @xmath5 being the rupture force for a base pair ) .
= ry_0_cosh_(n ) ( a6 ) from eq.(a3 ) we get f_c=2y_n/2 _
tanh_(n)+2 f_c ( a7 ) = 2f_c[^-1_tanh_(n)+1 ] ( a8 ) this equation differs from eq.(1 ) in the definition of @xmath9 and the mutiplying factor which is now @xmath95 instead of 2
. 199 w. saenger , _ principle of nucleic acid structure _ ( springer - verlag , berlin , 1984 ) .
t. strick , j .- f .
allemand , v. croquette and d. bensimon , phys . today * 54 * , 46 ( 2001 ) .
u. bockelmann , curr opin .
* 14 * , 368 ( 2004 ) ; u. bockelmann , b. essevaz - roulet , and f. heslot , phys .
* 79 * , 4489 ( 1997 ) .
s. b. smith , y. cui and c. bustamante , science * 271 * , 795 ( 1996 ) .
m. reif , h. clausen - schaumann and h. e. gaub , nature struct .
* 6 * , 346 ( 1999 ) .
r. s. convory and c. danilowicz , contemp phys . *
45 * , 277 ( 2004 ) .
s. kumar and m. s. li , phys . rep . * 486 * , 1 ( 2010 ) .
y. zhang , h. zhou and z. -c .
ou - yang , biophys .
j. * 81 * , 1133 ( 2001 ) .
i. rouzina and v. a. bloomfield , biophys . j.
* 80 * , 882 ( 2001 ) . h. clausen - schaumann , m. reif , c. tolksdorf and h. e. gaub , biophys .
j. * 78 * , 1997 ( 2000 ) . c. bustamante , s. b. smith , j. liphardt and d. smith , curr .
opin . struct
. biol . * 10 * , 279 ( 2000 ) .
j. morfill , f. k@xmath110hner , k. blank , r. a. lugmaier , j. sedlmair and h. e. gaub , biophys .
j. * 93 * , 2400 ( 2007 ) .
g. u. lee , l. a. chrisey , and r. j. colton , science * 266 * , 771 ( 1994 ) .
a. d. mc kerall and g. u. lee , eur .
j. * 28 * , 415 ( 1999 ) .
l. h. pope , m. l. davies , c. a. laughton , c. j. roberts , s. j. b. tendler and p. m. williams , eur .
j. * 30 * , 53 ( 2000 ) .
t. strunz , k. oroszlan , r. schafer and h .- j .
guntherodt , proc .
* 96 * , 11277 ( 1999 ) .
f. k@xmath110hner , j. morfill , r. a. neher , k. blank and h. e. gaub , biophys .
j. * 92 * , 2491 ( 2007 ) .
b. chakrabarti and d. r. nelson , j. phys .
b * 113 * , 3831 ( 2009 ) .
b. d. sattin , a. e. pelling , and m. c. goh , nucleic acids res .
* 32 * , 4876 ( 2004 ) .
w. grange , t. strunz , i. schumakovitch , h .- j .
guntherodt , and m. hegner , single mol .
* 2 * , 75 ( 2001 ) .
i. schumakovitch , w. grange , t. strunz , p. bertoncini , h .- j .
guntherodt and m. hegner , biophys .
j. * 82 * , 517 ( 2002 ) .
g. neuert , c. h. albrecht and h. e. gaub , biophys .
j. * 93 * , 1215 ( 2007 ) .
y. zhang , v. t. milam , d. j. graves and d. a. hammer , biophys .
j. * 90 * , 4128 ( 2006 ) .
m. v. wal , s. kamper , j. headley and k. sinniah , langmuir * 22 * , 882 ( 2006 ) . a. noy , d. v. vezenov , j. f. kayyem , t. j. meade and c. m. leiber , chem
. biol . * 4 * , 519 ( 1997 ) .
k. hatch , d. danilowicz , v. coljee and m. prentiss , phys .
e * 78 * , 011920 ( 2008 ) .
p. g. de gennes , soryushiron kenkyu * 2 * , 1505 ( 2001 ) .
r. k. mishra , g. mishra , m. s. li and s. kumar , arxiv:1104.3059 ( cond-mat.soft ) ( 2011 ) .
r. lavery and a. lebrun , genetica * 96 * , 75 ( 1999 ) .
d. keller , d. swigon and c. bustamante , biophys .
j. * 84 * , 733 ( 2003 ) .
this value quoted in ref.@xcite was found from m. fytas and e. kaxiras through private communication by the authors of @xcite .
[ fig3 ] as function of @xmath48 for four molecules of length 17 , 25 , 33 and 49 base pairs and for @xmath62 = 1.5@xmath46 ( full line ) , 2.0@xmath46 ( dashed line),2.38@xmath59 ( dotted line ) and 2.6@xmath59 ( dot - dashed line).,title="fig : " ] [ fig6 ] in pn as a function of number of base pairs of a dsdna molecule .
the curve shown by dotted line corresponds to constant force ensemble and dashed line to constant extension ensemble .
the curve shown by full line corresponds to values found by taking the effective length 10 and 15 base pairs for molecules of length 12 and 16 base pairs respectively .
the experimental values of hatch et al @xcite are shown by diamond when 5-ends are pulled and by star when 3-ends are pulled by shear force.,title="fig : " ] [ fig7 ] found from eq.(12 ) ( full line ) and the degennes relation eq.(24 ) ( dashed line ) when @xmath2=6.44 . for @xmath2=9.34 ,
the values found from eq.(24 ) overlap with the curve drawn by full line .
, title="fig : " ] [ fig8 ] as a function of the number of base pairs in a molecule .
the curves shown by dashed and dot - dashed lines represent the values found from eq.(25 ) when @xmath2=9.34 and 6.44 , respectively .
the dotted line represents corresponding curve of fig.6.,title="fig : " ] [ fig9 ] as a function of the number of base pairs in a molecule .
the curve shown by dot - dashed line is found from eq.(1 ) with @xmath2=6.8 and @xmath5=3.9pn @xcite , the curve shown by full line is found from eq.(25 ) with @xmath2=9.4 and @xmath5=4.1pn and the dotted line represents corresponding curve of fig.6.,title="fig : " ]
|
we use a simple nonlinear scaler displacement model to calculate the distribution of effect created by a shear stress on a double stranded dna ( dsdna ) molecule and the value of shear force @xmath0 which is required to separate the two strands of a molecule at a given temperature .
it is shown that for molecules of base pairs less than 21 , the entire single strand moves in the direction of applied force whereas for molecules having base pairs more than 21 , part of the strand moves in the opposite direction under the influence of force acting on the other strand .
this result as well as the calculated values of @xmath0 as a function of length of dsdna molecules are in very good agreement with the experimental values of hatch et al .
( phys .
rev .
e @xmath1 , 011920 ( 2008 ) ) .
|
the interacting binary hd170582 ( bd-14 5085 , asas i d 183048 - 1447.5 , @xmath6 = 18:30:47.5 , @xmath7 = -14:47:27.8 , @xmath8 = 9.66 mag , @xmath9 = 0.41 mag , spectral type a9v ) is a member of the class of double periodic variables , algol - related binaries showing a long photometric cycle lasting about 33 times the orbital period with a period ratio mostly between 25 and 45 ( mennickent et al .
2003 , mennickent et al .
2008 , poleski et al .
2010 , mennickent et al .
2012a , 2012b , mennickent 2013 , garrido et al .
2013 , barra et al .
2013 , 2014 , pawlak et al .
recently , mennickent et al . (
2016 ) showed that dpvs are those semi - detached algols with b - type primaries of radius slightly larger than the critical radius and also that they host stable accretion discs @xcite . in a recent study mennickent et al .
( 2015 , hereafter m15 ) disentangled the light curve of hd170582 into an orbital part , determining a period of 16.87 days and revealing orbital ellipsoidal variability with unequal maxima , and a long cycle of 587 days , showing quasi - sinusoidal changes with amplitude @xmath10= 0.1 mag .
they modeled the orbital light curve with a cool evolved star of @xmath11 = 1.9 @xmath12 0.1 @xmath13 , @xmath14 = 8000 @xmath12 100 @xmath15 and @xmath16 = 15.6 @xmath12 0.2 @xmath17 , and an early b - type dwarf of @xmath18 = 9.0 @xmath12 0.2 @xmath13 surrounded by a geometrically and optically thick accretion disc of radial extension 20.8 @xmath12 0.3 @xmath17 , contributing about 35% to the system luminosity in the @xmath8-band .
m15 fit the light curve asymmetries with two extended regions located at opposite sides of the disc rim , and hotter than the disc by 67% and 46% .
one of these regions can be associated to the impact with , and penetration of the gas stream into the accretion disc .
the origin of the long cycle in dpvs still is not understood , but it could be related to episodic mass loss from the system driven possibly by a bipolar wind ( mennickent et al .
2008 , 2012b ) .
this result shows the importance of imaging the circumstellar material with the aid of tomographic reconstructions , which are still scarce for the class of dpvs ( atwood - stone et al .
2012 , richards et al .
2014 , mennickent et al .
2012b ) .
in this paper we complement the work by m15 , studying the long - term variability of the balmer emission .
we also get insights on the line emissivity distribution at high and low stage with the aid of the technique of doppler tomography .
we also investigate for the first time the evolutionary stage of this binary determining its age and mass transfer rate and provide insights on the luminosity of the discs surrounding the primaries of dpvs .
we divide the paper as follows : in section 2 we present the spectroscopic data used in this paper , in section 3 we give insights on emission line profile variability . in the same section we build doppler tomograms for the h@xmath0 and h@xmath1 emission lines , discuss the system evolutionary stage , obtain a value for the mass transfer rate , and present a study of the luminosity of the dpv accretion discs .
we give a discussion in section 4 and our conclusions are presented in section 5 .
most of the observing material considered in this paper was described by m15 , it consists of 136 echelle optical spectra obtained between years 2008 and 2013 with resolving power @xmath19 40,000 , at the observatories las campanas with the echelle spectrograph ( dupont telescope ) , cerro tololo with the chiron spectrograph ( 1.5 m telescope ) and la silla with the coralie spectrograph ( euler telescope ) .
in addition to the aforementioned observing material , new spectra are studied in this paper .
the new spectroscopic data were obtained using the echelle spectrograph @xcite attached to the 2.12 m telescope of the observatorio astronmico nacional at san pedro mrtir ( oan spm ) , mexico .
this spectrograph provides spectra spread over 27 orders , covering the spectral range of 3500 - 7105 with a spectral resolution power of @xmath20 = 18000 .
a thar lamp was used for wavelength calibration .
the spectra were reduced by a standard way using the echelle package in iraf , including flat and bias correction , wavelength calibration and order merging .
all spectra discussed in this paper are normalized to the continuum and the rvs are heliocentric ones .
we use the orbital and long - cycle ephemerides for the binary provided by m15 .
the log of observation and characteristics of the new spectra are given in table1 .
.summary of new spectroscopic observations obtained with the echelle spectrograph at the 2.1 m telescope of the san pedro mrtir observatory . at each epoch
one exposure of 1200 s was secured .
the heliocentric julian day ( hjd ) at mid - exposure is given .
the @xmath21 ratio is calculated in the continuum around h@xmath0 . @xmath22 and
@xmath23 refer to the orbital and long - cycle phase , respectively , and are calculated according to the ephemerides given in m15 .
[ cols="<,^,^,^,^,^",options="header " , ] [ 1 ] [ 1 ] [ 1 ] represented in the inset graphs . ,
title="fig:",width=302 ] [ 1 ] represented in the inset graphs .
, title="fig:",width=302 ] [ 1 ]
we acknowledge an anonymous referee for useful comments regarding the first version of this paper . r.e.m .
acknowledges support by vrid - enlace 214.016.001 - 1.0 and the basal centro de astrofsica y tecnologas afines ( cata ) pfb06/2007 .
s. z. acknowledges support from dgapa / papiit project in100614 and conacyt grant car 208512 .
g. d. gratefully acknowledges the financial support of the ministry of education and science of the republic of serbia through the project 176004 , stellar physics .
frank j. , king a. , raine d. j. , 2002 , accretion power in astrophysics , by juhan frank and andrew king and derek raine , pp .
isbn 0521620538 .
cambridge , uk : cambridge university press , february 2002 .
fujiwara h. , makita m. , nagae t. , takuya , 2001 , progress of theoretical physics , vol .
4 , pp . 729 - 749 .
garrido h. e. , mennickent r. e. , djuraevi g. , koaczkowski z. , niemczura e. , mennekens n. , 2013 , mnras , 428 , 1594
|
hd170582 is an interacting binary of the double periodic variable ( dpv ) type , showing ellipsoidal variability with a period of 16.87 days along with a long photometric cycle of 587 days .
it was recently studied by mennickent et al .
( 2015 ) , who found a slightly evolved b - type star surrounded by a luminous accretion disc fed by a roche - lobe overflowing a - type giant . here
we extend their analysis presenting new spectroscopic data and studying the balmer emission lines .
we find orbitally modulated double - peak h@xmath0 and h@xmath1 emissions whose strength also vary in the long - term . in addition , doppler maps of the emission lines reveal sites of enhanced line emission in the 1st and 4th velocity quadrants , the first one consistent with the position of one of the bright zones detected by the light curve analysis .
we find a difference between doppler maps at high and low stage of the long cycle ; evidence that the emission is optically thicker at high state in the stream - disc impact region , possibly reflecting a larger mass transfer rate .
we compare the system parameters with a grid of synthetic binary evolutionary tracks and find the best fitting model .
the system is found to be semi - detached , in a conservative case - b mass transfer stage , with age 7.68 @xmath2 10@xmath3 yr and mass transfer rate 1.6 @xmath2 10@xmath4 @xmath5 . for 5 well - studied dpvs ,
the disc luminosity scales with the primary mass and is much larger than the theoretical accretion luminosity .
[ firstpage ] stars : early - type - stars : evolution - stars : mass - loss - stars : emission - line - stars : variables - others
|
as an interacting galaxy system , stephan s quintet ( hereafter sq ) is not as famous as arp 220 or m82 . taken as a popularity index , the number of references listed by ned ( until dec .
18th , 2005 ) under stephan s quintet is only 104 , far less than that under m82 ( 1355 ) and arp 220 ( 563 ) .
however , its fame as a very special case of multi - galaxy collision has been rising steadily recently , thanks mostly to the ever increasing number of observational windows opened one after another in this space astronomy era .
sq has been observed in almost all wavebands , and it keeps revealing surprises when being looked at by new instruments .
different from binary galaxies and mergers which have been thoroughly discussed elsewhere in this conference , sq belongs to a different class of interacting systems of galaxies : the compact groups ( hickson @xcite ) that are characterized by aggregates of 4 8 galaxies in implied space densities as high as those in cluster cores .
sq is unique among its peers because a seemingly very rare event is occurring to it : a high velocity ( @xmath0 km sec@xmath1 ) intruder is colliding into a debris field in the intragroup medium ( igm ) , the latter being a product of previous interactions among other member galaxies of the group . the complex web of galaxy - galaxy and galaxy - igm interactions have triggered all kinds of interaction related phenomena , including a large scale shock ( @xmath2 kpc ) , an igm starburst , long tidal tails ( @xmath3 kpc ) with tidal dwarf candidates , and a highly obscured type ii agn . in this talk i ll review the rich literature on the multi - frequency observations of sq and the constraints imposed on its interaction history and on its current star formation rate .
i ll also high light its role as a laboratory to study rare phenomena such as large scale shocks and high velocity collision induced starbursts .
inferences from these studies to the understanding of local ulirgs and high z galaxies will be briefly touched .
sq ( fig.1 ) was discovered as an aggregation of nebulae by stephan in late 19th century ( stephan @xcite ) , much earlier than anyone knew galaxies outside the milky way exist .
it raised strong interest in early years of modern astronomy because of its wide range of redshifts ( ambartsumian @xcite ; limber & mathews @xcite ; burbidge & burbidge @xcite ) .
in particular , arp(@xcite,@xcite ) argued that ngc 7320 , with a redshift about 5000 km sec@xmath1 less than the rest of the group ( burbidge & burbidge @xcite ) , is physically associated with other sq galaxies and is a prime case for the existence of non - dopler redshift .
the counter arguments against this claim include ( allen & sullivan @xcite ; moles et al .
@xcite ) : ( 1 ) no evidence for interaction with other sq galaxies was found in the hi or in the h@xmath4 emission associated with ngc 7320 . ( 2 ) results from redshift independent estimators yielded distances consistent with the redshifts . (
3 ) apparently ngc 7320 is a member of a foreground loose galaxy group which includes the large galaxy ngc 7331 at redshift @xmath5 km sec@xmath1 .
i shall assume that ngc 7320 is a foreground galaxy and omit it hereafter .
arp ( @xcite ) suggested that ngc 7320c , a galaxy about 4 to the east of ngc 7319 and linked to the group by optical tidal tails , should be included into sq .
therefore , dropping ngc 7320 and at same time picking up ngc 7320c , sq still consists of five galaxies as implied by the name .
the early deep optical images ( arp & kormendy @xcite ; arp @xcite ; arp & lorre @xcite ) show clearly two concentrating parallel tidal tails , both starting from ngc 7319 and extending to ngc 7320c , with the more diffuse one passing through the low redshift galaxy ngc 7320 .
however , contemplations on the interaction history of sq started only after two discoveries in the radio band , both made with the westerbork synthesis telescope .
the first is a huge ( @xmath2 kpc ) radio continuum ridge found in the igm between ngc 7319 and ngc 7318b ( allen & hartsuiker @xcite ) . among other possible interpretations , allen & hartsuiker suggested
this may be a large scale shock triggered by an on - going collision between ngc 7318b and the rest of the group .
the second ( allen & sullivan @xcite ; shostak et al .
@xcite ) is the revelation that nearly all hi gas associated with sq , amounting to @xmath6 m@xmath7 , is in the igm outside galaxy disks .
it is suggested that this is stripped gas from late type galaxy disks due to an earlier interaction ( a few 10@xmath8 yr ago ) either between ngc 7319 and ngc 7318a ( allen & sullivan @xcite ) , or between ngc 7319 and ngc 7320c ( shostak et al .
@xcite ) .
mole et al .
( @xcite , hereafter m97 ) put up a comprehensive picture for the interaction history of sq based on a two - intruders scenario ( see also sulentic et al .
@xcite , hereafter s01 ) : an old intruder ( ngc 7320c ) stripped most of the gas from group members , and a new intruder ( ngc 7318b ) is currently colliding with this gas and triggering the large scale shock .
the old intruder passed the core of sq twice , one in @xmath9 7 @xmath10 yr ago and pulled out the ` old tail ' , another about 2 @xmath10 yr ago and triggered the ` young tail ' ( the narrower one ) .
recently , xu et al .
( @xcite , hereafter x05 ) proposed a three - intruders scenario based on new observations .
they agreed with m97 on the origins of the old tail and of the shock .
but , different from m97 , they argued that it is unlikely that the young tail is also triggered by ngc 7320c .
this is because the recently measured redshift of ngc 7320c ( 6583 km sec@xmath1 , s01 ) is almost identical to that of ngc 7319 , indicating a slow passage ( s01 ) rather than a fast passage ( @xmath11 km sec@xmath1 , m97 ) . in order for ngc 7320c to move to its current position ,
the ngc 7319/7320c encounter must have occurred @xmath12 yr ago .
this is close to the age of the old tail , but older than that of the young tail .
they suggested that the young tail is triggered by a close encounter between the elliptical galaxy ngc 7318a and ngc 7319 .
the projected distance between ngc 7318a/7319 is only @xmath13 of that between ngc 7320c/7319 .
therefore the time argument is in favor of the new scenario .
also the morphology of the young tail revealed by the new uv observations , relative to ngc 7319 and ngc 7318a , looks very similar to the counter tidal tail found in dynamic simulations of equal mass galaxy - galaxy interactions ( see , e.g. , fig.2 of toomre & toomre @xcite , t=2 10@xmath8 yr ) . according to the k - band luminosities , ngc 7318a and ngc 7319 have nearly identical stellar mass .
moles et al .
( @xcite ; @xcite ) concluded that the disk of ngc 7318b is still intact because the time since it starts to interact with the group is too short ( @xmath14 yr ) for any tidal effect .
however , xu et al . ( @xcite ; @xcite ) speculated that the galaxy might have had a head - on collision with ngc 7318a @xmath15 yr ago , which means ngc 7318a must be @xmath16 kpc further away than ngc 7318b given the velocity difference .
this will help to explain the long outer arms ( looking like tidal tails ) , the peculiar hi gas distribution ( all outside the optical disk ) and the huge uv disk ( @xmath17 kpc ) of the ngc 7318b . also , in this scenario , the comma - like filaments in the region north of ngc7318 ( see , e.g. fig.1c of mendes de olivera et al .
@xcite ) can be interpreted in an analogy to the spoke - like filaments in the cartwheel galaxy , a prototype ring galaxy produced in a head - on galaxy - galaxy collision ( higdon @xcite ) .
the only sign for any interaction between ngc 7317 with other members of sq is a common halo linking it to the binary ngc 7318 , found in the deep x - ray and optical r - band images ( trinchieri et al .
since the discovery of allen & hartsuiker ( @xcite ) , the large scale radio ridge has been confirmed by later vla observations ( van der hulst & rots @xcite ; williams et al .
@xcite , hereafter w02 ; xu et al .
@xcite ) . the high resolution x - ray maps ( pietsch et al .
@xcite ; s01 ; trinchieri et al .
@xcite , @xcite ) and optical narrow - band h@xmath4/[nii ] maps ( vilchez & iglesias - pramo @xcite ; ohyama et al .
@xcite ; xu et al .
@xcite ; plana et al .
@xcite ; s01 ) show a similar structure , in both the hot gas and the ionized gas .
the optical spectroscopy of xu et al .
( @xcite ) at several positions along the ridge show spectra consistent with shock excitation .
this shock is truly unique to sq .
its size is second only to that of the radio relics caused by cluster mergers ( enlin & brggen @xcite ) . in the shock ,
the intruder / igm collision could have deposited as much as 10@xmath18 erg energy , equivalent to the kinematic energy of @xmath19 supernovae . using the formula of dopita & sutherland ( @xcite ) , xu et al .
( @xcite ) derived a total energy flux in the form of @xmath20 .
apparently , most of this flux comes out in the fir band .
based on iso observations , the total dust luminosity of the shock front is @xmath21 .
this is about an order of magnitude higher than the x - ray luminosity ( trinchieri et al .
@xcite ) , indicating that the hot gas is predominantly cooled by collisions with dust grains .
trinchieri et al .
( @xcite ) estimated the shock velocity to be 460 km sec@xmath1 , and the density of post - shock gas @xmath22 @xmath23 .
the corresponding sputtering time scale for @xmath24 grains is 3.7 @xmath25 yr , compared to the cooling time scale of 2.1 @xmath25 yr . hence the grains can indeed survive the shock long enough to be the major coolant . the energy density of relativistic particles plus the magnetic fields , derived from the minimum energy assumption , is @xmath26 erg @xmath23 ( xu et al .
this is significantly lower than the total energy density implied by the total energy flux , so the magnetic field may be dynamically insignificant for the shock .
recently , using spitzer , appleton et al .
( @xcite ) discovered strong mid - ir emission lines of molecular hydrogen ( and little else ! ) at position of one of the radio emission peaks along the shock .
the derived h@xmath27 luminosity is much higher than @xmath28 and only a factor of @xmath29 less than l@xmath30 .
this gives support to the idea ( rieke et al .
@xcite ) that the strong mir h@xmath27 lines often detected in luminous ir galaxies are primarily excited by shock waves .
the star formation activity in sq is apparently very much influenced by interactions .
the most spectacular star formation region in sq is the igm starburst sq - a , which is associated with a bright mir ( 15@xmath31 ) source ( xu et al .
@xcite ) just beyond the northern tip of the shock front .
the two velocity components of the starburst have same redshifts as those of the neutural gas , indicating that the starburst is occurring in the pre - shock gas .
the estimated star formation rate ( sfr ) is 1.45 m@xmath7/yr , and the age is @xmath14 yr . xu et al .
( @xcite ) advocated a scenario in which the starburst is triggered due to squeezing of pre - existing giant - molecular clouds ( gmcs ) by post - shock hi gas , as depicted in the model of jog & solomon ( @xcite ) .
the young tail is also active in star formation .
hunsberger et al .
( @xcite ) found 13 tidal dwarf galaxy candidates along this tail .
a bright star formation region ( sq - b ) is detected in both h@xmath4 ( arp @xcite ) and mir ( xu et al .
several discrete star formation regions near the western tip of the tail are found in both h@xmath4 ( s01 ; mendes de oliveira et al .
@xcite ) and fuv ( x05 ) .
the h@xmath4 images ( arp @xcite ; vilchez & iglesias - pramo @xcite 1998 ; plana et al .
@xcite ; s01 ) reveal numerous huge hii regions along several arms of ngc 7318b .
hunsberger et al .
( @xcite ) , iglesias - pramo & vilchez ( @xcite ) , and mendes de oliveira et al .
( @xcite ) classified them as tidal dwarf galaxy candidates , though s01 argued that the crossing time of ngc 7318b ( @xmath14 yr ) is too short for any tidal effects .
investigations based on optical colors ( schombert et al .
@xcite ; gallagher et al .
@xcite ) and uv colors ( x05 ) indicate several distinct epochs of star formation that appear to trace the history of dynamic interactions in sq . using the @xmath32 luminosity without correction for internal extinction ,
the results of iglesias - pramo & vilchez ( @xcite ) give a rather low net sfr of @xmath33 m@xmath7/yr for the whole group .
the extinction corrected fuv ( galex ) luminosities , with internal extinction constrained by the iso mir flux and the hi surface density , yield a net sfr of 6.7@xmath34 m@xmath7/yr ( x05 ) .
the sfr of the two sbc galaxies , ngc 7318b and 7319 , is consistent with that of normal sbc galaxies such as the milky way , confirming the general conclusion that late type galaxies in compact groups have their sfr indistinguishable from that of their field counterparts ( sulentic & de mello rabaca @xcite ; moles et al .
@xcite ; iglesias - pramo & vilchez @xcite ) .
this is in strong contrast with galaxy pairs where significant enhancement of star formation is very common .
* hi gas : * first noticed by allen & sullivan ( @xcite ) , and confirmed by later observations ( shostak et al . @xcite ; w02 ) , all of the hi in sq is in the igm outside galaxy disks .
w02 found the gas is located in five features : ( 1 ) arc - s ( 6641 km sec@xmath1 , 2.5 10@xmath35 m@xmath7 ) associated with the old optical tail , ( 2 ) arc - n ( 6604 km sec@xmath1 , 4.0 10@xmath35 m@xmath7 ) associated with the young tail , ( 3 ) nw - lv ( 6012 km sec@xmath1 , 2.2 10@xmath35 m@xmath7 ) and ( 4 ) nw - hv ( 6647 km sec@xmath1 , 0.9 10@xmath35 m@xmath7 ) both centered at the position of the igm starburst sq - a , and ( 5 ) sw ( 5699 km sec@xmath1 , 1.5 10@xmath35 m@xmath7 ) in the south of ngc 7318b .
these authors emphasized the uncertain origin of these hi features . on the basis that sw and
nw - lv features are spatially and kinematically separated with each other , w02 argued against the scenario that both features belong to ngc 7318b , each on the opposite side of a rotation disk ( m97 ) .
however , a comparison between the fuv image ( x05 ) and a new , higher resolution and higher sensitivity vla b - array hi map ( yun et al .
@xcite ) shows that sw and nw - lv indeed coincide very well with fuv arms belonging to a very large fuv disk ( @xmath36 kpc ) centered on the nucleus of ngc 7318b , lending support to their associations with ngc 7318b . * molecular gas : * the early millimeter co observations ( yun et al .
@xcite ; verdes - montenegro et al .
@xcite ; leon et al.@xcite ) detected molecular gas only in the disk of ngc 7319 , amounting to 4.8 10@xmath37 m@xmath7 ( smith & struck @xcite ) .
this is normal for its size and luminosity ( lisenfeld et al .
@xcite ) . the molecular gas associated with the igm starburst sq - a
is first detected by gao & xu ( @xcite ) using bima , later confirmed by single dish observations of smith & struck ( @xcite ) and lisenfeld et al .
( @xcite ) .
the co emission is in two separate velocity systems centered at 6000 km sec@xmath1 and 6600 km sec@xmath1 , respectively , in excellent agreement with the redshifts of the two hi features found in the same region ( nw - hv and nw - lv in w02 ) .
it is interesting to note that , according to lisenfeld et al .
( @xcite ) , there is more molecular gas ( 3.1 10@xmath35 m@xmath7 ) in sq - a than in hi gas ( 2.5 10@xmath35 m@xmath7 ) .
braine et al .
( @xcite ) detected co in sq - b , a ` tidal dwarf ' ( tdf ) candidate bright in mir ( xu et al .
@xcite ) , h@xmath4 ( arp @xcite ) and uv ( x05 ) . the follow - up observations of lisenfeld et al .
( @xcite ; @xcite ) found 7 10@xmath38 m@xmath7 molecular gas .
the ratio m@xmath39/m@xmath40 is consistent with the average of tdfs . from the close correspondence between the co , hi , h@xmath4 and mir emission , lisenfeld et al .
( @xcite ) argued that the hi feature arc - n ( w02 ) is linked to the young optical tail , therefore is very likely to be originated from ngc 7319 . a survey for co emission associated with other tdf candidates ( mendes de oliveira et al .
@xcite ) yielded no detection .
co emission associated with ngc 7318b nucleus and with several other regions in the disk was reported in gao & xu ( @xcite ) , smith & struck ( @xcite ) , petitpas & taylor ( @xcite ) .
metallicity has been determined for sq - a and sq - b , in both cases it is slightly higher than solar ( xu et al .
@xcite ; lisenfeld et al .
@xcite ) , suggesting that the igm gas is originated in the inner part of a galaxy disk ( or disks ) .
* hot gas ( x - ray ) : * sq has been observed by almost every x - ray satellite , including einstein ( bahcall et al .
@xcite ) , rosat ( sulentic et al . @xcite ; s01 ; pietsch et al .
@xcite ) , asca ( awaki et al .
@xcite ) , chandra ( trinchieri et al . @xcite ) and xmm ( trinchieri et al .
trinchieri et al .
( @xcite ) divide the x - ray emission in sq into 4 major features : ( 1 ) the shock , ( 2 ) ngc 7319 ( sy2 ) , ( 3 ) halo , ( 4 ) tail .
tail is associated with hot diffuse igm that can not be ascribed to the shock .
it has a size of 130 150 kpc and is in a region coinciding with the eastern end of the old optical tail , indicating a link to the early tidal interactions .
halo includes the diffuse emission in a region surrounding the shock , excluding the shock itself and emissions possibly related to individual galaxies .
these two diffuse features take @xmath41 of the total soft x - ray ( 0.5 2.0 kev ) luminosity of sq .
* dust : * sq was detected , but barely resolved , by iras ( allam et al .
@xcite ; yun et al .
@xcite ) . the higher resolution and more sensitive iso observations ( xu et al .
@xcite , @xcite ; s01 ) found most of the mir ( 11.3 and 15@xmath31 ) and fir ( 60 and 100@xmath31 ) emission in the disks of ngc 7319 and of the foreground galaxy ngc 7320 .
two igm starbursts sq - a and sq - b , both including several previously detected hii regions , stand out in the 15@xmath31 image due to their strong mir emission ( xu et al .
there is evidence for dust emission in the shock front , and the dust cooling is likely to be the dominant cooling mechanism for the shock ( xu et al .
@xcite ; trinchieri et al .
the isophot maps at 60@xmath31 and 100@xmath42 indicate that there might be diffuse dust emission in the igm , but the signal is marginal .
this has been followed up by spitzer observations at 70@xmath31 and 160@xmath31 ( xu et al.@xcite ) . with much improved sensitivity and angular resolution ,
the spitzer 160@xmath31 indeed detected significant diffuse dust emission outside individual galaxies , amounting as much as more than 50% of the total l@xmath43 of sq .
the origin of this emission is still unclear .
there are two competing possibilities : ( 1 ) emission of collisionally heated grains in the hot igm ; and ( 2 ) emission of radiative heated grains in the cold neutral igm .
the better correspondence between the morphology of the 160@xmath31 map and that of the x - ray map , compared to that between 160@xmath31 and hi , favors the possibility ( 1 ) .
the agn was first recognized by van der hulst and rots ( @xcite ) in the vla observations of sq at 20 cm continuum .
they detected a bright compact source with a jet - like extension , and suggested this might be related to those found in seyfert galaxies .
huchra et al .
( @xcite ) obtained the optical spectrum which shows clearly sy2-type emission - line features .
based on the stellar velocity dispersion , woo & urray ( @xcite ) estimated the black hole mass : m@xmath44 m@xmath7 .
aoki et al .
( @xcite ) made vla a - array radio continuum observations , complimented with an archival hst optical image .
they found a chain of 3 radio sources , interpreted as the nucleus and its two jets on opposite sides .
optical features are found in the hst image closely related to the radio jets , interpreted as gas compressed and excited by bow shocks driven into the ambient medium by the jets .
this is different from the so - called extended emission line regions ( eelr ) which are supposedly excited by the agn radiation ( aoki et al .
@xcite ) . even higher resolution ( 0.16 ) radio continuum observations at 1658 mhz , using merlin , were reported by xanthorouplos et al .
( @xcite ) .
they compared the data with an hst / acs u - band ( f330w ) image , and found extended uv emission around the nucleus and the northern jet .
they argued that this indicates star formation triggered by the jet / ism interaction . by assuming the diffuse radio emission outside the compact sources is due to the star formation
, they estimated that the sfr in the circum - nuclear region is 8.4 m@xmath7 .
however , all of the optical spectra of emission line regions in the circum - nuclear region show seyfert or liner line ratios and none is hii region - like ( aoki et al .
@xcite ) , indicating that most of the radio and uv radiation is related to the agn and/or the shocks . comparing the uv and mir observations , x05 found the agn and the surrounding region is highly obscured , with an fuv extinction of @xmath45 mag , consistent with the sy2 classification .
the fascination of sq is due to its rich history full of interactions , which we are just starting to unveil .
although the multi - wavelength observations in the literature have provided clues on how the different events may be related to each other , a clear - cut picture is yet to emerge .
many interpretations are based on dynamic simulations for binary galaxies , which may not be appropriate for multi - galaxy collisions such as sq .
dedicated simulations for sq are certainly desired .
such simulations will provide insights on how compact groups such as sq survive merging , at the same time turn late - type galaxies into early types without significantly increasing the sfr .
investigations on the difference between multi - galaxy collisions and two - galaxy collisions will shed light on how galaxy interactions may have contributed to the evolution of galaxies and galactic structures .
|
stephan s quintet ( sq ) , discovered more than 100 years ago , is the most famous and well studied compact galaxy group .
it has been observed in almost all wavebands , with the most advanced instruments including spitzer , galex , hst , chandra , vla , and various large mm / submm telescopes / arrays such as the iram 30 m and bima .
the rich multi - band data reveal one of the most fascinating pictures in the universe , depicting a very complex web of interactions between member galaxies and various constituents of the intragroup medium ( igm ) , which in turn trigger some spectacular activities such as a 40 kpc large scale shock and a strong igm starburst . in this talk
i will give a review on these observations .
|
it is frequently stated that we understand exoplanets only as well as we understand the host star .
such a statement is particularly true for low - mass dwarf stars , whose atmospheres often diverge from blackbody models . there has been a concerted effort in recent years to obtain observational constraints on the stellar models for low - mass stars @xcite , especially for those monitored by the _
mission @xcite .
a further challenge includes the confusion that can be caused by the stellar rotation period of low - mass stars since that can often coincide with the range of orbital periods of planets that may exist in the habitable zone ( hz ) of those stars @xcite .
even so , there have been several successful detections of terrestrial planets in or near the hz of low - mass stars , such as kepler-186 f @xcite , k2 - 3 d @xcite , and the recent discovery of proxima centauri b @xcite .
the low - mass m - dwarf star , wolf-1061 ( also designated as gj 628 ) , is one of our closest neighbors , located approximately 4.3 pc away @xcite .
the star was recently discovered to host three planets that lie within the super - earth mass regime , one of which may be located within the hz of the system @xcite .
the orbits of the planets were significantly updated by @xcite using additional data from the high accuracy radial velocity planet searcher ( harps ) spectrograph .
the orbital solutions largely agree with respect to the inner two planets , but the @xcite solution finds an orbital period of the outer planet that is a factor of @xmath33 larger than that found by @xcite . both of the solutions do agree that planet c is near or in the hz of the host star , the location of which is highly dependent on the star parameters of luminosity and effective temperature . both solutions also predict reasonably high transit probabilities and depths such that follow - up photometry during calculated transit windows are encouraged .
this paper presents a characterization of the wolf 1061 host star and the associated planets with new data and numerical simulations . in section [ fund ] , the fundamental stellar parameters of the host star are updated through interferometry data that provide measurements of the stellar radius , effective temperature , and luminosity .
section [ phot ] presents seven years of precise , automated photometry that reveals the correct stellar rotation period , supports the existence of the three purported planets , and finds no evidence for planetary transits .
the revised stellar parameters are utilized in section [ hab ] with a calculation of the system hz and a dynamical simulation that shows variation of eccentricities for the planetary orbits .
we provide a concluding discussion of the results in section [ conclusions ] .
wolf 1061 has been an object of interest for quite some time , primarily because of its high proper motion of @xmath4 mas / yr @xcite and its membership in the solar neighborhood . despite its relatively close position to our star
, the stellar properties of wolf 1061 have remained uncertain largely due to the fact that it is a very dim m3v late - type dwarf .
prior to our investigation , the values of wolf 1061 s temperature , luminosity , flux , and radius have varied in the literature .
previous estimates for the red dwarf s temperature have been reported as being as low as 2877 k @xcite to as high as 3400 k @xcite with a variety of values reported between these two extremes @xcite .
less extreme are the variations in radius , from 0.30 @xmath5 @xcite to @xmath6 @xmath5 @xcite , and luminosity , which was previously reported to be 0.007870 @xmath7 @xcite .
precise determination of these essential characteristics have taken on new importance in light of the early 2016 discovery that wolf 1061 is host to three exoplanets @xcite .
wolf 1061 was observed on june 30 and august 34 , 2016 , using the georgia state university center for high angular resolution astronomy ( chara ) interferometric array @xcite .
observations were conducted in @xmath8-band with the chara classic beam combiner @xcite in single - baseline mode . to remove the influence of atmospheric and instrumental systematics , our interferometric observations consist of bracketed sequences of object and calibrator stars .
calibrators were chosen using the aspro tool to be near - point - like sources of similar brightness as wolf 1061 , located at small angular distances from it , and observed directly before and after the target : hd 143459 , hd 146254 , hd 149013 , and hd 153229 .
this procedure follows our requirements that we use at least two calibrators , two baselines , and data obtained during at least two nights ( e.g. , * ? ? ?
* ; * ? ? ?
* and references therein ) .
.,width=309 ] the uniform disk and limb - darkened angular diameters ( @xmath9 and @xmath10 , respectively ) are calculated by fitting the calibrated visibility measurements ( figure [ fig : diameters ] ) to the respective functions for each relation .
these functions may be described as @xmath11-order bessel functions of the angular diameter of the star , the projected distance between the two telescopes , and the wavelength of observation @xcite .
we use the linear limb - darkening coefficient @xmath12 from the phoenix models in @xcite for stellar @xmath13 = 3000 k and @xmath14 = 4.5 to convert from @xmath15 to @xmath10 .
the uncertainties in the adopted limb darkening coefficient amount to 0.2% when modifying the adopted gravity by 0.5 dex or the adopted @xmath13 by 200k , well within the errors of our diameter estimate .
our interferometric measurements produce the following values for wolf 1061 : @xmath16 milliarcseconds ( mas ) and @xmath17 mas .
combined with the trigonometric parallax measurement of @xmath18 mas from @xcite , we obtain a stellar radius for wolf 1061 of @xmath19 @xmath1 , which is practically identical to the one estimated in @xcite of @xmath20 @xmath1 .
to calculate wolf 1061 s effective temperature and luminosity , we perform a spectral energy distribution ( sed ) fit based on spectrophotometry data obtained as part of the survey described in @xcite ; see in particular their section 3 .
these spectrophotometry data have no color terms and only require a zero point offset .
we use literature photometry from @xcite to scale the spectrophotometry data and obtain the bolometric flux by simply integrating over wavelength .
interstellar reddening is set to zero in the fit due to the close proximity of wolf 1061 . in the calculation of the bolometric flux , we use the modified filter profiles for the literature photometry from @xcite and use the 2% error correction described by @xcite to obtain realistic error estimates in @xmath21 .
we calculate the following for wolf 1061 : @xmath22 erg @xmath23 s@xmath24 , @xmath25 @xmath26 , and @xmath27 k. these values are consistent at @xmath28 with the ones in table 5 of @xcite that use interferometric data for calibration of their semi - empirical methods .
our stellar parameters for wolf 1061 , including the rotation period described in section [ rot ] , are summarized in table [ stellar ] .
lc @xmath29 & 10.07 + @xmath30 & 1.57 + distance ( pc ) & @xmath31 + @xmath32 ( erg @xmath23 s@xmath24 ) & @xmath33 + @xmath13 ( k ) & @xmath34 + @xmath35 ( @xmath5 ) & @xmath0 + @xmath36 ( @xmath7 ) & @xmath37 + @xmath38 ( days ) & @xmath2 +
: seven years of photometric observations of wolf 1061 , comprising 756 nightly measurements , acquired with the t11 0.8 m apt at fairborn observatory .
slow rotational modulation of dark spots on the star s surface , as well as year - to - year evolution of the spot distribution , account for the brightness variability .
the dotted line marks the mean brightness .
@xmath39 : the 756 observations are normalized so that all seasonal means are equal to the first , marked by the dotted line .
@xmath40 : frequency spectrum of the complete normalized data set revealing low - amplitude variability at 121 or 91 days.,width=309 ] we have observed wolf 1061 during its past seven observing seasons with the tennessee state university ( tsu ) t11 0.80 m automatic photoelectric telescope ( apt ) at fairborn observatory in arizona . between 2010 april and 2016 june , the apt acquired 756 brightness measurements of wolf 1061 on 464 different nights . like other tsu apts ,
t11 is equipped with a two - channel precision photometer designed and built by louis boyd at fairborn .
the photometer uses a dichroic filter and two emi 9124qb bi - alkali photomultiplier tubes to separate and simultaneously measure the strmgren @xmath41 and @xmath42 photometric pass bands .
wolf 1061 , designated here as the program star ( p , @xmath43 , @xmath44 , m3.5 v ) , was observed differentially with respect to three constant comparision stars hd 150177 ( c1 , @xmath45 , @xmath46 , f3 v ) , hd 147753 ( c2 , @xmath47 , @xmath48 , f2 v ) , and hd 148968 ( c3 , @xmath49 , @xmath50 , a0 v ) .
all differential magnitudes were corrected for extinction and transformed to the strmgren photometric system .
we computed final differential magnitudes of wolf 1061 against the mean brightness of all three comparison stars as @xmath51 , where the subscript @xmath52 indicates that we combined the strmgren @xmath41 and @xmath42 observations into a single @xmath53 passband .
the precision of a single observation from the t11 apt is typically @xmath54 mag , determined by intercomparison of the comparison stars .
further details of our automatic telescopes , precision photometers , and observing and data reduction procedures can be found in @xcite and @xcite and references therein .
note that the t11 apt is essentially identical to the t8 apt described in @xcite .
the final @xmath51 differential magnitudes are plotted in the top panel of figure [ photfig1 ] and show wolf 1061 to be varying over a range of @xmath55 mag with a timescale of @xmath56 days . to remove the effects of the small year - to - year variations in mean brightness
, we normalized the data shown in the second panel of figure [ photfig1 ] by adjusting the seven seasons so that each has the same mean as the first .
a frequency spectrum of the normalized data is shown in the bottom panel .
weak periodicity is found around 91 and 121 days , both with amplitudes of a few millimags .
the low amplitudes are due not only to the intrinsically low amplitude of wolf 1061 s photometric variability but also to year - to - year changes in the amplitude , shape , mean magnitude and phase of minimum of the lightcurve .
the 91 and 121 day periods are yearly aliases of each other caused by the large seasonal gaps in our lightcurve .
we take the variability in wolf 1061 to arise from the rotational modulation of a slowly evolving spot distribution on the photosphere of the star .
ccccccc 2010 & 92 & 5529755383 & 0.0061 & ( 103.9 ) & ( 0.015 ) & @xmath57 + 2011 & 117 & 5560155740 & 0.0070 & @xmath58 & @xmath59 & @xmath60 + 2012 & 136 & 5597656099 & 0.0068 & @xmath61 & @xmath62 & @xmath63 + 2013 & 115 & 5632556470 & 0.0055 & @xmath64 & @xmath65 & @xmath66 + 2014 & 111 & 5669856840 & 0.0050 & ( 108.3 ) & ( 0.011 ) & @xmath66 + 2015 & 93 & 5706257195 & 0.0042 & ( 58.7 ) & ( 0.008 ) & @xmath67 + 2016 & 92 & 5742857554 & 0.0045 & ( 199.2 ) & ( 0.015 ) & @xmath68 + : jd plot of the 2011 observing season of wolf 1061 .
@xmath39 : frequency spectrum of the 2011 data giving a photometric period of @xmath58 days .
@xmath40 : the 2011 data phased to the best period of 92.9 days and showing a peak - to - peak amplitude of 0.019 mag.,width=309 ] except for the 2012 observing season , giving a period of @xmath61 days and a peak - to - peak amplitude of 0.017 mag.,width=309 ] except for the 2013 observing season , giving a best period of @xmath64 days and a peak - to - peak amplitude of 0.015 mag.,width=309 ] to determine the correct rotation period , we attempted a periodogram analysis of the seven individual observing seasons .
only the 2011 , 2012 , and 2013 observing seasons cover the lightcurve sufficiently well to give reliable results .
frequency spectra for these three observing seasons are shown in figures [ photfig2 ] , [ photfig3 ] , & [ photfig4 ] .
the complete results of our seasonal photometric analysis of wolf 1061 are given in table [ phottab ] .
rotation periods and amplitudes that are poorly constrained are given in parentheses .
the weighted mean of the 2011 , 2012 , and 2013 photometric periods is @xmath2 days .
we identify this period as the true stellar rotation period and the 121-day period from figure [ photfig1 ] as its yearly alias .
this is consistent with the 93 day rotation period determined for wolf 1061 by @xcite from analyses of the ca ii h&k emission lines . using the kinematic work of @xcite for nearby m dwarfs , the rotation period suggests an age for wolf 1061 of @xmath69 gyrs .
however , as seen above in figure [ photfig1 ] , the rotational modulation of the lightcurve is not strictly sinusoidal over the seven year span of our observations . therefore
, stellar activity on the surface of wolf 1061 might still be responsible for radial velocity variability , as has been demonstrated in other moderately active stars ( see , e.g. , * ? ? ? * ; * ? ? ?
* ; * ? ? ?
: residuals from the individual sine fits to the seven observing seasons of wolf 1061 summarized in table [ phottab ] , plotted against julian date .
@xmath39 : residuals from the top panel phased with the 4.8870-day orbital period of planet b and time of conjunction derived from the radial velocities . a least - squares sine fit on the radial velocity period gives a semi - amplitude of just @xmath70 mag , establishing to high precision the lack of stellar activity on the radial velocity period and thus confirming the presence of stellar reflex motion caused by an orbiting planet .
@xmath40 : close - up of the observations near the time of planetary conjunction at phase 0.0 .
the solid line shows a model transit computed from the parameters of planet b. the vertical lines mark the uncertainty in the predicted transit window .
our current photometric observations provide no evidence for transits.,width=309 ] but for wolf 1061 c and d.,width=309 ] we remove most of the photometric variability in wolf 1061 by taking the residuals from the yearly sine fits specified in table [ phottab ] .
these are plotted in the top panel of figure [ photfig5 ] .
the standard deviation of the residuals from the mean ( marked by the dotted line ) is 0.0032 mag , roughly half the variability of the original lightcurve in the top panel of figure [ photfig1 ] .
periodogram analysis of the full set of residuals finds no significant periodicity .
the middle panel of figure [ photfig5 ] , shows the seasonal residuals phased with the 4.8870-day orbital period of wolf 1061 b and the epoch of mid transit from table 4 of @xcite .
a least - squares sinusoidal fit to the phased data gives a formal semi - amplitude of just @xmath71 mag , which limits any periodic brightness variability on the orbital period to a very small fraction of one milli - magnitude ( mmag ) .
this rules out the possibility that the 4.8870-day radial velocity variations are due to stellar activity .
instead , the lack of photometric variability confirms the radial velocity variations at 4.8870 day result from true planetary reflex motion .
the photometric observations within @xmath72 orbital phase of mid - transit are plotted with an expanded scale in the lower panel of figure [ photfig5 ] .
the solid curve shows the predicted transit phase ( 0.0 ) , depth ( @xmath73 mag ) , and duration ( @xmath74 days ) of a central transit , computed from the stellar and planetary radii and the orbital elements of wolf-1061 b. the vertical dotted lines give the @xmath75 uncertainty in the timing of the transit window , based on the uncertainties in the stellar radius provided in section [ sec : stellar_radius ] and the improved orbital elements from @xcite .
we find no evidence in our data for transits of planet b. results of similar analyzes for planets c and d are shown in figure [ photfig6 ] .
the low photometric amplitude of @xmath76 mag in the top panel confirms wolf-1061 c as a planet since it shows that the radial velocity variations are not due to stellar activity . in the second panel
, we find no evidence of photometric transits of planet c. the low amplitude of @xmath76 mag in the third panel confirms wolf-1061 d as a planet .
the bottom panel shows we have insufficient data to rule out transits of planet d.
using the stellar parameters described above , we calculate the boundaries of the hz and the long - term stability of the planets with respect to the hz . the concept of the hz as a target selection tool has been developed based on earth climate models for several decades @xcite .
specifically , the hz defines the locations around a host star where the climate of an earth - analog will remain cool enough to avoid a runaway greenhouse effect and warm enough to prevent a runaway snowball effect .
these calculations account for water absorption in the planetary atmosphere and stellar properties such as luminosity and effective temperature .
to calculate the extent of the hz , we use the methodology of @xcite .
the `` conservative '' and `` optimistic '' hz boundaries are calculated based on assumption regarding the timespan over which the atmospheric evolutionary history of venus and mars allowed liquid water to remain on the surface .
the catalog of confirmed planets and planetary candidates detected by the _
kepler _ mission @xcite describes the conservative and optimistic hz boundaries in more detail . .
the physical scale depicted is 1.0 au on a side .
the conservative hz is shown as light - gray and the optimistic extension to the hz is shown as dark - gray.,width=321 ] using the updated stellar parameters from this work ( see section [ sec : effective_temperature ] ) , we estimate the inner and outer boundaries of the conservative hz to be 0.11 au and 0.21 au respectively . allowing for the optimistic conditions for suface liquid water , the inner and outer hz boundaries are 0.09 au and 0.23 au respectively . shown in figure [ hzfig ]
is a top - down view of the planetary orbits in the wolf 1061 system , using the orbital solution of @xcite .
the conservative hz is shown as light - gray and the optimistic extension to the hz is shown as dark - gray .
the scale of the figure is 1.0 au on a side .
the orbital eccentricities of the planets are 0.13 , 0.13 , and 0.57 for the b , c , and d planets respectively .
the outer planet passes briefly through the hz during its periastron passage , spending 6% of the orbital period within the hz .
planet c spends 61% of its orbit duration within the hz , but that time remains constrained to the optimistic hz . in that respect ,
planet c is quite similar to the case of kepler-69 c , which was proposed to be a strong super - venus candidate by @xcite .
indeed , both of the inner two planets , terrestrial in nature according to the results of both @xcite and @xcite , lie within the venus zone of the host star @xcite and are thus possible runaway greenhouse candidates .
another factor that plays a role in the habitability of the system is the orbital dynamics between the planets as a function of time . to investigate orbital stability and dynamics
, we utilized the mercury integrator package @xcite , with the hybrid symplectic / bulirsch - stoer integrator and a jacobi coordinate system .
the initial conditions were set using the orbital solution of @xcite and the integration executed for a simulated duration of @xmath77 years .
the time resolution was set to 0.1 days in order to adequately meet the recommended minimum timestep criteria of @xmath78 of the shortest system orbital period @xcite .
the orbital architecture of the system was output in 100 year intervals .
for the coplanar scenario where the system is viewed approximately edge - on ( inclination of @xmath79 and the true planetary masses are equivalent to the minimum masses ) , the system was found to be stable over the full @xmath77 year simulation duration . although stable , the compact nature of the system combined with the relatively large orbital eccentricities results in an active dynamical evolution of key keplerian orbital parameters .
in particular , evidence of the angular momentum exchange between the planets can be observed in the oscillations of the eccentricity and argument of periastron .
these evolutions of eccentricity and argument of periastron are shown for a simulation duration of @xmath80 years in figures [ ecc ] and [ omega ] respectively .
+ + + + the bottom panel of figure [ ecc ] shows that the amplitude of the eccentricity variations for the outer planet is largely insensitive to the presence of the inner planets , and remains close to the initial value of @xmath81 .
the interaction between the inner two planets is more pronounced , with mean eccentricities substantially below the initial values of @xmath82 and @xmath83 .
similarly , the precession of the periastron arguments evolves on a rapid timescale for the inner two planets compared with the outer planet ( see figure [ omega ] ) .
the eccentricity of the inner planet reduces to the circular case at regular intervals , whereas the eccentricity of planet c drops as low as @xmath30.03 .
the eccentricity of a planet within the hz does not necessarily exclude the presence of liquid water on the surface , as the required conditions also depend upon such factors as atmospheric composition , scale height , and response to variations in incident flux @xcite .
however , it is worth noting that a zero eccentricity for planet c results in the orbit being entirely interior to the optimistic hz .
it is thus possible that planet c is more amenable to habitable conditions when near to peak eccentricity , since the planet moves slowly through the apastron passage in the hz .
the assessment of host star properties is a critical component of exoplanetary studies , at least for the realm of indirect detections through which exoplanet discoveries thus far have predominantly occurred .
this situation will remain true for the coming years during which the transit method will primarily be used from space missions such as the transiting exoplanet survey satellite ( tess ) , the characterising exoplanet satellite ( cheops ) , and the planetary transits and oscillations of stars ( plato ) mission .
of particular interest are the radius and effective temperature of the stars since the radius impacts the interpretation of observed transit events and the combination of radius and temperature is used to calculate the extent of the hz . here
, we have presented the results from direct measurements of stellar properties for one of the closest known exoplanet host stars ; wolf 1061 .
our direct measurement of the stellar radius from interferometric observations gives @xmath0 @xmath1 , which is remarkably close to the value previous calculated by @xcite , which can be considered a significant triumph for the empirical calibrations used in that work .
our sed fit resulted in determining a luminosity of @xmath84 @xmath26 for wolf 1061 and , after combination with the measured angular diameter , an effective temperature of @xmath27 k. we further provide seven years of wolf 1061 photometry based on obsservations acquired with tsu s t11 apt .
these data were sufficient to investigate periodic signals that are a measurement of the stellar rotation period .
our analysis was able to disentangle the various aliases and isolate a rotation period of @xmath2 days .
our photometric precision and observing cadence are able to rule out transits of the two inner planets in the system but the possibility of a transiting outer planet remains open .
finally , our measured stellar parameters were used to derive the hz boundaries of the system and investigate the location and dynamics of the planetary orbits with respect to the hz .
we find that , although the eccentric solution for planet c allows it to enter the optimistic hz , the two inner planets are consistent with possible super - venus planets @xcite .
long - term stability analysis shows that the system is stable in the current configuration , and that the eccentricity of the two inner planets frequently reduce to zero , at which times the orbit of planet c is entirely interior to the optimistic hz .
we thus conclude that the system is unlikely to host planets with surface liquid water .
g.w.h . acknowledges support from nasa , nsf , tennessee state university , and the state of tennessee through its centers of excellence program .
t.s.b . acknowledges support from nasa grant 14-k2go2_2 - 0075 and louisiana state university .
this work is based upon observations obtained with the georgia state university center for high angular resolution astronomy array at mount wilson observatory .
the chara array is supported by the national science foundation under grant no .
institutional support has been provided from the gsu college of arts and sciences and the gsu office of the vice president for research and economic development .
this research has made use of the jean - marie mariotti center ` aspro ` service , available at www.jmmc.fr/aspro .
this research has also made use of the following archives : the exoplanet orbit database and the exoplanet data explorer at exoplanets.org , the habitable zone gallery at hzgallery.org , and the nasa exoplanet archive , which is operated by the california institute of technology , under contract with the national aeronautics and space administration under the exoplanet exploration program .
the results reported herein benefited from collaborations and/or information exchange within nasa s nexus for exoplanet system science ( nexss ) research coordination network sponsored by nasa s science mission directorate .
anglada - escud , g. , amado , p.j . , barnes , j. , et al .
2016 , nature , 536 , 437 astudillo - defru , n. , delfosse , x. , bonfiles , x. , et al .
2016a , a&a , submitted astudillo - defru , n. , bonfils , x. , delfosse , x. , et al .
2016b , a&a , submitted avenhaus , h. , schmid , h.m . , meyer , m.r .
2012 , a&a , 548 , a105 bailey , j.i . ,
white , r.j . ,
blake , c.h .
2012 , apj , 749 , 16 beichman , c.a .
, neugebauer , g. , habing , h.j . , clegg , p.e . ,
chester , t.j .
1988 , infrared astronomical satellite ( iras ) catalogs and atlases .
volume 1 : explanatory supplement , nasa ref .
publ . , 1190 , 1 bessel , m.s . 1990 ,
a&a , 83 , 357 bohlin , r.c . ,
gordon , k.d . , tremblay , p .- e .
2014 , pasp , 126 , 711 boisse , i. , bonfils , x. , santos , n.c . 2012 , a&a , 545 , a109 boyajian , t.s . , von braun , k. , van belle , g. , et al .
2012 , apj , 757 , 112 boyajian , t. , von braun , k. , feiden , g.a . , et al .
2015 , mnras , 447 , 846 cantrell , j.r . ,
henry , t.j . ,
white , r.j .
2013 , aj , 146 , 99 chambers , j.e .
1999 , mnras , 304 , 793 claret , a. , bloemen , s. 2011 , a&a , 529 , a75 corben , p.m. , carter , b.s . ,
banfield , r.m . , harvey , g.m . 1972 ,
mnssa , 31 , 7 cousins , a.w.j . 1980a , saaoc , 1 , 166 cousins , a.w.j .
1980b , saaoc , 1 , 234 crossfield , i.j.m . ,
petigura , e. , schlieder , j.e . , et al .
2015 , apj , 804 , 10 cutri , r.m . ,
skrutskie , m.f . ,
van dyke , s. , et al .
2003 , the 2mass all sky catalog of point sources , ycat , 2246 , 0 davison , c.l . , white , r.j . , henry , t.j . , et al .
2015 , aj , 149 , 106 duncan , m.j . ,
levison , h.f .
, lee , m.h .
1998 , aj , 116 , 2067 eaton , j.a . ,
henry , g.w .
, fekel , f. c. 2003 , the future of small telescopes in the new millennium , vol .
ii , the telescopes we use , ed . t.d . oswalt ( dordrecht : kluwer ) , 189 gaidos , e. , mann , a.w . ,
kraus , a.l . ,
ireland , m. 2016 , mnras , 457 , 2877 gautier , t.n . ,
iii , rieke , g.h . , stansberry , j. et al .
2007 , apj , 667 , 527 hanbury brown , r. , davis , j. , lake , r.j.w . ,
thompson , r.j .
1974 , mnras , 167 , 475 henden , a.a . ,
levine , s.e .
, terrell , d. , smith , t.c . , welch , d. 2012 , javso , 40 , 430 henry , g.w . 1999 ,
pasp , 111 , 845 huber , d. , silva aguirre , v. , matthews , j.m .
, et al . 2014 , apjs , 211 , 2 jenkins , j.s . ,
ramsey , l.w . , jones , h.r.a .
, et al . 2009 , apj , 704 , 975 johnson , h.l .
1965 , apj , 141 , 170 johnson , h.l . ,
harris , d.l .
1954 , apj , 120 , 196 kilkenny , d. , koen , c. , van wyk , f. , marang , f. , cooper , d. 2007 , mnras , 380 , 1261 laing , j.d .
1989 , saaoc , 13 , 29 lger , a. , defrre , d. , malbet , f. , labadie , l. , absol , o. 2015 , apj , 808 , 194 lindgren , s. , heiter , u. , seifahrt , a. 2016 , a&a , 586 , a100 kane , s.r . , gelino , d.m .
2012 , asbio , 12 , 940 kane , s.r . ,
barclay , t. , gelino , d.m .
2013 , apj , 770 , l20 kane , s.r . ,
kopparapu , r.k . , domagal - goldman , s.d .
2014 , apj , 794 , l5 kane , s.r . , hill , m.l .
, kasting , j.f . , et al .
2016 , apj , 830 , 1 kasting , j.f . ,
whitmore , d.p . , reynolds , r.t .
1993 , icarus , 101 , 108 kasting , j.f .
, kopparapu , r. , ramirez , r.m . , harmen , c.e .
2014 , pnas , 111 , 12641 koen , c. , kilkenny , d. , van wyk , f. , cooper , d. , marang , f. 2002 , mnras , 334 , 20 koen , c. , kilkenny , d. , van wyk , f. , marang , f. 2010 , mnras , 403 , 1949 kopparapu , r.k . , ramirez , r. , kasting , j.f . , et al .
2013 , apj , 765 , 131 kopparapu , r.k . ,
ramirez , r. , schottelkotte , j. , et al .
2014 , apj , 787 , l29 mann , a.w . , von braun , k. 2015 , pasp , 127 , 102 mann , a.w . ,
feiden , g.a . , gaidos , e. , boyajian , t. , von braun , k. 2015 , apj , 804 , 64 mermilliod , j ,- c .
1986 , catalogue of eggen s ubv data mould , j.r . , hyland , a.r .
1976 , apj , 208 , 399 muirhead , p.s . ,
hamren , k. , schlawin , e. , et al .
2012 , apj , 750 , l37 muirhead , p.s . ,
becker , j. , feiden , g.a .
, et al . 2014 , apjs , 213 , 5 neves , v. , bonfils , x. , santos , n.c . , et al .
2014 , a&a , 568 , a121 newton , e.r . , irwin , j. , charbonneau , d. , berta - thompson , z.k . , dittmann , j.a .
2016a , apj , 821 , l19 newton , e.r . ,
irwin , j. , charbonneau , d. , et al .
2016b , apj , 821 , 93 niconov , v.b . ,
nekrasova , s.v . ,
polosuina , n.s . ,
rachkouvsky , n.d . ,
chuvajev , w.k .
1957 , izkry , 17 , 42 nehag , a. , heiter , u. , gustafsson , b. , et al .
2012 , a&a , 542 , a33 paulson , d.b . ,
saar , s.h . ,
cochran , w.d . ,
henry , g.w .
2004 , aj , 127 , 1644 queloz , d. , henry , g.w .
, sivan , j.p . , et al .
2001 , a&a , 379 , 279 quintana , e.v . ,
barclay , t. , raymond , s.n .
2014 , science , 344 , 277 reid , n. 1982 , mnras , 201 , 51 rojas - ayala , b. , covey , k.r . ,
muirhead , p.s . ,
lloyd , j.p .
2012 , apj , 748 , 93 sturmann , j. , ten brummelaar , t.a . ,
ridgway , s.t .
, et al . 2003 , spie , 4838 , 1208 ten brummelaar , t.a . , mcalister , h.a . ,
ridgway , s.t .
, et al . 2005 , apj , 628 , 453 turnbull , m.c .
2015 , exocat-1 : the nearby stellar systems catalog for exoplanet imaging missions ( arxiv:1510.01731 ) van leeuwen , f. 2007 , a&a , 474 , 653 vanderburg , a. , plavchan , p. , johnson , j.a . ,
2016 , mnras , 459 , 3565 veeder , g.j . 1974 , aj , 79 , 1056 von braun , k. , boyajian , t.s . , van belle , g.t . , et al . 2014 , mnras , 438 , 2413 weis , e.w . 1984 ,
apjs , 55 , 289 weis , e.w .
1986 , aj , 91 , 626 weis , e.w .
1987 , aj , 93 , 451 weis , e.w .
1996 , aj , 112 , 2300 williams , d.m . , pollard , d. 2002 , ijasb , 1 , 61 wright , d.j . , wittenmyer , r.a . ,
tinney , c.g . ,
bentley , j.s .
, zhao , j. 2016 , apj , 817 , l20
|
a critical component of exoplanetary studies is an exhaustive characterization of the host star , from which the planetary properties are frequently derived . of particular value
are the radius , temperature , and luminosity , which are key stellar parameters for studies of transit and habitability science . here
we present the results of new observations of wolf 1061 , known to host three super - earths .
our observations from the center for high angular resolution astronomy ( chara ) interferometric array provide a direct stellar radius measurement of @xmath0 @xmath1 , from which we calculate the effective temperature and luminosity using spectral energy distribution models .
we obtained seven years of precise , automated photometry that reveals the correct stellar rotation period of @xmath2 days , finds no evidence of photometric transits , and confirms the radial velocity signals are not due to stellar activity .
finally , our stellar properties are used to calculate the extent of the habitable zone for the wolf 1061 system , for which the optimistic boundaries are 0.090.23 au .
our simulations of the planetary orbital dynamics shows that the eccentricity of the habitable zone planet oscillates to values as high as @xmath30.15 as it exchanges angular momentum with the other planets in the system .
|
as a basis for the discussion which follows , i adopt the following version of the current `` experimental situation '' with regard to reported pentaquark resonances : * a significant number of medium - energy experiments report positive evidence for the @xmath0 .
its existence is thus plausible
. if these experiments are correct , the @xmath0 mass and width are @xmath3 mev and @xmath4 mev .
see ref .
@xcite for a critical assessment of both positive and negative experimental searches , as well as references to the experimental literature . * the na49 exotic @xmath5 , @xmath6 ( @xmath2 ) signal with @xmath7 mev and narrow width
@xcite has not been confirmed @xcite , and is hard to reconcile with high statistics @xmath6 production experiments @xcite .
though a @xmath2 partner is unavoidable if the @xmath0 lies in a @xmath1 , in view of the experimental situation , no assumption will be made about its mass . * the h1 report of an anticharmed baryon at @xmath8 mev @xcite is also unconfirmed @xcite , so no assumptions will be made about the masses of any possible heavy pentaquark states . a low - lying , relatively narrow ( @xmath9 ) , @xmath1 , @xmath10 , @xmath11 exotic baryon is natural in the chiral soliton model ( csm ) @xcite . in the quark model
, the ground state of the exotic system would have negative parity ( @xmath12 ) if all five particles were in a symmetric relative s - wave configuration .
whether such a configuration represents the true model ground state , however , is a dynamical question .
exciting one of the four light ( @xmath13 ) quarks into a p - orbit , reducing the @xmath14 spatial symmetry from @xmath15_l$ ] to @xmath16_l$ ] , allows spin - flavor - color configurations with significantly increased hyperfine attraction which were previously pauli - forbidden .
if the increased attraction is greater than the p - wave excitation cost , the ground state will have @xmath17 .
a dynamical calculation is needed to determine which of the two effects wins out .
two proposals of possible @xmath17 quark model configurations for the @xmath0 have received considerable attention , and are also useful for framing the discussions below .
the jaffe - wilczek ( jw ) ansatz for the @xmath0 @xcite involves two @xmath18 , @xmath19 diquark pairs , coupled to color @xmath20 , with only confinement forces between the diquarks and @xmath21 .
bose statistics for the diquarks imply a relative p - wave between them , hence @xmath22 .
the karliner - lipkin ( kl ) scenario for the @xmath0 @xcite involves one jw diquark and one @xmath23 triquark . in the triquark , the spin and color of a jw diquark
have been flipped and anti - aligned with those of the @xmath21 , producing triquark quantum numbers @xmath24 .
the ansatz is motivated by the color - magnetic - exchange model , in which the hyperfine attraction is stronger for the kl than for the jw correlation
. a relative diquark - triquark p - wave will yield @xmath22 . in both scenarios ,
the relative intercluster p - wave is assumed to make cross - cluster interaction and antisymmetrization ( asn ) effects negligible .
the `` gb model '' is a model with spin - dependent interactions generated by effective goldstone boson exchange .
the spin - dependent part of the hamiltonian is @xmath25 with @xmath26 the pauli spin matrices , @xmath27 the constituent quark mass , @xmath28 the @xmath29 gell - mann matrices , and @xmath30 as given in ref .
the jw diquark is by far the most attractive @xmath31 correlation for @xmath32 . if the sum in eq .
( [ hgb ] ) runs only over quark labels @xcite , the model also matches the jw assumption for the @xmath33 interactions .
assuming an orbital excitation cost similar to that in the ordinary baryon sector of the model , the gain in hyperfine energy in the @xmath16_l$ ] configuration leads to a @xmath17 ground state , one which should be dominated by the jw correlation .
the model thus provides a dynamical setting for studying ( i ) corrections to the strict jw scenario ( e.g. , cross - cluster asn / interaction effects , configuration mixing ) and ( ii ) the pattern of excitations above the jw - correlation - dominated ground state .
the `` cm model '' is a model with spin - dependent interactions generated by effective color magnetic exchange .
the spin - dependent part of the hamiltonian is @xmath34 with @xmath26 and @xmath27 as above , @xmath35 ( @xmath36 ) for quarks ( antiquarks ) the @xmath37 gell - mann matrices and @xmath38 a smeared delta function .
the jw diquark is again the most attractive @xmath31 correlation , but , as noted above , the more complicated kl configuration lies lower than the jw configuration .
the cm model thus provides a dynamical setting for studying ( i ) corrections to the kl scenario and ( ii ) the expected excitation pattern above the cm ground state .
an important correction results from the fact that the same @xmath33 interactions responsible for lowering the kl triquark hyperfine energy also mix the jw and kl corelations , producing a configuration significantly lower than either the jw or kl correlations separately .
this configuration turns out to be a roughly equal admixture of the two correlations @xcite .
space constraints prevent discussion of all points covered in the conference talk .
( see the conference website and refs .
@xcite for more details ; ref .
@xcite , in particular , gives an explicit definition of the dimensionless form used in quoting results below . ) _
( a ) the @xmath0 mass and pentaquark widths in the quark model : _ models such as the gb and cm models , with parameters fixed from the ordinary baryon sector , can not predict exotic pentaquark masses because of the absence of terms representing the response of the non - perturbative vacuum to the hadronic constituents . this response will differ depending on the number of constituents .
this effect will be absorbed by the additive constant used to set the overall scale in each sector ; the constant should thus be different for ordinary and pentaquark baryons , just as it is for mesons and baryons . in the bag model ,
where some modelling of vacuum response _ does _ occur , the difference in 1-body energies between a single @xmath39 bag and two isolated @xmath40 bags is an extremely sensitive function of the @xmath41 @xcite , suggesting it is impossible to anticipate in advance the size , or even sign , of this effect .
thus , the fact that quark model calculations for @xmath42 _ using parameter values fixed from the ordinary baryon spectrum _
yield masses @xmath43 mev or so too high does not rule out the validity of a pentaquark picture for such states . other observables ( widths , splittings , relative decay couplings )
must be used to test the models .
if the @xmath0 is well described as a pentaquark , the initial and final states in @xmath44 have the same number of constituents .
the decay thus proceeds by rearrangement , or `` fall - apart '' , and should have a width @xmath45 , where @xmath46 is the width for a @xmath47 potential scattering resonance with mass @xmath42 in an attractive potential of roughly hadronic size , and @xmath48 is a spectroscopic , or squared - overlap , factor @xcite .
a single - scale s - wave attractive potential too weak to bind does not produce resonant behavior @xcite ; hence a @xmath49 assignment for the @xmath0 is impossible . for a p - wave ( d - wave ) decay , @xmath50 ( respectively , 20 ) mev .
the spin - flavor - color ( jfc ) part of the overlap suppression is 1/24 for the jw correlation and @xmath51 for the optimized jw / kl combination in the cm model @xcite . with additional suppression from the spatial overlap , @xmath0
widths @xmath52 a few to several mev are expected for @xmath17 ground states involving such correlations . as for the csm
, some fine tuning is needed to reduce this to @xmath53 mev or less .
( a very narrow @xmath0 is natural for a @xmath54 assignment , where a d - wave @xmath55 decay is required .
however , at least in the gb and cm models , such a ground state assignment is impossible @xcite . )
the jfc `` overlap '' suppression effect is rather general in the quark model approach . in the soft-@xmath56/pcac approximation ,
the @xmath57 coupling is @xmath58 .
three of the light quarks in the @xmath0 must project onto the @xmath59 and the remaining constituent @xmath60 and @xmath21 be annihilated by the pseudoscalar density with @xmath61 quantum numbers .
the jfc factors are the same as would occur in a @xmath0 to @xmath55 overlap calculation using a constituent @xmath62 picture for the @xmath61 .
a model calculation @xcite confirms the reduction of @xmath63 to @xmath64 mev for the jw correlation , before inclusion of spatial overlap effects , and can be interpreted as a model - dependent version of the soft-@xmath56/pcac estimate . as stressed in ref .
@xcite , for @xmath17 pentaquarks with p - wave fall - apart decays , decay coupling ratios should be given by the corresponding overlap ratios squared .
_ ( b ) corrections to the jw and kl scenarios in the gb and cm models : _ for the gb model , @xmath65 in the strict jw limit is @xmath66 .
the lowest - lying @xmath17 state has jw quantum numbers ( which , up to a @xmath67 spin - orbit partner , are also the csm ground state quantum numbers ) .
cross - cluster interaction / asn and configuration mixing lower the actual ground state expectation to @xmath68 @xcite .
the jw approximation is reasonable , but not exact .
the ground state expectation _ is _ , however , the same in the @xmath0 and heavy @xmath69 ( @xmath70 ) sectors , as in the jw scenario . with the jw estimate for the 1-body contribution to @xmath71 @xcite
, the jw prediction of strong interaction stable @xmath72 is thus preserved . for the cm model , in the @xmath0 sector
, the optimized combination of jw and kl correlations provides a good approximation to the lowest @xmath17 eigenvalue @xcite , which again occurs for the channel with csm quantum numbers . in the heavy sector , the @xmath73 interactions are strongly suppressed by the @xmath74 factor in @xmath75 .
this ( i ) decreases the kl triquark hyperfine attraction , making the kl diquark - triquark configuration higher in energy than the jw correlation for both @xmath76 , and ( ii ) strongly suppresses kl - jw mixing .
the heavy pentaquark @xmath17 ground state should thus be dominated by the jw correlation .
the jw expectation is @xmath77 , c.f .
the actual ground state expectation @xmath78 @xcite .
the jw approximation is again reasonable , but not exact .
the kl scenario for the heavy quark analogues @xcite assumes the same triquark - diquark clustering as for the @xmath0 persists in the heavy system .
this assumption is _ not _ compatible with the cm model effective interactions , which motivated the original triquark - diquark proposal .
the kl heavy sector ansatz thus leads to an overestimate of the @xmath72 masses in the cm model .
modified estimates put the @xmath72 close to the @xmath79 , @xmath80 strong decay thresholds @xcite . _
( c ) splittings and relative decay couplings in the light and heavy @xmath17 @xmath69 sectors : _ though one can not predict absolute pentaquark masses in the quark model approach , splittings are independent of any additive constant and hence _ are _ predictable .
predictions for a given @xmath69 sector ( with @xmath81 fixed ) are particularly immune to uncertainties in treating 1-body energies , and are on the same footing as analogous predictions for the ordinary baryon sector .
the relative ordering of states is fixed by the jfc structure of the effective operators , with ( in a given sector ) an overall scale , dependent on the size of the pentaquark states .
_ ratios _ of overlap factors relevant to the decays to @xmath82 , with @xmath83 the relevant ground state baryon and @xmath84 the relevant vector or pseudoscalar meson , are also well - determined @xcite . for the gb model ,
the first excitation above the @xmath17 , @xmath69 ground state , for all of @xmath81 , is a degenerate pair with @xmath85 , where @xmath86 is the total intrinsic spin .
the excitation energy is @xmath87 mev .
no other states are expected within @xmath88 of the ground state . for the @xmath89 state
, the combination of phase space and squared - overlaps enhances the width to @xmath55 by a factor of at most @xmath90 c.f .
the @xmath0 , making this prediction potentially problematic . a similar , potentially problematic , state with @xmath91 , mass @xmath92 mev above the @xmath0 , and relatively narrow ( @xmath93 mev ) @xmath55 width , is also predicted in the csm @xcite . for the cm model , in the @xmath94 sector ,
a single low - lying @xmath95 excitation is predicted @xmath96 mev above the @xmath0 .
the next excitation lies @xmath97 mev higher .
phase space and the overlap factor for @xmath89 state s decay to @xmath55 are again such that a relatively narrow width is expected .
even more problematic are the predictions in the heavy sector , where _ five _ low - lying excitations are expected , with @xmath98 , @xmath89 , @xmath99 , @xmath100 , @xmath89 , the first @xmath101 mev above the ground state , and the last @xmath102 mev above the ground state @xcite .
such a dense , low - lying excitation spectrum seems unlikely , but would be a striking prediction of the cm model , if confirmed .
none of the overlap ratios are such as to allow any of these low - lying states to be experimentally undetectable in 2-body decay modes , provided at least one of either them , or the ground state , is experimentally accessible @xcite . for the gb model , the jw dominance of the ground state and decrease in jw diquark hyperfine energy under @xmath103 imply a reduction in @xmath2 hyperfine energy c.f .
the @xmath0 . an estimate of the splitting , using only the @xmath15_{fj}$ ] component in the @xmath2 wavefunction , yields @xmath104 mev @xcite
allowing other wavefunction components ( known to be small for the @xmath0 , but expected to be more important for the @xmath2 ) will lower this value . in the kl scenario ,
the @xmath0 and @xmath2 have almost identical @xmath105 values , leading to @xmath106 mev @xcite .
the lowest hyperfine eigenvalues in the @xmath0 and @xmath2 channels turn out to share this near equality , even though the ground state is not well approximated by the kl correlation .
the cm model thus produces a significantly smaller @xmath107 splitting than does the gb model .
16 hicks k 2004 hep - ph/0408001 alt c _ et al . _
( the na49 collaboration ) 2004 _ phys . rev . lett . _ * 92 * : 042003 knopfle k
_ et al . _ ( the hera - b collaboration ) 2004 hep - ex/0403020 fischer h and wenig s 2004 hep - ex/0401014 aktas a _ et al . _
( the h1 collaboration ) 2004 hep - ex/0403017 vazquez m
_ et al . _
( for the h1 , zeus , hermes and hera - b collaborations ) 2004 hep - ex/0411045 diakonov d , petrov v , and polyakov m 1997 _ z. phys .
a _ * 359 * 305 ellis j , karliner m and praszalowicz m 2004 _ jhep _ * 0405 * 002 jaffe r and wilczek f 2003 _ phys .
lett . _ * 91 * 232003 karliner m and lipkin h 2003 _ phys . lett .
b _ * 575 * 249 f. stancu 1998 _ phys . rev
. d _ * 58 * 111501 jennings b and maltman k 2004 _ phys .
d _ * 69 * 094020 maltman k 2004 hep - ph/0408144 maltman k 2004 _ phys .
b _ * 604 * 175 maltman k 1992 _ phys . lett .
b _ * 291 * 371 hosaka a , oka m and shinozaki t 2004 hep - ph/0409102 close f and zhao q 2004 _ phys
. lett . _ * b590 * 176 karliner m and lipkin h 2004 hep - ph/0307343 stancu f 2004 _ phys .
b _ * 595 * 269 [ erratum : _ phys . lett .
b _ * b598 * 295 ]
|
we discuss certain general features of the pentaquark picture for the @xmath0 , its @xmath1 partner , @xmath2 , and possible heavy quark analogues .
models employing spin - dependent interactions based on either effective goldstone boson exchange or effective color magnetic exchange are also used to shed light on possible corrections to the jaffe - wilczek and karliner - lipkin scenarios .
some model - dependent features of the pentaquark picture ( splitting patterns and relative decay couplings ) are also discussed in the context of these models .
|
two - level system ( tls ) has been an important paradigmatic model in many branches of contemporary physics ranging from radiation - matter interactions to collision physics@xcite .
it lies at the heart of modern applications such as quantum control and quantum information processing as well as constitutes the foundation for fundamental studies of quantum mechanics .
coherent manipulation of two - level systems can be achieved and optimized by introducing external fields , e.g. , the most commonly used types of periodic@xcite and pulse - shaped@xcite driving fields . in the past few decades , the two - level quantum system ( or a qubit ) involving single and/or combined modulations has proven a fertile ground for a variety of intriguing phenomena , including self - induced transparency@xcite , complete population inversion ( cpi)@xcite , dynamical stabilization@xcite , and coherent destruction of tunneling ( cdt)@xcite , to name only a few
. in certain situations , analytically exact solutions of driven two - level problems exist , offering many advantages in developing analytical approaches to the design of qubit control operations .
unfortunately , it is usually extremely difficult to acquire exactly soluble two - level evolutions so far .
started with the famous examples like the landau - zener model@xcite , rabi problem@xcite , rose - zener model@xcite , the problems of finding analytic solutions of driven tlss have been underway for a long time@xcite-@xcite .
however , most of the known exact solutions of tlss are expressed in terms of complicated special functions , such as the gauss hypergeometric function@xcite , weber function@xcite , associated legendre function@xcite , jacobi elliptic function@xcite , and heun function@xcite . up to now , only a very few examples of analytically solvable two - level evolutions described only by several of the simplest solutions@xcite , which renders the control strategies more transparent , have been reported . in recent years ,
parity - time ( @xmath0)-symmetric two - level system have attracted much attention from both theoretical and experimental studies@xcite . a surprising
finding of such @xmath0-symmetric non - hermitian hamiltonian systems is phase transition , where the spectrum changes from all real to complex when the gain - loss parameter exceeds a certain threshold .
it has been widely demonstrated that external driving fields can be effectively used for implementing coherent controls of @xmath0symmetry , and extension of driven two - level systems to non - hermitian case is currently of broad interest@xcite-@xcite .
for the proper control of the @xmath0-symmetric two - level system , different analytic methods are required .
for instance , approximate solutions to the @xmath0-symmetric rabi model have recently been proposed based on perturbation theory@xcite . given the vast literatures about analytical results for hermitian two - level systems
, it would be possible to find a few simple analytical exact solutions to the driven @xmath0-symmetric two - level problems , which will create the possibility of convenient controls of the non - hermitian systems . in this paper
, we have constructed a set of simple exact analytical solutions for the @xmath0-symmetric two - level systems under synchronous combined modulations , which helps illuminate some generalizations of the appealing concepts which originate in their hermitian counterparts . in the case of monochromatically synchronous combined modulations , as it is shown , for the nonzero constant modulation ( a periodic oscillating driving with a nonzero static component )
, the driven @xmath0-symmetric two - level system behaves like the non - driven one with a sudden @xmath0 symmetry breaking , whereas for the zero constant modulation ( a purely oscillating driving ) , the driven @xmath0-symmetric two - level system features all - real spectra and a type of `` generalized rabi oscillations '' . besides , a non - hermitian analog of coherent destruction of tunneling ( cdt ) , in which the tunneling can be brought to a standstill , has been observed . for a square sech - shaped synchronous modulation ,
the exact analytical solutions of simple forms enables us to precisely produce complete population inversion ( cpi ) between two states in @xmath0-symmetric systems . the pulse area theorem for such cpi
has also been derived analytically and confirmed numerically .
the central purpose of this article is to develop analytical approaches to the design of coherent control of @xmath0-symmetric two - level systems by constructing exact solutions of simple forms through synchronous combined modulations .
we consider a non - hermitian version of driven two - level system , in which the equation of motion for the probability amplitudes @xmath1 and @xmath2 for the two states @xmath3 and @xmath4 is given as @xmath5 with the following hamiltonian@xcite @xmath6 here @xmath7 and @xmath8 are the usual pauli matrices , the real rabi frequency @xmath9 describes the coupling between the two levels , and @xmath10 represents a pair of real gain - loss coefficients . the system ( [ eq1 ] ) can be realized in many physical contexts , such as coupled waveguides with a complex refractive index@xcite , dissipative two - state system of cold atom@xcite , and laser - driven atomic or molecular transitions with decay@xcite . in this work ,
we focus on how to design simple synchronous combined modulations to give analytic demonstrations of some intriguing and unexpected physics of non - hermitian two - level systems . here
synchronous modulations means that the driving terms @xmath10 and @xmath11 have the same time dependence , in other words , the modulation functions @xmath10 and @xmath9 satisfy the relation @xmath12 with @xmath13 being a proportional constant .
as we can easily demonstrate , a @xmath0-symmetric two - state problem is obtained if @xmath10 and @xmath9 ( e.g. , the cosine and square sech forms ) is temporally symmetric with respect to a certain point of time .
to obtain the exact solutions of model ( [ eq1 ] ) , we eliminate @xmath14 and arrive at a second - order linear differential equation for @xmath15 from eq .
( [ eq1 ] ) , @xmath16c_{1}(t)=0.\ ] ] in the case of synchronous modulations , @xmath12 , by introducing a new time scale @xmath17 , we can express the second - order differential equation of @xmath15 in a simple form , @xmath18 the general solution of eq .
( [ nsde ] ) , mathematically well known , falls into three cases as follows : _ case 1 _ : when @xmath19 , letting @xmath20 , the solution of eq .
( [ nsde ] ) is given by @xmath21 with @xmath22 and @xmath23 being the undetermined constants determined by the initial conditions . according to eq.(1 ) , @xmath14 is easily obtained as @xmath24 _ case 2 _ : when @xmath25 , eq .
( [ nsde ] ) gives the analytical solution of eq .
( [ eq1 ] ) as @xmath26 _ case 3 _ : when @xmath27 , letting @xmath28 , and combining eq .
( [ nsde ] ) with eq .
( [ eq1 ] ) , the exact solutions of @xmath29 and @xmath30 are obtained as @xmath31 any of the synchronous modulation functions @xmath9 and @xmath10 will produce an analytical solution to the schrdinger equation ( [ eq1 ] ) , which thus enables us to generate an infinite variety of analytically solvable @xmath0-symmetric two - state problems along with their explicit solutions ( [ s11])-([s33 ] ) . here
we will display some interesting physics by taking several specific cases as examples .
we present firstly the case of usual periodic modulation , @xmath32 where the new time scale is given by @xmath33 .
if the particle is initially populated at level @xmath3 , we have the initial condition @xmath34 . applying them to eqs .
( [ s11])-([s33 ] ) yields the undetermined constants @xmath35 and the exact population dynamics as follows \(i ) when @xmath19 , @xmath36 \(ii ) when @xmath25 , @xmath37 and ( iii ) when @xmath27 , @xmath38 from eqs .
( [ c1r1i1])-([c1r3i11 ] ) , we immediately have the following observations : ( 1 ) for zero constant modulation ( @xmath39 ) , the solution should be periodic irrespective of the value of proportional constant @xmath13 . ( 2 ) for the nonzero constant modulation ( @xmath40 ) , the solution does not have to be bounded .
indeed , as long as the proportional constant @xmath13 is smaller than the critical value , @xmath25 , the system will exhibit oscillations with bound , while as @xmath13 increases above the critical value , it will exhibit exponential growth . at the critical value , @xmath25
, the time - averaged population shows a parabolic growth .
a naive intuition might suggest that apart from a distinct time scale , periodic synchronous driving will bring no new physics to the non - hermitian system .
indeed , for the nonzero constant modulation ( @xmath40 ) , the driven @xmath0-symmetric two - level system acts like the undriven one , with a phase transition from full to broken @xmath0 symmetry .
however , for the zero constant modulation ( @xmath39 ) , stabilization of the non - hermitian system can be achieved by the periodic synchronous driving in the full range of system parameters , which is contrary to our naive intuition .
this is confirmed by numerical simulations in fig .
[ fig1 ] . in fig .
[ fig1 ] , we show the time evolutions of the populations for both @xmath40 and @xmath39 cases with the initial state @xmath34 . in our simulation , the numerical solutions of the equation ( [ eq1 ] ) are obtained through the use of the runge - kutta method .
as expected , for @xmath40 , like the non - driving situation , the system experiences a transition from a state with quasi - periodic oscillation ( @xmath0-symmetric state , @xmath19 , see fig . [ fig1](a ) ) to a state with unbounded growth ( broken @xmath0-symmetric state , @xmath27 , see fig .
[ fig1](c ) ) , whereas for @xmath39 , stable and coherent but non - norm - preserving population oscillations , referred to as generalized rabi oscillation in ref .
@xcite , are observed in fig .
[ fig1](d)-(f ) for all values of the proportional constant @xmath13 .
we have compared the analytical ( circles ) with the numerical ( red solid lines ) results in fig .
it is clearly seen that they agree well . for the system ( [ eq1 ] )
subject to a periodic modulation ( [ pm1 ] ) with @xmath41 ( left column ) and @xmath39 ( right column ) . from top to bottom : ( a ) and
( d ) @xmath42 ; ( b ) and ( e ) @xmath25 ; ( c ) and ( f ) @xmath43 . shown here are the numerical results of population @xmath44 in level @xmath3 ( blue dashed - dotted line ) , @xmath45 in level @xmath4 ( black dashed line ) , and the total population @xmath46 ( red solid line ) .
the open circles denote the exact analytical results of the total population .
the other parameters are @xmath47.,width=302 ] as is well known , floquet theory offers a general framework for treating periodically driven systems . the quasienergies and quasienergy states ( the so - called floquet states ) , as predicted by floquet theory , are two basic concepts for analyzing the periodically driven system .
for the periodic synchronous driving , we can write the floquet solution to the time - dependent schrdinger equation ( [ eq1 ] ) as @xmath48 , where the floquet state @xmath49 is periodic with the modulation period @xmath50 , and @xmath51 is the corresponding quasienergy .
in fact , when one of @xmath52 and @xmath53 equals zero , eqs .
( [ s11])-([s33 ] ) give the two linearly independent floquet states ( eigenstates ) and the corresponding quasienergies as \(i ) when @xmath19 , @xmath54 \\
\mp \exp[\mp i(\theta+\nu_{1}\cos\theta\sin\omega t/\omega)]\\ \end{array } \right),\\\varepsilon_{1,2}&=&\pm\nu_{0}\cos\theta , \label{e1}\end{aligned}\ ] ] and ( ii ) when @xmath27 , @xmath55 \\
i\exp[\pm ( -\phi+\nu_{1}\sinh\phi\sin\omega t/\omega)]\\ \end{array } \right),\\\varepsilon_{1,2}&=&\pm i\nu_{0}\sinh\phi .
\label{e2}\end{aligned}\ ] ] according to the above analysis , for the nonzero constant modulation ( @xmath40 ) , there exists a transition between unbroken @xmath0 symmetry ( the spectrum is all real , @xmath56 if @xmath19 ) and broken @xmath0 symmetry ( both eigenvalues are imaginary , @xmath57 if @xmath27 ) , however , for the zero constant modulation ( @xmath39 ) , the two eigenvalues ( quasienergies ) are degenerate and always equal to zero .
now we take a closer look at what happens at the transition point , @xmath25 . at this point , for @xmath40 , the two floquet modes coalesce to @xmath58 and the system has only one linearly independent eigenstate , leading to an exceptional point .
nevertheless , for @xmath39 , at the point @xmath25 , the two linearly independent floquet modes are given by @xmath59 and @xmath60 respectively , with corresponding degenerate eigenvalues equal to zero .
it should be emphasised that for @xmath39 , the general solution ( [ s11])-([s33 ] ) itself is also a kind of floquet state with a zero quasienergy , due to the fact that any linear combination of two degenerate floquet eigenstates ( the coefficients of linear combination @xmath61 are adjusted by the initial condition ) is also a floquet eigenstate of the hamiltonian operator corresponding to the same eigenvalue ( quasienergy ) . for the system ( [ eq1 ] ) subject to a periodic modulation ( [ pm1 ] ) with ( a)-(b ) @xmath41 ( upper row ) and ( c)-(d ) @xmath39 ( lower row ) .
the other parameters are @xmath47.,width=302 ] in fig .
[ fig2 ] , we exhibit the numerically obtained quasienergies @xmath51 of the system ( [ eq1 ] ) for both @xmath40 and @xmath39 cases . the quasienergies are numerically computed through direct diagonalization of the time evolution operator over one period of the driving , @xmath62 $ ] , where @xmath63 is the time - ordering operator . as we can see in figs .
[ fig2 ] ( a ) and ( b ) , for @xmath40 , the numerical eigenvalue ( quasienergy ) spectrum indeed changes from being real to complex , indicating a spontaneous @xmath0-symmetry - breaking transition .
however , for @xmath39 , the resulting periodically driven two - level system , although being non - hermitian , processes a pair of degenerate all - real quasienergies ( @xmath64 ) for the whole range of tuning parameters , as illustrated in figs .
[ fig2 ] ( c ) and ( d ) .
these numerical results are completely consistent with our theoretical analysis
. coherent destruction of tunneling ( cdt ) is one of the notable effects of the periodically driven system , a quantum interference effect discovered originally in the periodically driven hermitian two - level system , upon the occurrence of which the tunneling dynamics can be brought to a complete standstill .
the cdt is connected to the degeneracy of the quasienergies .
in fact , cdt requires in addition that the degenerate floquet states do not show appreciable amplitude oscillations within one oscillation cycle , which can be realized in the high frequency limit ( @xmath65 ) .
interestingly , in the non - hermitian case , cdt still exists if the quasienergies of the underlying time - periodic system are degenerate . )
subject to a periodic modulation ( [ pm1 ] ) with a zero static component @xmath39 , starting from the initial state @xmath34 .
( a ) time evolution of the probability @xmath44 , with @xmath66 and different values of the driving frequency ; ( b ) localization , defined as the minimum value of @xmath67 , as a function of @xmath68 , with the same parameters as in ( a ) ; ( c ) time evolution of the probability @xmath44 , with @xmath69 and different values of the proportional constant ; ( d ) localization as a function of @xmath13 , with the same parameters as in ( c).,width=302 ] next , we study further the dynamics of the system ( [ eq1 ] ) for the case of @xmath39 , where the two quasienergies are degenerate and equal to zero .
the results of the numerical simulation of the probability distribution @xmath44 are presented in fig .
[ fig3](a ) for different values of driving frequencies . for the low - frequency driving , the occupation of level @xmath3 oscillates and at certain times drops to zero , indicating no suppression of tunneling .
as the driving frequency @xmath68 is increased , the degree of localization is enhanced and the amplitude of oscillations of @xmath70 is smaller .
if @xmath68 is increased further , @xmath70 remains near unity and the tunneling dynamics is frozen , which is a direct signature of cdt . to quantify the stable but non - unitary dynamics more precisely ,
we have used localization , which is defined as the minimum value of @xmath67 , to measure the suppression of tunneling .
when cdt occurs , the localization is close to 1 ; when there is no suppression , localization is zero . in fig .
[ fig3](b ) we plot the minimum value of @xmath67 , or the localization , as a function of the driving frequency . for small values of @xmath68 ,
the minimum value of @xmath67 equals zero , and thus no cdt occurs . as @xmath68 is increased , however , the localization is close to one , and the oscillations in @xmath70 are quenched , producing cdt .
we have also examined the effects of the proportional constant @xmath13 on the localization as shown in figs .
[ fig3](c ) and ( d ) , where with the increase of @xmath13 the amplitude of oscillations of @xmath70 is larger and the localization is decreased accordingly .
since the time scale @xmath71 of the driven system is inversely proportional to the driving frequency , it follows from eqs .
( [ f1])-([e2 ] ) that the amplitude oscillations of the degenerate floquet doublet would be vanishingly small in the high frequency limit . so far , we have demonstrated the existence of cdt in the non - hermitian system , for which the basic two conditions , degeneracy of quasienergies and absence of appreciable amplitude oscillations of the two degenerate floquet states , must be simultaneously satisfied .
this is in full analogy to the hermitian case .
we now consider the synchronous modulation of a square sech - shaped ( a pulse - shaped ) form .
let @xmath72 such that the new time variable becomes @xmath73 when @xmath74 ( hence @xmath75 ) , the system ( [ eq1 ] ) reduces to the famous rose - zener model without detuning@xcite .
we first briefly overview the main results of the rose - zener model on resonance .
therein the transition probability is given as @xmath76 here @xmath77 is the area of pulse .
obviously , if the pulse area satisfies @xmath78 , this transition probability vanishes and the system completely returns its original state , while if @xmath79 , this pulse produces complete population inversion ( cpi ) between two states . note that the definition of pulse area used here differs by a factor 2 from the common usage in the literatures@xcite .
surprisingly , we find that there still exist cpi in the driven non - hermitian ( @xmath0-symmetric ) two - level system . hence ,
below we shall discuss non - hermitian generalization of cpi and formulate the pulse area theorem for such cpi based on the exact solutions ( [ s11])-([s33 ] ) .
assuming the initial condition @xmath80 i.e. , the system is in the level @xmath4 at @xmath81 , we look at the long - time properties of the solution and the final level population at @xmath82 .
three distinct cases for this problem are listed as follows .
_ the case @xmath19 .
_ inserting the initial condition ( [ is1 ] ) into eqs .
( [ s11])-([s12 ] ) and ( [ nt2 ] ) gives @xmath83 the final level populations at @xmath82 , according to the exact solutions ( [ s11])-([s12 ] ) , are thus given by @xmath84 from eqs .
( [ c2s1ni2 ] ) and ( [ c2s1ni22 ] ) , we readily observe the following circumstances : \(1 ) if @xmath85 then the population restores to its initial state at @xmath82 ; \(2 ) if @xmath86 then the population is completely inverted to the level @xmath3 at @xmath82 . it is clearly seen that when @xmath74 ( hence @xmath87=1 ) , eqs .
( [ apt1 ] ) and ( [ apt11 ] ) recover the characteristic of pulse area for the on - resonant rose - zener problem . _
the case @xmath25 .
_ at this point , starting with the initial state ( [ is1 ] ) , we have from eqs .
( [ s2])-([s22 ] ) and ( [ nt2 ] ) , @xmath88 it is obvious that if @xmath89 we get the complete population inversion ( cpi ) .
_ the case @xmath27 .
_ in that case , from eqs .
( [ s3])-([s33 ] ) and ( [ nt2 ] ) , the final populations in levels @xmath3 and @xmath4 corresponding to the initial condition ( [ is1 ] ) can be easily found to be @xmath90 if @xmath91 one finds that @xmath92 , such that a complete population inversion ( cpi ) occurs .
thus , three distinct modified pulse area conditions for cpi in the non - hermitian two - level system ( [ eq1 ] ) have been identified , which are dependent on the choice of the proportional constant @xmath13 and given by eqs .
( [ apt11 ] ) , ( [ apt2 ] ) and ( [ apt3 ] ) respectively . allowed for cpi in the system ( [ eq1 ] ) subjected to a synchronous modulation of the @xmath93 form given by eq .
( [ m2 ] ) .
the allowed parameter values of @xmath94 are obtained from eqs .
( [ apt11])(@xmath95 ) , ( [ apt2 ] ) and ( [ apt3 ] ) .
( b ) time evolution of the population distributions for the initial state @xmath96 .
the system parameters are set as @xmath97 and @xmath98.,width=302 ] the above - mentioned cpi conditions in the parametric plane of @xmath94 are visualized in fig .
[ fig4](a ) , which are directly obtained from eqs .
( [ apt11])(the @xmath95 case ) , ( [ apt2 ] ) and ( [ apt3 ] ) . as an example , in fig . [ fig4](b ) , we select @xmath97 and the pulse area @xmath98 given by eq .
( [ apt3 ] ) to show the numerical results for time evolutions of the population distributions with the initial state ( [ is1 ] ) .
clearly , a complete population inversion ( cpi ) has been fully confirmed by our numerical simulations .
in summary , we have presented a class of exact analytical solutions for a @xmath0-symmetric two - level system under synchronous combined modulations .
such exact solutions are expressible in terms of simple elementary functions and are applicable in the coherent control of quantum states . with these analytical solutions , we have demonstrated that for a cosine synchronous driving with a nonzero static component , the system behaves likes the undriven one with a phase transition from full to broken @xmath0 symmetry , while for a purely cosine synchronous driving , the system , although being non - hermitian , feature all - real quasienergy spectra for the whole range of tuning parameters .
surprisingly , we have found that coherent destruction of tunneling ( cdt ) and complete population inversion ( cpi ) still exist in the non - hermitian situation .
furthermore , we have given analytically the modified pulse area conditions for such non - hermitian analog of cpi , whose mathematical formulations are found to be dependent on the choice of the proportional constant between the two combined synchronous modulations .
these simple exact analytical solutions are numerically confirmed and may find applications in any and all physical contexts where the @xmath0-symmetric two - state problems arise .
the work was supported by the nsf of china under grants 11465009 , 11165009 , the doctoral scientific research foundation of jinggangshan university ( jzb15002 ) , and the program for new century excellent talents in university of ministry of education of china ( ncet-13 - 0836 ) .
m. o. scully and m. s. zubairy , _ quantum optics _ ( cambridge university press , cambridge , uk , 1997 ) . l. allen and j. h. eberly , _ optical resonance and two- level atoms _ ( cambridge university press , cambridge , uk , 1997 ) . b. w. shore , _ the theory of coherent atomic excitation _
( wiley , new york , 1990 ) .
i. i. rabi , physical review * 51 * , 652 ( 1937 ) .
n. rosen and c. zener , physical review * 40 * , 502 ( 1932 ) .
s. l. mccall and e. l. hahn , phys . rev . * 183 * , 457 ( 1969 ) .
y. v. bezvershenko , p. i. holod , and a. messina , phys .
* t143 * , 014005 ( 2011 ) .
f. grossmann , t. dittrich , p. jung , and p. hanggi , phys .
lett . * 67 * , 516 ( 1991 ) ; z. phys .
b * 84 * , 315 ( 1991 ) . l. landau , phys .
z. sowjetunion * 2 * , 46 ( 1932 ) a. bambini and p. r. berman , physical review a * 23 * , 2496 ( 1981 )
. f. t. hioe , physical review a * 30 * , 2100 ( 1984 ) .
e. j. robinson , physical review a * 31 * , 3986 ( 1985 ) .
f. t. hioe and c. e. carroll , physical review a * 32 * , 1541 ( 1985 ) .
a. m. ishkhanyan , journal of physics a : mathematical and general * 33 * , 5539 ( 2000 ) .
n. v. vitanov , new journal of physics * 9 * , 58 ( 2007 ) . c. zener , proc .
a * 137 * , 696 ( 1932 ) .
l. s. simeonov and n. v. vitanov , physical review a * 89 * , 043411 ( 2014 ) .
a. gangopadhyay , m. dzero and v. galitski , physical review b * 82 * , 024303 ( 2010 ) . p. k. jha and y. v. rostovtsev , physical review a * 82 * , 015801 ( 2010 ) .
p. k. jha and y. v. rostovtsev , physical review a * 81 * , 033827 ( 2010 ) .
q. xie and w. hai , physical review a * 82 * , 032117 ( 2010 ) .
zhang , k. jin , l .- l .
jin , and x .- t .
xie , physical review a * 93 * , 043840 ( 2016 ) .
e. barnes and s. das sarma , physical review letters * 109 * , 060401 ( 2012 ) .
e. barnes , physical review a * 88 * , 013818 ( 2013 ) .
a. messina and h. nakazato , journal of physics a : mathematical and theoretical * 47 * , 445302 ( 2014 ) .
w. hai , k. hai , and q. chen , physical review a * 87 * , 023403 ( 2013 ) . c. m. bender , m. berry , and a. mandilara , j. phys .
a * 35 * , l467 ( 2002 ) .
a. guo , g. j. salamo , d. duchesne , r. morandotti , m. volatier - ravat , v. aimez , g. a. siviloglou , and d. n. christodoulides , phys . rev
. lett . * 103 * , 093902 ( 2009 ) . c. e. ruter , k. g. makris , r. el - ganainy , d. n. christodoulides , m. segev , and d. kip , nat . phys . * 6 * , 192 ( 2010 ) .
n. moiseyev , phys .
a * 83 * , 052125 ( 2011 ) .
r. driben and b.
a. malomed , europhys .
lett . * 96 * , 51001 ( 2011 ) .
x. luo , j. huang , h. zhong , x. qin , q. xie , y. s. kivshar , and c. lee , phys .
lett . * 110 * , 243902 ( 2013 ) .
y. n. joglekar , r. marathe , p. durganandini , and r. k. pathak , phys .
a * 90 * , 040101 ( 2014 ) .
j. gong and q .- h .
wang , phys .
rev . a * 91 * , 042135 ( 2015 ) .
t. e. lee and y. n. joglekar , phys .
a * 92 * , 042103 ( 2015 ) .
b. yang , x. luo , q. hu , and x. yu , physical review a * 94 * , 043828 ( 2016 ) j. li , a. k. harter , j. liu , l. de melo , y. n. joglekar , and l. luo , arxiv preprint arxiv:1608.05061 ( 2016 ) .
|
we propose a simple method of combined synchronous modulations to generate the analytically exact solutions for a parity - time symmetric two - level system .
such exact solutions are expressible in terms of simple elementary functions and helpful for illuminating some generalizations of appealing concepts originating in the hermitian system . some intriguing physical phenomena , such as stabilization of a non - hermitian system by periodic driving , non - hermitian analogs of coherent destruction of tunneling ( cdt ) and complete population inversion ( cpi ) , are demonstrated analytically and confirmed numerically .
in addition , by using these exact solutions we derive a pulse area theorem for such non - hermitian cpi in the parity - time symmetric two - level system .
our results may provide an additional possibility for pulse manipulation and coherent control of the parity - time symmetric two - level system .
|
the calculus on time scales was introduced by bernd aulbach and stefan hilger in 1988 @xcite .
the new theory unify and extends the traditional areas of continuous and discrete analysis and the various dialects of @xmath0-calculus @xcite into a single theory @xcite , and is finding numerous applications in such areas as engineering , biology , economics , finance , and physics @xcite .
the present work is dedicated to the study of problems of calculus of variations on a generic time scale @xmath1 .
as particular cases , one gets the classical calculus of variations @xcite by choosing @xmath2 ; the discrete - time calculus of variations @xcite by choosing @xmath3 ; and the @xmath0-calculus of variations @xcite by choosing @xmath4 , @xmath5 .
the calculus of variations on time scales was born with the works @xcite and @xcite and seems to have interesting applications in economics @xcite .
currently , several researchers are getting interested in the new theory and contributing to its development ( see , , @xcite ) . here
we develop further the theory by proving necessary optimality conditions for more general problems of the calculus of variations with a lagrangian that may also depend on the unspecified end - point @xmath6 .
in section [ sec : prm ] we review the necessary concepts and tools on time scales ; our results are given in section [ sec : mr ] .
we begin section [ sec : mr ] by formulating the problem under study : to minimize a delta - integral functional subject to a given fixed initial - point @xmath7 and having no constraint on @xmath6 .
the novelty is the dependence of the integrand @xmath8 on the free end - point @xmath6 .
necessary optimality conditions for such problems , on a general time scale , are given using both lagrangian ( theorem [ thm : mr ] ) and hamiltonian formalisms ( theorem [ thm : mr : ham ] ) . under appropriate convexity and linearity assumptions ,
the conditions turn out to be sufficient for a global minimum ( theorem [ sc ] ) .
a number of important corollaries are obtained , and several examples illustrating the new results discussed in detail .
corollary [ cor : anws : az : cv ] ( see also corollary [ cor : t = r : hamilt ] ) give answer to a question posed to the second author by a. zinober in may 2008 during a visit to the university of aveiro , and again presented as an open question during the conference `` calculus of variations and applications from engineering to economy '' , held from 8th to 10th september 2008 in the new university of lisbon , monte de caparica , portugal : `` what are the necessary optimality conditions for the problem of the calculus of variations with a free end - point @xmath6 but whose lagrangian depends explicitly on @xmath6 ? ''
the new transversality condition ( or the equivalent natural boundary condition ) seems to have important implications in economics .
this question is under study by alan zinober , kim kaivanto , and pedro cruz and will appear elsewhere .
in this section we introduce basic definitions and results that will be needed for the rest of the paper . for a more general presentation of the theory of time scales ,
we refer the reader to @xcite .
a nonempty closed subset of @xmath9 is called a _ time scale _ and it is denoted by @xmath1 .
thus , @xmath9 , @xmath10 , and @xmath11 , are trivial examples of times scales .
other examples of times scales are : @xmath12 \bigcup \mathbb{n}$ ] , @xmath13 for some @xmath14 , @xmath15 for some @xmath5 , and the cantor set .
we assume that a time scale @xmath1 has the topology that it inherits from the real numbers with the standard topology .
the _ forward jump operator _
@xmath16 is defined by @xmath17 while the _ backward jump operator _
@xmath18 is defined by @xmath19 with @xmath20 ( , @xmath21 if @xmath1 has a maximum @xmath22 ) and @xmath23 ( , @xmath24 if @xmath1 has a minimum @xmath25 ) .
a point @xmath26 is called _ right - dense _ , _ right - scattered _ , _ left - dense _ and _ left - scattered _ if @xmath27 , @xmath28 , @xmath29 and @xmath30 , respectively .
the _ graininess function _
@xmath31 is defined by @xmath32 if @xmath2 , then @xmath33 and @xmath34 .
if @xmath3 , then @xmath35 , @xmath36 , and @xmath37 . on the other hand , if @xmath38 , where @xmath5 is a fixed real number , then we have @xmath39 , @xmath40 , and @xmath41 .
@xcite a time scale @xmath1 is called _ regular _ if the following two conditions are satisfied simultaneously : * @xmath42 , for all @xmath43 ; * @xmath44 , for all @xmath43 .
following @xcite , let us define @xmath45 $ ] .
[ def : de : dif ] we say that a function @xmath46 is _ delta differentiable _ at @xmath47 if there exists a number @xmath48 such that for all @xmath49 there is a neighborhood @xmath50 of @xmath51 ( , @xmath52 for some @xmath53 ) such that @xmath54 we call @xmath48 the _ delta derivative _ of @xmath8 at @xmath51 and say that @xmath8 is _ delta differentiable _ on @xmath55 provided @xmath48 exists for all @xmath47 .
if @xmath56 , then @xmath48 is not uniquely defined , since for such a point @xmath51 , small neighborhoods @xmath50 of @xmath51 consist only of @xmath51 and , besides , we have @xmath57 .
for this reason , maximal left - scattered points are omitted in definition [ def : de : dif ] .
note that in right - dense points @xmath58 , provided this limit exists , and in right - scattered points @xmath59 , provided @xmath8 is continuous at @xmath51 .
[ ex : der : qc ] if @xmath2 , then @xmath60 , , the delta derivative coincides with the usual one . if @xmath3 , then @xmath61 . if @xmath38 , @xmath5 , then @xmath62 , , we get the usual derivative of quantum calculus @xcite . a function @xmath46 is called _ rd - continuous _ if it is continuous at right - dense points and if its left - sided limit exists at left - dense points .
we denote the set of all rd - continuous functions by c@xmath63 and the set of all delta differentiable functions with rd - continuous derivative by c@xmath64 .
it is known that rd - continuous functions possess an _ antiderivative _ , , there exists a function @xmath65 with @xmath66 , and in this case the _ delta integral _ is defined by @xmath67 for all @xmath68 .
let @xmath69 with @xmath70 .
if @xmath2 , then @xmath71 , where the integral on the right - hand side is the classical riemann integral .
if @xmath3 , then @xmath72 . if @xmath38 , @xmath5 , then @xmath73 .
the delta integral has the following properties : * if @xmath74 and @xmath75 , then @xmath76 * if @xmath68 and @xmath77 , then + @xmath78_{t = c}^{t = d}-\int_{c}^{d}f^{\delta}(t)g(t)\delta t \ , , \ ] ] + @xmath79_{t = c}^{t = d}-\int_{c}^{d } f^{\delta}(t)g(\sigma(t))\delta t.\ ] ] the dubois - reymond lemma of the calculus of variations on time scales will be useful for our purposes .
( lemma of dubois - reymond @xcite ) [ lemma : dr ] let @xmath80 , @xmath81^k\rightarrow\mathbb{r}$ ] .
then , @xmath82 if and only if @xmath83^k$ for some $ c\in\mathbb{r}$}$ ] .
let @xmath1 be a bounded time scale .
throughout we let @xmath84 with @xmath85 . for an interval @xmath86\cap \mathbb{t}$ ]
we simply write @xmath86 $ ] .
we also abbreviate @xmath87 by @xmath88 .
now let @xmath89 $ ] with @xmath90 be a subinterval of @xmath91 $ ] .
the problem of the calculus of variations on time scales under consideration has the form @xmath92=\int_{a}^{t}f(t , x^{\sigma}(t),x^{\delta}(t),x(t))\delta t , \ ] ] over all @xmath93 satisfying the boundary condition @xmath94 where the function @xmath95 from @xmath89\times \mathbb{r } \times \mathbb{r } \times \mathbb{r } $ ] to @xmath9 has partial continuous derivatives with respect to @xmath96 for all @xmath97 $ ] , and @xmath98 and its partial derivatives are rd - continuous for all @xmath51 .
a function @xmath93 is said to be admissible if it is satisfies condition .
let us consider the following norm in @xmath99 : @xmath100}|x^{\sigma}(t)|+sup_{t\in[a , t]}|x^{\delta}(t)|.\ ] ] an admissible function @xmath101 is said to be a _ weak local minimum _ for if there exists @xmath102 such that @xmath103\leq \mathcal{l}[x]$ ] for all admissible @xmath104 with @xmath105 .
next theorem gives necessary optimality conditions for problem . [ thm : mr ] if @xmath106 is a solution of the problem
, then @xmath107 for all @xmath108^\kappa$ ] .
moreover , @xmath109 suppose that @xmath110 $ ] has a weak local minimum at @xmath106 .
we can proceed as lagrange did , by considering the value of @xmath111 at a nearby function @xmath112 , where @xmath113 is a small parameter , @xmath114 , and @xmath115 . because @xmath6 is free
, we do not require @xmath116 to vanish at @xmath117 .
let @xmath118 = \int_a^t f(t,\tilde{x}^{\sigma}(t)+\varepsilon h^{\sigma}(t),\tilde{x}^{\delta}(t)+ \varepsilon h^{\delta}(t),\tilde{x}(t)+\varepsilon h(t ) ) \delta t.\ ] ] a necessary condition for @xmath106 to be a minimum is given by @xmath119\delta t = 0 \ , , \ ] ] where @xmath120 .
integration by parts gives @xmath121 because @xmath115 , the necessary condition can be written as @xmath122 for all @xmath123 such that @xmath115 .
in particular , equation holds for the subclass of functions @xmath123 that do vanish at @xmath124 .
thus , by the dubois - reymond lemma [ lemma : dr ] , we have @xmath125 for some @xmath126 and all @xmath97 $ ] .
equation must be satisfied for all @xmath123 with @xmath115 , which includes functions @xmath116 that do not vanish at @xmath117 .
consequently , equations and imply that @xmath127 from the properties of the delta integral and from , it follows that @xmath128 hence , we can rewrite as .
[ thm : mr : reg ] let @xmath1 be a regular time scale . if @xmath106 is a solution of the problem , then @xmath129 for all @xmath108^\kappa$ ] .
moreover , @xmath130 by theorem [ thm : mr ] we need only to show that on a regular time scale equation can be written in the form .
indeed , from the properties of the delta integral it follows that @xmath131 choosing @xmath132 in theorem [ thm : mr : reg ] we immediately obtain the corresponding result in the classical context of the calculus of variations .
we were not able to find a reference , in the vast and rich literature of the calculus of variations , to the result given by corollary [ cor : anws : az : cv ] .
[ cor : anws : az : cv ] if @xmath106 is a solution of the problem @xmath133 = \int_a^t f(t , x(t),x'(t),x(t ) ) dt \\
x(a ) = \alpha \ , , \quad \alpha\in \mathbb{r } \quad ( x(t ) \text { free } ) , \end{gathered}\ ] ] where @xmath134 , @xmath135 , @xmath136 , then the euler - lagrange equation @xmath137 holds for all @xmath108 $ ] .
moreover , @xmath138 in the classical setting @xmath8 does not depend on @xmath6 , , @xmath139 .
then , reduces to the well known natural boundary condition @xmath140 .
similarly , we can obtain other corollaries by choosing different time scales .
corollary [ cor : z ] is obtained from theorem [ thm : mr : reg ] letting @xmath141 .
[ cor : z ] if @xmath106 is a solution of the discrete - time problem @xmath133 = \sum _ { t = a}^{t-1}f(t , x(t+1),\delta x(t),x(t))\\ x(a ) = \alpha \ , , \quad \alpha\in \mathbb{r } \quad ( x(t ) \text { free } ) , \end{gathered}\ ] ] where @xmath142 , @xmath135 , then @xmath143 for all @xmath144 $ ] .
moreover , @xmath145 in the case @xmath8 does not depend on @xmath6 , reduces to the natural boundary condition for the discrete variational problem ( see ( * ? ? ?
* theorem 8.3 ) ) .
let now @xmath146 , @xmath5 .
we then obtain the analogous result for the @xmath0-calculus of variations . in what follows we use the standard notation @xmath147 for the @xmath0-derivative : @xmath148 ( example [ ex : der : qc ] ) .
the @xmath0-derivative is also known in the literature as the jackson derivative @xcite .
if @xmath106 is a solution of the problem
@xmath149 = \sum _ { t = a}^{tq^{-1}}(q-1)t f \left(t , x(qt),d_q x(t),x(t)\right)\\ x(a ) = \alpha \ , , \quad \alpha\in \mathbb{r } \quad ( x(t ) \text { free } ) \ , , \end{gathered}\ ] ] where @xmath150 , @xmath135 , then @xmath151 for all @xmath152 $ ] .
moreover , @xmath153 we illustrate the application of our theorem [ thm : mr ] with an example .
[ fromb ] consider the problem @xmath154=\int_{0}^{1}\left(\sqrt{1+(x^{\delta}(t))^2}+\beta(x(1)-1)^2 \right)\delta t , \ ] ] where @xmath155 , subject to the boundary condition @xmath156 since @xmath157 we have @xmath158 if @xmath106 is a local minimizer of
, then conditions must hold , , @xmath159 @xmath160 equation implies that there exists a constant @xmath161 such that @xmath162 solving the latter equation with initial condition @xmath163 we obtain @xmath164 , where @xmath165 . in order to determine @xmath166 we use the natural boundary condition , which can be rewritten as @xmath167
the real solution of equation is @xmath168 hence , @xmath169 is a candidate to be a minimizer with @xmath170 = \sqrt { 1 + \left ( { \frac { 2\,{\beta}^{2}+\sqrt { 4\,{\beta}^{4}+ { \beta}^{2}}}{{4 \beta}^{2}}}- { \frac { \sqrt { -8\,{\beta}^{2}+4\ , \sqrt { 4\,{\beta}^{4}+{\beta}^{2}}+1}}{4 \beta } } \right ) ^{2}}\\ + \beta\ , \left ( { \frac { 2\,{\beta}^{2}+\sqrt { 4\,{\beta}^{4}+{\beta}^{2 } } } { { 4 \beta}^{2}}}-{\frac { \sqrt { -8\,{\beta}^{2}+4\,\sqrt { 4\ , { \beta}^{4}+{\beta}^{2}}+1}}{4 \beta}}-1 \right ) ^{2}.\end{gathered}\ ] ] the extremal @xmath171 is represented in figure [ fig:1 ] for different values of @xmath172 . [ fig:1 ] we note that @xmath173 and in the limit , when @xmath174 , the solution of coincides with the solution of the following problem with fixed initial and terminal points ( @xcite ) : @xmath133 = \int_{0}^{1}\left(\sqrt{1+(x^{\delta}(t))^2}\right)\delta t \\
x(0 ) = 0 \ , , \quad x(1 ) = 1 \ , . \end{gathered}\ ] ] hamiltonian systems on time scales were introduced in @xcite and have a central role in the study of optimal control problems on time scales @xcite .
let us consider now the more general problem @xmath175=\int_{a}^{t}f(t , x^{\sigma}(t),u^{\sigma}(t),x(t))\delta t\ ] ] subject to @xmath176 @xmath177 where @xmath178\times \mathbb{r } \times \mathbb{r } \times \mathbb{r } \rightarrow \mathbb{r}$ ] and @xmath179\times \mathbb{r } \times \mathbb{r } \times \mathbb{r } \rightarrow \mathbb{r}$ ] have partial continuous derivatives with respect to @xmath96 for all @xmath97 $ ] , and @xmath98 , @xmath180 and their partial derivatives are rd - continuous for all @xmath51 . in the particular case @xmath181
problem reduces to . [ thm : mr : ham ] if @xmath182 is a weak local minimizer for the problem , then there is a function @xmath183 such that the triple @xmath184 satisfies : ( i ) the hamiltonian system @xmath185 ( ii ) the stationary condition @xmath186 for all @xmath108^\kappa$ ] ; and ( iii ) the transversality condition @xmath187 where the hamiltonian @xmath188\times \mathbb{r } \times \mathbb{r } \times \mathbb{r } \times \mathbb{r}\rightarrow \mathbb{r}$ ] is defined by @xmath189 in theorem [ thm : mr : ham ] we are assuming to have a time scale @xmath1 for which @xmath190 is delta - differentiable on @xmath89^\kappa$ ]
. examples of time scales for which @xmath191 is not delta - differentiable are easily found @xcite . using the lagrange multiplier rule we can form an expression @xmath192 for each value of @xmath51 .
the replacement of @xmath8 by @xmath193 in the objective functional give us the following new problem : @xmath194 = \int_{a}^{t}\bigl\{f(t , x^{\sigma}(t),u^{\sigma}(t),x(t))\\ + \lambda^{\sigma}(t)\left[g(t , x^{\sigma}(t),u^{\sigma}(t),x(t))-x^{\delta}(t)\right]\bigr\}\delta t , \end{gathered}\ ] ] subject to @xmath195 suppose that @xmath182 is a weak local minimizer for the problem . then the triple @xmath184 should be a weak local minimizer for the problem . using in we write the functional in the form @xmath196= \int_{a}^{t}[h(t , x^{\sigma},u^{\sigma},\lambda^{\sigma},x(t ) ) -\lambda^{\sigma}(t)x^{\delta}(t)]\delta t .\ ] ] applying theorem [ thm : mr ] to the problem
, in view of , gives conditions . if @xmath1 is a regular time scale , then by theorem [ thm : mr : reg ] the transversality condition can be written in the form @xmath197 [ ham : nc ] consider the problem @xmath198=\int_{0}^{3}(u^{\sigma}(t))^2+t^{2}(x(3)-1)^2 \delta t , \\
x^{\delta}(t)=u^{\sigma}(t ) , \end{gathered}\ ] ] @xmath199 to find candidate solutions for the problem , we start by forming the hamiltonian function @xmath200 candidate solutions @xmath182 are those satisfying the following conditions : @xmath201 @xmath202 @xmath203 @xmath204 from we conclude that @xmath205 . in order to determine @xmath206 we use the transversality condition which we can write as @xmath207 the value of the delta integral in depends on the time scale .
let us compute , for example , this delta integral on @xmath141 and on @xmath208 for @xmath209 .
for @xmath141 @xmath210 equations and yield @xmath211 .
therefore , the extremal of the problem on @xmath141 is @xmath212 .
on the other hand , for @xmath213 we have @xmath214 and @xmath215 . when @xmath2 we immediately obtain from theorem [ thm : mr : ham ] the following corollary .
[ cor : t = r : hamilt ] if @xmath182 is a solution of the problem @xmath216 = \int_a^t f(t , x(t),u(t),x(t ) ) dt \\
x'(t)=g(t , x(t),u(t),x(t))\\ x(a ) = \alpha \quad ( x(t ) \text { free } ) , \end{gathered}\ ] ] where @xmath166 , @xmath217 , @xmath218 , @xmath135 , then there exists a function @xmath183 such that the triple @xmath184 satisfies the hamiltonian system @xmath219 the stationary condition @xmath220 for all @xmath108 $ ] , and the transversality condition @xmath221 where the hamiltonian @xmath222 is defined by @xmath223 . in the classical context @xmath8 and @xmath224
do not depend on @xmath6 . in that case
the transversality condition coincides with the standard one ( @xmath225 ) and corollary [ cor : t = r : hamilt ] coincides with the hestenes theorem @xcite ( a weak form of the pontryagin maximum principle @xcite ) .
we were not able to find a single reference to the transversality condition in the literature of optimal control .
we illustrate the use of corollary [ cor : t = r : hamilt ] with an example : [ ham : nc : r ] consider the problem @xmath226=\int_{-1}^{1}(u(t))^2dt , \\
x'(t)=u(t)+x(1)t , \end{gathered}\ ] ] @xmath227 we begin by writing the hamiltonian function @xmath228 candidate solutions @xmath182 are those satisfying the following conditions : @xmath229 @xmath230 @xmath231 @xmath232 the equation has solution @xmath233 , @xmath234 , which upon substitution into yields @xmath235 from the stationary condition we get @xmath236 . finally , substituting the optimal control candidate back into yields @xmath237 integrating the latter equation with the initial condition @xmath238 ,
we obtain @xmath239 substituting @xmath240 into we get @xmath241 .
therefore , the candidate to solution for the problem is @xmath242 . in certain cases
it is easy to show that the extremal candidates obtained from theorem [ thm : mr ] , theorem [ thm : mr : ham ] , or one of the corollaries are indeed minimizers .
[ sc ] let @xmath243 be jointly convex in @xmath244 and @xmath245 be linear in @xmath244 for each @xmath51 . if @xmath184 is a solution of system , then @xmath182 is a global minimizer of . since @xmath8 is convex in @xmath244 for any admissible pair @xmath246
, we have @xmath247-\mathcal{l}[\tilde{x},\tilde{u}]=\int_{a}^{t}\left[f(t , x^{\sigma}(t),u^{\sigma}(t),x(t ) ) -f(t,\tilde{x}^{\sigma}(t),\tilde{u}^{\sigma}(t),\tilde{x}(t))\right]\delta t\\ \geq \int_{a}^{t}\bigl[f_{x^{\sigma}}(t,\tilde{x}^{\sigma}(t ) , \tilde{u}^{\sigma}(t),\tilde{x}(t))(x^{\sigma}(t)-\tilde{x}^{\sigma}(t))\\ + f_{u^{\sigma}}(t,\tilde{x}^{\sigma}(t),\tilde{u}^{\sigma}(t),\tilde{x}(t))(u^{\sigma}(t)-\tilde{u}^{\sigma}(t))\\ + f_{z}(t,\tilde{x}^{\sigma}(t),\tilde{u}^{\sigma}(t),\tilde{x}(t))(x(t)-\tilde{x}(t))\bigr]\delta t.\end{gathered}\ ] ] because the triple @xmath184 satisfies equations , , and , we obtain @xmath247-\mathcal{l}[\tilde{x},\tilde{u } ] \geq \int_{a}^{t}\bigl[-\tilde{\lambda}^{\sigma}(t)g_{x^{\sigma}}(\cdots)(x^{\sigma}(t)-\tilde{x}^{\sigma}(t))\\ -(\tilde{\lambda}^{\sigma}(t))^{\delta}(x^{\sigma}(t ) -\tilde{x}^{\sigma}(t))-\tilde{\lambda}^{\sigma}(t ) g_{u^{\sigma}}(\cdots)(u^{\sigma}(t)-\tilde{u}^{\sigma}(t))\\ -\tilde{\lambda}^{\sigma}(t)g_{z}(\cdots)(x(t)-\tilde{x}(t))\bigr]\delta t\\ + \left[\tilde{\lambda}^{\sigma}(\rho(t))-\int_{\rho(t)}^{t}h_{x^{\sigma}}(\cdots\cdot)\delta t\right](x(t)-\tilde{x}(t)),\end{gathered}\ ] ] where @xmath248 and @xmath249 .
integrating by parts the term in @xmath250 we get @xmath247-\mathcal{l}[\tilde{x},\tilde{u}]\geq \int_{a}^{t}\bigl[-\tilde{\lambda}^{\sigma}(t)g_{x^{\sigma}}(\cdots)(x^{\sigma}(t)-\tilde{x}^{\sigma}(t))\\ + \tilde{\lambda}^{\sigma}(t)(x^{\delta}(t)-\tilde{x}^{\delta}(t))-\tilde{\lambda}^{\sigma}(t ) g_{u^{\sigma}}(\cdots)(u^{\sigma}(t)-\tilde{u}^{\sigma}(t ) ) -\tilde{\lambda}^{\sigma}(t)g_{z}(\cdots)(x(t)-\tilde{x}(t))\bigr]\delta t\\ + \left[\tilde{\lambda}^{\sigma}(\rho(t))- \int_{\rho(t)}^{t}h_{x^{\sigma}}(\cdots\cdot)\delta t\right](x(t)-\tilde{x}(t))-\tilde{\lambda}^{\sigma}(t)(x(t)-\tilde{x}(t)).\end{gathered}\ ] ] but from and properties of the delta integral , we have @xmath251 rearranging the terms we obtain : @xmath252-\mathcal{l}[\tilde{x},\tilde{u}]\geq \int_{a}^{t}\bigl\{-\tilde{\lambda}^{\sigma}(t)\bigl[\tilde{x}^{\delta}(t)-x^{\delta}(t ) + g_{x^{\sigma}}(\cdots)(x^{\sigma}(t)-\tilde{x}^{\sigma}(t))\\ + g_{u^{\sigma}}(\cdots)(u^{\sigma}(t)-\tilde{u}^{\sigma}(t ) ) + g_{z}(\cdots)(x(t)-\tilde{x}(t))\bigr]\bigr\}\delta t.\end{gathered}\ ] ] the intended conclusion follows from and linearity of @xmath224 in @xmath244 : the right hand side of inequality is equal to zero , that is , @xmath253\geq\mathcal{l}[\tilde{x},\tilde{u}]\ ] ] for each admissible pair @xmath246 .
consider the problem in example [ ham : nc : r ] .
the integrand is independent of @xmath254 and convex in @xmath255 .
the right - hand side of the control system is linear in @xmath256 and independent of @xmath104 .
hence , theorem [ sc ] asserts that the extremal @xmath257 found in example [ ham : nc : r ] gives the global minimum to the problem .
consider again the problem from example [ fromb ] .
replacing @xmath258 by @xmath259 we can rewrite problem in the following form : @xmath260=\int_{0}^{1}\left(\sqrt{1+(u^{\sigma}(t))^2}+\beta(x(1)-1)^2 \right)\delta t\ ] ] subject to @xmath261 the function @xmath8 is independent of @xmath104 and convex in @xmath256 .
the right - hand side of the control system is linear in @xmath255 and independent of @xmath254 .
therefore , by theorem [ sc ] @xmath171 is the global minimizer for the problem .
the first author was supported by biaystok technical university , via a project of the polish ministry of science and higher education `` wsparcie miedzynarodowej mobilnosci naukowcow '' ; the second author by the r&d unit ceoc , via fct and the ec fund feder / poci 2010 .
abm is grateful to the good working conditions at the department of mathematics of the university of aveiro where this research was carried out .
\(1062633 ) b. aulbach and s. hilger , a unified approach to continuous and discrete dynamics , in _ qualitative theory of differential equations ( szeged , 1988 ) _ , colloq . math .
jnos bolyai , 53 , north - holland , amsterdam , 1990 , 3756 .
r. hilscher and v. zeidan , _ weak maximum principle and accessory problem for control problems on time scales _ , nonlinear analysis : theory , methods & applications , in press , doi : 10.1016/j.na.2008.04.025 .
\(0166037 ) l. s. pontryagin , v. g. boltyanskii , r. v. gamkrelidze , e. f. mishchenko , `` the mathematical theory of optimal processes '' , interscience publishers john wiley & sons , inc.new york - london , 1962 .
|
we prove necessary optimality conditions for problems of the calculus of variations on time scales with a lagrangian depending on the free end - point . agnieszka b. malinowska delfim f. m. torres
|
the aim of this paper is to introduce and compute a function , called the
renormalized energy " , for some specific random point processes that arise in random matrix models , and in this way to associate to each of these processes a unique number , which is expected to measure its disorder " .
our renormalized energy " , that we denote @xmath3 , is defined over configurations of points lying either on the real line or on the plane , as the limit as @xmath4 of @xmath5 } \log \left|2\sin \frac{\pi(a_i - a_j ) } n\right| + \log n\qquad \text{in dimension 1},\\ \wc(\{a_i\ } ) & = \frac{1}{2\pi n^2 } \sum_{i\neq j , a_i , a_j \in [ 0,n]^2 } e_n(a_i - a_j ) + \log \frac{n}{2\pi \eta(i)^2}\qquad \text{in dimension 2},\end{aligned}\ ] ] where @xmath6 is an explicit eisenstein series , and @xmath7 is the dedekind eta function .
this definition is inspired by that of the renormalized energy " , denoted @xmath8 , introduced by sandier and the second author in @xcite in the case of points in the plane and in @xcite in the case of points on the real line .
the definitions for @xmath8 and @xmath3 coincide when the point configuration has some periodicity ( this is where our new definition originates ) , and in that case they amount to computing a sum of pairwise interactions @xmath9 where @xmath10 are the points and @xmath11 is a suitable logarithmic kernel ( the green s function on the underlying torus ) ; however they are not identical in general
. we will give more details on the connection in section [ secdefw ] .
in @xcite it is shown that in dimension 1 , @xmath8 is bounded below and its minimum is achieved at the perfect lattice @xmath12 . in dimension 2 , the situation is more complex ; it is also shown in @xcite that the minimum of @xmath8 is achieved , but it is only conjectured that this minimum value is achieved at the perfect triangular lattice or abrikosov lattice " according to the terminology of the physics of superconductors ( which was the first main motivation for @xmath8 in @xcite where it was introduced ) .
this conjecture is supported by the result that , among configurations of points which form a perfect lattice ( of fixed volume ) , the renormalized energy is minimal if and only if the lattice is the perfect triangular lattice , i.e. with 60@xmath13 angles ( that result is shown in @xcite based on the use of modular functions and results in number theory ) .
it is thus natural to think of @xmath8 or @xmath3 as a way to measure the disorder of a configuration of points . with this outlook , in dimension 1
the lattice @xmath12 is the most ordered configuration of points with prescribed density 1 , while in dimension 2 , it is expected to be the triangular lattice ( which is better ordered than any other lattice , say the square lattice ) .
in addition , due to its logarithmic nature , @xmath3 has some nice scaling and additive properties , which we believe make it a very good object .
a further motivation for choosing to study @xmath3 as opposed to any other total pairwise interaction function is that , as seen in @xcite and @xcite , @xmath8 arises very naturally from the statistical mechanics of coulomb or log gases , which contain as particular cases the ginibre and goe / gue / gse ensembles of random matrices . in @xcite , @xmath8
was introduced and derived in the context of the minimization of the ginzburg - landau model of superconductivity . in @xcite it was derived as a sort of limiting interaction energy for two dimensional coulomb gases , and similarly in @xcite with log gases .
these works are based on analysis and energy estimates ( upper and lower bounds ) .
both the questions we pursue here and the methods we use are quite different : they aim at obtaining explicit formulas for specific random matrix models . in particular , this is a way to compute some interesting statistics over random matrix eigenvalues , as initiated by dyson - mehta @xcite .
we will comment more on this just below .
let us now briefly introduce the notion of a random point process .
a ( simple ) random point process is a probability measure on the set of all locally finite collections of ( mutually distinct ) points in the space , cf .
e.g. @xcite .
it can also be viewed as a random measure of the form @xmath14 , with the points @xmath15 distinct and @xmath16 discrete .
random point processes are essentially characterized by their
@xmath17-point correlation functions " @xmath18 , which give the probability densities of finding @xmath17 points at the locations @xmath19 .
we will normalize our processes so that the average number of points per unit volume is always @xmath20 , which is equivalent to @xmath21 .
perhaps the most famous random point process is the poisson process , characterized by the facts that the number of points in disjoint subsets are independent , and the number of points in any finite volume subset of the space follows a poisson distribution with parameter equal to the volume of the set with respect to a reference measure . an important class of point processes is that of determinantal point processes , see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and references therein . that class is characterized by the fact that the @xmath17-point correlation functions are given by symmetric minors of a ( correlation ) kernel , cf .
section [ correl ] .
it is easy to see that @xmath22 for the translation invariant poisson process , which means that it is `` too chaotic '' from the point of view of the renormalized energy .
let us list the ( stationary ) processes for which we show that @xmath23 provides more meaningful information .
we will be interested in one dimension in the @xmath0-sine processes ( @xmath1 ) which arise as the local limit of the law of eigenvalues in random matrix ensembles with orthogonal , unitary , and symplectic symmetry groups ( they are determinantal for @xmath2 and pfaffian otherwise ) . in two dimensions we will examine the ginibre " point process , which is also a determinantal process arising as the local limit of the law of eigenvalues of matrices from the complex ginibre ensemble ( i.e. square matrices with complex gaussian iid entries ) , for further reference , see @xcite ; as well as the random process of zeros of random gaussian analytic functions , often denoted gaf , whose description can be found in @xcite
. as our processes are always translation invariant , the @xmath24-point correlation function can always be written in the form @xmath25 for the 2-point _ cluster function _
@xmath26 ( we will come back to this notation in sections [ correl ] and [ sec4 ] ) .
the main results we obtain are the following : * for general stationary processes we identify sufficient conditions on the process and its @xmath24-point correlation function @xmath27 for the existence of @xmath28 , and give an explicit formula in terms of @xmath27 which is ( up to constants ) @xmath29 with @xmath30 or @xmath24 according to the dimension . that we check for the above mentioned processes .
we expect it to hold for general @xmath0-sine processes but we do not see an _ a priori _ reason for that . it is also not clear to us whether this condition has a physical meaning . ] * we apply this formula to the specific point processes mentioned above .
* for the specific point processes above , we explicitly compute the limit of the variance of @xmath31 and obtain that it is @xmath32 .
this implies that for such processes , @xmath31 concentrates around its expected value , and converges in probability to @xmath33 as @xmath4 .
* we prove that in the class of determinantal point processes with translation invariant kernels in dimensions 1 and 2 , @xmath34 is minimized by the processes whose correlation kernel is the fourier transform of the characteristic function of the ball .
a complete physical interpretation of this fact seems to be missing . in dimension 1
the optimization gives the @xmath2 sine process , which can be seen as a heuristic explanation of the fact that this process is the universal local limit of @xmath2 random matrix ensembles .
indeed , such a local limit has to be translation invariant and determinantal , and it is natural to expect that it also minimizes an appropriate energy functional . in other dimensions ,
point processes with such specific kernels called fermi - sphere point processes " appear in @xcite as higher - dimensional analogues of the sine process . for our set of specific processes , we thus show that we can attribute to the process a unique number , which we compute explicitly .
whenever these numbers are distinct this implies that the processes are mutually singular ( as measures on the space of point configurations ) .
moreover , we check that the processes that are expected to have the highest level of order " or rigidity indeed have a lower value of @xmath35 . for example for the @xmath0-sine processes we find @xmath36 where @xmath37 is the euler constant . these three numbers
form a decreasing sequence , as can be expected . in two dimensions
, we obtain that the ginibre process has a higher @xmath38 , hence less rigidity , than that of zeros of gaussian analytic functions , in agreement with the results and philosophy of @xcite .
the values of @xmath39 for the @xmath1 sine processes above equal twice the thermodynamic `` energy per particle '' or `` mean internal energy '' for the log - gas with infinitely many particles , as obtained by dyson in 1962 ( * ? ? ?
* i.ix , iii.vi ) .
this is not surprizing as our definition of @xmath31 essentially coincides with that of dyson ( * ? ? ? *
i.vi ) once the points on the interval @xmath40 $ ] are identified with points on a circle of same length .
furthermore , one can use this fact to infer finer asymptotic properties of @xmath31 as follows ( we are very grateful to peter forrester for the idea ) .
if , instead of considering growing windows , one approximates the @xmath0-sine processes by the _
circular ensembles _ of particles on the unit circle with joint probability density @xmath41 , one observes that in rescaled variables , @xmath31 is simply @xmath42 , and its characteristic function can be immediately obtained from selberg s formula for the partition function : @xmath43 see e.g. ( * ? ? ?
* section 4.7.2 ) for the formula for @xmath44 . using stirling s formula
it is not hard to see that for any @xmath45 and @xmath46 @xmath47 with @xmath48 and the expression of @xmath49 through the partition function @xmath44 coincides with that of the thermodynamic energy per particle . by lvy s continuity theorem
this implies the central limit theorem @xmath50 it is natural to conjecture that the same central limit theorem holds for @xmath31 on general @xmath0-sine processes ( constructed in @xcite ) as well , in particular providing asymptotic values @xmath51 and @xmath52 for any @xmath45 . at the moment
we do not know how to prove it , although in the @xmath1 cases one may hope to obtain a proof via controlling the asymptotics of moments of @xmath31 through explicit formulas for the correlation functions . in the follow - up paper @xcite , prompted by the wish to analyze experimental data , dyson and
mehta looked for a way of approximating the energy per particle , in the case @xmath53 , by averaging a statistic of the form @xmath54 over windows of increasing length in the random matrix spectrum .
the concentration requirement of the asymptotically vanishing variance led them to the conclusion that with their choice of @xmath55 , the statistic had to be corrected with further interaction terms ( * ? ? ? * eq .
( 109 ) ) .
our renormalized energy @xmath31 seems to be a better solution to dyson - mehta s problem , although one should note that dyson - mehta s statistic is purely local , while @xmath31 involves long range interaction ( between leftmost and rightmost particles in @xmath56 $ ] ) .
the paper is organized as follows : in section [ secdefw ] we give the precise definitions of @xmath8 , the context about it and its connection to coulomb energy , from @xcite .
this leads us to the definition of @xmath31 . in section [ sec3 ]
we compute limits of expectations of @xmath31 : first on the real line and for general processes , then for the @xmath1 sine processes , then in the plane and for general processes , finally for the explicit ginibre and zeros of gaf processes . in section [ optim ] we find minimizers of @xmath57 among determinantal processes with translation - invariant correlation kernels . in section [ sec4 ]
we compute the limit variance of @xmath31 for our specific processes ( and show it is @xmath32 ) . in section [ sc : misc ] we gather some miscellaneous computations : the effect on @xmath58 of superposition and decimation of processes and the computation of expectation of @xmath58 for the @xmath2 discrete sine processes .
.5 cm * acknowledgements : * we are very grateful to peter forrester for his idea of using circular ensembles for the heuristics above , and also for drawing our attention to @xcite .
we also thank the anonymous referee for a careful reading and very interesting comments .
a. b. thanks alan edelman for help with numerics and s. s. thanks c. sinan gntrk for helpful discussions related to section [ optim ] .
a. b. was partially supported by nsf grant dms-1056390 , s. s. was supported by a euryi award .
we also thank the msri for support to attend a workshop where this work was initiated .
the aim of this section is to present a definition of @xmath8 for a configuration of points either on the line or in the plane , which is directly deduced from that of @xcite , but depends only on the data of the points .
we do not attempt here to fully and rigorously connect the next definition with that of @xcite ( since we believe it presents some serious technical difficulties ) but the link will be readily apparent .
we start by recalling the definitions from @xcite , but rather in the form presented in @xcite . in @xcite ,
@xmath8 was associated to a discrete set of points ( with asymptotic density ) @xmath60 in the plane , via a vector field @xmath61 : if @xmath61 is a vector field in @xmath62 satisfying @xmath63 where @xmath60 is a positive number ( corresponding to the average point density ) and @xmath64 has the form @xmath65 then for any function @xmath66 we define @xmath67 [ defa ] let @xmath60 be a nonnegative number . let @xmath61 be a vector field in @xmath62 .
we say @xmath61 belongs to the admissible class @xmath68 if , hold and @xmath69 the renormalized energy " @xmath59 relative to the family of balls ( centered at the origin ) @xmath70 ( and the number @xmath60 ) in @xmath62 is defined for @xmath71 by @xmath72 where @xmath73 denotes positive cutoff functions satisfying , for some constant @xmath74 independent of @xmath75 , @xmath76 note that by scaling , the density of points can be changed to any , and the rule for that change of scales is @xmath77 where @xmath64 has density @xmath60 and @xmath78 ( hence the set @xmath79 has density @xmath20 ) .
this function was first introduced in @xcite and derived as a limiting interaction energy for vortices of ginzburg - landau configurations .
independently of ginzburg - landau it can be viewed as a coulombian interaction energy for an infinite number of points in the plane , computed via a renormalization .
many of its properties are stated in @xcite . in @xcite
it was directly connected to 2d log gases .
we will give more details and properties of @xmath8 in section [ sec : background ] . in @xcite a one - dimensional analogue to the above definition is introduced for the study of 1d log gases , we now present it . the renormalized energy between points in 1d
is obtained by embedding " the real line in the plane and computing the renormalized energy in the plane , as defined in @xcite .
more specifically , we introduce the following definitions : @xmath80 denotes the set of real numbers , but also the real line of the plane @xmath62 i.e. points of the form @xmath81 . for the sake of clarity ,
we denote points in @xmath80 by the letter @xmath82 and points in the plane by @xmath83 . for a function @xmath84 on @xmath80
, we define its natural extension @xmath85 to a function on @xmath62 by @xmath86 .
@xmath87 denotes the interval @xmath88 $ ] in @xmath80 .
@xmath89 denotes the measure of length on @xmath80 seen as embedded in @xmath62 , that is @xmath90 for any smooth compactly supported test function @xmath91 in @xmath62 .
this measure can be multiplied by bounded functions on the real - line . for any function @xmath84 on @xmath80 , and
any function @xmath92 in @xmath62 such that @xmath93 where @xmath60 is a nonnegative number and @xmath64 has the form @xmath94 we denote @xmath95 and latexmath:[\[\label{wr1}w(j , \chi ) : = \lim_{\eta\to 0 } \
( \hal\int_{\mr^2 \backslash \cup_{p\in\lambda } b(p,\eta ) } \bar{\chi } @xmath85 is the natural extension of @xmath66 .
[ defa ] let @xmath97 be a nonnegative number .
we say @xmath98 belongs to the admissible class @xmath68 if , hold and @xmath99 we use the notation @xmath73 for positive cutoff functions over @xmath80 satisfying , for some constant @xmath74 independent of @xmath75 , @xmath100 the renormalized energy @xmath8 is defined , for @xmath71 , by @xmath101 note that while @xmath8 in 2d can be viewed as a renormalized way of computing @xmath102 , in 1d it amounts rather to a renormalized computation of @xmath103 . in one dimension ,
the formula for change of scales is @xmath104 where @xmath61 corresponds to density @xmath60 and @xmath105 to @xmath106 .
we recall here some properties and background from @xcite . * since in the neighborhood of @xmath107 we have @xmath108 , @xmath109 , we have near @xmath15 the decomposition @xmath110 where @xmath111 is locally bounded , and it easily follows that the limits , exists .
it also follows that @xmath61 belongs to @xmath112 for any @xmath113 . *
because the number of points is infinite , the interaction over large balls needs to be normalized by the volume as in a thermodynamic limit , and thus @xmath8 does not feel compact perturbations of the configuration of points . even though the interactions are long - range , this is not difficult to justify rigorously . * the cut - off function @xmath114 can not simply be replaced by the characteristic function of @xmath115
because for every @xmath107 @xmath116 * in dimension 1 , there is a unique @xmath92 satisfying and for which @xmath117 is finite .
thus @xmath8 amounts to a function of @xmath64 only . * in dimension 2 , there is not a unique @xmath61 for which @xmath118 and it would be tempting to define @xmath8 as a function of @xmath64 only and not @xmath61 by minimizing over @xmath61 , however we do not know how to prove that the resulting object has good properties such as measurability .
an important question is that of characterizing the minimum and minimizers of @xmath8 . for the case of dimension 1 , it is proven in @xcite
that the value of @xmath8 does not depend on the choice of @xmath114 satisfying , that @xmath8 is borel - measurable over @xmath119 for @xmath113 , and that @xmath120 is achieved for @xmath121 .
for the case of dimension 2 , it is proven in theorem 1 of @xcite that the value of @xmath8 does not depend on @xmath122 as long as it satisfies , that @xmath8 is borel - measurable over @xmath119 , for @xmath113 and that @xmath123 is achieved and finite . moreover
, there exists a minimizing sequence consisting of periodic vector fields .
the value of the minimum of @xmath8 is not known , it is conjectured ( see @xcite ) that it is asymptotically equal to the value at the perfect triangular lattice .
this conjecture is supported by the fact that the triangular lattice can be proved to be the minimizer among lattice configurations of fixed volume .
in addition to arising from the study of ginzburg - landau @xcite , @xmath8 is naturally connected , as shown in @xcite , to vandermonde factors such as @xmath124 with @xmath125 where @xmath126 with @xmath30 or @xmath24 , for any potential @xmath127 ( for example @xmath127 quadratic ) . to explain this ,
let us introduce some more notation .
we set @xmath128 it is well known in potential theory ( see @xcite ) that , provided @xmath129 as @xmath130 , @xmath131 has a unique minimizer among probability measures , called the equilibrium measure let us denote it @xmath132 .
it is characterized by the fact that there exists a constant @xmath133 such that @xmath134 where for any @xmath135 , @xmath136 is the potential generated by @xmath135 , defined by @xmath137we may then set @xmath138 where @xmath133 is the constant in .
it satisfies @xmath139 it is easy to check that @xmath140 grows like @xmath127 at infinity .
the connection between @xmath141 and @xmath8 is given by the following exact splitting formula " , valid in both 1 and 2 dimensions .
[ lem26 ] for any @xmath142 , @xmath143 or @xmath144 , the following holds @xmath145 where @xmath8 is defined in or respectively , and where @xmath146 where the equation is solved in @xmath62 and @xmath147 is naturally extended into a measure on @xmath62 if @xmath143 .
( note that @xmath148 is the convolution with @xmath149 . )
we start by writing @xmath150 where @xmath151 denotes the diagonal of @xmath152 and @xmath153 .
the idea is to compute the right - hand side by splitting @xmath64 as @xmath154 . this way , using the fact that @xmath155
, we obtain @xmath156 since @xmath157 and since @xmath64 and @xmath158 have same mass @xmath159 , we have @xmath160 where we used that @xmath161 on the support of @xmath132 .
thus @xmath162 we then claim that the last term in the right - hand side is equal to @xmath163 .
we define @xmath164 .
we have @xmath165 , with @xmath166 , and near @xmath167 , @xmath168 is @xmath169 .
it is then not difficult to deduce @xmath170 on the other hand , by definition , and using green s formula , we have @xmath171 using the decomposition @xmath172 near each @xmath167 , adding @xmath173 and then letting @xmath174 , we arrive at the same result as the right - hand side of .
this establishes the claim , and the final result then follows from the change of scales @xmath175 in @xmath8 .
the reason why we need another definition here is that , as we already mentioned , we wish to define @xmath8 based on the knowledge of @xmath64 , i.e. a set of points , alone . for a given @xmath64 ,
there is no uniqueness of the @xmath61 satisfying or ( in fact the indetermination is exactly a constant , see ( * ? ? ?
* lemma 1.4 ) ) , and this makes it problematic to define @xmath8 for points only in a measurable manner .
however , if the configuration of points is periodic , then @xmath8 can be computed nonambiguously from @xmath64 only .
the formula for this in dimension 2 is given in @xcite .
if the periodicity is with respect to a torus @xmath176 , and the number of points in that torus is @xmath159 ( denote them @xmath177 ) then there exists a unique ( up to a constant ) solution to @xmath178 ( i.e. a periodic solution ) and we let @xmath179 it was proved in @xcite that @xmath180 is the smallest of the @xmath59 over all @xmath61 satisfying which are @xmath176-periodic , and it was established ( see formula ( 1.10 ) ) that @xmath181 where @xmath133 is a constant , and @xmath11 is the green s function associated to the torus in the following way : @xmath182 in @xcite , ( 1.11 ) , an explicit formula is given for this : latexmath:[\[\label{formexpl } w(j_{\{a_i\ } } ) = \hal \sum_{i \neq j } \sum_{p \in ( \mz^2)^ * \sm \{0\ } } \frac{e^{2i\pi p \cdot ( a_i- a_j ) } } { 4\pi^2 |p|^2 } + \frac{n}{2 } \lim_{x\to 0 } \
( \sum_{p \in ( \mz^2)^*\backslash \{0\ } } \frac { e^{2i\pi p \cdot x } } { 4\pi^2 |p|^2 } + \log for simplicity we prefer to work with density @xmath20 in a square torus , which is then necessarily @xmath184 where @xmath185 .
compared to @xcite , this will change the constants in the formulae .
so for the sake of clarity we will include below a complete proof of the following : [ wnper ] let @xmath186 be @xmath185 points on @xmath187 .
let @xmath188 be the @xmath187-periodic vector field associated through to the configuration of points @xmath189 , extended by periodicity to the whole plane .
then @xmath180 as defined in is a function of @xmath177 only , which is equal to @xmath190 where @xmath11 is the unique solution of @xmath191 with @xmath192 ; and @xmath75 is a constant , equal to @xmath193 . first ,
arguing as in @xcite proposition 3.1 , we have @xmath194 which means we reduce the computation in the plane to a computation on the torus .
the next step is a renormalized computation la @xcite , as skeched in the proof of lemma [ lem26 ] . solving for via the green function , in view of the translation invariance of the equation , we can choose @xmath195 let us denote @xmath196 it is well - known that @xmath197 is a continuous function .
we denote @xmath198 . inserting into ,
we have @xmath199 using green s formula for integration by parts , we have @xmath200 inserting and , we have @xmath201 with @xmath159 the inner unit normal to the circles . using for the first term the fact that @xmath11 is chosen to have average zero , and for the second term splitting @xmath202 as @xmath203 , using the continuity of @xmath197 and @xmath204 ( from ) , and letting @xmath174 we find @xmath205 inserting into yields .
it turns out that @xmath11 can be expressed through an eisenstein series .
the eisenstein series with parameters @xmath206 and @xmath207 ( cf .
@xcite ) is defined by @xmath208 let us now define , for @xmath209 , @xmath210 as in @xcite we will also need another classical modular function : the dedekind @xmath211 function .
it is defined , for @xmath212 in the upper - half of the complex plane by @xmath213 the following holds [ form2d ] let @xmath186 be @xmath185 points on @xmath187 , and let @xmath214 be defined as in .
@xmath6 being defined in , we have @xmath215 where points in the plane are identified with complex numbers . as in @xcite , @xmath11 ,
introduced in lemma [ wnper ] , can be computed from using fourier series and this yields : @xmath216 we recognize here an eisenstein series : @xmath217 note that the fact that @xmath11 has zero average implies that @xmath218 there remains to compute the value of the constant @xmath219 .
for that we use the second kronecker limit formula " ( see @xcite ) that asserts that @xmath220 where @xmath221 , @xmath222 , @xmath223 , and @xmath224 we use this formula with @xmath225 , @xmath226 , @xmath227 . in that case @xmath228 ,
@xmath229 and @xmath230 ( where the bar denotes complex conjugation ) , and the formula yields @xmath231 as @xmath232 we have @xmath233 and @xmath234 and also ( see ) @xmath235
so as @xmath236 , we have @xmath237 combining with , it follows that @xmath238 inserting this and into we get the result .
we emphasize here that this formula is the exact value for @xmath8 as defined in provided we assume full periodicity with respect to @xmath187 .
we do here the analogue in dimension 1 , i.e. we compute @xmath8 given by assuming that the point configuration is periodic with respect to @xmath239 .
we assume that there are @xmath240 points @xmath241 in @xmath187 , hence @xmath8 is computed with respect to the average density 1 , i.e. @xmath242 . there exists a unique ( up to a constant ) @xmath243 satisfying @xmath244 in @xmath62 and which is @xmath245-periodic with respect to the first variable , i.e @xmath246 for every @xmath247 .
we then set @xmath248 .
then , as in @xcite we have let @xmath249 be @xmath245 points on @xmath250 .
let @xmath188 be as above .
then @xmath180 as defined in is a function of @xmath241 only , which is equal to @xmath251 where @xmath252 is a restriction of the green function of @xmath253 , more precisely the solution to @xmath254 with @xmath255 , and @xmath75 is a constant , equal to @xmath256 here @xmath257 is the distribution defined over @xmath258 by @xmath259 the proof is analogous to that of lemma [ wnper ] .
first we have @xmath260 we again write @xmath261 with @xmath262 , @xmath75 a continuous function .
using green s formula we have @xmath263 from we have @xmath264 by choice of @xmath11 .
the rest of the proof is exactly as in lemma [ wnper ] , inserting the splitting of @xmath11 and .
on the other hand we shall see below that we have the following explicit formula for @xmath11 restricted to the real axis : @xmath265 note that a consequence of this and the zero average condition on @xmath11 is that @xmath266 the previous lemma and lead us to the following let @xmath241 be @xmath245 points on @xmath267 and let @xmath214 be defined as in .
we have @xmath268 the first step is to prove .
this is done by solving in fourier series / transform .
we choose the following normalization for fourier transforms and series : @xmath269 @xmath270 then the fourier inversion formula is @xmath271 and respectively @xmath272 since @xmath11 is @xmath187-periodic in its first variable , we may define its fourier transform as a fourier series in the first variable and a fourier transform in the second , i.e. , for @xmath273 and @xmath274 , @xmath275 if @xmath11 solves then by fourier transform , @xmath276 has to satisfy @xmath277 it is direct to establish that @xmath278 with @xmath279 by definition equal to @xmath20 if @xmath280 and @xmath32 otherwise . combining these facts ,
we obtain @xmath281 the undetermination of @xmath276 at @xmath282 corresponds to the fact that @xmath11 is only determined by up to a constant . by fourier inversion
, it follows that @xmath283 using the formula @xmath284 ( cf .
@xcite ) with @xmath285 and @xmath286 , we arrive at @xmath287 we next particularize to the case @xmath288 , and use clausen s formula ( cf .
4 ) ) @xmath289 and thus we find @xmath290 the constant @xmath133 should be chosen so that @xmath255 , which imposes @xmath291 , in view of the relation @xmath292 , a direct consequence of .
this establishes .
in addition the value of @xmath75 follows : @xmath293 and inserting into , the result follows . in order to define @xmath8 for a general configuration of points , without referring to the current they generate , we choose to base ourselves on the formulas and of the previous subsection .
to get rid of some constants , we use however a different normalization and , for a given family of points @xmath189 , we define @xmath294 } \log \left|2\sin \frac{\pi(a_i - a_j ) } n\right| + \log n\qquad \text{in dimension 1}\ ] ] respectively @xmath295 ^ 2 } e_n(a_i - a_j ) + \log \frac{n}{2\pi \eta(i)^2}\qquad \text{in dimension 2}.\ ] ] the results of the previous sections suggest trying to define @xmath3 as the limit as @xmath296 of these quantities .
more precisely , for a given random point process , @xmath31 becomes a random variable , and we try to see whether it has a limit as @xmath4 .
again , we do not claim a complete rigorous connection between such a quantity ( which only depends on the point configuration ) and the original @xmath8 , which we recall , was defined via a vector field @xmath61 .
we also emphasize that in and the number of points in @xmath56 $ ] resp .
@xmath56 ^ 2 $ ] is no longer necessarily equal to @xmath245 resp .
@xmath297 .
let us comment a little on the minimization of @xmath31 .
since our definition has been relaxed , the question of whether @xmath31 achieves a minimum becomes unclear .
however , in dimension 1 , observing that still @xmath298 , we have the following statement ( in dimension 2 no such result is available ) : assume @xmath299 are @xmath17 points in @xmath56 $ ] .
then @xmath300 thus , for any point configuration @xmath301 , where @xmath17 is the number of points in @xmath56 $ ] .
the proof is a simple adaptation of the proof of minimality of the perfect lattice in @xcite .
let @xmath302 $ ] , and assume @xmath303 .
let us also denote @xmath304 , with the convention @xmath305 .
we have @xmath306 .
similarly , let @xmath307 , with the convention @xmath308 .
we have @xmath309 . by periodicity of @xmath310
, we may view the points @xmath10 as living on the circle @xmath311 . when adding the terms in @xmath312 in the sum of , we can split it according to the difference of @xmath313 but modulo @xmath245 .
this way , there remains @xmath314 } \sum_{i=1}^k \log \left| 2\sin\frac { \pi u_{p , i}}{n } \right|+ \log n,\ ] ] where @xmath315 $ ] denotes the integer part .
but the function @xmath316 is stricly concave on @xmath317 $ ] .
it follows that @xmath318 inserting into we obtain @xmath319 } \log \left|2\sin\frac { p\pi}{k}\right|+ \log n.\ ] ] on the other hand , we know that @xmath298 which means that @xmath320}n \log \left|2\sin \frac{p \pi } n\right| + \log n= 0 $ ] , but also , since this is true for arbitrary integers @xmath245 , @xmath321 } \log \left|2\sin \frac{p \pi } k\right| + \log k=0 .\ ] ] inserting into we are led to @xmath322
we now turn to evaluating @xmath31 for random point processes . in view of its form , the expectation of the random variable @xmath31 can be computed from the sole knowledge of the second correlation function of the process . indeed ,
recall that for any @xmath323 , the @xmath17-point correlation function @xmath324 of a random point process in @xmath325 is characterized by the property that @xmath326 where the expectation is with respect to our measure on locally finite subsets @xmath327 , and @xmath55 ranges over a suitable space of test functions , see e.g. @xcite .
we note here that determinantal processes are a particular class of processes characterized by the fact that the correlation functions can be expressed as @xmath328}\ ] ] for some kernel @xmath329 , see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and references therein .
this will be used later . here
we need specifically the two - point correlation function .
in addition , for our function the formula simplifies when the process is assumed to be stationary ( i.e. translation invariant ) . from now on we make the basic assumption that we are dealing with a translation invariant multiplicity - free random point process in @xmath330 or @xmath62 , of density @xmath20 ( i.e @xmath331 ) with absolutely continuous correlation functions ( hence , holds ) .
if @xmath332 is the two - point correlation function of such a process , it is of the form @xmath333 since the process is stationary .
it is more convenient to work with the second cluster function " @xmath334 ( we will give a general definition of the cluster functions in section [ sec4 ] ) .
our basic assumptions thus imply @xmath335 by definition of @xmath27 , the expectation of @xmath31 is , in dimension 1 , @xmath336 ^ 2 } \log \left|2 \sin \frac{\pi ( x - y)}{n } \right|t_2(x - y)\ , dx\ , dy + \log n\ ] ] ( where we have used and ) and respectively in dimension 2 @xmath337 ^ 2 \times [ 0 , n]^2 } e_n(x - y ) t_2(x - y ) \ , dx \ , dy + \log \frac{n}{2\pi \eta(i)^2}\ ] ] ( where we have used and ) . the question is then whether these quantities have a limit as @xmath4 .
as we show below , this will only be true under additional assumptions which in particular ensure a sufficient decay of @xmath26 . finally , it would be most interesting to find natural conditions on the random points behavior ( their spacing etc ) that would guarantee the existence of a limit to @xmath338 .
[ th1 ] consider a random point process @xmath339 on the real line with the one - point correlation function @xmath340 and a two - point correlation function @xmath332 , satisfying . under the following assumptions 1 .
@xmath341 ; 2 .
there exist a sequence @xmath342 such that @xmath343 as @xmath4 ( for some @xmath344 ) and uniformly in @xmath345 $ ] we have @xmath346 the following holds : + - if @xmath347 and @xmath348 , then @xmath349 as @xmath296 ; should be bounded below or at least that the value @xmath350 is in fact not taken . ]
+ - if @xmath351 and @xmath352 for @xmath342 as above , then @xmath353 exists and is finite if and only if @xmath354 converges , and if so then @xmath355 [ 2.1 ] condition 2 ) is satisfied by the stronger one : @xmath356 to see this it suffices to observe that on @xmath357 $ ] we have @xmath358 . in view of we
need to compute @xmath359 ^ 2 } \log \left|2\sin \frac{\pi ( x - y)}{n}\right| t_2(x - y)\ , dx\ , dy + \log n.\ ] ] for @xmath360
^ 2 $ ] we denote @xmath361 and @xmath362 .
we then split @xmath56 ^ 2 $ ] into the disjoint union of the following domains , see figure [ fig1 ] , where @xmath363 is in condition 2 ) : @xmath364 ^ 2 , |v| \le \a_n , \a_n \le u \le 2n- \a_n \}\\ & \cdot & d_{1 } = \{(x , y)\in [ 0,n]^2 , v\ge n- \a_n \ } \\ & \cdot & d_{1'}=\ { ( x , y)\in [ 0,n]^2 , v\le -n+ \a_n \ } \\ & \cdot & d_{2 } = \{(x , y ) \in [ 0,n ] ^2 , u \ge 2n - \a_n\}\\ & \cdot & d_{2 ' } = \{(x , y ) \in [ 0,n]^2 , u \le \a_n\}\\ & \cdot & d_3= \{(x , y ) \in [ 0,n ] ^2 ,
\a_n \le v \le n-\a_n\}\\ & \cdot & d_{3 ' } = \{(x , y ) \in [ 0,n ] ^2 , - n + \a_n \le v \le -\a_n\}\end{aligned}\ ] ] [ fig1 ] we evaluate the integral in over each of these domains successively .
we start with the contribution of @xmath365 . making the change of variables
@xmath366 we have @xmath367 using assumption 1 ) and noting that in @xmath368 , @xmath369 , and thus @xmath370 , we deduce @xmath371 using @xmath372 , it follows that @xmath373 using @xmath374 the estimate for the domains @xmath375 is similar . for the contribution over @xmath376 ,
using the change of variables @xmath377 we have @xmath378 this is @xmath379 when assumption 2 ) holds .
the estimate on @xmath380 is completely analogous .
we have thus found that all contributions over @xmath381 are negligible .
the behavior of the integral will thus be determined by the contribution of @xmath382 .
changing again the variables @xmath383 into @xmath384 , we have @xmath385 but in @xmath382 we have @xmath386 as @xmath4 , hence @xmath387 therefore , @xmath388 using @xmath389 , we easily deduce that if @xmath390 and @xmath391 , then @xmath392as @xmath4 , and we conclude as desired .
if @xmath393 , then we may proceed and find @xmath394 and we conclude ( returning to and , and using the assumptions ) that holds . in this subsection , we use the result from the previous theorem to compute explicit asymptotic values for @xmath395 for some well - known point processes , namely the homogeneous poisson process , and the @xmath0-sine processes with @xmath1 .
( recall that @xmath31 was defined in such a way that for the lattice @xmath12 we have @xmath396 . )
the homogeneous poisson process satisfies @xmath397 and @xmath398 , hence @xmath399 .
we immediately deduce from that @xmath400 .
hence , the poisson process can be viewed as having the ` value of @xmath8 ' equal to @xmath401 .
the @xmath0-sine processes for @xmath1 arise in random matrices as the local limits for random matrix ensembles with orthogonal , hermitian , and symplectic symmetries , see @xcite and references therein .
these are stationary processes whose correlations can be computed as follows : introduce the kernels @xmath402 \d\frac 1\pi\int_0^{\pi(x - y ) } \frac{\sin t}{t}\ , dt - \d\hal sgn(x - y ) & \d\frac{\sin\pi(x - y)}{\pi(x - y)}\end{array}\right ) & \quad \text{for } \
\beta=1\\ \label{k4 } & k^{(4)}(x , y)= \d\left(\begin{array}{lr}\d\frac{\sin 2\pi(x - y)}{2\pi(x - y ) } & \d \frac{\p } { \p x } \frac{\sin 2\pi(x - y)}{2\pi(x - y)}\\[2 mm ] \d\frac{1}{2\pi}\int_0^{2\pi(x - y ) } \frac{\sin t}{t}\ , dt & \d \frac{\sin 2\pi(x - y)}{2\pi(x - y)}\end{array}\right ) & \quad \text{for } \
\beta=4\end{aligned}\ ] ] where all the indeterminacies @xmath403 at @xmath404 are resolved by continuity .
+ the @xmath2 sine process is a determinantal process with kernel @xmath405 , thus from , its @xmath24-point correlation function is given by @xmath406}.\ ] ] the correlation functions for the @xmath407 sine processes have the form @xmath408},\qquad n=1,2,\dots\ ] ] where @xmath409 denotes the quaternion determinant , see e.g. ( * ? ? ?
* section 6.1.1 ) for a definition .
alternatively , the right - hand side of can be expressed as the pfaffian of a closely related matrix , cf .
* proposition 6.1.5 ) .
random point processes with correlation functions of such form are often called _ pfaffian _ , see ( * ? ? ?
* section 10 ) and references therein .
the three processes above satisfy @xmath331 , and their second cluster functions can be easily seen to be given by @xmath410 [ pro23 ] the @xmath0-sine processes for @xmath1 satisfy the assumptions of theorem [ th1 ] , and @xmath411 where @xmath37 is the euler constant . before stating the proof , we recall some integrals of classical functions that we will need .
we state them without a proof and refer to @xcite . @xmath412
these formulas can be found in @xcite , they are respectively ( 2.5.3.12 ) , ( 2.6.32.3 ) , ( 2.5.6.9 ) , and ( 2.6.32.7 ) .
finally we need a few more integrals that are based on @xmath413 and @xmath414 these are @xmath415 cf . respectively ( 2.6.4.11 ) , ( 2.6.2.10 ) and ( 2.6.4.16 ) in @xcite .
we start with the simplest case .
+ _ case @xmath2 : _ first it is easy to check that assumptions as well as assumption 1 ) of theorem [ th1 ] are verified .
also , we have @xmath416 as @xmath417 , and @xmath418 , thus assumption 2 ) is implied by . according to theorem [ th1 ] we then have @xmath419 from , we obtain @xmath420 hence the result .
+ _ case @xmath53 : _ similarly to the case of @xmath2 , one checks that @xmath416 .
indeed , note that @xmath421 .
assumptions and 1 ) , 2 ) are then verified as in the @xmath422 case .
next we check that @xmath423 . by evenness and integration by parts
@xmath424 the convergence is also fast enough , since @xmath425 . according to theorem [ th1 ] we
thus have @xmath426 integrating by parts , we are led to @xmath427 in view of the first term on the right - hand side is equal to @xmath428 .
the second is equal to @xmath429 with and , @xmath430 on the other hand , near @xmath431 one has @xmath432 hence the above right - hand side limit is @xmath32 .
the second term in thus vanishes . from and
, the third one is equal to @xmath433 this concludes the case @xmath53 .
+ _ case @xmath434 : _ assumptions and 1 ) are easily verified , we proceed to 2 ) .
we may write @xmath435 the first part is @xmath436 just as in the case above , so it remains to check that @xmath437 uniformly in @xmath438.$ ] integrating by parts , we have @xmath439 in @xmath440 $ ] we may bound from below @xmath441 by @xmath442 , which is asymptotically equivalent to @xmath443 as @xmath4 .
hence the integral on the right hand side may by bounded by @xmath444 in view of assumption 4 ) .
the other terms are also easily found to be @xmath445 by a similar argument .
we next check that @xmath446 with fast enough convergence . by evenness and integration by parts
@xmath447 according to theorem [ th1 ] we thus have @xmath448 using evenness and integration by parts as above , we find @xmath449 by the first term on the right - hand side is equal to @xmath450 . by change of variables , the second term is equal to @xmath451 by . combining with we find that the second term is equal to @xmath452 and we conclude the proof . as expected @xmath453 decreases as @xmath1 increases . in the plane ,
the computations are easier because we can take advantage of the fast ( exponential ) decay of the correlation kernels .
[ th2 ] consider a random point process @xmath339 in the plane , with a one point correlation function @xmath340 and a two - point correlation function @xmath332 , satisfying .
we identify the plane with the complex plane @xmath454 . under the assumption @xmath455
the following holds : + - if @xmath456 , we have @xmath457 as @xmath296 ; + - if @xmath458 and @xmath459 ^ 2 } t_2(v)\ , dv= o((\log n)^{-1})$ ] , then @xmath353 exists and is finite if and only if @xmath460 converges , and if so , then @xmath461 returning to we have to compute @xmath462 ^ 2 \times [ 0,n]^2 } e(x - y ) t_2(x - y)\ , dx \ , dy + \log \frac{n}{2\pi \eta(i)^2}.\ ] ] making the change of variables @xmath463 , we have @xmath464 ^ 2 \times [ 0,n]^2}e_n(x - y ) t_2(x - y)\ , dx \ , dy = \frac{1}{4 } \int_{v\in [ -n , n]^2}\int _ { u \in s_n(v)}e(v ) t_2(v)\,du \ , dv,\ ] ] where @xmath465,\ , y\in [ 0,n]^2,\ , x - y = v\}$ ]
. we may compute that @xmath466 so @xmath467 ^ 2\times [ 0,n]^2 } e_n(x - y ) t_2(x - y)\ , dx \ , dy \\
= \frac{1}{4}\int_{[-n , n]^2 } \left(4n^2 - 2n|v_1|- 2n|v_2|+|v_1||v_2|\right ) e_n(v ) t_2(v)\ , dv.\end{gathered}\ ] ] next we return to where @xmath111 is given by and perform an asymptotic analysis as @xmath296 .
we have @xmath468 while , since @xmath469 we may write ( with @xmath228 ) @xmath470 hence @xmath471 inserting into and combining with @xmath472 we obtain @xmath473 inserting this into , we are led to @xmath474 ^ 2\times [ 0,n]^2 } e_n(x - y ) t_2(x - y)\ , dx \ , dy \\ = \int_{[-n ,
n]^2 } \left(n^2 - \hal n|v_1|- \hal n|v_2|+\frac{1}{4}|v_1||v_2|\right ) \
( -2\pi \log |v|-2\pi \log \frac{2\pi \eta(i)^2 } { n}+o\left ( \frac{|v|}{n}\right)\ ) t_2(v ) \ , dv.\end{gathered}\ ] ] therefore @xmath475 ^ 2 \times [ 0,n]^2 } e_n(x - y ) t_2(x - y ) \ , dx\ , dy + \log \frac{n}{2\pi \eta(i)^2 } \\= \log \frac{n}{2\pi \eta(i)^2 } \
( 1-\int_{[-n , n]^2 } \(1 + o\left(\frac{|v|}{n}\right ) + o\left(\frac{|v|^2}{n^2}\right)\ ) t_2(v)\,dv \)\\
+ \int_{[-n , n]^2 } \
( \log |v| + o\left(\frac{|v|}{n}\right ) \)\(1 + o\left(\frac{|v|}{n}\right ) + o\left(\frac{|v|^2}{n^2}\right ) \ ) t_2(v ) \ , dv.\end{gathered}\ ] ] using the assumption we have @xmath476 ^ 2 } \frac{|v|}{n}t_2(v)\ , dv = o(1 ) , \\ &
\int_{[-n , n]^2 } \frac{|v|^2}{n^2 } t_2(v)\ , dv = o(1),\\ & \int_{[-n , n]^2 } \frac{|v|^3}{n^3}t_2(v)\ , dv = o(1),\\ & \int_{[-n , n]^2 } \frac{|v|}{n } \log |v|t_2(v)\ , dv = o(1 ) , \\ & \int_{[-n , n]^2 } \frac{|v|^2}{n^2 } \log |v|t_2(v)\ , dv = o(1).\end{aligned}\ ] ] it follows that , as @xmath4 , @xmath475 ^ 2\times [ 0,n]^2 } e_n(x - y ) t_2(x - y ) \ , dx\ , dy + \log \frac{n}{2\pi \eta(i)^2 } \\= \log \frac{n}{2\pi \eta(i)^2 } \
( 1-\int_{[-n , n]^2 } t_2(v)\,dv \ ) + \int_{[-n , n]^2 }
\log |v| t_2(v ) \ , dv + o(1).\end{gathered}\ ] ]
the result then easily follows .
we turn to computing that limit for two specific processes .
the first one is the determinantal random point process with correlation kernel @xmath477 this process arises in random matrices as the local limit of the complex ginibre ensemble , see e.g. ( * ? ? ?
* proposition 15.2.3 ) , and thus it is sometimes called the ginibre point process . from the determinantal structure of the correlation functions , cf . , we have @xmath478 and @xmath479 .
this easily satisfies all the assumptions of theorem [ th2 ] ( in particular @xmath446 ) and we obtain the determinantal process with kernel satisfies @xmath480 this statement can be compared to a computation done in @xcite , see also ( * ? ? ?
* ex.15.3.1(iv ) ) .
according to theorem [ th2 ] it suffices to compute @xmath481 using the change of variables @xmath482 .
we have @xmath483 , and the result follows . the second one is the process of the zeros of a gaussian analytic function ( often denoted gaf ) .
it consists of the random zeros of the analytic function @xmath484 when the @xmath485 are i.i.d gaussians suitably normalized , and it is a stationary process in the plane .
the general background can be found e.g. in @xcite . the second cluster function for the process , when the density @xmath486 is taken to be @xmath20 , is given according to @xcite by @xmath487 where @xmath488 it is easy to check that the assumptions of theorem [ th2 ] are satisfied , and we deduce the zeros of gaussian analytic functions " process satisfies @xmath489 to check that we may apply theorem [ th2 ] we first compute @xmath490_0^\infty \\ & = & - \left [ 2x(\coth x-1 ) + x^2 ( 1-\coth^2 x)\right]_0^\infty \\ & = & 1 \end{aligned}\ ] ] where we have used the change of variables @xmath491 and the asymptotic relation @xmath492 as @xmath236 .
it is also easy to check that the convergence of @xmath493 is exponential , hence fast enough , and we may apply theorem [ th2 ] .
this yields @xmath494 let us now compute @xmath495_{\ep}^\infty - \int_{\ep}^\infty \frac{1}{u } \frac{d}{du } ( u^2 ( \coth u -1))\ , du \\ = - \log \frac{2\ep}{\pi } \
( 2\ep ( \coth \ep
-1 ) + \ep^2 ( 1- \coth^2 \ep)\ ) - \left [ x ( \coth x -1 ) \right]_{\ep}^\infty - \int_{\ep}^\infty ( \coth u - 1 ) \ , du\\ = - \log \frac{2\ep}{\pi } ( 1+o(\ep ) ) + 1+o(\ep )
- \left[\log \sinh x - x \right]_{\ep}^\infty.\end{gathered}\ ] ] taking the limit @xmath496 , we conclude @xmath497 it is well known ( cf . @xcite and references therein )
that the gaf process " is more rigid " than the ginibre point process " ( cf . also recent work @xcite ) .
we just demonstrated here it has more order via the renormalized energy .
as explained in the introduction and section [ sec : background ] , the question of minimizing @xmath8 is an important one , and open in dimension 2 .
it thus seems interesting to try to minimize @xmath28 as expressed in and , over a subclass of processes . in this section
we show that we can characterize the minimizer of this expression over the class of determinantal random point processes whose correlation kernel @xmath329 is hermitian and translation invariant , i.e. @xmath498 for some function @xmath17 . for those processes ,
we have @xmath499 . note , however , that the important determinantal process with kernel is not in this class : while all its correlation functions are translation invariant , the correlation kernel is not .
we prove the following statement .
the proof relies on a rearrangement inequality .
[ th3 ] let @xmath500 be the class of determinantal processes on the real line , respectively the plane , with self - adjoint translation - invariant kernels @xmath501 , and @xmath502 , @xmath143 or @xmath144 , such that 1 . @xmath503 and @xmath504 ; 2 .
@xmath505 .
let @xmath506 .
then for any process from @xmath507 with correlation kernel @xmath501 , we have @xmath508 where @xmath509 is the ball centered at @xmath32 of volume one , and @xmath510 is the fourier transform .
thus , on the real line @xmath511 is minimized over @xmath507 by the @xmath2 sine process , while on the plane the minimizing determinantal process has the kernel given by @xmath512 , where @xmath513 is a bessel function of the first kind .
condition 1 ) says that our processes have density 1 , which we have assumed throughout , and that @xmath514 is finite , cf .
theorems [ th1 ] and [ th2 ] .
the functional @xmath515 coincides with @xmath516 , given that the decay assumptions are satisfied .
numerical integration shows that in dimension 2 , @xmath517 , which is greater than @xmath514 for the ginibre ensemble @xmath518 and for the zeroes of the gaussian analytic function @xmath519 .
thus , the two latter processes are `` more rigid '' .
let us denote by @xmath111 the inverse fourier transform of @xmath17 , and by @xmath520 the integral operator corresponding to @xmath521 via @xmath522 we have @xmath523 by macchi - soshnikov s theorem , see @xcite , any self - adjoint translation - invariant correlation kernel of a determinantal process gives rise to an integral operator with spectrum between @xmath32 and @xmath20 .
hence , the spectrum of @xmath520 is in @xmath524 $ ] .
since @xmath520 is also the convolution by @xmath17 , this implies that @xmath525 takes values in @xmath524 $ ] .
moreover @xmath526 with @xmath527 , and @xmath528 from assumption 1 )
. a function @xmath529 with values in @xmath524 $ ] which satisfies @xmath530 can only be a characteristic function of a set @xmath531 , denoted @xmath532 , with @xmath531 measurable of measure 1 .
writing @xmath533 for the corresponding @xmath17 , we may write @xmath534 there remains to optimize over @xmath531 , measurable set of measure 1 , @xmath535 let @xmath531 be measurable of measure 1 , such that @xmath536 .
noting that @xmath537 is an @xmath538 ( and also continuous ) function , for every @xmath539 the integrals @xmath540 then also converge , by comparison . given any @xmath541
, @xmath542 converges to @xmath543 monotonically as @xmath544 ( it suffices to check that this function is increasing in @xmath92 ) .
it then follows that in each of the domains @xmath545 and @xmath546 , we have @xmath547 as @xmath548 , monotonically . splitting the integrals as sums over these two regions , it follows by monotone convergence theorem that @xmath549 we then remark ( in view of the formula @xmath550 ) that @xmath551 is the fourier transform of @xmath552 we next claim that @xmath553 is a continuous function .
first , consider the case where @xmath531 is an open set .
then , as @xmath554 we have @xmath555 almost everywhere , by openness of @xmath531 , while @xmath556 and @xmath557 .
the claim is thus true by dominated convergence theorem .
second , if @xmath531 is a general measurable set , by outer regularity of the measure we may approximate it by an open set @xmath558 such that @xmath559 and @xmath560 .
then it is immediate that for any @xmath82 , @xmath561 .
since @xmath562 is continuous and @xmath563 is arbitrary , it follows that @xmath553 has to be continuous too as a uniform limit of continuous functions .
the claim is proved .
we also note that @xmath564 .
the next ingredient is that in dimension 1 @xmath565 while in dimension 2 @xmath566 for a reference see e.g. ( * ? ? ?
* chapter 5 ) , or @xcite .
let us continue with the one - dimensional case .
we deduce from the above facts that @xmath567 the relation can be justified by convoluting @xmath568 with a gaussian kernel approximating @xmath569 at scale @xmath570 , using the fact that @xmath571 in the schwartz class , the continuity of @xmath553 and then letting @xmath496 on both sides .
moreover , this argument shows that @xmath572 is convergent . using @xmath573 and fubini s theorem , we may rewrite as @xmath574 notice that @xmath575 as @xmath576 so for @xmath577 small enough , @xmath578 is increasing in @xmath579 .
now riesz s rearrangement inequality ( see ( * ? ? ?
* theorem 3.7 ) ) asserts that a quantity of the form of the right - hand side of is always decreased by changing @xmath580 into its symmetric rearrangement @xmath581 .
this means that for all @xmath582 small enough , @xmath583 . but
@xmath584 hence the same is true also for @xmath131 i.e. @xmath585 the symmetric rearrangement @xmath586 of @xmath531 is the ball centered at @xmath32 and of volume @xmath587 .
we have thus found that @xmath588 is minimal when @xmath531 is the ball centered at @xmath32 and of volume 1 . in dimension one , the fourier transform of @xmath589}$ ] is @xmath590 , which corresponds to the determinantal process with @xmath591 , that is the sine process ( for @xmath2 ) . in dimension 2 , the argument is exactly parallel , starting again from .
in section [ sec3 ] we dealt with the expectation of @xmath8 . in this section
we turn to examining its variance in the sense of computing @xmath592 for the same specific random point processes . in what follows
we will need the formalism of ( higher ) cluster functions to efficiently deal with the @xmath17-point correlation functions for @xmath593 ; we refer to @xcite for details and further references on this formalism . for any nonempty subset @xmath594 of @xmath595 we write @xmath596 , where @xmath597 is the @xmath17-point correlation function , and define the @xmath159-point cluster function as @xmath598 with the sum running over all partitions of @xmath599 into nonempty subsets @xmath600 . from the @xmath601
, the @xmath602 can be recovered through the reciprocal formula @xmath603 if a random point process is determinantal ( cf . ) with correlation kernel @xmath521 , then ( see e.g. @xcite ) for any @xmath323 @xmath604 where @xmath605 denotes the symmetric group on @xmath17 symbols .
if the correlation functions of a point process are given by quaternion determinants ( cf . ) then ( see e.g. @xcite ) for any @xmath323 @xmath606 [ prform ] we have @xmath607 where @xmath608 ^ 2 } g_n(x - y)^2 \ , dx\ , dy\\ \label{t2 } & i_2= - 4 \int_{[0,n]^3 } g_n(x - y)g_n(x - z ) t_2(y , z)\ , dx\ , dy\ , dz\\ \label{t3 } & i_3 = 2 \int_{[0,n]^4 } g_n(x - y)g_n(z - t ) t_2(x , z)t_2(y , t ) \ , dx\ , dy\ , dz\ , dt \\
\label{t4 } & i_4= - 2\int_{[0,n]^2 } g_n(x - y)^2 t_2(x , y)\ , dx\ , dy\\ \label{t5 } & i_5= 4 \int_{[0,n]^3 } g_n(x - y)g_n(x - z)t_3(x , y , z)\ , dx\ , dy\ , dz\\ \label{t6 } & i_6= - \int_{[0,n]^4 } g_n(x - y)g_n(z - t)t_4(x , y , z , t)\ , dx\ , dy\ , dz\ , dt,\end{aligned}\ ] ] where @xmath609 in dimension 1 , resp .
@xmath610 defined in in dimension 2 .
expanding the square , we have @xmath611 } g_n(a_i- a_j ) \)^2 = \sum_{i , j , k , l \ \mathrm{p.d . } } g_n(a_i- a_j)g_n(a_k- a_l ) \\
\label{li2 } & + \sum_{i , j , l\ \mathrm{p.d . } } g_n(a_i- a_j)g_n(a_i- a_l ) + \sum_{i , j , k \ \mathrm{p.d . } } g_n(a_i- a_j ) g_n(a_k- a_i)\\ \label{li3 } & + \sum_{i , j , l \ \mathrm{p.d . } } g_n(a_i- a_j ) g_n(a_j - a_l)+ \sum_{i , j , k \ \mathrm{p.d . } } g_n(a_i- a_j ) g_n(a_k - a_j)\\ \label{li4 } & + \sum_{i\neq j } g_n(a_i - a_j)^2 + \sum_{i\neq j } g_n(a_i-
a_j ) g_n(a_j - a_i),\end{aligned}\ ] ] where the sums are still taken over points in @xmath56 $ ] , and p.d .
stands for `` pairwise distinct '' .
since @xmath612 is even , it is clear that all the sums in and are equal , and the sums in as well . using @xmath17-point correlation functions ( cf . ) , we thus may write @xmath613 } g_n(a_i- a_j)\)^2\\ = \int_{[0,n]^4 } g_n(x - y)g_n(z - t ) \ro_4(x , y , z , t ) \ , dx\ , dy\ , dz\ , dt \\
+ 4\int_{[0,n]^3 } g_n(x - y)\ro_3(x , y , z)\ , dx\ , dy\ , dz+ 2\int_{[0,n]^2 } g_n(x - y)^2 \ro_2(x - y)\ , dx\ , dy .
\end{gathered}\ ] ] it is now convenient to express this in terms of the cluster functions @xmath614 , using , which yields @xmath615 substituting these relations into and using that @xmath616 , we obtain ( writing the terms in the same order ) @xmath617 } g_n(a_i- a_j)\)^2 = \ ( \int_{[0,n]^2 } g_n(x - y)t_2(x , y ) \ , dx\ , dy \)^2 \\ \nonumber & + 2 \int_{[0,n]^4 ] } g_n(x - y)g_n(z - t ) t_2(x , z)t_2(y , t ) \ , dx\ , dy\ , dz\ , dt \\
\nonumber & - \int_{[0,n]^4 } g_n(x - y)g_n(z - t)t_4(x , y , z , t)\ , dx\ , dy\ , dz\ , dt\\ \nonumber & + 4 \int_{[0,n]^3 } g_n(x - y)g_n(x - z)t_3(x , y , z)\ , dx\ , dy\ , dz\\ \nonumber & -4 \int_{[0,n]^4 } g_n(x - y)g_n(x - z)t_2(y , z)\ , dx\ , dy\ , dz\\ \nonumber & + 2 \int_{[0,n]^2 } g_n(x - y)^2 \ ,
dx\ , dy- 2\int_{[0,n]^2 } g_n(x - y)^2 t_2(x , y)\ , dx\ , dy.\end{aligned}\ ] ] similarly ( and as we have seen in the proof of theorem [ th1 ] ) we have @xmath618 ^ 2 } g_n(x - y)t_2(x , y)\ , dx\ , dy\ ] ] and the result follows .
[ th4 ] for the sine-@xmath0 processes with @xmath1 as described above , we have @xmath619 since we already know from proposition [ pro23 ] that @xmath35 exists , and in view of , it suffices to show that @xmath620 } g_n(a_i- a_j)\)^2 - \(\e\sum_{i\neq j , a_i ,
a_j\in [ 0,n ] } g_n(a_i- a_j)\)^2\ ) = 0.\ ] ] we apply lemma [ prform ] and now deal with all the terms @xmath621 to @xmath622 in . first , we have @xmath623 ^ 2 } g_n(x - y)^2 \ , dx\ , dy= 2n^2 \int_{[0,1]^2 } \
( \log \left|2\sin \pi ( x - y)\right|\)^2\ , dx\ , dy.\ ] ] for @xmath624 , using the explicit expression for @xmath612 , making the change of variables @xmath625 , and recalling that @xmath26 is translation - invariant since the process is , as for theorem [ th1 ] ] we find @xmath626 ^ 4 } g_n(x - y)g_n(x - z ) t_2(y - z)\ , dx\ , dy\ , dz\\ = - 4 n^2 \int_{[0,1]^2}\int_{[n(y-1 ) , ny ] } \log |2\sin \pi ( x - y)|\log \left|2\sin \pi \left(x - y+\frac{t}{n}\right)\right| t_2(t)\ , dx\ , dy\ , dt.\end{gathered}\ ] ] one may like to think that @xmath627 since @xmath446 .
however , @xmath628 is not regular enough to apply this reasoning and @xmath493 may converge only conditionally .
first notice that the cluster functions given in -- satisfy @xmath629 @xmath630 pick two exponents @xmath631 with @xmath632 .
let us examine the @xmath633 integral in the right - hand side of .
assume first that @xmath634\subset [ n(y-1 ) , ny]$ ] and that @xmath635 .
note that for this to be satisfied it suffices that @xmath636 by the mean value formula , we may write for some @xmath637 @xmath638 since we assumed @xmath635 .
thus @xmath639 } \log |2\sin \pi ( x - y)|\log \left|2\sin \pi \left(x - y+\frac{t}{n}\right)\right| t_2(t)\ , dt\\ = \int_{[-n^b , n^b ] } \log^2|2\sin \pi ( x - y)| t_2(t)\ , dt + o(n^{a-1 } \log
n ) \\ = \log |2\sin \pi(x - y)| \bigl(1- o(n^{-b})\bigr ) + o(n^{a-1 } \log n)= \log |2\sin \pi(x - y)| + o(n^{-b } \log n).\end{gathered}\ ] ] where we have used , then and @xmath635 .
we then claim that if @xmath635 we have @xmath640 assuming this , and combining with we obtain that @xmath641 ^ 2\cap s_{n , a } } \int_{[n(y-1 ) , ny ] } \log |2\sin \pi ( x - y)|\log \left|2\sin \pi \left(x - y+\frac{t}{n}\right)\right| t_2(t)\ , dx\ , dy\ , dt \\= \int_{(x , y)\in [ 0,1]^2\cap s_{n , a}}\log^2 |2\sin \pi ( x - y)|\ , dx\ , dy+o(1).\end{gathered}\ ] ] but it is easy to check , since the integrals converge and holds , that the contributions of the set where does not hold are @xmath445 as @xmath4 .
we may thus conclude that @xmath642 ^ 2 } \log^2 |2\sin \pi ( x - y)|\ , dx\ , dy + o(n^2).\ ] ] to finish with this @xmath624 term , it remains to prove . for @xmath643
this is immediately true since @xmath416 . for @xmath644
, we notice that the same argument that was used above to restrict to @xmath635 can be used to restrict to @xmath645 ( note that the initial integral is symmetric in @xmath646 and @xmath647 ) . inserting the formula for @xmath26 , and neglecting the @xmath648 part of @xmath26
, we thus have to prove that @xmath649 we integrate by parts and find that the boundary terms are negligible , and there remains to show that @xmath650 if the derivative falls on the second factor , we are back to the @xmath648 situation which gives a negligible term , and for the other term we use @xmath651 by explicit computation , which gives that the integral is @xmath652 .
this completes the treatment of @xmath624 .
we turn to @xmath653 . using a similar change of variables , we may write this term @xmath654 ^ 2 } \int_{[n(x-1),nx]}\int_{[n(y-1 ) , ny ] } \log \left|2\sin\pi ( x - y)\right|\log \left|2\sin \pi \left(x - y+\frac{v - u}{n}\right)\right| \\ \times
t_2(u ) t_2(v)\ , dx\ , dy\ , du\ , dv.\end{gathered}\ ] ] very similar manipulations to those above show that @xmath655 can be replaced by @xmath656 with a @xmath657 correction .
this leads us to @xmath658 ^ 2 } \log^2 |2\sin \pi(x - y)|\ , dx\ , dy+o(n^2).\ ] ] for @xmath659 , we have @xmath660 ^ 2 } \log^2 \left|2\sin \frac{\pi(x - y)}{n}\right|t_2(x - y)\ , dx\ , dy = o(n^2).\ ] ] indeed , we may take away a @xmath661-neighborhood of the diagonal , outside of which @xmath662 is bounded by @xmath663 and @xmath664 is controlled by @xmath665 , using .
thus the whole integral is controlled by @xmath666 .
adding the above results we find that @xmath667 it remains to show that @xmath668 and @xmath622 also give @xmath657 contributions .
the expressions @xmath668 and @xmath622 are estimated using explicit formulas for cluster functions of the sine-@xmath0 processes .
first returning to we see that for @xmath669 the entries of @xmath670 and @xmath671 are @xmath672 . combining with
, it follows that @xmath673 in both these cases . for
, we have @xmath674 we thus obtain @xmath675 in we may first ( as above ) remove a small neighborhood of the diagonals , off of which @xmath676 and @xmath677 are bounded by @xmath678 .
it then remains to estimate @xmath464 ^ 3 } |t_3(x , y , z)|\ , dx\ , dy\ , dz.\ ] ] replacing @xmath679 by its above estimates , and changing variables to @xmath680 and successively two out of @xmath681 , @xmath682 , and @xmath683 , we find @xmath684 ^ 3 } |t_3(x , y , z)|\ , dx\ , dy\ , dz\le o(n \log^3 n)$ ] , and @xmath685 .
we finally turn to @xmath622 .
the formula for @xmath686 is given by . comparing to , we see that for @xmath643 , we have @xmath687 the same reasoning as for @xmath668 gives @xmath688 .
for @xmath434 , in the formula for @xmath689 obtained with , in view of , there are terms which a priori have insufficient decay : they are terms of the form @xmath690 where @xmath691 is a permutation of the variables @xmath692 .
this leads to two different types of integrals @xmath693 ^ 4 } \log \left|2\sin \frac{\pi(x - y)}{n}\right| \log \left|2\sin \frac{\pi(z - t)}{n}\right| \\ \times \int_0^{2\pi(x - y)}\frac{\sin s}{s}\ , ds\int_0^{2\pi(z - t ) } \frac{\sin s}{s}\,ds \cdot \frac{\p } { \p x } \frac{\sin 2\pi(x - z)}{2\pi(x - z)}\frac{\p } { \p y } \frac{\sin 2\pi(y - t)}{2\pi(y - t)}\ , dx\,dy\,dz\,dt\end{gathered}\ ] ] and @xmath694 ^ 4 } \log \left|2\sin \frac{\pi(x - y)}{n}\right| \log \left|2\sin \frac{\pi(z - t)}{n}\right| \\
\times \int_0^{2\pi(x - z)}\frac{\sin s}{s}\ , ds\int_0^{2\pi(y - t ) } \frac{\sin s}{s}\,ds \cdot \frac{\p } { \p x } \frac{\sin 2\pi(x - y)}{2\pi(x - y)}\frac{\p } { \p z } \frac{\sin 2\pi(z - t)}{2\pi(z - t)}\ , dx\,dy\,dz\,dt.\end{gathered}\ ] ] for , we may again restrict the domain to @xmath695 with @xmath696 . then , integrating by parts in @xmath82 gives boundary terms which are negligible , and a new integrand with extra decay , involving @xmath697 or @xmath698 .
this leads again to @xmath657 contributions . for
, we may first restrict the integral to @xmath699 and @xmath700 , using arguments as above .
then , we may replace @xmath701 and @xmath702 by @xmath703 and @xmath704 respectively , making only a @xmath657 error . then note that the integrand in @xmath681 is a locally odd function , so we can remove domains @xmath705 , @xmath706 from the integration domain .
finally , integration by parts in @xmath82 gives additional decay , yielding @xmath657 contribution .
we conclude that @xmath688 , and the result follows . for all point processes with finite @xmath707 and @xmath708 , @xmath709 in @xmath710 and thus in probability .
the proof is immediate .
processes satisfying these assumptions and having different values for @xmath34 are thus mutually singular , such as @xmath0-sine processes with different @xmath711 . for the determinantal random point process with kernel we have @xmath712
the starting point is again lemma [ prform ] .
we note that in view of , and , all the cluster functions for that process are exponentially decreasing when viewed as functions of pairwise distances between arguments .
we start with the term . replacing @xmath612 by @xmath713 , using the translation invariance , and changing variables as before ( @xmath714 ) , we find @xmath715 ^ 2}\int_{ny-[0,n]^2}e_n(n(x - y))\ ( e_n(n(x - y))- e_n(n(x - y+u / n))\ ) t_2(u)dx\ , dy\ , du.\end{gathered}\ ] ] noting that in view of the definition of @xmath716 ( cf .
) we have @xmath717 and from , @xmath718 behaves like @xmath719 near @xmath431 ( and similarly near points of the lattice @xmath720 ) . given @xmath721 there
thus exists @xmath722 such that @xmath723 ^ 2 \cap \{|x - y- \mz^2|<\delta\ } } \int_{ny-[0,n]^2}e_1(x - y)\ ( e_1(x - y))- e_1(x - y+u / n)\ ) t_2(u)dx\ , dy\ , du<\eta.\ ] ] on the other hand , still in view of the definition , @xmath718 is uniformly continuous away from @xmath720 , hence we may write , as @xmath4 , @xmath723 ^ 2 \backslash \{|x - y- \mz^2|<\delta\ } } \int_{ny-[0,n]^2}e_1(x - y)\ ( e_1(x - y))- e_1(x - y+u / n)\ ) t_2(u)dx\ , dy\ , du = o(1).\ ] ] since this is true for any @xmath7 , it follows that @xmath724 similarly , @xmath725 ^ 2 } \int_{nx-[0,n]^2}\int_{ny-[0,n]^2 } e_1(x - y)\(e_1\left(x - y+\frac{u - v}{n}\right ) - e_1(x - y)\ ) t_1(u)t_2(v)\ , dx\ , dy\ , du\ , dv.\end{gathered}\ ] ] the same reasoning shows that this is @xmath657 . for @xmath659 , the change of variables @xmath726 and @xmath727 yields @xmath728 ^ 2 } e_1(x - y)^2 t_2(n(x - y))\ , dx\ , dy.\ ] ] we may take out a @xmath569 neighborhood of the diagonal and its translates by @xmath720 , off of which @xmath729 can be bounded by @xmath730 and @xmath731 by @xmath732 .
the whole term is thus @xmath657 .
we turn to . from
and we find that @xmath733 as above we have @xmath734 ^ 3}e_1(x - y)e_1(x - z)t_2(nx , ny , nz)\ , dx\ , dy\ , dz\ ] ] and as above we may take out @xmath569-neighborhoods @xmath735 or @xmath736 or @xmath737 ( and their translates by @xmath720 ) , outside of which the @xmath718 terms are bounded by @xmath738 s and @xmath679 by @xmath732 .
the whole term is thus @xmath657 .
a very similar reasoning applies to @xmath622 . combining all these
, we find the result . in view of the proof above
, the same result holds for any process such that the cluster functions decay sufficiently fast away from diagonals , for example exponentially .
in this section we gather various additional computations of expectations and additional facts . assume that @xmath742 for @xmath743 .
let @xmath339 denote the superposition of independent processes @xmath744 , where @xmath744 denotes the image of the process @xmath745 under the dilation by factor @xmath740 of the line .
then , with obvious notation , @xmath746 let @xmath747 be the second cluster functions corresponding to @xmath745 .
let @xmath748 now denote the second correlation function for the process @xmath749 , which has density @xmath750 .
we have @xmath751 .
the process @xmath339 clearly has density @xmath20 , and its second correlation function is @xmath752 } \ro_1^{(i ) } \ro_1^{(j ) } + \sum_{i=1}^m \bar{\ro}_2 ( x , y ) = m(m-1 ) \frac{1}{m^2}+ \frac{1}{m^2 } \sum_{i=1}^m\ro_2^{(i)}\left(\frac{x}{m } , \frac{y}{m}\right).\ ] ] we also denote by @xmath26 the corresponding second cluster function .
we thus have @xmath753 and it easily follows , with a change of variables , that @xmath754 in addition , @xmath755 the result follows easily using theorem [ th1 ] . for example
, superposing two independent processes with the same second correlation function leads to an increase of @xmath453 by @xmath758 in dimension 1 and @xmath759 in dimension 2 .
superposition can thus be seen as increasing the disorder " .
one can define a ` random decimation ' of a process by erasing points at random with probability 1/2 .
the second correlation function then transforms into @xmath760 but the space needs to be rescaled by a factor 2 in order to maintain a density one , so the correlation function after that is @xmath761 it is clear in view of theorems [ th1 ] and [ th2 ] that if a process has finite @xmath34 , its decimation will not , since the condition @xmath446 will be destroyed by this operation . on the other hand , one can define a ` deterministic decimation ' by erasing every even ( or odd ) point of an ( ordered ) random point configuration followed by rescaling of the space to keep the density at @xmath20 .
while we can not say anything about this operation in general , one can observe what it does in a couple of cases .
it is known , see e.g. @xcite , that the @xmath2 sine process is the deterministic decimation of superposition of two @xmath53 sine processes , or symbolically @xmath762 from proposition [ pro23 ] we know that @xmath35 is @xmath763 for the left - hand side , and proposition [ pro26 ] says that @xmath35 is @xmath764 for @xmath765 . thus , the deterministic decimation decreased the value of @xmath35 by 1 .
the @xmath2 discrete sine process was first obtained in @xcite as the bulk scaling limit of the plancherel measure for symmetric groups , and it was shown in @xcite to be the a universal local scaling limit for a broad family of discrete probabilistic models of random matrix type with @xmath2 .
the goal of this section is to compute @xmath769 for the suitably scaled discrete sine process embedded into the real line . by the construction
, this provides an interpolation between the case of the perfect lattice , for which @xmath770 , and the case of the continuous @xmath2 sine process treated in the previous sections .
embed @xmath774 into @xmath80 via @xmath775 ; this turns the discrete sine process of density @xmath776 into a random point process on @xmath80 with density @xmath20 . for the latter process
we have @xmath777 for the calculation , we assume that @xmath778 is an integer ( the same argument however should hold without this assumption by examining more carefully error terms ) .
the calculation is then can be viewed as a discrete version of that of theorem [ th1 ] .
first , by definition of @xmath31 , we have @xmath779 } \ro_2 ( i , j ) \log \left|2\sin \frac{\pi \ro ( i - j)}{n } \right|+\log n\ ] ] and since @xmath780 we find @xmath781 } \ ( \frac{\sin \pi \ro ( i - j)}{\pi(i - j ) } \)^2 \log \left|2\sin \frac{\pi \ro ( i - j)}{n}\right| + \log n \\ - \frac{\ro^2}{n } \sum_{i \neq j \in [ 1 , n/\ro ] } \log \left|2\sin \frac{\pi \ro ( i - j)}{n}\right| .\end{gathered}\ ] ] let us first examine the contribution of the last sum . from the knowledge of @xmath298 , we know that for @xmath782 we have @xmath783 } \log \left|2\sin \frac { \pi ( i - j)}{k}\right|+\log k=0.\ ] ] applying this to @xmath784 we find that @xmath785 } \log \left|2\sin \frac{\pi ( i - j)}{n/\ro}\right| = - \ro \log n/\ro + o(1).\ ] ] we next turn to the first two terms in .
as in theorem [ th1 ] , we expect only the near diagonal terms to contribute , so that @xmath786 inserting this and into we find @xmath787 } \left ( \frac{\sin \pi \ro ( i - j)}{\pi(i - j ) } \right)^2 ( \log \left|2\pi \ro ( i - j)\right| -\log n ) - \ro \log
n/\ro + \log n + o(1).\ ] ] we first focus on @xmath788 } \
( \frac{\sin \pi \ro ( i - j)}{\pi(i - j ) } \)^2 & = & - \frac{\log n}{n } \sum_{u=1}^{n/\ro -1 } ( n/\ro - u ) \frac{2\sin^2
( \pi \ro u)}{\pi^2 u^2}\\ \nonumber & = & - \frac{\log n}{\ro } \sum_{u=1}^{n/\ro -1 } \frac{1-\cos 2\pi \ro u}{\pi^2 u^2}+o(1).\end{aligned}\ ] ] indeed , we can bound @xmath789 by @xmath790 and this multiplied by @xmath791 is negligible as @xmath792 . the last sum
then appears as a fourier series and can be computed explicitly , which leads to @xmath793 we next turn to @xmath794 } \ ( \frac{\sin \pi \ro ( i - j)}{\pi(i - j ) } \)^2 \log |2\pi \ro ( i - j)|= \frac{2}{n } \sum_{u=1}^{n/\ro-1 } ( n/\ro -u ) \(\frac { \
( \sin \pi \ro u\)^2}{\pi^2 u^2}\)^2\log ( 2\pi \ro u).\ ] ] again , the term containing @xmath795 can be neglected since it is bounded by @xmath796 . combining and letting @xmath4
, we finally arrive at .
j. baik , t. kriecherbauer , k. t .-
mclaughlin , p. d. miller , _ discrete orthogonal polynomials .
asymptotics and applications . _
annals of mathematics studies * 164 * , princeton university press , princeton , nj , 2007 .
a. borodin , _ determinantal point processes _ , arxiv:0911.1153 .
a. borodin , a. okounkov , g. olshanski , _ asymptotics of plancherel measures for symmetric groups _ , j. amer .
* 13 * ( 2000 ) , no . 3 , 481 - 515 .
d. j. daley , d. verey - jones , _ an introduction to the theory of point processes_. springer , 1988 .
f. dyson , _ statistical theory of the energy levels of a complex system _ , part i , j. math .
phys . * 3 * , 140156 ( 1962 ) ; part ii , ibid .
157165 ; part iii , ibid .
166175 f. j. dyson , m. l. mehta , _ statistical theory of the energy levels of complex systems _ , part iv , j. mathematical phys . * 4 * ( 1963 ) , 701712 .
w. g. faris , _ combinatorics and cluster expansions _ , probability surveys * 7 * , ( 2010 ) , 157206 .
j. forrester , _ log - gases and random matrices_. london mathematical society monographs series * 34 * , princeton university press , 2010 .
p. j. forrester , g. honner , _ exact statistical properties of the zeros of complex random polynomials _ , j. phys .
a * 32 * ( 1999 ) , no .
16 , 2962981 .
s. ghosh , f. nazarov , y. peres , m. sodin , in preparation .
j. b. hough , m. krishnapur , y. peres and b. virg , _ zeros of gaussian analytic functions and determinantal point processes_. university lecture series * 51 * , american mathematical society , providence , ri , 2009 .
b. jancovici , _ exact results for the two - dimensional one - component plasma _ , phys .
lett . * 46 * ( 1981 ) , 263280 .
k. johansson , _ random matrices and determinantal processes _ , mathematical statistical physics , 155 , elsevier b. v. , amsterdam , 2006 .
s. lang , _
elliptic functions_. springer - verlag , new york , 1987 .
l. lewin , _ polylogarithms and associated functions _
north - holland , 1981 .
e. lieb , m. loss , _
analysis_. graduate studies in mathematics * 14 * , american mathematical society , 1997 .
w. knig , _ orthogonal polynomial ensembles in probability theory _ , probab .
surveys 2 ( 2005 ) 385447 , arxiv : math/0403090 .
r. lyons , _ determinantal probability measures _ , publ .
hautes etudes sci .
* 98 * ( 2003 ) 167212 , arxiv : math/0204325
. m. l. mehta , _
random matrices_. third edition , elsevier / academic press , 2004 .
f. nazarov , m. sodin , _ random complex zeroes and random nodal lines , _ proceedings of the international congress of mathematicians .
volume iii , ( 2010 ) , 1450?1484 .
a. p. prudnikov , y. a. brychkov , o. i. marichev , _ integrals and series , vol . 1 : elementary functions_. gordon and breach sci .
publ . , 1986 .
a. p. prudnikov , y. a. brychkov , o. i. marichev , _ integrals and series , vol . 2 : special functions_. gordon and breach sci .
publ . , 1986 .
e. saff , v. totik , _ logarithmic potentials with external fields_. springer - verlag , 1997 .
e. sandier , s. serfaty , _ from the ginzburg - landau model to vortex lattice problems _ ,
e. sandier , s. serfaty , _
2d coulomb gases and the renormalized energy _ , arxiv:1201.3503 .
e. sandier , s. serfaty , _
1d log - gases and the renormalized energy _ , in preparation .
e. sandier , s. serfaty , _ vortices in the magnetic ginzburg - landau model _ ,
birkhuser , 2007 .
a. soshnikov , _ determinantal random point fields _ , russian math .
surveys * 55 * ( 2000 ) 923975 , arxiv : math/0002099 .
a. soshnikov , _ determinantal random fields _ , in : encyclopedia of mathematical physics , pp . 4753 , oxford : elsevier , 2006 . s. torquato , a. scardicchio , c. e. zachary , _ point processes in arbitrary dimension from fermionic gases , random matrix theory , and number theory , _
theory exp .
* 11 * ( 2008 ) , p11019 . c. tracy , h. widom , _ correlation functions , cluster functions , and spacing distributions for random matrices _ , j. stat
* 92 * ( 1998 ) , no 5 - 6 , 809835 .
b. valk , b. virg , _ continuum limits of random matrices and the brownian carousel _ , invent
* 177 * ( 2009 ) , no .
3 , 463508 .
alexei borodin + massachusetts institute of technology + department of mathematics + 77 massachusetts avenue , cambridge , ma 02139 - 4307 , usa .
+ [email protected] + + sylvia serfaty + upmc univ paris 06 , umr 7598 laboratoire jacques - louis lions , + paris , f-75005 france ; + cnrs , umr 7598 ljll , paris , f-75005 france + & courant institute , new york university + 251 mercer st , ny ny 10012 , usa + [email protected]
|
we define a renormalized energy " as an explicit functional on arbitrary point configurations of constant average density in the plane and on the real line .
the definition is inspired by ideas of @xcite .
roughly speaking , it is obtained by subtracting two leading terms from the coulomb potential on a growing number of charges .
the functional is expected to be a good measure of disorder of a configuration of points .
we give certain formulas for its expectation for general stationary random point processes .
for the random matrix @xmath0-sine processes on the real line ( @xmath1 ) , and ginibre point process and zeros of gaussian analytic functions process in the plane , we compute the expectation explicitly .
moreover , we prove that for these processes the variance of the renormalized energy vanishes , which shows concentration near the expected value .
we also prove that the @xmath2 sine process minimizes the renormalized energy in the class of determinantal point processes with translation invariant correlation kernels .
* keywords : * renormalized energy , random point processes , random matrices , sine processes , ginibre ensemble , zeros of gaussian analytic functions . +
* msc classification : * 60g55 , 60g10 , 15b52
|
in this paper we introduce a system of finite number of one - dimensional brownian motions which kill each other , and evaluate long - term asymptotics of the probability that all particles survive .
then we define a process of mutually killing brownian motions _ conditioned that all particles survive forever_. we show that this conditional process is equivalent to a special case of the process recently introduced by oconnell in order to analyze a directed polymer model in 1 + 1 dimensions @xcite . as an introduction of our study of many - particle systems ,
here we consider a family of one - particle systems with a parameter @xmath1 .
let @xmath2 be such a one - dimensional brownian motion that its survival probability @xmath3 decays following the equation @xmath4 with a decay rate function @xmath5 the function ( [ eqn : v1 ] ) implies that , if the brownian particle moves in the positive region far from the origin @xmath6 , decay of survival probability is negligible , while as it approaches the origin the decay becomes large .
note that the brownian particle is able to penetrate a negative region @xmath7 , but there the decay rate of survival probability grows exponentially as a function of @xmath8 .
the parameter @xmath9 is the _ characteristic length of the interaction _ to kill a particle acting from the origin and it is also the _ penetration length _ of a particle into the negative region .
see fig.[fig : fig1](a ) . for @xmath10 ,
if a path of the brownian motion up to time @xmath11 is given as @xmath12 , the survival probability of the particle at the time @xmath11 is given by an integration of ( [ eqn : pt1 ] ) , @xmath13 provided that @xmath14 , in the limit @xmath15 , the process becomes an absorbing brownian motion in the positive region with an absorbing wall at the origin , in which the survival probability ( [ eqn : pt2 ] ) is zero if the particle hits the origin at any time @xmath16 , and it is one if it stays in the positive region in the time period @xmath17 $ ] .
see fig.[fig : fig1](b ) . given by ( [ eqn : v1 ] ) .
the parameter @xmath9 is the characteristic length of the interaction to kill a particle acting from the origin .
it also indicates the penetration length of a particle into the negative region .
( b ) an illustration of a path of surviving brownian particle in the absorbing brownian motion with an absorbing wall at the origin .
it is regarded as the limit @xmath15 of the system shown in ( a ) . [
fig : fig1 ] ] let @xmath18 be the transition probability density of this killing brownian motion from a position @xmath19 to a position @xmath20 during time @xmath21 .
it will be obtained by taking an average of ( [ eqn : pt2 ] ) over all realizations of brownian path under the conditions @xmath22 .
( note that the time ordering is reversed , since the equation of @xmath23 given by ( [ eqn : q2 ] ) below is the _
backward _ kolmogorov equation . ) that is , @xmath24 , \label{eqn : q1}\ ] ] where @xmath25 $ ] denotes an expectation with respect to brownian motion , and @xmath26 is the indicator function of a condition @xmath27 ; @xmath28 if @xmath27 is satisfied , and @xmath29 otherwise .
it is the feynman - kac formula ( see , _
e.g. _ , @xcite ) for the unique solution of the diffusion equation @xmath30 under the initial condition @xmath31 .
let @xmath32 be the modified bessel function of the first kind @xmath33 and @xmath34 be macdonald s function @xcite defined by @xmath35 and for integers @xmath36 , @xmath37 which are both analytic functions of @xmath38 for all @xmath38 in the complex plane @xmath39 cut along the negative real axis , and entire functions of @xmath40 .
we see that @xmath41 and @xmath42 are linearly independent solutions of the differential equation @xmath43 for @xmath44 and @xmath45 , @xmath46 is a positive function which increases monotonically as @xmath47 , while @xmath48 is a positive function which decreases monotonically as @xmath47 .
then an integral representation of @xmath49 is given by @xcite @xmath50 where @xmath51 and @xmath52 is the gamma function .
for each initial position @xmath19 , the survival probability of this killing brownian motion at time @xmath53 is obtained by integrating @xmath54 over @xmath20 , @xmath55 if we condition the process to survive up to time @xmath56 , the transition probability density from the state @xmath57 to @xmath58 of the killing brownian motion is given by @xmath59 if we use the integral representations of @xmath42 @xcite , @xmath60 and @xmath61 we can evaluate the long - term asymptotics of the survival probability for @xmath62 , @xmath63 with @xmath64 as shown in appendix a. then we can take the limit @xmath65 of ( [ eqn : pt1 ] ) as @xmath66 this is the transition probability density of the killing brownian motion _ conditioned to survive forever_. we should note that this conditional brownian motion is the diffusion process with infinitesimal generator @xmath67 matsumoto and yor @xcite have studied the exponential functionals of brownian motion @xmath68 which is the diffusion process whose infinitesimal generator is ( [ eqn : my1 ] ) .
we can say that the matsumoto - yor process ( [ eqn : my2 ] ) is realized as the present killing brownian motion conditioned to survive forever @xcite .
now we consider the limit @xmath15 of the formulas obtained above . since @xmath41 is defined as the series expansion ( [ eqn : inu1 ] )
, we can see @xmath69 in @xmath15 for @xmath70 . then ( [ eqn : knu1 ] ) gives @xmath71 in @xmath15 .
therefore the integral of ( [ eqn : q3 ] ) can be performed in the limit and we have @xmath72 it is the transition probability density of the absorbing brownian motion with an absorbing wall at the origin , which is easily obtained by applying the reflection principle of brownian motion @xcite .
applying the result ( [ eqn : k0lima ] ) shown in appendix a , we see that @xmath73 and then ( [ eqn : p1 ] ) gives in this limit @xmath74 which is a harmonic transform ( @xmath75-transform ) of ( [ eqn : q1 ] ) and is identified with the transition probability density of the three - dimensional bessel process , bes(3 ) ( see , for instance , @xcite ) . as a matter of fact @xcite , ( [ eqn : my2 ] ) gives @xmath76 , which is indeed distributed as bes(3 ) ( pitman s @xmath77 theorem @xcite ) . in the present paper
, we consider an @xmath78-particle system of one - dimensional brownian motions with @xmath79 , @xmath80 , with a positive parameter @xmath81 , such that the probability that all @xmath78 particles survive up to time @xmath11 , @xmath82 , decays following the equation @xmath83 with a decay rate function @xmath84 we study an integral representation of the transition probability density @xmath85 , which is a multivariate extension of ( [ eqn : q3 ] ) .
the extensions of the formulas ( [ eqn : nlimit ] ) and ( [ eqn : p1 ] ) are shown .
we prove that the @xmath78-particle system of the mutually killing brownian motions following ( [ eqn : pn1 ] ) and ( [ eqn : vn1 ] ) _ conditioned that all @xmath78 particles survive forever _ is equivalent to a special case ( without drift , @xmath86 ) of the oconnell process @xcite .
in the context of quantum mechanics , the decay rate functions of survival probability ( [ eqn : v1 ] ) and ( [ eqn : vn1 ] ) in the killing brownian motions are considered to giving potential energy of the systems .
they are called the _
yukawa potential _ and the _ toda - lattice potential _ @xcite , respectively .
the multivariate extensions of macdonald s function ( [ eqn : knu1 ] ) with ( [ eqn : knu2 ] ) are the whittaker functions ( see @xcite and references therein ) , which have been extensively studied in mathematical physics as eigenfunctions of the quantum toda lattice @xcite .
see also @xcite . in sect.2 , as preliminaries , useful integral representations of the whittaker functions
are given for the @xmath87-quantum toda lattice .
the transition probability density of the system of mutually killing brownian motions is given , in the case that killing of particles does not occur during an observing time period , as an integral of a product of the whittaker functions over the sklyanin measure .
then the main results are given .
we also demonstrate that when the characteristic length @xmath9 of the interaction killing other particles goes to zero , the oconnell process is reduced to the noncolliding brownian motion , which is known to be equivalent to the dyson model @xcite for the eigenvalue process of hermitian matrix - valued diffusion process ( _ i.e. _ dyson s brownian motion model with the parameter @xmath88 @xcite ) .
the proofs of lemmas and proposition are given in sect.3 .
appendix is given for proving ( [ eqn : nlimit ] ) and ( [ eqn : klimb ] ) used above .
in this subsection we set @xmath89 in eq.([eqn : vn1 ] ) and consider @xmath90 as a potential energy of a quantum @xmath78-particle system in one dimension .
then we have the hamiltonian of the @xmath87-quantum toda lattice @xmath91 for @xmath92 .
( in this paper , we set the mass of particle @xmath93 and @xmath94 for simplicity of notation . )
the generalized eigenvalue problem @xmath95 is solved by setting @xmath96 with the eigenfunction @xmath97 , where @xmath98 is the @xmath87-whittaker function @xcite ( the class - one @xmath99-whittaker function @xcite ) .
when @xmath100 , it is expressed by using macdonald s function ( [ eqn : knu1 ] ) as @xmath101 in this sense , the whittaker functions @xmath102 are regarded as multivariate extensions of macdonald s function .
corresponding to the two kinds of integral representations ( [ eqn : kint1 ] ) and ( [ eqn : kint2 ] ) for @xmath42 , two integral representations are known for the whittaker function @xmath103 .
the integral expression corresponding to ( [ eqn : kint1 ] ) is the classical one originally given by jacquet @xcite and was rewritten as the following form by stade @xcite .
( here we use the notation given by @xcite . )
let @xmath104 denote an upper triangular @xmath105 matrix with unit diagonal ; @xmath106 , @xmath107 we write the integral of a function @xmath108 of @xmath104 over all real @xmath104 as @xmath109 the transpose of @xmath104 is denoted by @xmath110 and the principal minor of size @xmath111 of matrix @xmath112 is written as @xmath113 . for @xmath114 , @xmath115 d \z .
\label{eqn : wint1}\end{aligned}\ ] ] another integral representation , which corresponds to ( [ eqn : kint2 ] ) , was given by givental @xcite .
see also @xcite .
let @xmath116 denote a lower triangular array with size @xmath78 , @xmath117 .
for a given @xmath118 , let @xmath119 be the space of all real triangular arrays @xmath116 with size @xmath78 conditioned @xmath120 we write the integral of a function @xmath108 of @xmath116 over @xmath119 as @xmath121 then @xmath122 d \t .
\label{eqn : wint2}\end{aligned}\ ] ] 0.3 cm * remark 1 .
* there is the ` third ' kind of integral representation of the whittaker function , @xmath123 where @xmath124 , @xmath125 for @xmath126 , and the domain of integration @xmath127 is defined by the conditions @xmath128 for all @xmath129 .
this is called the mellin - barnes integral representation , since it can be regarded as the multivariate extension of the mellin - barnes representation of macdonald s function @xcite @xmath130 where the path of integration @xmath131 being a vertical line to the right of any poles of the integrand .
the derivation of the mellin - barnes integral representation ( [ eqn : wint3 ] ) from the classical one ( [ eqn : wint1 ] ) is shown in @xcite ( see also @xcite ) .
the equivalence between the givental integral representation ( [ eqn : wint2 ] ) and the mellin - barnes integral representation ( [ eqn : wint3 ] ) is fully discussed in @xcite .
0.3 cm ; @xmath132 .
the ordering of particle positions can be changed , but the risk of pair annihilation becomes large exponentially , when @xmath133 .
( b ) in the limit @xmath15 , the present system converges to the vicious brownian motion , in which the process survives if and only if paths are not intersecting .
[ fig : fig2 ] ] for @xmath92 , we consider the @xmath78-particle system of mutually killing brownian motions , in which the probability that all @xmath78 particles survive decays in time following ( [ eqn : pn1 ] ) with ( [ eqn : vn1 ] ) depending on realization of paths . for @xmath134 , the transition probability density from the state @xmath135 to @xmath136 during time interval @xmath11
is then given by @xmath137 \nonumber\\ & & \quad = \be \left [ \exp \left\ { - \frac{1}{\xi^2 } \sum_{j=1}^{n-1 } \int_0^t e^{-(b_{j+1}(s)-b_j(s))/\xi } ds \right\ } \1(\b(0)=\y , \b(t)=\x ) \right ] .
\label{eqn : qn1}\end{aligned}\ ] ] it is the feynman - kac formula for the solution of the diffusion equation ( the backward kolmogorov equation ) @xmath138 with @xmath139 under the initial condition @xmath140 0.3 cm
* remark 2 . *
if we regard ( [ eqn : diff1 ] ) as a diffusion equation describing an @xmath78-dimensional brownian motion in @xmath141 with a killing term @xmath142 , ( [ eqn : qn1 ] ) gives a transition probability density for a particle assumed to be at a position @xmath118 such that it survives during time @xmath11 and it arrives at a position @xmath143 after the time duration @xmath11 ( the feynman - kac formula , see , for instance , @xcite ) . in the present paper , on the other hand , we would like to consider an @xmath78-particle system of one - dimensional brownian motions , such that the @xmath111-th and @xmath144-th particles will be pair annihilated with high probability if @xmath145 , and then the probability that all @xmath78 particles survive avoiding from any mutual killing decays in time following ( [ eqn : pn1 ] ) with ( [ eqn : vn1 ] ) .
in order to discuss processes , in which mutual killing of particles actually occurs and total number of particles decreases in time , we have to specify the stochastic rule to determine which pair of particles is annihilated , when the survival probability conditional on a path , @xmath146 , becomes small . here
we are interested in , however , the situation that mutual killing of particles does not occur at all and all @xmath78 particles survive , following the notion of vicious walker models @xcite . the feynman - kac formula ( [ eqn : qn1 ] ) gives the transition probability density between @xmath78-particle configurations @xmath147 .
figure 2(a ) illustrates three paths of surviving particles , in which change of ordering of particle positions occurs within the spatial scale @xmath148
. 0.3 cm by the fact that the whittaker function @xmath149 solves ( [ eqn : h2 ] ) with ( [ eqn : lambda1 ] ) , we can see that @xmath150 solves the diffusion equation ( [ eqn : diff1 ] ) .
the density function of the sklyanin measure @xcite is defined by @xmath151 for @xmath152 . by euler
s reflection formula @xmath153 , if @xmath154 , @xmath155 since the whittaker functions satisfy the completeness relation with respect to the sklyanin measure @xmath156 @xcite , @xmath157 the integral of the solution ( [ eqn : sol1 ] ) multiplied by @xmath158 on this measure , @xmath159 satisfies the initial condition ( [ eqn : diff2 ] ) .
that is , ( [ eqn : qn2 ] ) is an integral representation of the transition probability density ( [ eqn : qn1 ] ) .
it should be noted that the orthogonality relation @xcite @xmath160 is established , where @xmath161 is a set of all permutation of @xmath78 indices and @xmath162 .
it guarantees the chapman - kolmogorov equation for the transition probability density @xmath163 for @xmath164 , which should hold , since the present process is markovian
. 0.3 cm * remark 3 .
* the completeness relation ( [ eqn : complete1 ] ) of the whittaker functions is a multivariate extension of the relation for macdonald s functions , @xmath165 which will define the kontorovich - lebedev transform @xcite
. we can use ( [ eqn : complete2 ] ) to confirm that ( [ eqn : q3 ] ) satisfies the initial condition @xmath31 .
0.3 cm by using the integral representation ( [ eqn : wint1 ] ) , we can prove the following asymptotics of the transition probability density .
[ thm : asymq ] for @xmath166 , as @xmath167 , @xmath168 d \vmu \times
\left\ { 1 + { \cal o } \left ( \frac{|\x|}{\sqrt{t } } \right ) \right\}. \label{eqn : asymq}\end{aligned}\ ] ] 0.5 cm on the other hand , by using givental s integral representation ( [ eqn : wint2 ] ) , we have the following @xcite .
let @xmath169 be the weyl chamber of type a@xmath170 , @xmath171 and set @xmath172 [ thm : asympsi0 ] for @xmath118 , @xmath173 0.3 cm the probability that all @xmath78 particles survive up to time @xmath11 in this system of the mutually killing brownian motions is given by @xmath174 for each initial state @xmath118 .
we call it simply the _ survival probability _ in this paper .
the following long - term asymptotics of the survival probability is obtained .
[ thm : asymn ] let @xmath175 with @xmath176 .
\label{eqn : an}\ ] ] then for @xmath118 , @xmath177 0.3 cm let @xmath178 and we consider the @xmath78-particle system of the mutually killing brownian motions conditioned that all particles survive up to time @xmath179 . for @xmath180 , the transition probability density from the state @xmath181 to @xmath182 of this conditional process is given by @xmath183 the fact that ( [ eqn : pnt1 ] ) depends not only @xmath184 but also @xmath185 and @xmath185 implies that this conditional process is inhomogeneous in time .
proposition [ thm : asymn ] enables us to take the limit @xmath65 of ( [ eqn : pnt1 ] ) . as a result
we obtain the temporally homogeneous process with the transition probability density @xmath186 it is easy to confirm the following @xcite . 0.3 cm [ thm : diffeq ] the transition probability density ( [ eqn : pn2 ] ) of the @xmath78-particle system of the mutually killing brownian motions conditioned that all particles survive forever solves the following diffusion equation , @xmath187 under the initial condition @xmath188 .
0.3 cm recently oconnell @xcite introduced a diffusion process in @xmath141 with the infinitesimal generator of the process @xmath189 with drift @xmath190 .
then we can state the following .
0.3 cm [ thm : oconnell ] the oconnell process without drift , @xmath86 , is realized as the system of mutually killing brownian motions conditioned that all particles survive forever
. 0.5 cm in an earlier paper @xcite , we considered the system of mutually killing brownian motions with drifts @xmath191 conditioned that all particles survive forever and took the limit @xmath192 to realize the oconnell process without drift @xmath86 . in the present paper
we have shown that an introduction of drift @xmath193 is not necessary to deriving ( [ eqn : pn2 ] ) and concluding theorem [ thm : oconnell ] .
it is known that the functions @xmath194 called the fundamental whittaker functions are defined so that the ( class - one ) whittaker functions discussed so far are expressed by the following alternating sum of them @xcite , @xmath195 where @xmath196 for each permutation @xmath197 .
the functions @xmath194 are normalized as @xmath198 note that ( [ eqn : alt1 ] ) is a multivariate extension of the definition ( [ eqn : knu1 ] ) of @xmath42 , which is given by an alternating sum of @xmath41 and @xmath199 .
therefore , when @xmath200 , we can see that ( [ eqn : qn2 ] ) with ( [ eqn : skl1 ] ) gives @xmath201 .
\label{eqn : km1}\end{aligned}\ ] ] this is the karlin - mcgregor determinant giving a total mass of nonintersecting paths of brownian motions from @xmath202 to @xmath203 during time @xmath11 @xcite . as a matter of fact , when @xmath200 , the limit @xmath15 of ( [ eqn : qn1 ] ) should give the transition probability density of the @xmath78-dimensional absorbing brownian motion in the weyl chamber @xmath169 ( the vicious brownian motion @xcite ) .
see fig.[fig :
fig2](b ) . lemma [ thm : asympsi0 ] implies that , if @xmath200 , @xmath204 which is an @xmath75-transform of @xmath205 by the vandermonde determinant ( [ eqn : hn1 ] ) .
it shows that the limit @xmath15 of the oconnell process is identified with the dyson model ( dyson s brownian motion model with @xmath88 @xcite ) , which is the @xmath78-particle system of brownian motions conditioned never to collide with each other ( the noncolliding brownian motion ) @xcite .
0.3 cm * remark 4 .
* the survival probability up to time @xmath206 of the one - dimensional absorbing brownian motion with an absorbing wall at the origin is given by @xmath207 where the transition probability density @xmath208 is given by ( [ eqn : q1 ] ) .
the long - term behavior is readily obtained as @xmath209 with @xmath210 .
therefore , the exponent @xmath211 is the same with that found in the asymptotics ( [ eqn : nlimit ] ) of the survival probability of the killing brownian motion discussed in sect.1 ; @xmath212 if we take the limit @xmath15 , ( [ eqn : klimb ] ) gives @xmath213 with @xmath214 . for @xmath79 , proposition
[ thm : asymn ] implies @xmath215 with @xmath216 given by ( [ eqn : cn ] ) and with the _ survival probability exponent _
@xmath217 for the present @xmath78-particle system of mutually killing brownian motion .
this exponent is the same as that for the vicious brownian motion ( the absorbing brownian motion in @xmath169 ) @xcite ; @xmath218 with @xmath219 lemma [ thm : asympsi0 ] gives then @xmath220 where @xmath221 with ( [ eqn : an ] ) .
the above results @xmath222 and @xmath223 are consequences of the facts that for the transition probability densities @xmath49 and @xmath224 , the long - term limit @xmath65 and the limit @xmath15 are noncommutable .
by the integral representation ( [ eqn : wint1 ] ) , we put @xmath225 .
then by changing the integral variables as @xmath226 , ( [ eqn : qn2 ] ) with ( [ eqn : skl1 ] ) is written as @xmath227 where @xmath228 and @xmath229 . for @xmath230 , in the limit @xmath167 , @xmath231 , and thus @xmath232 . since @xmath233 as @xmath234 , @xmath235 in this limit . then ( [ eqn : asymq ] ) is obtained .
the following argument is found in section 6 of oconnell @xcite .
for @xmath79 , the _ gelfand - tsetlin pattern _
( _ complete branching _ ) is defined as @xcite @xmath236 where @xmath237 . by the givental integral representation ( [ eqn : wint2 ] ) , @xmath238 d \t \nonumber\\ & = & \beta^{n(n-1)/2 } \int_{\gamma_n(\x ) } \exp \left [ - \sum_{j=1}^{n-1 } \sum_{k=1}^j \big\ { e^{-\beta({t'}_{j , k}-{t'}_{j+1,k } ) } + e^{-\beta({t'}_{j+1 , k+1}-{t'}_{j , k } ) } \big\ } \right ] d \t ' , \nonumber\end{aligned}\ ] ] where we set @xmath239 .
let @xmath240 . as @xmath241 ,
if @xmath242 then the integrand converges to 1 , and otherwise it becomes zero .
then the integral is the volume of @xmath243 under the condition @xmath244 ( the gt - polytope ) , which is given by @xmath245 . then the proof is completed . by definition ( [ eqn : nn1 ] ) , lemma [ thm : asymq ] gives @xmath246 \times \left\ { 1+{\cal o } \left ( \frac{|\x|}{\sqrt{t } } \right ) \right\ } \nonumber\end{aligned}\ ] ] in @xmath167 .
if we put @xmath247 , it is written as @xmath248 \times \left\ { 1 + { \cal o } \left ( \frac{|\x|}{\sqrt{t } } \right ) \right\ } , \nonumber\end{aligned}\ ] ] in @xmath167 . by lemma [ thm : asympsi0 ] ,
@xmath249 in @xmath250 , and then the proposition is proved .
* appendix *
by ( [ eqn : kint1 ] ) , we can put @xmath251 with @xmath252 @xmath253 . changing the integral variable in ( [ eqn : q3 ] ) by @xmath254
, the transition probability density can be written as @xmath255 with @xmath256 . for @xmath62 , as @xmath257 , @xmath258 , and @xmath259 .
then @xmath260 in @xmath257 , since @xmath261 by definition .
it gives through ( [ eqn : n1 ] ) @xmath262 with @xmath263 by ( [ eqn : kint2 ] ) , for @xmath264 , @xmath265 is written as @xmath266 where we have set @xmath267 . in the limit @xmath241 , @xmath268 , if @xmath269 and @xmath270 , and @xmath271 , otherwise .
therefore @xmath272 0.5 cm * acknowledgements * the present author would like to thank t. sasamoto and t. imamura for useful discussion on the present work .
a part of the present work was done during the participation of the present author in cole de physique des houches on vicious walkers and random matrices " ( may 16 - 27 , 2011 ) .
the author thanks g. schehr , c. donati - martin , and s. pch for well - organization of the school .
this work is supported in part by the grant - in - aid for scientific research ( c ) ( no.21540397 ) of japan society for the promotion of science .
givental , a. : stationary phase integrals , quantum toda lattices , flag manifolds and the mirror conjecture . in : topics in singular theory , ams trans .
180 , pp.103 - 115 , ams , rhode island ( 1997 ) kostant , b. : quantisation and representation theory . in : representation theory of lie groups , proc .
src / lms research symposium , oxford 1977 , lms lecture notes 34 , pp .
287 - 316 , cambridge university press , cambridge ( 1977 ) semenov - tian - shansky , m. : quantisation of open toda lattices .
in : dynamical systems vii : integrable systems , nonholonomic dynamical systems .
edited by v. i. arnold and s. p. novikov .
encyclopedia of mathematical sciences , vol.16 , springer , berlin ( 1994 ) .
|
recently oconnell introduced an interacting diffusive particle system in order to study a directed polymer model in 1 + 1 dimensions .
the infinitesimal generator of the process is a harmonic transform of the quantum toda - lattice hamiltonian by the whittaker function . as a physical interpretation of this construction ,
we show that the oconnell process without drift is realized as a system of mutually killing brownian motions conditioned that all particles survive forever .
when the characteristic length of interaction killing other particles goes to zero , the process is reduced to the noncolliding brownian motion ( the dyson model ) . 0.5 cm * keywords * mutually killing brownian motions @xmath0 survival probability @xmath0 quantum toda lattice @xmath0 whittaker functions @xmath0 the dyson model
|
one of the important tasks of quantum mechanics is to find exact solutions of the wave equations ( nonrelativistic and relativistic ) for certain potentials of physical interest since they contain all the necessary information regarding the quantum system under consideration .
it is well known that the exact solutions of these wave equations are only possible in a few simple cases such as the coulomb , the harmonic oscillator , pseudoharmonic and mie - type potentials [ 1 - 8 ] . for an arbitrary @xmath3-state
, most quantum systems could be only treated by approximation methods . for the rotating morse potential
some semiclassical and/or numerical solutions have been obtained by using pekeris approximation [ 9 - 13 ] . in recent years , many authors have studied the nonrelativistic and relativistic wave equations with certain potentials for the @xmath6- and @xmath3-cases .
the exact and approximate solutions of these models have been obtained analytically [ 10 - 14 ] .
many exponential - type potentials have been solved like the morse potential [ 12,13,15 ] , the hulthn potential [ 16 - 19 ] , the pschl - teller [ 20 ] , the woods - saxon potential [ 21 - 23 ] , the kratzer - type potentials [ 12,14,24 - 27 ] , the rosen - morse - type potentials [ 28,29 ] , the manning - rosen potential [ 29 - 33 ] and other multiparameter exponential - type potentials [ 34,35 ] .
various methods are used to obtain the exact solutions of the wave equations for this type of exponential potentials ( cf . [ 36 ] and the references therein@xmath7 .
recently , the nu method [ 37 ] has shown its power in calculating the exact energy levels of all bound states for some solvable quantum systems . in this work ,
we attempt to apply this method to study another exponential - type potential proposed by manning and rosen [ 29 - 33 ] . with an approximation to centrifugal term ,
we solve the @xmath0-dimensional schrdinger equation to its bound states energies and wavefunctions . this potential is defined as [ 29 - 33 ] @xmath8 where @xmath9 and @xmath10 are
two - dimensionless parameters [ 27,28 ] but the screening parameter @xmath11 has dimension of length which has a potential range @xmath12 the potential ( 1 ) may be further put in the following simple form @xmath13 which is usually used for the description of diatomic molecular vibrations [ 38,39 ] .
it is also used in several branches of physics for their bound states and scattering properties .
the potential in ( 1 ) remains invariant by mapping @xmath14 and has a relative minimum value @xmath15 at @xmath16 $ ] for @xmath17 to be obtained from the first derivative @xmath18 the second derivative which determines the force constants at @xmath19 is given by @xmath20 ^{2}}{8b^{4}\alpha ^{3}(\alpha -1)^{3}}.\ ] ] the contents of this paper are as follows : in section ii we outline the nikiforov - uvarov ( nu ) method . in section iii , we derive @xmath21 bound state eigensolutions of the @xmath0-dimensional schrdinger equation for the manning - rosen potential by this method . in section iv , we present our numerical calculations in @xmath22 and @xmath23 systems for various quantum numbers @xmath24 and @xmath25 section v , is devoted to for two special cases , namely , @xmath26 and the hulthn potential . the concluding remarks are given in section vi .
the nu method is based on solving the second - order linear differential equation by reducing it to a generalized equation of hypergeometric type [ 37 ] . in this method after employing an appropriate coordinate transformation @xmath27 the schrdinger equation can be written in the following form : @xmath28 @xmath29 where @xmath30 and @xmath31 are the polynomials with at most of second - degree , and @xmath32 is a first - degree polynomial .
the special orthogonal polynomials [ 37 ] reduce eq .
( 4 ) to a simple equation of the following hypergeometric type : @xmath33 where @xmath34 @xmath35 and @xmath36 is a constant given in the form @xmath37 @xmath38 it is worthwhile to note that @xmath36 or @xmath39 are obtained from a particular solution of the form @xmath40 which is a polynomial of degree @xmath41 further , @xmath42 is the hypergeometric - type function whose polynomial solutions are given by rodrigues relation @xmath43 , \ ] ] where @xmath44 is the normalization constant and the weight function @xmath45 must satisfy the condition [ 37 ] @xmath46 in order to determine the weight function given in eq .
( 9 ) , we must obtain the following polynomial : @xmath47 in principle , the expression under the square root sign in eq .
( 10 ) can be arranged as the square of a polynomial .
this is possible only if its discriminant is zero . in this case ,
an equation for @xmath48 is obtained . after solving this equation , the obtained values of @xmath48
are included in the @xmath49 method and here there is a relationship between @xmath36 and @xmath48 given in eq .
in this section , we follow closely the approach of ref . [ 36 ] .
we begin by considering the se , in arbitrary dimension @xmath50 as [ 40 - 42 ] @xmath51 \right\ } \psi _ { l_{1}\cdots l_{d-2}}^{(l_{d-1}=l)}({\bf x})=0,\ ] ] @xmath52 @xmath53 , \ ] ] @xmath54 where the potential @xmath55 is taken as the manning - rosen form in ( 1 ) . in addition , @xmath56 @xmath57and @xmath58 denote the reduced mass and energy of two interacting particles , respectively .
@xmath59 is a @xmath0-dimensional position vector with the hyperspherical cartesian components @xmath60 given as follows [ 43 - 47 ] : @xmath61 @xmath62 @xmath63 @xmath64 @xmath65 @xmath64 @xmath66 @xmath67 for @xmath68 we have @xmath69 @xmath70 for @xmath71 and @xmath72 @xmath73 @xmath74 for @xmath75 the laplace operator @xmath76 is defined by [ 48 ] @xmath77 the volume element of the configuration space is given by @xmath78 where @xmath79 @xmath80 $ ] and @xmath81,$ ] @xmath82.$ ] equation ( 13 ) permits a solution via separation of variables , if one writes generalized spherical harmonics @xmath83 [ 49 ] : @xmath84 @xmath85 as a simultaneous eigenfunction of @xmath86@xmath87 @xmath88,\ ] ] @xmath89,\ ] ] @xmath90 @xmath91 the unit vector along @xmath59 is usually denoted by @xmath92 additionally , the angular momentum operators @xmath93 are defined as [ 43 - 47,49]@xmath94@xmath95 @xmath96 @xmath97 .\ ] ] the substitution of eqs .
( 13 ) and ( 15)-(17 ) into eq .
( 11 ) allows us to obtain , via the method of separation of variables , the following equation : @xmath98 @xmath99 @xmath100 @xmath101 @xmath102 @xmath64 @xmath103,\ ] ] @xmath64 @xmath104 where @xmath105 @xmath106 $ ] are the angular operators .
thus , with the aid of eqs .
( 15 ) and ( 16 ) , the last wave equation can be easily separated into the following radial and angular parts as [ 50,51]:- angular equations for the angular parameters @xmath107 and one radial equation for the radial parameter @xmath108 with calculated separation constants @xmath109 where @xmath110.$ ] ] @xmath111 @xmath112 + \frac{% ( d+2l-2)^{2}-1}{4r^{2}},\ ] ] @xmath113 @xmath114 h(\theta _ { d-1})=0,\ ] ] @xmath64 @xmath115 h(\theta _ { j})=0,\text { \ } j\in \lbrack 2,d-2],\ ] ] @xmath64 @xmath116 h(\theta _ { 1})=0,% \text { } \lambda _ { 1}=l_{1}^{2}=m^{2},\ ] ] where @xmath117 @xmath110 $ ] are separation constants and @xmath118 is defined in eq . ( 11 ) .
the solution in ( 22 ) is periodic and must satisfy the periodic boundary condition @xmath119 from which we obtain [ 42 ] @xmath120 further , eqs .
( 20 ) and ( 21 ) representing the angular wave equation become @xmath121 @xmath122 h(\theta _ { d-1})=0,\ ] ] @xmath123 with @xmath124,$ ] @xmath125 and @xmath109 which is well - known in three - dimensional space [ 48]@xmath126 and @xmath127 for @xmath75 ] hence , eqs .
( 24 ) and ( 25 ) are to be solved in the following subsection . in order to apply nu method
, we introduce a new variable @xmath129 hence , eq . ( 25 ) is then rearranged in the form of the universal associated - legendre differential equation @xmath130 where @xmath124,$ ] @xmath131 by comparing eqs .
( 26 ) and ( 4 ) , the corresponding polynomials are obtained @xmath132 inserting the above expressions into eq .
( 10 ) and taking @xmath133 , one obtains the following function : @xmath134 s^{2}+k-\lambda _
{ j}+\lambda _ { j-1}}.\ ] ] following the method , the polynomial @xmath135 is found to have the following four possible values : @xmath136 where @xmath137 with @xmath138 @xmath139 and @xmath140,$ ] @xmath141 imposing the condition @xmath142 in eq .
( 6 ) , one selects the following physically valid solutions : @xmath143 which yields from eq .
( 6 ) that @xmath144 making use from eq .
( 7 ) , the following expressions for @xmath36 are obtained as follows : @xmath145 @xmath146 upon comparing eqs .
( 32 ) and ( 33 ) , we obtain @xmath147 in addition , using eqs . ( 6 ) and ( 8)-(9 ) , we obtain the following useful parts of the wavefunctions : @xmath148 where @xmath124,$ ] @xmath131 besides , substituting the weight function @xmath149 given in ( 35 ) into eq .
( 8) , we obtain @xmath150 where @xmath151 is the normaliation factor .
finally the angular wavefunction is @xmath152,\text { } d>3\ ] ] with @xmath153 given in eq .
( 34 ) becomes @xmath154,d>3.\ ] ] likewise , using @xmath155 we can rewrite eq .
( 24 ) in the associated legendre form @xmath156 @xmath157 it s worth to note that , eq . ( 39 ) has been recently solved in @xmath158 by the nu method in [ 40,41,51 ] . however , our aim is to solve it in @xmath0-dimensions .
hence , comparing eqs .
( 39 ) and ( 4 ) , the corresponding polynomials are obtained @xmath159 inserting the above expressions into eq .
( 10 ) and taking @xmath133 , one obtains : @xmath160 s^{2}+k-\nu + \lambda _ { d-2}}.\ ] ] following the method , the polynomial @xmath135 is found to have the following four possible values : @xmath161 where @xmath162 imposing the condition @xmath142 in eq .
( 6 ) , one selects the following physically valid solutions : @xmath163 giving @xmath164 making use from eq .
( 7 ) , we obtain @xmath165 @xmath166 we compare eqs .
( 46 ) and ( 47 ) , the angular momentum @xmath3 values are obtained from @xmath167 which can be easily reduced to the well - known solution @xmath168 in @xmath158 [ 51 ] .
using eqs ( 6 ) and ( 8)-(9 ) , we obtain the following useful parts of the wavefunctions : @xmath169 besides , the rodrigues relation ( 8) gives @xmath170 where @xmath171 is the normaliation factor . finally the angular wavefunctions are @xmath172 where @xmath173 is given by eq . ( 48 ) . since the radial part of the @xmath0-dimensional schrdinger equation with above manning - rosen effective potential has no analytical solution for @xmath174 states@xmath175 an approximation to the centrifugal term has to be made . the good approximation for @xmath176 in the centrifugal barrier is taken as [ 18,33 ] it includes a coulomb term . ]
@xmath177 in a short potential range . to solve it by the present method
, we need to recast eq .
( 19 ) with the aid of eq .
( 53 ) , into the form of eq .
( 4 ) changing the variables @xmath178 through the mapping function @xmath179 and energy transformation given by @xmath180 to obtain the following hypergeometric equation : @xmath181 @xmath182 ^{2}}\left\ { -\varepsilon ^{2}+\frac{1}{4}% \left [ 4a+8\varepsilon ^{2}-(d+2l-2)^{2}+1\right ] z-\left [ a+\varepsilon ^{2}+\alpha ( \alpha -1)\right ] z^{2}\right\ } g(z)=0.\ ] ] we notice that for bound state ( real ) solutions , the last equation requires that @xmath183 and thus the finite radial wavefunctions @xmath184 to apply the nu method , it is necessary to compare eq . (
15 ) with eq .
subsequently , the following value for the parameters in eq . ( 4 ) are obtained as @xmath185 @xmath186 z^{2}+\frac{1}{4}\left [ 4a+8\varepsilon ^{2}-(d+2l-2)^{2}+\right ] z-\varepsilon ^{2}.\ ] ] if one inserts these values of parameters into eq .
( 10 ) , with @xmath187 the following linear function is achieved @xmath188 @xmath189 -k\right\ } z^{2}+\left\ { 4k-\left [ 4a+8\varepsilon ^{2}-(d+2l-2)^{2}+1\right ] \right\ } z+4\varepsilon ^{2}}.\ ] ] according to this method the expression in the square root has to be set equal to zero , that is , @xmath190 -k\right\ } z^{2}+\left\ { 4k-\left [ 4a+8\varepsilon ^{2}-(d+2l-2)^{2}+1\right ] \right\ } z+4\varepsilon ^{2}=0.$ ] thus the constant @xmath48 can be obtained as @xmath191 in view of that , we can find four possible functions for @xmath192 as @xmath193 we must select @xmath194 in order to obtain the polynomial , @xmath195 having negative derivative as @xmath196 we can also write the values of @xmath197 and @xmath198 @xmath199 as @xmath200 , \ ] ] @xmath201 respectively . additionally , using the definition of @xmath202 and solving the resulting equation for @xmath203
allows one to obtain @xmath204 from which we obtain the discrete energy levels @xmath205 ^{2},\text { \ } 0\leq n , l<\infty\ ] ] where @xmath2 denotes the radial quantum number .
it is found that @xmath206 remains invariant by mapping @xmath207 so do the bound state energies @xmath208 an important quantity of interest for the manning - rosen potential is the critical coupling constant @xmath209 which is that value of @xmath9 for which the binding energy of the level in question becomes zero .
( 26 ) , in atomic units @xmath210
@xmath211 let us now find the corresponding radial part of the wave function .
using @xmath212 and @xmath192 in eqs ( 57 ) and ( 61 ) , we obtain @xmath213 @xmath214 @xmath215 .\ ] ] the functions @xmath216 are , up to a numerical factor , are in the form of jacobi polynomials , i.e. , @xmath217 valid physically in the interval @xmath218 @xmath219 @xmath220 [ 52 ] . therefore , the radial part of the wave functions can be found by substituting eqs .
( 68 ) and ( 70 ) into @xmath221 as @xmath222 where @xmath223 and @xmath206 are given in eq .
( 65 ) and @xmath224 is a normalization constant .
this equation satisfies the requirements ; @xmath225 as @xmath226 @xmath227 and @xmath228 as @xmath229 @xmath230 therefore , the wave functions , @xmath231 in eq . (
71 ) is valid physically in the closed interval @xmath232 $ ] or @xmath233 further , the wave functions satisfy the normalization condition @xmath234 where @xmath224 can be determined via @xmath235 ^{2}dz.\ ] ] following ref . ]
36 } , we find the normalization constant @xmath236 where @xmath237 @xmath238 therefore , we may express the normalized total wave functions as @xmath239 @xmath240 to show the accuracy of our results , we calculate the energy eigenvalues for various @xmath2 and @xmath3 quantum numbers with three different values of the parameters @xmath4 as seen n table 1 , the energy eigenvalues for different quantum numbers are obtained numerically for @xmath71 and @xmath241 cases .
> from eq .
( 66 ) , we have seen that two interdimensional states are degenerate whenever [ 53 ] @xmath242 thus , a knowledge of @xmath243 for @xmath71 and @xmath244 provides the information necessary to find @xmath243 for other higher dimensions .
for example , @xmath245 this is the same transformational invariance described for bound states of free atoms and molecules [ 54,55 ] and demonstrates the existence of interdimensional degeneracies among states of the confined manning - rosen potential . as an example of incidental degeneracy , table 1 presents the results of the confined @xmath22 and @xmath23 manning - rosen energies at several radii of confinement for various @xmath2 and @xmath3 states .
in this work , we have utilized @xmath1 method to solve the @xmath0-dimensional @xmath246 for the manning - rosen model potential with the angular momentum @xmath21 states@xmath126 we have derived the binding energy spectra in eq . ( 66 ) and their corresponding wave functions in eq . ( 71 ) .
let us study special cases .
we have shown that for @xmath247 @xmath248 , the present solution reduces to the one of the hulthn potential [ 16,18,19 ] : @xmath249 where @xmath250 is the strength and @xmath251 is the screening parameter and @xmath252 is the range of potential .
if the potential is used for atoms , the @xmath253 is identified with the atomic number .
this can be achieved by setting @xmath254 hence , the energy for @xmath21 states @xmath255 ^{2}\hbar ^{2}}{32\mu b^{2}(2n+d+2l-1)^{2}},\text { \ } 0\leq n , l<\infty .\ ] ] and for @xmath6-wave ( @xmath256 states @xmath257 ^{2}\hbar ^{2}}{8\mu b^{2}(n+1)^{2}},% \text { \ } 0\leq n<\infty\ ] ] essentially , these results coincide with those obtained by the feynman integral method [ 31 ] and the standard way [ 32,33 ] , respectively . furthermore , if taking @xmath258 and identifying @xmath259 as @xmath260 we are able to obtain @xmath261 ^{2},\ ] ] which coincides with those of refs .
[ 16,18 ] . with natural units
@xmath262 we have @xmath263 ^{2},\ ] ] which coincides with refs .
[ 16,33 ] .
the corresponding radial wave functions are expressed as @xmath264 where @xmath265 , \text { } 0\leq n , l<\infty , \ ] ] which coincides for the ground state with that given in eq .
( 6 ) by gnl _
et al _ [ 18 ] .
in addition , for @xmath266 ( i.e. , @xmath267 the hulthn potential turns to become a coulomb potential : @xmath268 with energy levels and wavefunctions : @xmath269 @xmath270 where @xmath271 @xmath272 and @xmath273 is bohr radius for the hydrogen atom .
the wave functions are @xmath274 r^{(d+2l-1)/2}p_{n}^{\left ( \frac{2\mu ze^{2}}{% \hbar ^{2}\delta \left ( 2n+d+2l-1\right ) } , d+2l-2\right ) } ( 1 + 2\delta r)\ ] ] which coincide with refs .
[ 3,16,22 ] .
in this work , we have extended the approximate solutions of the @xmath3-wave schrdinger equation with the manning - rosen potential to @xmath0-dimensions .
the special cases for @xmath275 are discussed .
the results are found to be in good agreement with those obtained by other methods in @xmath158 for short potential range , small @xmath10 and @xmath3 [ 36]@xmath126 these numerical solutions have also extended to various dimensional space , @xmath71 and @xmath241 systems .
we have also studied two special cases for @xmath276 @xmath21 and hulthn potential .
the results we have ended up show that the nu method constitute a reliable alternative way in solving the exponential potentials .
ikhdair and r. sever , int .
c 18 ( 10 ) ( 2007 ) 1571 ; centr .
j. phys . 5 ( 4 ) ( 2007 ) 516 ; preprint quant - ph/0702052 to appear in the centr .
j. phys . ;
ikhdair , preprint quant - ph/0703042 , to appear in the chinese j. phys .
|
the schrdinger equation in @xmath0-dimensions for the manning - rosen potential with the centrifugal term is solved approximately to obtain bound states eigensolutions ( eigenvalues and eigenfunctions ) .
the nikiforov - uvarov ( @xmath1 ) method is used in the calculations .
we present numerical calculations of energy eigenvalues to two- and four - dimensional systems for arbitrary quantum numbers @xmath2 and @xmath3 with three different values of the potential parameter @xmath4 it is shown that because of the interdimensional degeneracy of eigenvalues , we can also reproduce eigenvalues of a upper / lower dimensional sytem from the well - known eigenvalues of a lower / upper dimensional system by means of the transformation @xmath5 .
this solution reduces to the hulthn potential case .
keywords : bound states ; manning - rosen potential ; nikiforov - uvarov method .
pacs number(s ) : 03.65.-w ; 02.30.gp ; 03.65.ge ; 34.20.cf [ theorem]acknowledgement [ theorem]algorithm [ theorem]axiom [ theorem]claim [ theorem]conclusion [ theorem]condition [ theorem]conjecture [ theorem]corollary [ theorem]criterion [ theorem]definition [ theorem]example [ theorem]exercise [ theorem]lemma [ theorem]notation [ theorem]problem [ theorem]proposition [ theorem]remark [ theorem]solution [ theorem]summary
|
corot ( convection , rotation and transits ) is a modest scale satellite funded primarily by the french space agency cnes , with additional contributions from the european space agency ( esa ) , belgium , austria , germany , spain and brazil .
the project was first proposed to cnes as an asteroseismology mission in 1993 .
an exoplanet programme was added to the mission soon after the discovery of the first exoplanet around a sun - like star @xcite , and it soon became clear that a long baseline , high - precision stellar photometry mission had major potential in exo - planetology as well as in stellar structure and evolution ( see @xcite ) .
the new mission concept was formally selected by cnes in 2001 , and corot was launched on the @xmath1 of december 2006 after a successful development and integration phase .
the main technical characteristics of the mission , and in particular of the instrument , are described in detail in the corot instrument handbook and the reference esa publication on the pre - launc status of the corot mission ( see section [ aig_concl ] ) and are thus only briefly sketched out here .
the payload consists of a 27 cm aperture afocal telescope equipped with an extremely high performance baffle to minimise straylight on the detector .
the telescope feeds a focal plane unit of four @xmath2 pixel ccds arranged in a square pattern , with a pixel scale of @xmath3 m , corresponding to @xmath4arcsec on the sky .
two of the ccds are dedicated to asteroseismology , and two to planet finding , giving rectangular fields of @xmath5 for each program ( see figure [ aig_focal ] , left ) . the asteroseismology ccds are set slightly forwards of the focal plane to create a defocussed image , and are read in frame transfer mode every second .
up to 5 stars with apparent magnitudes in the range @xmath6 are monitored on each ccd , and their light curves are transmitted to the ground with 32s sampling ( 1s on request ) . in addition , up to six @xmath7 pixel windows can be transmitted to the ground every 32s . the exoplanet ccds are read every 32s , and up to 6000 stars per ccd in the range @xmath8 are monitored . because of telemetry limitations , aperture photometry is performed on - board and light curves with 512s sampling are downloaded ( 32s sampling is available for up to 500 stars per ccd ) . an objective prism inserted in the light path of the exoplanet ccds gives rise to a very low - resolution spectrum at the location of each star ( see figure [ aig_focal ] , right ) , allowing the generation of 3-colour light curves for up to 5000 stars with @xmath9 using appropriately positioned apertures .
small windows spread over the ccd are used in both fields to monitor the background level .
the pointing of the satellite is adjusted based on astrometry from the asteroseismology channel .
corot uses the multi - mission proteus platform developed by cnes , which allowed for significant cost savings in the development phase .
it was launched on the @xmath1 of december 2006 by a soyuz ii 1-b launcher from ba " ikonour in kazakhstan , into a polar orbit with an altitude of 896 km and a period of 6714s . to keep scattered light to a minimum
implies a sun angle constraint which results in two 6-month continuous viewing zones of roughly @xmath10 diameter on opposite sides of the sky along the plane of the ecliptic . to maximise the number of targets available for planet finding
, two ` corot eyes ' were selected at the intersections of the ecliptic and galactic planes , one towards the galactic centre ( @xmath11 , @xmath12 ) and one towards the galactic anti - centre ( @xmath13 , @xmath14 ) .
corot is restricted to point within these two eyes only . after a commissioning phase lasting roughly one month
, the first science observations will be in the galactic centre direction , with an initial run lasting one month to six weeks .
after that , corot will start its regular observing pattern of observing one target field for 5 months continuously ( long run ) , then switching to another field in the same eye for approximately 21 days ( short runs ) , and then turning to face the opposite direction .
this cycle will be repeated a minimum of 5 times during the mission lifetime , starting with the galactic anti - centre direction in the spring of 2007 .
the choice of each of the target fields for the long runs was a compromise between the needs of the asteroseismology and the exoplanet science programmes . using gaudi @xcite , a database of all stars brighter than @xmath15 in the corot eyes ,
promising targets for asteroseismology were identified as possible ` primary targets ' .
the potential planet finding fields around each possible primary target ( allowing for all rotations and translations of the detectors while keeping the primary target on the asteroseismology ccds ) were surveyed from the ground using multi - band optical photometry @xcite , combined with 2mass , to evaluate the amount of reddening and the relative fraction of dwarf and giant or early type stars ( the latter , which make an estimated 20% and 50% of the stellar population in the galactic anti - centre and centre fields respectively , are not useful targets for transit finding ) .
also taken into account when selecting target fields were potential secondary targets for the asteroseismology fields and potential targets for additional programmes ( see section [ aig_ap ] ) in both fields .
the fields of the initial run and the first two long runs are listed in table [ aig_fields ] .
lccc name & ra & dec & roll angle + & [ h m s ] & [ @xmath16 ] & [ @xmath16 ] + ir & @xmath18 & @xmath19 & @xmath20 + lrc1 & @xmath21 & @xmath22 & @xmath23 + lra1 & @xmath24 & @xmath25 & @xmath26 + the choice of the target fields for the short runs is primarily driven by the requirements of the asteroseismology program ( to ensure coverage of all the hr diagram ) , though a small number of short runs will be driven by the needs of the additional program .
the expected scientific performance of corot from the exoplanet point of view is reported in @xcite .
this section summarises the methods that are foreseen to be used to produce light curves and analyse them , and the results of simulations of the corot fields which aim to give an estimate of the planet yield of the mission .
the point spread function ( psf ) in the exoplanet field is tear - shaped , owing to the presence of the prism in the optical path .
it is strongly peaked at the red end and extended in the x- ( east - west ) direction .
the exact psf depends on the star s temperature and position on the ccd , and the on - board software selects the most appropriate mask for each target from a set of 256 templates , based on an initial long - exposure image taken at the beginning of each run .
aperture photometry is then carried out by summing the flux in the pixels within the mask ( see figure [ aig_focal ] , bottom right ) . to generate three - colour photometry ,
the mask is divided into three sub - masks and the columns falling into each of the sub - masks are summed separately .
binned averages from small background windows spread over each ccd are used to monitor the background level . except for the 500 oversampled targets per ccd
, the aperture photometry from 16 32s exposures is summed on - board and individual light curve points are downlinked to earth every 512s .
the raw data received from the satellite are labelled n0 data .
the data downlinked from the satellite ( labelled n0 data ) is further processed on the ground , first by a generic pipeline which corrects for anticipated astrophysical and instrumental noise sources and concatenates the photometry of each object into light curves ( n1 data ) , and then by a pipeline specific to the exoplanet channel which is designed to correct for unanticipated noise sources identified from an ensemble analysis of the n1 light curves ( n2 data ) .
the n2 data are the science grade data on which all scientific analysis , including the transit detection , is performed .
these data will typically be available a few months after the end of each run .
however , a preliminary real - time analysis of the n1 light curves is carried out to detect interesting events ( for example , events which could be transits of giant planets ) .
the purpose of this ` alarm mode ' is to provide a weekly update of the list of targets to be oversampled ( i.e. targets for which data is downlinked to earth every 32 rather than 512s . ) the detection of transits in the corot light curves , and particularly of shallow transits caused by terrestrial planets , is a challenging task .
first one must minimise the impact of a number of high- and low - frequency noise sources affecting the light curves , which become important at the level of precision achieved with corot .
data taken during passages through the south atlantic anomaly ( saa ) unusable , and the light curves therefore contain a few gaps lasting approximately 10 min during each 24h period .
the data around these gaps are also affected by higher background noise level .
additionally , all stars are expected to display some degree of variability ( micro - variability , associated with the rotational modulation and intrinsic evolution of surface features of magnetic or convective origin ) , which must be treated before the shallowest and most interesting transits can be detected . in order to prepare for the analysis of the corot light curves , the corot exoplanet science team ( cest ) has conducted two blind tests . in both cases ,
a ` game master ' generated a set of light curves including as many of the foreseen noise sources as possible , using a detailed model of the corot instrument and the micro - variability models of @xcite and @xcite .
simulated transit events and a range of astrophysical contaminants ( events which mimic transits but are not of planetary origin ) were then added some of the light curves before distributing them to the participating analysis teams , but the true content of the light curves was known only to the game master .
the first blind test ( bt1 , @xcite ) was focussed on detection , and so a relatively large sample of light curves was generated with few containing transit - like events .
only white light light curves were simulated for this first exercise .
five teams took part ( two from the laboratoire dastrophysique de marseille , one from geneva observatory , one from the institute for planetary research in berlin and one from the institute of astronomy in cambridge ) , and the main results of the exercise were the following : * micro - variability can be filtered out relatively effectively without strongly affecting the transits by exploiting the distinct frequency signatures of the two types of signal : micro - variability is strongest on timescales of days and longer , while transits occur over timescales of minutes ( ingress / egress ) to hours . * a number of detection algorithms were tested , including matched filters , and cross - correlation methods , all of which were relatively successful , but the best performances in terms of sensitivity and false alarm rejection were obtained with algorithms based on least - squares fitting of box - shaped transits , such as the well - known bls @xcite . * no false alarms were reported by more than one team , highlighting the advantage of several teams working with different methods but sharing their results . * by tracing a line on a diagram of transit depth @xmath27 versus number @xmath28 of observed transits which separates events detected by most teams from events missed by most teams , a rough detection limit of @xmath29 was established .
this dependence on @xmath28 is as expected if the number of observations per transit and the noise level per observation is roughly the same for all cases , although in reality a number of other factors , in particular star magnitude and activity level , also affect the detection limit .
the second blind test ( bt2 , @xcite ) was focussed mainly on discriminating between real transit events and transit - like events of stellar origin , often referred to as astrophysical false positives , and therefore a smaller number of light curves were generated but all contained a signal transit or false positive .
the transit or eclipse light curve alone does not enable the companion s mass to be measured , and spectroscopic follow - up is always necessary to confirm the planetary nature of any candidate , but a significant fraction of the astrophysical false positives can be eliminated by a close inspection of the light curve , including the colour information .
an important aspect of bt2 was testing the use of the information which will be known about each target in advance as a result of the preliminary ground - based observations ( see section [ aig_strategy ] ) .
these observations provide a list of background stars which contribute some flux to the aperture of each target in the corot fields , or ` contaminants ' .
this information is stored , along with the results of other preliminary observations and , in the long run , the results of follow - up observations aimed at obtaining precise parameters for the host stars of the best planet candidates , in a dedicated database called exodat @xcite , which will be accessible to the scientist analysing the corot light curves .
the exercise is still ongoing , but preliminary conclusions include : * direct application of simple diagnostics to distinguish between planetary transits and stellar eclipses based on event duration ( such as the @xmath30 parameter proposed by @xcite ) or colour ( assuming planetary transits are grey , whereas most stellar eclipses are not ) can lead to the rejection of real transit events , and should be used with care .
* diagnostics using the presence of secondary eclipses or ellipsoidal variability are robust and can be used to distinguish some stellar binaries from planets .
* knowledge of the positions and colours of the ` contaminants ' can be used along with the light curve ( white light alone or including colour information ) to identify many likely blended eclipsing binaries ( stellar binaries whose signal is diluted by a third , constant light source and therefore resembles a planetary transit ) . *
the most difficult type of false positive to identify from the light curves and contaminants information alone is the case of stellar eclipses with a small ( m - type ) secondary , but this type of system has a clear radial velocity signature easily identified from a few spectroscopic observations .
the samples of planets and eclipsing binaries included in the blind tests were designed to contain examples of all the types of events expected , but were not intended to reproduce the real relative frequencies of the different events , or the incidence of transit - like event in the entire sample of light curves .
thus , they tell us what kind of events are detectable and characterisable , but not how many of them to expect .
corotlux @xcite is a tool to evaluate the yield of transit surveys which simulates the entire stellar population observed ( including background stars ) based on standard galactic models @xcite and generates a population of eclipsing binaries and of planets using the most up - to date binarity and planet incidence statistics .
it simulates light curves with realistic astrophysical and instrumental noise sources , after which a matched filter transit detection algorithm is used to evaluate the number of candidate transits expected .
it can be applied to a number of ground or space - based surveys . in the case of corot
, the galactic model used is adjusted to match the star counts in exodat .
two sets of corotlux simulations were carried out : one to investigate the yield of giant planets , for which the incidence and orbital period distribtions are relatively well known , and one to investigated the yield of ` terrestrial ' planets ( @xmath31@xmath32 ) , for which the incidence is unknown , but was assumed to be 5 times higher than that of the giants .
the results will be described in detail in a forthcoming paper ( fressin et al . in prep . ) , but preliminary indications are that corot should detect @xmath33 giant planets per run and @xmath344 terrestrial planets per long run , i.e. a total of @xmath35 giant planets and @xmath36 terrestrial planets .
approximately 100 transit candidates per long run are foreseen , of which the bt2 results to date suggest 50% may be identifiable as stellar events from the light curves and exodat information alone .
the shallowest transits , which are potentially the most interesting , carry the highest follow - up cost , as the relative fraction of true transits to stellar mimics decreases strongly with decreasing transit depth .
three main types of follow - up observations are foreseen : * photometric follow - up , to obtain high spatial resolution images in- and out - of transit in order to identify which of the stars falling in the corot aperture is being eclipsed , and thus what is the true ( undiluted ) depth of the event . as the psf of corot is large , 1 m class telescopes should suffice to obtain a significant resolution improvement . among the facilities foreseen for this follow - up
are the 80 cm telescope at observatorio del teide in tenerife , and the euler telescope in la silla . *
radial velocity follow - up , to measure the companion mass . to make optimal use of the available facilities ,
a progressive strategy is foreseen , involving initial follow - up wit flames in paranal for the fainter objects and , for the brighter candidates , with coralie in la silla , sophie at the observatoire de haute provence , and the 2 m telescope at the thringer landessternwarte tautenburg .
the best low radial velocity amplitude candidates will then be targeted with harps in la silla . as illustrated on figure [ aig_follow ] , corot s detection capabilities
are particularly well - matched to the world - leading radial velocities currently available ; * high - resolution spectra of the host stars , using e.g. uves on the vlt , to derive accurate physical parameters ( for confirmed candidates only ) . to ensure prompt confirmation of transit candidates , the activities of the four cest teams in charge of light curve analysis
, photometric follow - up , radial velocity follow - up and stellar parameter determination will be co - ordinated by a small team which will regularly re - analyse the available data for each candidate in real time and review the candidate priority rankings for each type of follow - up .
once candidates are confirmed , further follow - up of the giant planets is foreseen using space based facilities , with the goal of obtaining extended baseline and improved precision light curves ( hst : improved radii , search for other non - transiting planets by transit timing , search for rings and moons , oblateness measurements ) and detecting secondary eclipses in the infrared ( spitzer : surface temperature measurement and atmospheric studies ) .
although primarily driven by asteroseismology and exoplanet detection , corot also accomodates an additional program , that includes all scientific use of corot data for purposes outside the core program areas of exoplanet detection in the exoplanet field and asteroseismology in the asteroseismology field .
thus , exoplanet detection in the asteroseismology field and asteroseismology in the exoplanet field , as well as any other analysis of corot light curves ( rotation , activity , flaring , eclipsing binaries ) fall under the additional program . through yearly announcements of opportunity ( aos )
, members of the astronomical community at large in the participating countries ( including all esa member states ) can propose either to systematically analyse core program data for scientific purposes outside the core program , or to observe specific targets in the corot fields .
additionally , a small number of short runs will be driven primarily by the additional program .
one such run is foreseen on the ngc2264 star forming region .
the ao for the first year of corot operations is past , but an announcement for the second year is foreseen early in 2007 .
corot is the first space mission dedicated to exoplanet detection , and the first project capable of detecting transits of terrestrial planets . while stopping short of detecting habitable planets
, its capabilities will enable it to explore for the first time the uncharted transition regime between the giant and terrestrial planets found in our own system .
prompt follow - up is expected to enable the confirmation of the first candidate planets from the initial run and the first long run ( alarm mode ) by the summer of 2007 .
as shallow transits will be detectable only in the full long run n2 data , the first of which are expected around august 2007 , the first confirmed terrestrial planets are expected in the spring of 2008 .
the corot light curves will be come public one year after their release to the co - is , i.e. approximately 15 months after the end of each run . * the official corot website : http://corot.oamp.fr ; * _ the corot mission : pre - launch status
stellar seismology and planet finding _ : esa - sp 1306 , eds .
m. fridlund , a. baglin , l. conroy & j. lochard ( 2006 ) ; * for information on the additional program : + http://ams.astro.univie.at/?s=space;corotapwg .
barge , p. , lger , a. , ollivier , m. , rouan , d. , schneider , j. and the exoplanet corot team 2006 , in _ the corot mission : pre - launch status _ , eds .
m. fridlund , a. baglin , l. conroy & j. lochard , esa - sp 1306 , 83 barge , p. , moutou , c. , lger , a. , bord , p. , rouan , d. , pont , f. , guillot , t. , aigrain , s. , deeg , h. , erikson , a. , queloz , d. , rauer , h. and the exoplanet corot team 2006 , in _ the corot mission : pre - launch status _ , eds .
m. fridlund , a. baglin , l. conroy & j. lochard , esa - sp 1306 , 419 deleuil , m. , moutou , c. , deeg , h.j . , meunier , j.c . ,
surace , c. , guterman , p. , almenara , j.m . , alonso , r. , barge , p. , bouchy , f. , erikson , a. , lger , a. , loeillet , b. , ollivier , m. , pont , f. , rauer , h. , rouan , d. & queloz , d. 2006 , in _ the corot mission : pre - launch status _ , eds
. m. fridlund , a. baglin , l. conroy & j. lochard , esa - sp 1306 , 341 moutou , c. , pont , f. , barge , p. , aigrain , s. , auvergne , m. , blouin , d. , cautain , r. , erikson , a. r. , guis , v. , guterman , p. , irwin , m. , lanza , a. f. , queloz , d. , rauer , h. , voss , h. & zucker , s. 2005 , a&a , 437 , 355 solano , e. , catala , c. , garrido , r. , poretti , e. , janot - pacheco , e. , gutirrez , r. , gonzlez , r. , mantegazza , l. , neiner , c. , fremat , y. , charpinet , s. , weiss , w. , amado , p. j. , rainer , m. , tsymbal , v. , lyashko , d. , ballereau , d. , bouret , j. c. , hua , t. , katz , d. , lignires , f. , lftinger , t. , mittermayer , p. , nesvacil , n. , soubiran , c. , vant veer - menneret , c. , goupil , m. j. , costa , v. , rolland , a. , antonello , e. , bossi , m. , buzzoni , a. , rodrigo , c. , aerts , c. , butler , c. j. , guenther , e. & hatzes , a. 2005 , aj , 129 , 547
|
corot , which was launched successfully on the 27@xmath0 of december 2006 , is the first space mission to have the search for planetary transits at the heart of its science programme .
it is expected to be able to detect transits of planets with radii down to approximately two earth radii and periods up to approximately a month .
thus , corot will explore the hereto uncharted area of parameter space which spans the transition between the gaseous giant planets discovered in large numbers from the ground , and terrestrial planets more akin to our own .
this papers briefly sketches out the main technical characteristics of the mission before summarising estimates of its detection potential and presenting the data analysis and follow - up strategy .
|
the investigation of ultrahigh energy ( uhe ) cosmic neutrinos provides an opportunity for study particle physics beyond the reach of the lhc @xcite . as an example , nowadays the pierre auger observatory is sensitive to neutrinos of energy @xmath0 gev @xcite . a crucial ingredient in the calculation of attenuation of neutrinos traversing the earth and the event rate in high energy neutrino telescopes is the high energy neutrino - nucleon cross section , which provides a probe of quantum chromodynamics ( qcd ) in the kinematic region of very small values of bjorken-@xmath1 .
the typical @xmath1 value probed is @xmath2 , which implies that for @xmath3 gev one have @xmath4 at @xmath5 gev@xmath6 .
this kinematical range was not explored by the hera measurements of the structure functions @xcite .
the description of qcd dynamics in such very high energy limit is a subject of intense debate @xcite .
theoretically , at high energies ( small bjorken-@xmath1 ) one expects the transition of the regime described by the linear dynamics , where only the parton emissions are considered , to a new regime where the physical process of recombination of partons becomes important in the parton cascade and the evolution is given by a non - linear evolution equation .
this regime is characterized by the limitation on the maximum phase - space parton density that can be reached in the hadron wavefunction ( parton saturation ) , with the transition being specified by a typical scale , which is energy dependent and is called saturation scale @xmath7 . moreover
, the growth of the parton distribution is expected to saturate , forming a color glass condensate ( cgc ) , whose evolution with energy is described by an infinite hierarchy of coupled equations for the correlators of wilson lines @xcite . in the mean field approximation , the first equation of this hierarchy decouples and boils down to a single non - linear integro - differential equation : the balitsky - kovchegov ( bk ) equation @xcite .
experimentally , possible signals of parton saturation have already been observed both in @xmath8 deep inelastic scattering at hera and in deuteron - gold collisions at rhic @xcite .
currently , there are predictions of the neutrino nucleon cross sections with structure functions constrained by hera data are based on linear dynamics @xcite , using dglap or an unified dglap / bfkl evolution , or phenomenological models that resembles the expected behavior predicted by the non - linear qcd dynamics @xcite ( i.e. , the proton structure function saturating the froissart bound at asymptotic energies , @xmath9 ) . as a general feature , the nonlinear qcd dynamics predicts sizable suppression of uhe neutrino cross section in comparison with standard approaches .
here , we summarize the main results of works presented in refs .
@xcite . in ref . @xcite the contribution of non - linear effects was estimated considering distinct phenomenological models based on saturation physics .
an update on those calculations have been done recently @xcite . in ref .
@xcite , the geometric scaling property ( which is a natural consequence of the asymptotic solutions of the nonlinear qcd evolution equations ) is considered to obtain an analytical parameterization for the uhe neutrino cross sections . in what follows we introduce the theoretical and phenomenological tools and present the main results and predictions
deep inelastic neutrino scattering is described in terms of charged current ( cc ) and neutral current ( nc ) interactions , which proceed through @xmath10 and @xmath11 exchanges , respectively .
the total cross sections are given by : @xmath12 , \label{difcross } \label{total}\end{aligned}\ ] ] where @xmath13 is the neutrino energy , @xmath14 with @xmath15 the nucleon mass , @xmath16 and @xmath17 is the minimum value of @xmath18 which is introduced in order to stay in the deep inelastic region . in what follows
we assume @xmath19 gev@xmath6 .
moreover , @xmath20 is the fermi constant and @xmath21 denotes the mass of the charged of neutral gauge boson . in
the qcd improved parton model the structure functions @xmath22 are expressed in terms of the parton distributions on the nucleon , which satisfy the dglap @xcite linear dynamics . on the other hand ,
an efficient way of introducing non - linear effects is the description of the structure functions considering the color dipole approach in which the dis to low @xmath1 can be viewed as a result of the interaction of a color @xmath23 dipole which the gauge boson fluctuates @xcite . in this approach the @xmath24 structure function is expressed in terms of the transverse and longitudinal structure functions , @xmath25 which are given by @xmath26 where @xmath27 denotes the transverse size of the dipole , @xmath28 is the longitudinal momentum fraction carried by a quark and @xmath29 are proportional to the wave functions of the virtual charged of neutral gauge bosons corresponding to their transverse or longitudinal polarizations .
explicit expressions for @xmath29 are given , e.g. , in ref .
furthermore , @xmath30 describes the interaction of the color dipole with the target . in next section
we will discuss some models for @xmath30 , based on the non - linear qcd dynamics , which describe the current hera data . as a comment , the desy - hera measurements of the structure function at low - @xmath1 ( @xmath31 ) are for very low values of @xmath18 ( @xmath32 gev@xmath6 ) , which implies that small @xmath1 extrapolations of the parton distributions are necessary to estimate @xmath33 above @xmath34 gev .
as mentioned above , uhe neutrino - nucleon cross section accesses very large values of @xmath18 and very small values of bjorken @xmath1 .
the usual linear perturbative qcd evolution equations predictions are based on the linear dglap and/or bfkl equations which implies a power increase with the energy of the neutrino - nucleon cross section that eventually violate the froissart bound .
a easy way to introduce consistently the unitarity constraints on the theoretical estimate of the ultrahigh energy behavior of the neutrino - nucleon cross section is to express the structure functions in the dipole approach [ eq .
( [ funcs ] ) ] and to consider the state - of - art of the non - linear qcd dynamics : the color glass condensate formalism . in this formalism , the dipole - target cross section @xmath30 can be computed in the eikonal approximation , resulting @xmath35 where @xmath36 is the dipole - target forward scattering amplitude for a given impact parameter @xmath37 which encodes all the information about the hadronic scattering , and thus about the non - linear and quantum effects in the hadron wave function .
it is useful to assume that the impact parameter dependence of @xmath38 can be factorized as @xmath39 , so that @xmath40 , with @xmath41 being a free parameter related to the non - perturbative qcd physics .
the balitsky - jimwlk hierarchy describes the energy evolution of the dipole - target scattering amplitude @xmath42 . in the mean field approximation ,
the first equation of this hierarchy decouples and boils down to the balitsky - kovchegov ( bk ) equation @xcite .
the dipole - target cross section can also be calculated considering phenomenological parameterizations for @xmath42 based on saturation physics , which provide an economical description of a wide range of data with a few parameters .
this approach has been considered in refs .
@xcite . in general
, the dipole scattering amplitude is modeled in the coordinate space in terms of a simple glauber - like formula as follows @xmath43 \,\ , , \label{ngeral}\end{aligned}\ ] ] where @xmath44 is the anomalous dimension of the target gluon distribution .
the main difference among the distinct phenomenological models comes from the predicted behavior for the anomalous dimension , which determines the transition from the non - linear to the extended geometric scaling regimes , as well as from the extended geometric scaling to the dglap regime @xcite .
one important property of the nonlinear perturbative qcd approaches for high energy deep inelastic @xmath45 scattering is the prediction of the geometric scaling .
namely , the total @xmath46 cross section at large energies is not a function of the two independent variables @xmath1 and @xmath18 , but is rather a function of the single variable @xmath47 . as usual , @xmath18 is the photon virtuality and @xmath1 the bjorken variable .
geometric scaling is the exact asymptotic solution of a general class of nonlinear evolution equations and it appears as a universal property of these kind of equations .
the saturation momentum @xmath48 ( @xmath49 , @xmath50 ) is connected with the phenomenon of gluon saturation . in principle , geometric scaling is predicted to be present only on process dominated by low momenta . however , it is known that the geometric scaling is preserved by the qcd evolution up to relatively large virtualities @xcite , within the kinematical window @xmath51
. that is , the scaling property extends towards very large virtualities provided one stays in low-@xmath1 .
this kinematical window is further enlarged due to the nuclear enhancement of the saturation scale .
these facts have direct consequences in the behavior of uhe neutrino cross section . as similar scaling holds for the massive boson - nucleon cross section ,
we are able to compute analytically the neutrino cross sections .
this feature was verified to be true in ref .
@xcite , where geometric scaling was shown to occur in the small-@xmath1 charged current neutrino data . in this sense , we are able to construct a theoretical prediction which is model independent .
they are as follows @xcite , @xmath52\ , , \label{resgs}\end{aligned}\ ] ] where @xmath53 are overall normalizations , @xmath54 are numerical constants with @xmath55 , @xmath56 and @xmath57 .
this implies in a mild power - like rise @xmath58 for the neutrino cross section in contrast with other theoretical approaches .
the nuclear dependence is approximately linear , @xmath59 , once @xmath60 and hence @xmath61 .
the remaining constants are given by ( see ref .
@xcite for details ) , @xmath62 where one uses the notation @xmath63 , @xmath64 and @xmath65 .
numerically , this gives a total cross section @xmath66 cm@xmath6 .
the cross section above can have implications for neutrino observatories because experiments are planned to detect uhe by observation of the nearly horizontal air showers in earth coming from neutrino - air interactions @xcite .
a reduced cross section produces a smaller event rate for such neutrino - induced showers and could compromise the detection signal .
however , the rate of up - going air showers initiated by muon and tau leptons produced in neutrino - nucleon reactions just below the surface would increase , being possibly larger than the horizontal air shower rate .
let us present the comparison between the predictions of the linear approaches ( nlo dglap and unified dglap / bfkl ) and non - linear qcd approaches ( a phenomenological model for the color glass condensate and the geometric scaling prediction ) . in fig .
[ fig:1 ] the energy dependence of the neutrino nucleon cc cross section predicted by the linear and nonlinear perturbative qcd approaches are compared . as expected from the solution of the dglap equation at small-@xmath1 ,
the nlo dglap ( dot - dashed curve ) and unified dglap / bfkl ( long - dashed curve ) results predict a strong increase of the cross section at ultrahigh energies .
such estimates are largely used to estimate the event rates in neutrinos telescopes .
clearly , the cgc calculation predicts an increasingly suppressed cross section compared to its linear qcd counterpart . in fig .
[ fig:1 ] the solid line represents the result of geometric scaling parameterization , eq .
( [ resgs ] ) .
the suppression in the cross section is even higher than for the cgc phenomenological models reaching a factor two compared the the linear qcd approaches .
the present results demonstrate that the determination of @xmath33 can be useful to constrain the underlying qcd dynamics . in principle
, this cross section could be constrained at high energies by studying the ratio between quasi - horizontal deeply penetrating air showers and earth - skimming tau showers @xcite . as a summary ,
the detection of uhe neutrinos may shed light on the observation of air showers events with energies in excess of @xmath68 gev , reveal aspects of new physics as well as of the qcd dynamics at high energies .
one of the main ingredients for estimating event rates in neutrino telescopes and cosmic ray observatories ( e.g. auger ) is the neutrino - nucleon cross section .
we have examined to what extent the cross section is sensitive to the presence of new dynamical effects in the qcd evolution .
we compare the predictions of several approaches based on different assumptions for the qcd dynamics .
in particular , we have compared the phenomenological models based on nonlinear perturbative qcd with the usual nlo dglap evolution .
moreover , we compute a model independent cross section , using the experimental verification of the geometric scaling property .
the results demonstrate that the current theoretical uncertainty for the neutrino - nucleon cross section reaches a factor two or higher for neutrino energies above @xmath68 gev .
|
the ultrahigh energy neutrino cross section is a crucial ingredient in the calculation of the event rate in high energy neutrino telescopes .
currently there are several approaches which predict different behaviors for its magnitude for ultrahigh energies . in this contribution
is presented a summary of current predictions based on the non - linear qcd evolution equations , the so - called perturbative saturation physics . in particular ,
predictions are shown based on the parton saturation approaches and the consequences of geometric scaling property at high energies are discussed .
the scaling property allows an analytical computation of the neutrino scattering on nucleon / nucleus at high energies , providing a theoretical parameterization .
|
-5 mm we will survey recent progress and describe open problems in the area of accurate floating point computation , in particular for matrix computations .
a very short bibliography would include @xcite .
we consider the evaluation of multivariate rational functions @xmath8 of floating point numbers , and matrix computations on rational matrices @xmath9 , where each entry @xmath10 is such a rational function .
matrix computations will include computing determinants ( and other minors ) , linear equation solving , performing gaussian elimination ( ge ) with various kinds of pivoting , and computing the singular value decomposition ( svd ) , among others .
our goals are _ accuracy _ ( computing each solution component with tiny relative error ) and _ efficiency _ ( the algorithm should run in time bounded by a polynomial function of the input size ) .
we consider three models of arithmetic , defined in the abstract , and for each one we try to classify rational expressions and matrices as to whether they can be evaluated or factored accurately and efficiently ( we will say `` compute(d ) accurately and efficiently , '' or `` cae '' for short ) . in the traditional `` @xmath11 '' model ( tm ) , we have @xmath12 , @xmath13 and @xmath14 , where @xmath15 is called _ machine precision_. it is the conventional model for floating point error analysis , and means that every floating point result is computed with a relative error @xmath16 bounded in magnitude by @xmath17
. the values of @xmath16 may be arbitrary real ( or complex ) numbers satisfying @xmath18 , so that any algorithm proven to cae in the tm must work for arbitrary real ( or complex ) number inputs and arbitrary real ( or complex ) @xmath14 .
the size of the input in the tm is the number of floating point words needed to describe it , independent of @xmath17 .
the long exponent ( lem ) and short exponent ( sem ) models , which are implementable on a turing machine , make errors that may be described by the tm , but their inputs and @xmath16 s are much more constrained . also , we compute the size of the input in the lem and sem by counting the number of bits , so that higher precision and wider range take more bits .
this will mean that problems we can provably cae in the tm are a strict subset of those we can cae in the lem , which in turn we conjecture are a strict subset of those we can cae in the sem .
in all three models we will describe the classes of rational expressions and rational matrices in terms of the factorization properties of the expressions , or of the minors of the matrices .
the reader may wonder why we insist on accurately computing tiny quantities with small relative error , since in many cases the inputs are themselves uncertain , so that one could suspect that the inherent uncertainty in the input could make even the signs of tiny outputs uncertain .
it will turn out that in the tm , the class we can cae appears to be identical to the class where all the outputs are in fact accurately determined by the inputs , in the sense that small relative changes in the inputs cause small relative changes in the outputs .
we make this conjecture more precise in section 3 below .
there are many ways to formulate the search for efficient and accurate algorithms @xcite .
our approach differs in several ways .
in contrast to either conventional floating point error analysis @xcite or the model in @xcite , we ask that even the tiniest results have correct leading digits , and that zero be exact . in @xcite the model of arithmetic allows a tiny absolute error in each operation , whereas in tm we allow a tiny relative error .
unlike @xcite our lem and sem are conventional turing machine models , with numbers represented as bit strings , and so we can take the cost of arithmetic on very large and very small numbers ( i.e. those with many exponent bits ) into precise account . for these reasons
we believe our models are closer to computational practice than the model in @xcite .
in contrast to @xcite , we ( mostly ) consider the input as given exactly , rather than as a sequence of ever better approximations . finally , many of our algorithms could easily be modified to explicitly compute guaranteed interval bounds on the output @xcite .
-5 mm we show here how to reduce the question of accurate and efficient matrix computations to accurate and efficient rational expression evaluation .
the connection is elementary , except for the svd , which requires an algorithm from @xcite .
[ prop_caenecessity ] being able to cae the absolute value of the determinant @xmath19 is _ necessary _ to be able to cae the following matrix computations on @xmath9 : lu factorization ( with or without pivoting ) , qr factorization , all the eigenvalues @xmath20 of @xmath9 , and all the singular values of @xmath9 .
conversely , being able to cae _ all _ the minors of @xmath21 is _ sufficient _ to be able to cae the following matrix computations on @xmath9 : @xmath22 , lu factorization ( with or without pivoting ) , and the svd of @xmath9 .
this holds in any model of arithmetic .
* proof * first consider necessity .
@xmath19 may be written as the product of diagonal entries of the matrices @xmath23 , @xmath24 and @xmath25 in these factorizations , or as the product of eigenvalues or singular values .
if these entries or values can be cae , then so can their product in a straightforward way .
now consider sufficiency .
the statement about @xmath22 is just cramer s rule , which only needs @xmath26 different minors .
the statement about lu factorization depends on the fact that each nontrivial entry of @xmath23 and @xmath24 is a quotient of minors .
the svd is more difficult @xcite , and depends on the following two step algorithm : ( 1 ) compute a _
rank revealing _ decomposition @xmath27 where @xmath28 and @xmath29 are `` well - conditioned '' ( far from singular in the sense that @xmath30 is not too large ) and ( 2 ) use a bisection - like algorithm to compute the svd from @xmath31 .
we believe that computing @xmath32 is actually necessary , not just @xmath19 . the sufficiency proof can be extended to other matrix computations like the qr decomposition and pseudoinverse by considering minors of matrices like @xmath33}$ ] . furthermore ,
if we can cae the minors of @xmath34 , and @xmath35 and @xmath36 are well - conditioned , then we can still cae a number of matrix factorizations , like the svd .
the svd can be applied to get the eigendecomposition of symmetric matrices , but we know of no sufficient condition for the accurate and efficient calculation of eigenvalues of nonsymmetric matrices .
-5 mm we begin by giving examples of expressions and matrix computations that we can cae in the tm , and then discuss what we can not do .
the results will depend on details of the axioms we adopt , but for now we consider the minimal set of operations described in the abstract . as long as we only do _ admissible operations _ , namely multiplication , division , addition of like - signed quantities , and addition / subtraction of ( exact ! ) input data ( @xmath37 ) , then the worst case relative error only grows very slowly , roughly proportionally to the number of operations .
it is when we subtract two like - signed approximate quantities and significant cancellation occurs , that the relative error can become large .
so we may ask which problems we can cae just using only admissible operations , i.e. which rational expressions factor in such a way that only admissible operations are needed to evaluate them , and which matrices have all minors with the same property . here
are some examples , where we assume that the inputs are arbitrary real or complex numbers .
( 1 ) the determinant of a cauchy matrix @xmath38 is cae using the classical expression @xmath39 , as is every minor . in fact , changing one line of the classical ge routine will compute each entry of the lu decomposition accurately in about the same time as the original inaccurate version . ( 2 ) we can cae all minors of sparse matrices , i.e. those with certain entries fixed at 0 and the rest independent indeterminates @xmath40 , if and only if the undirected bipartite graph presenting the sparsity structure of the matrix is _ acyclic _ ; a one - line change to ge again renders it accurate . an important special case
are bidiagonal matrices , which arise in the conventional svd algorithm .
( 3 ) the eigenvalue problem for the second centered difference approximation to a sturm - liouville ode or elliptic pde on a rectangular grid ( with arbitrary rectilinear boundaries ) can be written as the svd of an `` unassembled '' problem @xmath41 where @xmath42 and @xmath43 are diagonal ( depending on `` masses '' and `` stiffnesses '' ) and @xmath24 is _ totally unimodular _ , i.e. all its minors are @xmath44 or 0 .
again , a simple change to ge renders it accurate .
in contrast , one can show that it is impossible in the tm to add @xmath7 accurately in constant time ; the proof involves showing that for _ any _ algorithm the rounding errors @xmath16 and inputs @xmath45 can be chosen to have an arbitrarily large relative error .
this depends on the @xmath16 s being permitted to be arbitrary real numbers in our model .
vandermonde matrices @xmath46 are more subtle .
since the product of a vandermonde matrix and the discrete fourier transform ( dft ) is cauchy , and we can compute the svd of a cauchy , we can compute the svd of a vandermonde .
this fits in our tm model because the roots of unity in the dft need only be known approximately , and so may be computed in the tm model . in contrast , one can use the result in the last paragraph to show that the inverse of a vandermonde can not be computed accurately .
similarly , polynomial vandermonde matrices with @xmath47 , @xmath48 a ( normalized ) orthogonal polynomial , also permit accurate svds , but probably not inverses .
-5 mm if we further restrict the domain of ( some ) inputs to be nonnegative , then much more is possible , @xmath7 as a trivial example .
a more interesting example are weakly diagonally dominant m - matrices , which arise as discretizations of pdes ; they must be represented as offdiagonal entries and the row sums .
more interesting is the class of _ totally positive ( tp ) matrices _ , all of whose minors are positive .
numerous structure theorems show how to represent such matrices as products of much simpler tp matrices .
accurate formulas for the ( nonnegative ) minors of these simpler matrices combined with the cauchy - binet theorem yield accurate formulas for the minors of the original tp matrix , but typically at an exponential cost . an important class of tp matrices where we can do much better are the tp generalized vandermonde matrices @xmath49 , where the @xmath50 form an increasing nonnegative sequence of integers
. @xmath51 is known to be the product of @xmath52 and a _ schur function _
@xcite @xmath53 , where the sequence @xmath54 is called a _
partition_. schur functions are polynomials with nonnegative integer coefficients , so since their arguments @xmath55 are nonnegative , they can certainly be computed accurately .
however , straightforward evaluation would have an exponential cost @xmath56 , @xmath57 .
but by exploiting combinatorial identities satisfied by schur functions along with techniques of divide - and - conquer and memoization , the cost of evaluating the determinant can be reduced to polynomial time @xmath58 .
the cost of arbitrary minors and the svd remains exponential at this time .
note that the @xmath20 are counted as part of the size of the input in this case . here is our conjecture generalizing all the cases we have studied in the tm .
we suppose that @xmath59 is a homogeneous polynomial , to be evaluated on a domain @xmath60 .
we assume that @xmath61 , to avoid pathological domains .
typical domains could be all tuples of the real or complex numbers , or the positive orthant .
we say that @xmath6 satisfies condition @xmath62 ( for _ accurate _ ) if @xmath6 can be written as a product @xmath63 where each factor @xmath64 satisfies * @xmath64 is of the form @xmath55 , @xmath65 or @xmath66 , or * @xmath67 is bounded away from 0 on @xmath60 .
let @xmath6 and @xmath60 be as above . then condition @xmath62 is a necessary and sufficient condition for the existence of an algorithm in the tm model to compute @xmath6 accurately on @xmath60 .
note that we make no claims that @xmath6 can be evaluated efficiently ; there are numerous examples where we only know exponential - time algorithms ( doing ge with complete pivoting on a totally positive generalized vandermonde matrix ) .
-5 mm so far we have considered the simplest version of the tm , where ( 1 ) we have only the input data , and no additional constants available , ( not even integers , let alone arbitrary rationals or reals ) , ( 2 ) the input data is given exactly ( as opposed to within a factor of @xmath68 ) , and ( 3 ) there is no way to `` round '' a real number to an integer , and so convert the problem to the lem or sem models .
we note that in @xcite , ( 1 ) integers are available , ( 2 ) the input is rounded , and ( 3 ) there is no way to `` round '' to an integer .
changes to these model assumptions will affect the classes of problems we can solve .
for example , if we ( quite reasonably ) were to permit exact integers as input , then we could cae expressions like @xmath69 , and otherwise presumably not .
if we went further and permitted exact rational numbers , then we could also cae @xmath70 .
allowing algebraic numbers would make @xmath71 cae .
if inputs were not given exactly , but rather first multiplied by a factor @xmath68 , then we could no longer accurately compute @xmath37 where @xmath72 and @xmath73 are inputs , eliminating cauchy matrices and most others .
but the problems we could solve with exact inputs in the tm still have an attractive property with inexact inputs : small relative changes in the inputs cause only a small relative change in the outputs , independent of their magnitudes .
the output relative errors may be larger than the input relative error by a factor called a _ relative condition number _ @xmath74 , which is at most a polynomial function of @xmath75 . here
@xmath76 is the _ relative gap _ between inputs @xmath55 and
@xmath77 , and the maximum is taken over all expressions @xmath78 where appearing in @xmath63 .
so if all the input differ in several of their leading digits , all the leading digits of the outputs are determined accurately .
we note that @xmath74 can be large , depending on @xmath6 and @xmath60 , but it can only be unbounded when a relative gap goes to zero .
if a problem has this attractive property , we say that it possesses a relative perturbation theory . in practical situations ,
where only a few leading digits of the inputs @xmath55 are known , this property justifies the use of algorithms that try to compute the output as accurately as we do .
we state a conjecture very much like the last one about when a relative perturbation theory exists .
let @xmath6 and @xmath60 be as in the last conjecture . then condition @xmath62 is a necessary and sufficient condition for @xmath6 to have a relative perturbation theory .
-5 mm now we consider standard turing machines , where input floating point numbers @xmath3 are stored as the pair of integers @xmath4 , so the size of @xmath72 is @xmath79 .
we distinguish two cases , the long exponent model ( lem ) where @xmath6 and @xmath5 may each be arbitrary integers , and the short exponent model ( sem ) , where the length of @xmath5 is bounded depending on the length of @xmath6 . in the simplest case , when @xmath80 ( or lies in a fixed range )
then the sem is equivalent to taking integer inputs , where the complexity of problems is well understood .
this is more generally the case if @xmath81 grows no faster than a polynomial function of @xmath82 .
in particular it is possible to cae the determinant of an integer ( or sem ) matrix each of whose entries is an independent floating point number @xcite .
this is not possible as far as we know in the lem , which accounts for a large complexity gap between the two models .
we start by illustrating some differences between the lem and sem , and then describe the class of problems that we can cae in the lem .
first , consider the number of bits in an expression with lem inputs can be exponentially larger than the number of bits in the same expression when evaluated with sem inputs .
for example , @xmath83 when @xmath72 and @xmath73 are integers , but @xmath84 when @xmath72 and @xmath73 are lem numbers : @xmath85 has up to @xmath86 different bit positions to store , each @xmath87 , not @xmath88 . in other words
, lem arithmetic can encode symbolic algebra , because if @xmath89 and @xmath90 have no overlapping bits , then we can recover @xmath89 and @xmath90 from the product @xmath91 .
second , the error of many conventional matrix algorithms is typically proportional to the condition number @xmath92 .
this means that a conventional algorithm run with @xmath93 extra bits of precision will compute an accurate answer .
it turns out that if @xmath9 has rational entries in the sem model , then @xmath94 is at most a polynomial function of the input size , so conventional algorithms run in high precision will cae the answer .
however @xmath95 for lem matrices can be exponentially larger , so this approach does not work .
the simplest example is @xmath96 . on the other hand
@xmath97 is a lower bound on the complexity of any algorithm , because this is a lower bound on the number of exponent bits in the answer .
one can show that @xmath97 grows at most polynomially large in the size of the input .
finally , we consider the problem of computing an arbitrary bit in the simple expression @xmath98 .
when the @xmath55 are in the sem , then @xmath99 can be computed exactly in polynomial time .
however when the @xmath55 are in the lem , then one can prove that computing an arbitrary bit of @xmath99 is as hard as computing the permanent , a well - known combinatorially difficult problem .
here is another apparently simple problem not known to even be in np : testing singularity of a floating point matrix . in the sem , we can cae the determinant .
but in the lem , the obvious choice of a `` witness '' for singularity , a null vector , can have exponentially many bits in it , even if the matrix is just tridiagonal .
we conjecture that deciding singularity of an lem matrix is np - hard .
so how do we compute efficiently in the lem ?
the idea is to use _ sparse arithmetic _ , or to represent only the nonzero bits in the number .
( a long string of 1s can be represented as the difference of two powers of 2 and similarly compressed ) . in contrast ,
in the sem one uses _ dense arithmetic _ , storing all fraction bits of a number .
for example , in sparse arithmetic @xmath100 takes @xmath101 bits to store in sparse arithmetic , but @xmath5 bits in dense arithmetic .
this idea is exploited in practical floating point computation , where extra precise numbers are stored as arrays of conventional floating point numbers , with possibly widely different exponents @xcite .
now we describe the class of rational functions that we can cae in the lem .
we say the rational function @xmath8 is in factored form if @xmath102 , where each @xmath103 is an integer , and @xmath104 is written as an explicit sum of nonzero monomials .
we say @xmath105 is the number of bits needed to represent it in factored form .
then by ( 1 ) computing each monomial in each @xmath106 exactly , ( 2 ) computing the leading bits of their sum @xmath106 using sparse arithmetic ( the cost is basically sorting the bits ) , and ( 3 ) computing the leading bits of the product of the @xmath107 by conventional rounded multiplication or division , one can evaluate @xmath8 accurately in time a polynomial in @xmath105 and @xmath108 .
in other words , the class of rational expression that we can cae are those that we can express in factored form in polynomial space . now we consider matrix computations .
it follows from the last paragraph that if each minor @xmath8 of @xmath9 can be written in factored form of a size polynomial in the size of @xmath9 , then we can cae all the matrix computations that depend on minors .
so the question is which matrix classes @xmath9 have all their minors ( or just the ones needed for a particular matrix factorization ) expressible in a factored form no more than polynomially larger than the size of @xmath9 .
the obvious way to write @xmath8 , with the laplace expansion , is clearly exponentially larger than @xmath9 , so it is only specially structured @xmath9 that will work .
all the matrices that we could cae in the tm are also possible in the lem .
the most obvious classes of @xmath9 that we can cae in the lem that were impossible in the tm are gotten by replacing all the indeterminates in the tm examples by arbitrary rational expressions of polynomial size .
for example , the entries of an m - matrix can be polynomial - sized rational expressions in other quantities .
another class are green s matrices ( inverses of tridiagonals ) , which can be thought of as discretized integral operators , with entries written as @xmath109 .
the obvious question is whether @xmath110 each of whose entries is an independent number in the lem falls in this class .
we conjecture that it does not , as mentioned before .
-5 mm our goal has been to identify rational expressions ( or matrices ) that we can evaluate accurately ( or on which we can perform accurate matrix computations ) , in polynomial time . accurately means that we want to get a relative error less than 1 , and polynomial time means in a time bounded by a polynomial function of the input size .
we have defined three reasonable models of arithmetic , the traditional model ( tm ) , the long exponent model ( lem ) and the short exponent model ( sem ) , and tried to identify the classes of problems that can or can not be computed accurately and efficiently for each model .
the tm can be used as a model to do proofs that also hold in the implementable lem and sem , but since it ignores the structure of floating point numbers as stored in the computer , it is strictly weaker than either the lem or sem . in other words , there are problems ( like adding @xmath7 ) that are provably impossible in the tm but straightforward in the other two models .
we also believe that the lem is strictly weaker than the sem , in the sense that there appear to be computations ( like computing the determinant of a general , or even tridiagonal , matrix ) that are possible in polynomial time in the sem but not in the lem . in the sem ,
essentially all problems that can be written down in polynomial space can be solved in polynomial time . for the lem ,
only expressions that can be written in _ factored form _ in polynomial space can be computed efficiently in polynomial time .
a number of open problems and conjectures were mentioned in the paper .
we mention just one additional one here : what can be said about the nonsymmetric eigenvalue problem ?
in other words , what matrix properties , perhaps related to minors , guarantee that all eigenvalues of a nonsymmetric matrix can be computed accurately ?
* acknowledgements * the author acknowledges benjamin diament , zlatko drma , stan eisenstat , ming gu , william kahan , ivan slapniar , kresimir veselic , and especially plamen koev for their collaboration over many years in developing this material .
j. demmel and p. koev .
necessary and sufficient conditions for accurate and efficient singular value decompositions of structured matrices . in v. olshevsky , editor , _ special issue on structured matrices in mathematics , computer science and engineering _ , volume 281 of _ contemporary mathematics _ , 117145 , ams , 2001 .
d. priest .
algorithms for arbitrary precision floating point arithmetic . in p.
kornerup and d. matula , editors , _ proceedings of the 10th symposium on computer arithmetic _ , 132145 , grenoble , france , june 26 - 28 , 1991 .
ieee computer society press .
|
our goal is to find accurate and efficient algorithms , when they exist , for evaluating rational expressions containing floating point numbers , and for computing matrix factorizations ( like lu and the svd ) of matrices with rational expressions as entries .
more precisely , _ accuracy _ means the relative error in the output must be less than one ( no matter how tiny the output is ) , and _ efficiency _
means that the algorithm runs in polynomial time .
our goal is challenging because our accuracy demand is much stricter than usual .
the classes of floating point expressions or matrices that we can accurately and efficiently evaluate or factor depend strongly on our model of arithmetic : 1 . in the `` traditional model '' ( tm ) ,
the floating point result of an operation like @xmath0 is @xmath1 , where @xmath2 must be tiny .
2 . in the `` long exponent model '' ( lem )
each floating point number @xmath3 is represented by the pair of integers @xmath4 , and there is no bound on the sizes of the exponents @xmath5 in the input data .
the lem supports strictly larger classes of expressions or matrices than the tm .
3 . in the `` short exponent model '' ( sem ) each floating point number
@xmath3 is also represented by @xmath4 , but the input exponent sizes are bounded in terms of the sizes of the input fractions @xmath6 .
we believe the sem supports strictly more expressions or matrices than the lem .
these classes will be described by factorizability properties of the rational expressions , or of the minors of the rational matrices . for each such class , we identify new algorithms that attain our goals of accuracy and efficiency .
these algorithms are often exponentially faster than prior algorithms , which would simply use a conventional algorithm with sufficiently high precision . for example , we can factorize cauchy matrices , vandermonde matrices , totally positive generalized vandermonde matrices , and suitably discretized differential and integral operators in all three models much more accurately and efficiently than before .
but we provably can not add @xmath7 accurately in the tm , even though it is easy in the other models .
* 2000 mathematics subject classification : * 65f , 65g50 , 65y20 , 68q25 . * keywords and phrases : * roundoff , numerical linear algebra , complexity .
|
the late time evolution of stars influenced by strong gravitational pull has been a largely anticipated field in astrophysics and gravitational theories .
it facilitates in investigations of diverse characteristics of gravitating sources via various physical phenomenons .
baade and zwicky ( 1934 ) proposed the inception of massive compact stellar objects , establishing the argument that supernova may result in a small and super dense star .
this eventually came in to reality in 1967 , when bell and hewish ( longair 1994 , ghosh 2007 ) discovered pulsars that are highly magnetized and rotating neutron stars .
so , in reality , we come across a fundamental revolutionary shift from normal stars to compact stars , with a wide range in the form of stars to neutron stars , quarks , dark stars , gravastars and finally black holes
. in our present work , we are more specifically interested in study of compact stars category .
generally the homogeneity of the spherically symmetric matter configuration is emphasized while theoretical modeling of a compact star , satisfying the tolman oppenheimer - volkoff ( tov ) equation .
ruderman ( 1972 ) was the first one to argue that the nuclear matter density becomes anisotropic at the core of compact object .
analytic solutions of the field equations for various static spherically symmetric configurations for anisotropic compatible to interior of compact stellar modeling have been obtained in numerous works ( maurya and gupta 2012,2013 2014 , maharaj et al.2014 , pant et al .
2014a , 2014b ) .
the pressure inside the fluid sphere disintegrates in two parts , namely radial and tangential pressure .
it has been investigated that anisotropy gave rise to the repulsive force that assists to construct the compact objects . in this context , it is established in ( kamal et al.2012 ) that the krori and barua ( henceforth kb)(1975 ) metric provides an effective and realistic approach in modeling of compact stars .
the numerical simulations can be taken into account to explore the characteristics of compact stars from integrated tov equations , if equation of state ( eos ) is known .
rahaman et al .
( 2012 ) extended the kb models by using the chaplygin gas eos and discussed their physical features .
mak and harko ( 2004 ) established standard models of spherically symmetric compact stars via exact solution of the field equations .
they determined the impressions of physical parameters such as energy density , tangential and radial pressure , concluding that these parameters remain finite and positive inside the stars .
the anisotropic exact models for compact objects with a barotropic eos are discussed in ( 2006 ) .
hossein et al.(2012 ) studied the impact of cosmological constant on anisotropic compact stars .
general relativity ( gr ) being fundamental theory of gravity is successful in weak field limit , but insufficient to explore the strong field .
the expected description of gr in strong field regime can be done by its modifications .
the modifications in einstein - hilbert ( eh ) action are induced to arrive at alternative theories of gravity . among modified theories of gravity @xmath7 gravity being most elementary modification of gr
is extensively studied in context of existence and stability of neutron stars and compact stars ( arapoglu et al.2012 , alavirad and weller 2013 , astashenok et al . 2014 , 2015 , yazadjiev et al .
2014 , kausar and noureen 2014 , noureen et al.2015 ) .
abbas and his collaborators ( 2014 , 2015a , 2015b , 2015c ) analyzed a class compact stars in gr , @xmath8,(where @xmath2 is torsion scalar ) with different equation of state .
the issue of accelerated expansion of the universe can be explained by taking into account the modified theories of gravity such as @xmath0 gravity ( harko et al.2011 ) .
the @xmath0 gravity provides an alternative way to explain the current cosmic acceleration with no need of introducing either the existence of extra spatial dimension or an exotic component of dark energy .
harko et al.(2011 ) generalized @xmath7 gravity by introducing an arbitrary function of the ricci scalar @xmath1 and the trace of the energy - momentum tensor @xmath2 .
the dependence of @xmath2 may be introduced by exotic imperfect fluids or quantum effects ( conformal anomaly ) . as a result of coupling between matter and geometry , motion of test particles is nongeodesic and an extra acceleration is always present . in @xmath0 gravity ,
cosmic acceleration may result not only due to geometrical contribution to the total cosmic energy density but it also depends on matter contents .
soon after the origination of @xmath0 gravity , its cosmological and thermodynamic implications including the energy conditions and dynamical analysis were extensively discussed ( shabani and farhoudi 2014 , harko and lobo 2010 , harko 2010 , azizi 2013 , sharif and zubair 2012a , 2012b , 2013 , jamil et al .
2012a , 2012b , momeni et al .
2015a , 2015b , momeni and myrzakulov 2015 , barrientos and rubilar 2014 ) . however , the explorations regarding compact stars in @xmath0 gravity are yet to be done . herein , we are interested to study the structure of a class of compact stars model 1 , model 2 and model 3 in @xmath0 gravity .
the modified action in @xmath0 is as follows(harko et al .
2011 ) @xmath9,\ ] ] where @xmath10 is matter lagrangian and @xmath11 denote the metric tensor .
different choices of @xmath10 can be considered , each of which directs to a specific form of fluid . in our present work
the viable @xmath6 model we have chosen is of following type @xmath12 where @xmath13 is a function of ricci scalar and @xmath14 is some positive constant value .
we have analyzed above mentioned @xmath6 model with @xmath15 , @xmath16 being a positive scalar . in our discussion , we have considered the model @xmath17 which does not imply the direct non - minimal gravitational coupling between scalar curvature @xmath1 and trace of the energy - momentum tensor @xmath2 in the lagrangian level .
but there may be coupling between matter and geometry which becomes apparent in the study of thermodynamics ( sharif and zubair 2012a ) .
the @xmath18 can be considered as matter correction term to @xmath7 gravity and in this particular study we choose @xmath19 .
the main reason behind the difference on cosmology in ordinary @xmath7 gravity and in the above @xmath0 model is the non - trivial coupling between matter and geometry . in our previous work ( sharif and zubair 2012b )
, we have reconstructed some explicit models of this @xmath0 gravity for anisotropic universe and explored the phantom era of dark energy .
we have investigated that only for @xmath20 and @xmath21 , the energy density of all the considered models remains positive and energy conditions are only valid for these values of @xmath14 and @xmath16 .
so , through out the analysis , we have used the above values of @xmath14 and @xmath16 , it will not be mentioned again explicitly .
this paper is arranged as : in the following section anisotropic matter distribution and expressions physical parameters for energy density @xmath22 , radial and tangential pressure are established .
section * 3 * constitutes the analysis of physical features of compact stars and their stability analysis . in the last section ,
we have summarized our results .
the line element for particular spherically symmetric metric describing the compact star stellar configuration is @xmath23 where @xmath24 , @xmath25 ( krori and barua 1975 ) , @xmath26 , @xmath27 and @xmath28 are arbitrary constant that will be calculated by using some physical assumptions .
the above set of functions are introduced to arrive at singularity free structure for compact star .
clearly , this set of functions leads to non singular density and curvature setting . taking @xmath29 and upon variation of modified eh action in @xmath0 ( [ 1 ] ) with respect to metric tensor @xmath30 ,
the following modified field equations are formed as @xmath31,\end{aligned}\ ] ] where @xmath32 , @xmath33 and @xmath34 denotes the usual matter energy momentum tensor that is considered to be anisotropic , is given by @xmath35 where @xmath22 , @xmath36 and @xmath37 denote energy density , radial and transverse stresses respectively .
the four velocity is denoted by @xmath38 and @xmath39 to be the radial four vector satisfying @xmath40 when @xmath41 , the expression for @xmath22 , @xmath36 and @xmath37 can be extracted from modified field equations as follows @xmath42 , \\\nonumber p_r&= & \frac{-1}{2(1 + 2\lambda)}\left[\frac{\lambda}{e^b(1+\lambda ) } \left\{\left(\frac{a'}{r}-\frac{a'b'}{4}+\frac{a''}{2 } + \frac{a'^2}{4}\right)f_{1r}- f''_{1r}+\left(\frac{b'}{2}-\frac{f_1}{2}e^b\right.\right.\right.\\\nonumber&&\left.\left.\left .
-\frac{2}{r}\right)f'_{1r}\right\}- \frac{(2 + 3\lambda)}{e^b(1+\lambda)}\left\{\left(\frac{a'b'}{4}+\frac{b'}{r}-\frac{a''}{2 } -\frac{a'^2}{4}\right)f_{1r } + \left(\frac{a'}{2}+\frac{2}{r}\right)f'_{1r}\right.\right.\\\nonumber&&\left.\left.+\frac{f_1}{2}e^b\right\}+ \frac{2\lambda}{e^b(1+\lambda)}\left\{\frac{f_{1r}}{r^2}\left(\frac{(b'-a')r}{2}-e^b+1\right ) + \left(\frac{a'-b'}{2}+\frac{1}{r}\right)f'_{1r}\right.\right.\\\label{8}&&\left.\left.+f''_{1r } + \frac{f_1}{2}e^b\right\}\right ] , \\\nonumber p_t&= & \frac{-1}{2(1 + 2\lambda)}\left[\frac{\lambda}{e^b(1+\lambda ) } \left\{\left(\frac{a'}{r}-\frac{a'b'}{4}+\frac{a''}{2 } + \frac{a'^2}{4}\right)f_{1r}- f''_{1r}+\left(\frac{b'}{2}-\frac{f_1}{2}e^b\right.\right.\right.\\\nonumber&&\left.\left.\left
. -\frac{2}{r}\right)f'_{1r}\right\}- \frac{\lambda}{e^b(1+\lambda)}\left\{\left(\frac{a'b'}{4}+\frac{b'}{r}-\frac{a''}{2 } -\frac{a'^2}{4}\right)f_{1r } + \left(\frac{a'}{2}+\frac{2}{r}\right)f'_{1r}\right.\right.\\\nonumber&&\left.\left .
+ \frac{f_1}{2}e^b\right\}- \frac{2}{e^b}\left\{\frac{f_{1r}}{r^2}\left(\frac{(b'-a')r}{2}-e^b+1\right ) + \left(\frac{a'-b'}{2}+\frac{1}{r}\right)f'_{1r}\right.\right.\\\label{9}&&\left.\left.+f''_{1r } + \frac{f_1}{2}e^b\right\}\right].\end{aligned}\ ] ] here @xmath43 and prime denotes the derivatives with respect to radial coordinate . substituting eq.([2 ] ) in eqs.([7])-([9 ] ) and inserting value of ricci scalar in the form of metric functions together with the kb metric coefficients , we arrive at @xmath44 , \\\nonumber p_r&=&\frac{1}{r^4(1 + 3\lambda+2\lambda^2)}\left[e^{-2ar^2}\left\{e^{2ar^2}(r^2 + 2\alpha ( \lambda-1)+e^{ar^2}(4\alpha(\lambda-1)\right.\right.\\\nonumber&&\left.\left .
+ abr^6\lambda+r^2(1 + 8b\alpha ( \lambda-1)-8a\alpha\lambda)+br^4(2+\lambda-4\alpha\lambda(2a - b ) ) ) \right.\right.\\\nonumber&&\left.\left.- 2\alpha(1 + 3br^2+b^3r^4-ar^2(2+br^2))(-1+br^2(\lambda-1)-ar^2(2(1 \right.\right.\\\label{11}&&\left.\left.+\lambda ) + br^2(1 + 3\lambda)))\right\}\right ] , \\\nonumber p_t&=&\frac{1}{r^4(1 + 3\lambda+2\lambda^2)}\left[e^{-2ar^2}\left\{2\alpha(1 + 3br^2+b^2r^4 -ar^2(2+br^2))(-3\right.\right.\\\nonumber&&\left.\left .
-b^2r^4 - 2\lambda+br^2(ar^2 - 1)(1 + 2\lambda))+e^{2ar^2}(r^2(2+\lambda)-2\alpha ( 3 + 2\lambda))\right.\right.\\\nonumber&&\left.\left .
+ e^{ar^2}(4\alpha(3 + 2\lambda)+br^6(a - b+a\lambda)-r^2(2+\lambda+4a\alpha(3 + 2\lambda ) \right.\right.\\\nonumber&&\left.\left.-4b\alpha(5 + 4\lambda))+r^4(b(4b\alpha-1 ) ( 2+\lambda)+a(1 - 8b\alpha(1+\lambda ) ) ) ) \right\}\right].\\\label{12}\end{aligned}\ ] ] the schwarzschild solution is considered to be most suitable choice for matching conditions in exterior regime ( noureen and zubair 2015 , 2014 , goswami et al.2014 cooney et al.2010 , ganguly et al .
the interior metric of the boundary surface will be the same for the internal or external geometry of the star .
herein , the exterior metric of the star is described by the schwarzschild solution , given by @xmath45 smooth matching of the interior metric ( [ 6 ] ) to the vacuum exterior solution at the boundary surface @xmath46 , yield @xmath47 where superscript @xmath48 and @xmath49 stands for interior and exterior solutions .
matching of the interior and exterior spacetime leads to following @xmath50 the values of the constants @xmath26 and @xmath27 are evaluated by using approximate values of @xmath51 and @xmath1 ( lattimer and steiner 2014 , li et al.1999 ) of the considered compact stars , provided in the table * 1 * . the compactness of a star can be defined by @xmath52 and surface redshift @xmath53 can be determined by using the result @xmath54 .
the values of @xmath53 for the considered models have been given in table * 1*. .approximate values of the model parameters for considered compact stars [ cols="^,^,^,^,^,^,^",options="header " , ]
this section covers the physical constraints required for interior solution , incorporating anisotropic behavior , matching and energy conditions together with the stability analysis of considered compact stars .
prior to the discussion of anisotropy measure , we discuss the evolution of energy density @xmath22 and anisotropic stresses @xmath36 and @xmath37 respectively , shown in figures * 1 - 3*. radial derivative of equation ( [ 7 ] ) leads to the following expression @xmath55 ) + 2 a b r^3 ( 1 + 3 \lambda ) + 2 a r ( -2 + b r^2 ) ( 1 + 3 \lambda ) + ( 3 \\\nonumber&&+ 7 \lambda)4 b^2 r^3 ) + 2 ( -6 b r + 2 a b r^3 - 4 b^2 r^3 + 2 a r ( 2 + b r^2 ) ) \lambda ( 1 - \lambda - 3 b r^2 ( 1 \\\nonumber&&+ 3 \lambda ) + a r^2 ( -2 + b r^2 ) ( 1 + 3 \lambda ) + b^2 r^4 ( 3 + 7 \lambda ) ) - e^{a r^2 } ( 2 r ( 1 + 24 b \alpha \lambda + ( 1 \\\nonumber&&+ \lambda)8 a \lambda ) + 6 b r^5 ( a \lambda + b ( 2 + 4 \lambda ) ) - 4 r^3 ( a ( 2 + ( 4 + 8 b \alpha ) \lambda ) + b ( 3 \lambda + ( 2 \\\nonumber&&+ 3 \lambda)4 b \alpha ) ) ) - 2 a e^{a r^2 } r ( 4 \alpha ( -1 + \lambda ) + r^2 ( 1 + 24 b \alpha \lambda + 8 a \alpha ( 1 + \lambda ) ) + b r^6 \\\nonumber&&\times(a \lambda + b ( 2 + 4 \lambda ) ) - r^4 ( a ( 2 + ( 4 + 8 b \alpha ) \lambda ) + b ( 3 \lambda + 4 b \alpha ( 2 + 3 \lambda)))))<0,\\\label{15}\end{aligned}\ ] ] and also eq.([8 ] ) reveals that @xmath56 , depicting decrease in energy density and radial pressure with increasing radius of the compact object , well supported by the results shown in figures * 4 * and * 5*. maximality of central density and pressure is achievable at @xmath57 , indicating @xmath58 using the eos @xmath59 and @xmath60 , we obtain the following form of eos parameters @xmath61 ( 2 + 3\lambda))))),\\\nonumber \omega_t&=&(-2\alpha(-1 - 3br^2-b^2r^4+ar^2(2+br^2))(-3-b^2r^4 - 2\lambda + br^2(-1\\\nonumber&&+ar^2)(1 + 2\lambda))+e^{2ar^2}(r^2(2+\lambda ) -2\alpha(3 + 2\lambda))+e^{ar^2}(4\alpha(3 + 2\lambda)\\\nonumber & & + br^6(a - b+a\lambda)-r^2(2+\lambda+4a\alpha(3 + 2\lambda)-4b\alpha ( 5 + 4\lambda))+r^4(b(-1\\\nonumber&&+4b\alpha)(2+\lambda)+a(1 -8b\alpha(1+\lambda)))))/(e^{2ar^2}(r^2 + 2\alpha(-1+\lambda ) ) + 2\alpha(-1\\\nonumber&&-3br^2-b^2r^4+ar^2(2+br^2))(1-\lambda -3br^2(1 + 3\lambda)+ar^2(-2+br^2)(1\\\nonumber&&+3\lambda)+b^2r^4 ( 3 + 7\lambda))-e^(ar^2)(4\alpha(-1+\lambda)+r^2(1 + 24b\alpha\lambda ] + 8a\alpha(1+\lambda))\\\nonumber&&+br^6(a\lambda+b(2 + 4\lambda ) ) -r^4(a(2+(4 + 8b\alpha)\lambda)+b(3\lambda+4b\alpha(2 + 3\lambda))))).\\\label{14}\end{aligned}\ ] ] it is interesting to mention here that the eos parameters depend on radius rather than a constant quantity as in ordinary matter distribution . this non - constant behavior of eos parameters is constituted by usual matter and exotic matter contributions ( see fig . * 6 * and * 7 * ) .
+ the measure of anisotropy @xmath62 , for considered @xmath0 model takes following form @xmath63.\end{aligned}\ ] ] figure * 8 * describes the evolution of anisotropy measure , @xmath64 i.e. , @xmath65 corresponds to the outward drawn anisotropy and its directed inward when @xmath66 . in our model , we find that @xmath65 for the different compact stars as shown in figure * 8*. it can be seen from fig .
* 8 * that @xmath65 at most of the points for compact stars model 1 and model 3 , indicating that that a repulsive anisotropic force occurs , allowing the construction of more massive distributions . in our model
anisotropy measure for model 2 decreases with the increase in radius and becomes negative beyond @xmath67 .
energy bounds are of significant importance including null energy condition ( nec ) , weak energy condition ( wec ) , strong energy condition ( sec ) and dominant energy condition ( dec ) , defined as @xmath68 the considered anisotropic sphere satisfy the energy conditions , exhibited graphically in figure * 9 * for compact star model 1 .
the radial and transverse sound speeds denoted by @xmath69 and @xmath70 should be less than speed of light i.e. , @xmath71 , @xmath72 , where @xmath73 and @xmath74 .
we plot the evolution of radial and transverse sound speeds for compact stars and found that above mentioned conditions hold , as shown in fig . *
10 * and * 11*. one can undergo with the stability analysis of compact objects considering sound speeds ( herrera 1992 , herrera and barreto 2013 , herrera and santos 1997 , herrera et al.2008 ) .
the the potentially stable and unstable regimes can be estimated by considering the difference of the sound propagation within the matter distribution .
the variation of @xmath75 of different strange stars is shown in figure 12 .
figure 12 shows that difference of the two sound speeds , i.e. , @xmath75 retain similar sign within the specific configuration and it satisfies the inequality @xmath76 .
thus , our proposed compact stars models are stable .
the modified theory of gravity can provide the explanation of accelerated expansion of universe .
the modified gravity theories ( such as the f(r , t ) gravity ) have been the center of attention for many researchers in the recent past , because this type of theories seems to provide a viable explanation for dark energy .
the conformal relation of @xmath0 to general theory of relativity with a self - interacting scalar field has been examined ( astashenok et al .
2014 ) .
this paper deals with the interior solutions for anisotropic fluids , which have been used to model compact stars in the context of modified gravity theory @xmath0 .
the modeling has been completed by taking compact stars as anisotropic in @xmath0 gravity . for the exact solution of the governing differential equations , we have used the krori and barua ( 1975 ) form of the metric function , i.e. , @xmath24 , @xmath25 , @xmath26 , @xmath27 and @xmath28 are arbitrary constant that have been calculated by using some matching conditions .
the smooth matching of interior and exterior regions of a star lead us to express the unknown constants in terms of masses and radii of the compact stars . using the observed values of the masses and radii of the compact stars ,
we get the values of the model constants that are used to discuss the nature of the stars . for the calculated values of the constant
, we found that the energy density , radial and transverse pressure decreases for given class of compact strange stars .
for the particular choice @xmath0 , we have found that eos parameters behave like normal matter distribution . on the basis of this fact
, we may conclude that compact stars are composed of ordinary matter and effect of @xmath0 term .
the regularity analysis of the proposed model implies that density and pressure are regular every where and attain the maximum value at the center . thus radial pressure@xmath36 and matter density @xmath22 have maximum values at the center and it decreases from the center to the surface of the star , where density becomes constant and pressure reduces to zero .
it has been found that the anisotropy will be directed outward when @xmath64 this implies that @xmath65 .
we have found that @xmath65 for compact stars model 1 and model 3 , indicating that that a repulsive anisotropic force occurs , allowing the construction of more massive distributions .
while the anisotropy measure for model 2 decreases with the increase in radius and becomes negative beyond @xmath67 .
the the potentially stable and unstable regimes have been estimated by considering the difference of the sound propagation within the matter distribution .
the compact stars remain potentially stable in the regions where difference of radial and sound speeds remain positive .
it is found that @xmath76 for all three considered compact stars .
so , our considered model is potentially stable in @xmath0 gravity , as shown in fig . * 12*. we would like to mention that tov equations for modified theories of gravity like @xmath77 ( momeni and myrzakulov 2015 ) , non - local @xmath7 gravity ( momeni et al . 2015 ) and @xmath0 ( moraes et al.2015 ) have been studied analytically and numerically .
authors declare that there is no conflict of interest regarding the publication of this paper .
we are thankful to the unknown referee for his / her deep study of the paper and fruitful comments , that have improved the paper a lot . 40 arapoglu , s. , deliduman , c. and eksi , k.y . :
jcap * 07*,020(2011 ) + azizi , t. : int .
phys . * 52*(2013)3486.// alavirad , h. and weller , j.m .
d * 88*,124034(2013 ) + astashenok , a.v . , capozziello , s. and odintsov , s.d .
d * 89 * , 103509(2014 ) + astashenok , a.v . , capozziello , s. and odintsov , s.d .
: jcap * 01 * , 001(2015 ) .
+ abbas , g. , nazeer , s. , meraj , m.a .
: astrophys .
space sci .
354 , 449 ( 2014 ) + abbas , g. , kanwal , a. , zubair , m. : astrophys .
space sci .
357 , 109 ( 2015a ) + abbas , g. , et al .
: astrophys .
space sci .
357 , 158 ( 2015b ) + abbas , g. , qaisar , s. , meraj , m.a .
: astrophys .
space sci .
357 , 156(2015c ) .
+ baade w and zwicky , f. : physical review * 46*,76(1934 ) + barrientos o.j . ,
rubilar , f. g. : phys .
d 90(2014)028501 cooney , a. et al .
d * 83*,064033(2010 ) + goswami , r. , nzioki , a.m. , maharaj , s.d . and ghosh s.g .
d * 90*,084011(2014 ) + ganguly , a. et al .
d * 89 * , 064019(2014 ) + ghosh , p. : rotation and accretion powered pulsars ( world scientific , 2007 ) + gokhroo , m.k . and
mehra , a.l .
. grav . * 26*,75(1994 ) + herrera , l. : phys .
a 165 , 206 ( 1992 ) + herrera , l. , barreto , w. : phys . rev .
d 88 , 084022 ( 2013 ) + herrera , l. , santos , n.o .
286 , 53 ( 1997 ) + herrera , l. , ospino , j. , di prisco , a. : phys . rev .
d 77 , 027502 ( 2008 ) + harko , t. , lobo , f.s.n . , nojiri , s. and odinstov , s.d .
d * 84*(2011)024020 .
+ harko , t. and lobo , f.s.n .
j. c * 70*(2010)373 .
+ hossein s.k.m .
: et al , int . j. mod .
d * 21*(2012)1250088 + jamil , m. momeni , d. , myrzakulov , r. : chin .
* 29*(2012)109801 + jamil , m. momeni , d. , raza , m. and myrzakulov , r. : eur .
j. * c72*(2012)1999 + krori , k.d . and barua , j. : j. phys .
a. : math .
gen.*8*(1975)508 + kalam , m. et al .
j. c * 72*(2012)2248 + kausar , h.r . , and noureen , i. : eur .
j. c * 74 * , 2760(2014 ) + longair , m.s . :
high energy astrophysics ( cambridge university press , 1994 ) + lobo , f.s.n . : class .
. grav . * 23*,1525(2006 ) + mak , m.k . and harko , t. : int .
d * 13*,149(2004 ) + lattimer , j.m . and steiner , a.w .
: astrophys .
j. * 784*(2014)123 + li , x.d . ,
bombaci , i. , dey , m. , dey j. and van den heuvel , e.p.j .
: phys . rev .
lett . * 83*(1999)3776 .
+ moraes , p.h.r.s . ,
jose , arbanil , d.v .
, malheiro m. : arxiv .
arxiv:1511.06282 [ gr - qc ] .
+ maurya , s.k . ,
gupta , y.k .
: astrophys space sci 344 , 243 ( 2013 ) + maurya , s.k . ,
gupta , y.k .
: astrophys .
space sci .
86 , 025009 ( 2014 ) + maharaj , s.d . ,
sunzu , j.m . , ray , s. : eur .
j. plus 129 ( 2014 ) + momeni , d. , myrzakulov , r. : int .
j. geomrtic metod in modern phys . * 12*(2015)1550014 + momeni , d. , myrzakulov , r. , gudekli , e. : int .
j. geomrtic metod in modern phys .
* 12*(2015)1550101 + momeni , d. , h. gholizade , raza , m. , myrzakulov , r. : int
. j. modern phys . *
a30*(2015)1550093 + jamil , m. momeni , d. , myrzakulov , r. : chin
. phys . lett .
* 29*(2012)109801 + jamil , m. momeni , d. , raza , m. and myrzakulov , r. : eur .
j. * c72*(2012)1999 + noureen , i. , bhatti , a.a . and
zubair , m. : jcap**1502 * * , 033(2015 ) + noureen , i. and zubair , m. : eur . phys .
j. c * 75*,62(2015 ) + noureen , i. and zubair , m. : astrophys .
space sci . * 355 * , 2202(2014 ) + pant .
n , pradhan , n . , murad h.m .
: astrophys .
space sci .
( accepted ) ( 2014a ) , pant.n . ,
pradhan , n .
, ksh newton singh .
: journal of gravity.doi.org/10.1155/2014/380320 2014b + ruderman , r. : rev .
. astrophys . * 10 * , 427(1972 ) + rahaman , f. et al .
relativ . grav . * 44 * , 107(2012 ) + rahaman , f. et al .
phys . j. c * 72 * , 2071(2012 ) + shabani , h. and farhoudi , m. : phys .
d. * 90 * , 044031(2014 ) + sharif , m. and zubair , m. : jcap * 03*,028(2012a ) + sharif , m. and zubair , m. : j. phys .
. jpn . * 82*,014002(2013 ) + sharif , m. and zubair , m. : phys .
jpn . * 81 * , 114005(2012b ) + yazadjiev , s.s . ,
doneva , d.d . ,
kokkotas , k.d . and
staykov , k.v .
: jcap * 06*,003(2014 )
|
this paper constitutes the investigations regarding possible formation of compact stars in @xmath0 theory of gravity , where @xmath1 is ricci scalar and @xmath2 is trace of energy momentum tensor . in this connection , we use analytic solution of krori and barua metric ( krori and barua 1975 ) to spherically symmetric anisotropic star in context of @xmath0 gravity .
the masses and radii of compact stars models namely model @xmath3 , model@xmath4 and model @xmath5 are employed to incorporate with unknown constants in krori and barua metric .
the physical features such as regularity at center , anisotropy measure , casuality and well - behaved condition of above mentioned class of compact starts are analyzed .
moreover , we have also discussed energy conditions , stability and surface redshift in @xmath0 gravity .
* keywords : * compact stars , @xmath6 gravity . +
|
in many application domains , software systems need to perpetually and rapidly evolve to cope with new user and technology requirements .
being able to modify existing systems or redesign new systems to rapidly take in account new functionalities or preferences has led to the proposition of several software engineering approaches such as the agile software development methodology @xcite .
one of the key principles of agile software development is to build software through an incremental and iterative process .
each iteration adds a new feature and produces a fully working system by going through the whole the software lifecycle , _
i.e. , _ the analyze , develop and test phases .
another particularity of agile development is that the testing activity is not just confined to the classical test phase but rather integrated throughout the entire lifecycle , meaning that the software is continuously tested throughout its development , from its specifications to the final running system , in order to augment the overall software system quality .
another software engineering approach that contributes to facilitating the rapid development of software systems is the use of component - based software architectures . in this context , the overall structure of the application is first described with an architecture description language ( adl ) @xcite .
such description highlights the needed components and their assembly , which facilitates the understanding and analysis of the application s properties , such as behavioral or quality of service properties . if the specifications are coherent , the application is eventually instantiated , deployed and executed to be tested . although agile software development and component - based software engineering ( cbse ) may appear quite different approaches , some works
@xcite have identified that both approaches could benefit to each other , cbse bringing for example the capability of building large software and enhancing reusability , and agile development offering more flexible development processes for shorter time - to - market products .
nevertheless we believe that there is still a bridge between these two approaches , one reason being the lack of component frameworks that allow incremental and iterative development processes , as well as throughout - lifecycle testing . to address this problematic ,
we have developed a model - based framework , named calico , that enables architects to design and test component - based systems in an iterative and uniformed process @xcite .
calico allows software architects to specify their architectures as models , and to analyze them with respect to application and platform constraints .
our approach enables the testing of the system throughout the system lifecycle .
more concretely , calico analyses architecture models and creates contracts by composing contractual application properties , _
e.g. _ , behavioral , dataflow , qos properties .
this composition allows compatible and incompatible interaction to be identified , as well as partially compatible interactions , which require runtime checking @xcite .
when runtime checking is needed , calico automatically instruments the application to reify runtime information to complete the resolution of the partially compatible interaction contract and thus detects if the given interaction may lead to an error . by using this framework in iterative software design processes ,
architects get design feedback , _
i.e. , _ information on identified interaction errors , and can then modify the models accordingly .
each modification performed on the model is propagated to the running system since calico ensures the synchronization between the model and the runtime system , both of which thus coexist during the whole application development .
furthermore , the solution offered by calico is generic regarding underlying platforms , allowing component platforms to benefit from all the analyses integrated into calico .
the rest of this paper is organized as follows .
section [ sec : presentation ] gives an overview of the calico iterative and incremental development process .
section [ sec : example ] illustrates with a concrete scenario the calico approach .
finally , section [ sec : conclusion ] provides some information about the current status of our framework implementation .
calico is composed of two levels : a model level and a platform level as shown in figure [ fig : calico ] .
the model level is independent of any component - based or service - oriented platform .
it contains the calico architecture description metamodels that enable an architect to describe the structure and the properties , _
i.e. _ , structural , behavioral , dataflow and qos properties , of an application .
it is also possible to specify some contextual adaptation rules , independently of any platform , in order to allow the debugging of autonomic systems .
the platform level holds the running system on a target platform .
the iterative and incremental development process of calico , illustrated in figure [ fig : calico ] , is as follows : * ( 1 ) design :* the architect specifies the design of the desired application using the calico metamodels .
the system structure metamodel enables architects to describe the structure of their architecture , independently of any component platform .
calico provides also four contract metamodels to allow architects to specify structural , behavioral , dataflow and qos properties for each component . *
( 2 ) static analysis :* the interaction analysis tool checks the coherence of the system architecture . for each partially compatible interaction , a test to be performed at execution time
is automatically inserted into the calico debug metamodel . for each incompatible interaction ,
the architect is notified of the problem and he / she may thus provide some modification of the application design . as long as some incompatible interactions remain , the next steps of the development process can not be reached .
once all of the problems are fixed , the architect specifies the runtime platform on which the application is to be executed and calico verifies that the specifications do not go against the platform constraints in order to make sure that the application can be indeed deployed on that specific platform . *
( 3 ) code generation :* if a component or service does not already exist , then the generation tool generates code skeleton such that only business code needs to be provided by the developers . *
( 4 ) instrumentalisation :* this step makes the link between the static analysis and the dynamic checks of the application at runtime .
the instrumentation tool takes the debug model as input and automatically instruments the application code to enable the capture of the needed runtime information to complete the resolution of the partially compatible interaction .
this instrumentalisation relies on an aspect - oriented approach and is independent of the underlying platform . *
( 5 ) instantiation :* the loader instantiates the application on the target platform as described by the architect s structural model . concretely ,
the running system is created incrementally by calling the appropriate sequence of system construction operations , such as creating / removing components and connectors . * ( 6 ) reification :* as the testers run the application in different execution contexts , the instrumented application automatically reifies any context changes and monitored information . *
( 7 ) dynamic debugging :* during the debugging phase , the debug tool analyzes the information reified by the running system and triggers when needed the tests contained in the debugging model .
the architect is notified each time an error is detected , allowing him / her to correct the application design .
other debugging action rather than the notification action maybe chosen , such as logging the information into a file , or executing a reconfiguration script that will automatically modify the design and trigger the step 2 of the process
. this latter case may be useful to tune / test adaptation policies for autonomic system . *
( 8 ) evolution of the design :* the architect can modify the design with respect to the debugging information if problems have occurred
. they can also adapt at anytime the design of the application to address new user or application requirements .
after any modification , the development cycle iterates again starting at step 2 .
to illustrate the agile development process offered by calico , we use an example of architecture in the context of the french personal health record system ( phr ) @xcite .
phr is the french personal health record system that is intended to provide health - care professionals with the information needed for their patients care .
figure [ fig : dmp ] represents a possible architecture of the phr system . all medical information , ( such as biological analyses , x - rays , medications , _ etc . _ ) , will be stored in distributed databases and will be made accessible through an on - line interface client . in order to build a robust phr application , architects need to be able to express several application properties .
a first requirement of this system architecture is related to authentication issues since not everybody should have access to anybody s health records .
the architecture of this system must thus provide some authentication mechanism .
the authentication architecture element logs a health - care professional in and returns a session ticket through the functionality getticket that is offered by sessionserver . for security reason
, the functionality getticket can be used only by the element authentication to avoid that an unauthenticated user get a session ticket .
finally the session ticket must be validated by the sessionserver before retrieving any medical data from the database .
another requirement is the high reliability of the system .
such system has to be able to handle very heterogeneous medical information , going from light - weight text records to gigabytes of echographies .
furthermore , the devices used to display this information are also heterogeneous .
they range from desktop computers with high - quality large - screen monitors and gigabyte network connexions to simple pdas with small screens and low - bandwidth gprs network connexion .
handling such data in a reliable way is critical because the system must be able to determine if a given data can be displayed appropriately with no loss of information , as well as in which time - frame , depending on the available resources and amount of information to display .
for example , a dataflow constraint may express that medical data received on a terminal of type pda nokia n800 should be less than 10 megaoctets .
another constraint can also specify that only text or jpg documents can be read on that terminal .
overall the constraints may evolve in time or just not reflect exactly one given execution context .
there is thus a need to iterate the whole process to check if the declared application constraints can be all checked , statically or dynamically . *
( 1 ) * the architect specifies the architecture and the properties of the phr application . for example , the first requirement mentioned above can be specified using the structural contract metamodel . * ( 2 ) * the overall coherence of the constraints is statically verified .
a partially compatible interaction is detected between the medicalserver and the pds since data sent by the server could be greater than 10 megaoctets or in a different format than txt or jpg .
accordingly calico adds some rules to validate in the debug model .
these rules specify that the size and the data type must be captured at runtime . *
( 3 ) * code skeletons are generated and developers can provide the business code of each components . *
( 4 ) * following the information contained in the debug model , the application is automatically modified to capture the size and the type of the medical data that enters the pda . * ( 5 ) * the application is deployed on the target application . * ( 6 ) * at this step different execution contexts are tested .
one may consists in the use of the phr application by a druggist , who typically uses the phr only to consult text documents .
another test scenario considers a radiologist . during the test scenario execution
, monitored informations are reified . * ( 7 ) * the debug tool resumes the interaction compatibility checks that were partially compatible . in the case of the druggist ,
no error is detected , whereas for the radiologist , the analyse indicates that the data are too large for the pda . *
( 8) * the architect can accordingly modify the application design by inserting a new component dataconverter between the pda and the globalsearch component in order to reduce the size of a too large radiography .
the whole process is then iterated again .
if no error is detected statically , the new component dataconverter is automatically integrated into the already deployed application , and new test scenarios may be executed .
calico is a model - based framework that enables the design and debug of systems in an iterative and incremental way , bridging a little more the gap between component - based software engineering and agile development approaches . our framework is generic and highly extensible .
all metamodels for specifying the structure , the application properties and the adaptation rules are independent of any underlying platform .
this enables architects to perform various architecture analyses on their applications even if the underlying component or service framework does not provide any verification tools .
the current implementation of calico is developed in java .
all calico metamodels are implemented with the eclipse modeling framework ( emf ) .
a graphical editor , implemented with the graphical modeling framework ( gmf ) , enables the architect to edit the model during the whole development cycle .
we have integrated several existing tools to verify the coherence of the component interactions in term of structural , behavioral , dataflow and quality of service properties .
structural constraints are expressed in ocl @xcite , using the emf - ocl library .
behavioral specifications are based on existing process algebra , such as csp @xcite , fsp @xcite , sfsp .
the current implementation uses the fractal behavioral protocol checker @xcite to verify that a given component composition does not introduce a deadlock . we have developed a dataflow analysis based on the algorithm of constant propagation in partial program validation @xcite .
the qos metamodel has been inspired by the qml @xcite and wsla @xcite approaches .
the associated analysis is based on prediction of quality property in a workflow of web services @xcite .
furthermore , application instrumentation has been implemented with spoon @xcite .
the sensor framework wildcat @xcite has been integrated in calico .
our current implementation supports four component platforms ( fractal @xcite , openccm @xcite , opencom @xcite and frascati @xcite ) and one service - oriented platform(web services @xcite ) .
calico has been carefully designed to allow new extensions in terms of support for new platforms , new qos sensors and new kinds of debugging actions .
we have performed benchmarks on our implementation and showed that calico is usable to design reliable large systems up - to 10000 components , which is the maximum load of most runtime platforms .
calico is still being developed , to support more extensions .
the current implementation is freely available at http://calico.gforge.inria.fr .
10 stojanovic , z. , dahanayake , a. , sol , h.g . : component - oriented agile software development . in marchesi , m. , succi , g. ,
volume 2675 of lecture notes in computer science . , springer ( 2003 ) 315318 radinger , w. , goeschka , k.m . : agile software development for component based software engineering . in : proceedings of 18th annual acm sigplan conference on object - oriented programming , systems , languages , and applications , new york ,
ny , usa , acm ( october 2003 ) 300301 waignier , g. , sriplakich , p. , le meur , a.f . , duchien , l. : . in : proceedings of the acm / ieee 11th international conference on model driven engineering languages and systems ( models08 ) .
volume 5301 of lecture notes in computer science .
, toulouse , france , springer - verlag ( october 2008 ) 371385 waignier , g. , le meur , a.f . ,
duchien , l. : a model - based framework to design and debug safe component - based autonomic systems . in mirandola , r. , gortona , i. , hofmeiste , c. , eds . :
proceedings of the 5th international conference on the quality of software - architectures ( qosa 2009 ) .
volume 5581 of lecture notes in computer science . ,
pennsylvania , usa , springer - verlag ( june 2009 ) 117 waignier , g. , le meur , a.f . , duchien , l. : . in becker , s. , plasil , f. , reussner , r. ,
eds . : proceedings of the 4th international conference on the quality of software - architectures ( qosa08 ) .
volume 5281 of lecture notes in computer science . , karlsruhe ( th ) , germany , springer - verlag ( october 2008 ) 152170 bruneton , e. , coupaye , t. , leclercq , m. , quma , v. , stefani , j.b .
: an open component model and its support in java . in : proceedings of the 7th international symposium component - based software engineering .
volume 3054 of lecture notes in computer science . ,
edinburgh , scotland , springer - verlag ( may 2004 ) 722 briclet , f. , contreras , c. , merle , p. : penccm : une infrastructure composants pour le dploiement dapplications base de composants corba . in imag
/ lsr , ed . : proceedings of the 2004 dploiement et ( re ) configuration de logiciels ( decor04 ) . ( 2004 ) 101112 http://openccm.ow2.org .
seinturier , l. , merle , p. , fournier , d. , dolet , n. , schiavoni , v. , stefani , j.b . :
reconfigurable sca applications with the frascati platform . in : 6th ieee international conference on service computing ( scc09 ) .
( september 2009 )
|
agile development processes and component - based software architectures are two software engineering approaches that contribute to enable the rapid building and evolution of applications .
nevertheless , few approaches have proposed a framework to combine agile and component - based development , allowing an application to be tested throughout the entire development cycle . to address this problematic ,
we have built calico , a model - based framework that allows applications to be safely developed in an iterative and incremental manner .
the calico approach relies on the synchronization of a model view , which specifies the application properties , and a runtime view , which contains the application in its execution context .
tests on the application specifications that require values only known at runtime , are automatically integrated by calico into the running application , and the captured needed values are reified at execution time to resume the tests and inform the architect of potential problems .
any modification at the model level that does not introduce new errors is automatically propagated to the running system , allowing the safe evolution of the application . in this paper
, we illustrate the calico development process with a concrete example and provide information on the current implementation of our framework .
|
we would like to think prof .
shih - chang lee , prof . tsz - king wong and prof .
anatoly v. borisov for useful discussions .
this work is supported in part by the national science council of r.o.c . under grant nos : nsc- 97 - 2112-m-006 - 001-my3 and nsc-95 - 2112-m-007 - 059-my3 .
d. hooper , p. blasi and p. d. serpico , jcap * 0901 * , 025 ( 2009 ) [ arxiv:0810.1527 ] ; h. yuksel , m. d. kistler and t. stanev , arxiv:0810.2784 [ astro - ph ] ; s. profumo , arxiv:0812.4457 [ astro - ph ] .
d. malyshev , i. cholis and j. gelfand , arxiv:0903.1310 [ astro - ph ] ; v. barger _
_ , arxiv:0904.2001 [ hep - ph ] .
l. bergstrom , t. bringmann and j. edsjo , phys .
rev . d*77 * , 103520 ( 2008 ) ; m. cirelli _ et al .
_ , arxiv:0808.3867 [ astro - ph ] ; 0809.2409 [ hep - ph ] ; v. barger _ et al .
b*672*,141 ( 2009 ) ; j. liu , p. f. yin and s. h. zhu , arxiv:0812.0964 [ astro - ph ] ; d. hooper , a. stebbins and k. m. zurek , arxiv:0812.3202 [ hep - ph ] .
x. j. bi , p. h. gu , t. li and x. zhang , jhep * 0904 * , 103 ( 2009 ) ; d. hooper and k. m. zurek , arxiv:0902.0593 [ hep - ph ] ; k. cheung , p. y. tseng and t. c. yuan , arxiv:0902.4035 [ hep - ph ] ; d. s. m. alves _ et al .
_ , arxiv:0903.3945 [ hep - ph ] ; a. a. e. zant , s. khalil and h. okada , arxiv:0903.5083 [ hep - ph ] ; j. mcdonald , arxiv:0904.0969 [ hep - ph ] ; m. kuhlen and d. malyshev , arxiv:0904.3378 [ hep - ph ] ; f. y. cyr - racine , s. profumo and k. sigurdson , arxiv:0904.3933 [ astro-ph.co ] . c. r. chen , f. takahashi and t. t. yanagida , phys .
b*671 * , 71 ( 2009 ) ; phys .
b*673 * , 255 ( 2009 ) ; c. r. chen and f. takahashi , jcap * 0902 * , 004 ( 2009 ) ; p.
f. yin _ et al .
rev . d*79 * , 023512 ( 2009 ) ; k. hamaguchi , e. nakamura , s. shirai , t. t. yanagida , arxiv:0811.0737 [ hep - ph ] ; a. ibarra and d. tran , jcap * 0902 * , 021 ( 2009 ) [ arxiv:0811.1555 [ hep - ph ] ] ; c. r. chen , m. m. nojiri , f. takahashi and t. t. yanagida , arxiv:0811.3357 [ astro - ph ] ; e. nardi , f. sannino and a. strumia , jcap * 0901 * , 043 ( 2009 ) [ arxiv:0811.4153 [ hep - ph ] ] ; a. arvanitaki _
_ , arxiv:0812.2075 [ hep - ph ] ; k. hamaguchi , s. shirai and t. yanagida , phys .
b*673 * , 247 ( 2009 ) [ arxiv:0812.2374 [ hep - ph ] ] ; k. hamaguchi , f. takahashi and t. t. yanagida , arxiv:0901.2168 [ hep - ph ] ; k. ishiwata , s. matsumoto and t. moroi , arxiv:0811.0250 [ hep - ph ] ; 0903.0242 [ hep - ph ] ; 0903.3125 [ hep - ph ] ; s. l. chen , r. n. mohapatra , s. nussinov and y. zhang , arxiv:0903.2562 [ hep - ph ] ; a. arvanitaki , s. dimopoulos , s. dubovsky , p. w. graham , r. harnik and s. rajendran , arxiv:0904.2789 [ hep - ph ] .
j. hisano , s. matsumoto and m. m. nojiri , phys . rev
* 92 * , 031303 ( 2004 ) [ arxiv : hep - ph/0307216 ] ; n. arkani - hamed , d. p. finkbeiner , t. r. slatyer and n. weiner , phys .
rev . d*79 * , 015014 ( 2009 ) [ arxiv:0810.0713 [ hep - ph ] ] .
m. pospelov and a. ritz , phys .
d*78 * , 055003 ( 2008 ) [ arxiv:0803.2251 [ hep - ph ] ] ; d. feldman , z. liu and p. nath , phys .
rev . d*79 * , 063509 ( 2009 ) [ arxiv:0810.5762v3 [ hep - ph ] ] ; j. d. march - russell and s. m. west , arxiv:0812.0559 [ astro - ph ] ; w. l. guo and y. l. wu , arxiv:0901.1450 [ hep - ph ] ; x. j. bi , x. g. he and q. yuan , arxiv:0903.0122 [ hep - ph ] .
|
a long - lived decaying dark matter as a resolution to fermi , pamela and atic anomalies is investigated in the framework of split supersymmetry ( susy ) without r - parity , where the neutralino is regarded as the dark matter and the extreme fine - tuned couplings for the long - lived neutralino are naturally evaded in the usual approach .
the energy spectra of electron and positron are from not only the direct neutralino decays denoted by @xmath0 , but also the decaying chains such as @xmath1 . we find that with a proper lifetime of the neutralino , slepton - mediated effects could explain the atic and pamela data well , but a inconsistence occurs to it the fermi and pamela data without considering the atic one .
however , by a suitable combination of @xmath0 and @xmath2 , the sneutrino - mediated effects could simultaneously account for the fermi and pamela data .
it is mysterious that the direct measured matter in the standard model ( sm ) only occupies @xmath3 of our universe while dark matter and dark energy have the occupancies of @xmath4 and @xmath5 , respectively .
it becomes a very important issue to understand what the dark stuff is . through high energy colliders such as the large hadron collider ( lhc ) at cern
, we may directly observe dark matter . on the other hand , by the study of the high energy cosmic - ray
, we may have the chance to probe dark matter indirectly .
recently , pamela @xcite and atic @xcite collaborations have published some astonished events in cosmic - ray measurements , in which the former finds the excess in the positron flux ratio with energies in the @xmath6 gev range , while the later observes anomaly in the electron+positron flux in the @xmath7 gev range .
intriguingly , these data are consistent with the measurements of the high energy electron and positron fluxes in the cosmic ray spectra by ppb - bets @xcite , heat @xcite , ams @xcite and hess @xcite , respectively .
inspired by the pamela / atic anomalies , various interesting possible mechanisms to generate the high energy positrons and electrons are proposed , such as pulsars @xcite , dark matter annihilations @xcite and dark matter decays @xcite . very recently , a more precision measurement on the electron+positron flux by the fermi collaboration @xcite with the similar energy range as atic has shown some enhancement in flux in the higher energy range .
however , the fermi s result indicates smaller fluxes than the atic data in the 500 gev range .
nevertheless , in our study we consider both data .
other relevant studies could be referred to ref . @xcite .
although the mechanism of dark matter annihilations could provide the source for the fermi , pamela and atic ( fpa ) anomalies , it is inevitable that an enhanced boost factor of a few orders of magnitude , such as sommerfeld enhancement @xcite , near - threshold resonance and dark - onium formation @xcite , has to be introduced . in this paper , we investigate other sources for the excess of positrons and electrons in fpa without a large boost factor .
we will concentrate on the mechanism of dark matter decays .
as known that in order to make the unstable dark matter be long - lived , say @xmath8 , usually we have to fine - tune either couplings to be tiny or the scale of the intermediate state to be as large as the gut scale @xcite .
therefore , our purpose is to explore what kind of the long - lived dark matter in the extension of the sm that could satisfy one of two criteria naturally , at least in technique . in the literature ,
one of the popular sm extensions is supersymmetry ( susy ) .
it has been known that the effects of susy at the scale @xmath9 of @xmath10 can solve not only the hierarchy problem , but also the problem of the unified gauge coupling @xcite . moreover , the predicted lightest neutralino in supersymmetric models could also provide the candidate of dark matter @xcite . in spite of
the above successes , models with susy still suffer some difficulties from phenomenological reasons , such as the problems on small cp violating phases , large flavor mixings and proton decays , as well as they predict too large cosmological constant . as a result
, a fine tuning always comes up in the low energy physics . in order to interpret the cosmological constant problem and maintain the beauty of the ordinary low - energy susy models ,
the scenario of split susy is suggested @xcite , in which the susy breaking scale is much higher than the electroweak scale . in this split susy scenario , except the sm higgs which could be as light as the current experimental limit , the scalar particles are all ultra - heavy , denoted by @xmath11 gev . on the other hand , by the protection of approximate chiral symmetries
, the masses of sfermions , such as gauginos and higgsinos , could be at the electroweak scale @xcite . clearly , split susy not only supplies the candidate of fermionic dark matter at the tev scale but also provides a large scale for the susy breaking , so that the second criterion mentioned early for model - searches is satisfied automatically .
thus , if the ultraheavy scalar sparticles could play the role of messenger to deliver the dark matter decay , the decay rate of the dark stuff is suppressed and the lifetime of the unstable dark matter could be still long enough to explain the fpa puzzle . in split supersymmetric models with r - parity ,
the neutralino regarded as the lightest susy particle ( lsp ) is a stable particle . to study the dark matter
decays , we extend our consideration to the framework of split susy with the violation of r - parity , in which the conservations of lepton or / and baryon numbers are broken . since fpa anomalies only involve leptons , it is plausible to only consider lepton number violating effects by requiring a good symmetry on the baryon number . as usual ,
the superpotential for the minimal supersymmetric standard model ( mssm ) with bilinear and trilinear terms that violate the lepton number is expressed by w&=&w_mssm + w_r , + w_mssm&= & h^_ij l_i h_d e^c_j + h^d_ij q_i h_d d^c_j + h^u_ij q_i h_u u^c_j + h_d h_u , + w_r&= & _ [ ij]k l_i l_j e^c_k + _ijk l_i q_j d^c_k + _ i l_i
h_u . we note that the bilinear term @xmath12 could be rotated away by redefining the superfield as @xmath13 . in the ordinary susy model
, the bilinear operator could be reinduced by loop effects .
however , in the scenario of split susy , since all bosonic sparticles are very heavy , the loop induced effects are highly suppressed and negligible @xcite .
in addition , with the measurement on the antiproton flux by pamela that shows no exotic events , it is conceivable to set the quarks related effects governed by the parameters @xmath14 as small as possible .
thus , in our analysis , we will only focus on trilinear interactions of @xmath15k } l_{i } l_j e^c_k $ ] . consequently , the lagrangian for the r - parity violation is found to be _
r&= & _ [ ij]k ( |_k p_l _ j _ il + |_k p_l _ il _ jl + |_j p_r ^c_il _ kr . + & - & .
|_k p_l _ i _
jl - |_k p_l _ j _ il - |_i p_r ^c_jl _ kr ) + h.c .
[ eq : rp ] with @xmath16 .
the bracket [ ij ] denotes that the @xmath17 parameters are antisymmetry in the first two indices . since in mssm neutralinos
are composed of bino , wino and higgsinos , for simplicity , we take the bino as the neutralino denoted by @xmath18 and the couplings to leptons are given by @xcite _ ^0_1 & = & _ w | ( c^_l p_r _ l + c^_r p_l _ r ) ^0_1 , + l_^0_1 & = & _ w c^_l ( |p_r ^0_1)_l [ eq : int_neutralino ] with @xmath19 being weinberg angle , @xmath20 , @xmath21 and @xmath22 . according to eqs .
( [ eq : rp ] ) and ( [ eq : int_neutralino ] ) , we see that split susy with the r - parity violation has two channels to decay to the charged leptons : one is via the charged sleptons and another one is mediated by sneutrinos , where we sketch the feynman diagrams in fig .
[ fig : dmdecay01 ] .
before we make the detailed analysis for the energy spectra of electron and positron in neutralino decays , it is worthwhile to understand the allowed range of free parameters to satisfy the condition of the long - lived dark matter .
since there are 6 ( 3 ) species of sleptons ( sneutrinos ) and many free parameters in @xmath23k}$ ] , to simplify the estimation and without loss of generality , we use one single decay channel , associated with the slepton as the mediator , to illustrate what the lifetime of the neutralino could be .
thus , based on the introduced interactions , we obtain the lifetime of the neutralino to be _ & & . with @xmath24 tev , the lifetime as a function of the slepton mass and trilinear coupling
is shown in fig .
[ fig : dmlife ] .
k}$ ] in log scale . ] in the figure , we have used @xmath17 instead of @xmath23k}$ ] .
the typical values of @xmath25 are chosen according to the necessity to fit the data of fpa .
clearly , in split susy , with the susy breaking scale of @xmath26 gev and @xmath27 , @xmath25 is much longer than the age of our universe with @xmath28 s. since multi - indices are involved in @xmath23k}$ ] , there are many possible channels to produce the high energy positrons and electrons . in order to display the characters of neutralino
decays in split susy efficiently , we will study the cases in which the final states involve an electron - positron pair or one electron ( positron ) associated with one anti - muon ( muon ) .
for the latter case , anti - muon ( muon ) will further decay to positron ( electron ) due to weak interactions of the sm .
in short , we will focus on the decays @xmath29 and @xmath30 , where the relevant r - parity violating couplings are @xmath311}$ ] , @xmath312}$ ] and @xmath321}$ ] .
in addition , since for each slepton flavor , we have two sparticles @xmath33 and @xmath34 , corresponding to the superpartners of left - handed and right - handed leptons , respectively , in weak states . for simplicity , in our following analysis
, we will concentrate on the contributions of @xmath35 .
expectably , the effects of @xmath36 on the energy spectra of electron and positron should be very similar to those of @xmath33 . according to our setup on the conditions for split susy without r - parity ,
the corresponding feynman diagrams are displayed in fig .
[ fig : dmdecay ] . since figs .
[ fig : dmdecay](a ) and ( b ) are mediated by the selectron , while figs .
[ fig : dmdecay ] ( c ) and ( d ) are exchanged by the sneutrino , we separately study the energy spectrum of electron induced by these diagrams . moreover , to be more clear , we discuss the contributions in terms of the individual diagram in fig .
[ fig : dmdecay ] . following the interactions in eqs .
( [ eq : rp ] ) and ( [ eq : int_neutralino ] ) , the decay amplitude for fig .
[ fig : dmdecay](a ) can be written as m_a&= & ic^e_l _ w |u_e p_r ^0_1 spinor of electron ( positron ) .
we note that since the emitted lepton from the first vertex connected to the neutralino is electron or positron , eq .
( [ eq : ampa ] ) could be also applied to the case where an electron is emitted at the second vertex associated with the neutrino . by combining both possibilities and including the phase space for the three - body decay , the differential rate as a function of the electron energy is obtained to be & = & |_[i1]1|^2 e^2 ( 9 m^2_-16 m_e ) , + c_&= & .
[ eq : dif_a ] the allowed range of e is known as @xmath37
. this energy spectrum should be also suitable for positron .
accordingly , the normalized spectrum is defined by = [ eq : slepa ] with @xmath38 . for fig . [
fig : dmdecay](b ) , due to muon involved , the situation is a little bit complicated . according to the previous analysis ,
the spectrum for electron directly emitted from the neutralino decay can be easily read by & = & 2c_|_[i1]2|^2 e^2 ( m^2_-2 m _ e ) .
[ eq : gammab1 ] however , if the electron is produced via the muon decay , we have to deal with the decaying chain , i.e. @xmath39 .
we find that indeed the differential decay rate for the chain can be simplified to be d^2_b2&= & d^ _ [ eq : double_diff ] where @xmath40 denotes the decay rate for @xmath41 , @xmath42 has combined the amplitude square for @xmath43 and the three - body phase space and @xmath44 is the total width of muon . if we further integrate out all the angles , we get d^_(e _ ) & = & c _ |_[i1]2|^2 e^2 _
( m^2_- m_e _ ) de _ , + d|m_|^2(e _ , e ) & = & \ { . + & & . -m^2 _ } de .
[ eq : diff_slepmu ] here , the allowed energy range for muon and electron are found to be @xmath45 and @xmath46 .
however , for an energetic muon , the upper value becomes @xmath47 .
we adopt the normalized energy spectrum of electron ( positron ) for fig .
[ fig : dmdecay](b ) to be = .
[ eq : slepb ]
similarly , we can apply the calculations for slepton - mediated effects to figs .
[ fig : dmdecay](c ) and ( d ) mediated by sneutrinos . utilizing the results in eq .
( [ eq : diff_slepmu ] ) by replacing suitable coefficients and parameters , the differential decay rate for fig .
[ fig : dmdecay](c ) is easily obtained as & = & c_||^2 |_[i1]1|^2e^2 ( m^2_- m_e ) .
[ eq : snu_c ] following the definition of eq .
( [ eq : slepa ] ) , we can get the normalized energy spectrum for fig .
[ fig : dmdecay](c ) . since fig .
[ fig : dmdecay](d ) involves muon in the final state , similar to the figure ( b ) , we have to deal with the decaying chain such as @xmath48 . by comparing with eq .
( [ eq : double_diff ] ) , the differential decay rate for the decaying chain is given by d^2_d = d^ _ with d^_&= & d_c(e _ ) .
the corresponding normalized spectrum is = .
[ eq : slepd ] to study the measured cosmic - ray , we have to know the behavior of the number density of particles per unit energy , governed by the transport equation @xcite = ( k(e , r)f ) + ( b(e , r)f ) + q_e ( e , r ) , [ eq : tranport ] where @xmath49 denotes the number density of cosmic - ray per unit energy , @xmath50 is the diffusion coefficient , @xmath51 describes the rate of the energy loss by the inverse compton scattering and synchrotron radiation etc . , and @xmath52 represents the source of cosmic - ray from the decay of dark matter , given by q(e , r)= , with @xmath53 ( @xmath54 ) the lifetime ( mass ) of dark matter , @xmath55 the energy spectrum of cosmic - ray and @xmath56 the density profile of dark matter . in our following analysis
, we will adopt the so - called navarro - frenk - white ( nfw ) profile , given by @xcite _ = with @xmath57 gev/@xmath58 and @xmath59 kpc . by using the green function method , the steady solution to eq .
( [ eq : tranport ] ) for electron and positron can be expressed by f(e)=_0^m_/2de^g(e , e^ ) .
since the experiments measure the flux of cosmic - rays , the relation to the number density is given by @xmath60 with @xmath61 being the speed of light .
for numerical estimations , we adopt the result parametrized by @xcite g(e , e^ ) ( e^-e ) with @xmath62 , @xmath63 and @xmath64 .
we note that these values of parameters should depend on the modeling for the profile of dark matter in our galaxy .
the detailed analysis on the dependence of various profiles can be referred to ref .
@xcite .
besides the new source for the fluxes of electron and positron , we also need to understand the contributions of primary and secondary electrons and secondary positrons , in which the former comes from supernova remnants and the spallation of cosmic rays in the interstellar medium , respectively , while the latter could be generated by primary protons colliding with other nuclei in the interstellar medium . in our numerical calculations , we use the parametrizations , given by @xcite _ e^-^prim(e)&= & [ gev^-1cm^-2s^-1sr^-1 ] , + _ e^-^sec(e)&= & [ gev^-1cm^-2s^-1sr^-1 ] , + _ e^+^sec(e)&= & [ gev^-1cm^-2s^-1sr^-1 ] [ eq : bg ] where @xmath65 denotes the primary ( secondary ) cosmic ray . accordingly ,
the total electron and positron fluxes are defined by _
e^-&=&^prim_e^-+^sec_e^-+^_e^- , + _ e^+&=&^sec_e^++^_e^+ .
[ eq : flux_tot ] here , according to refs .
@xcite and @xcite , we have regarded the normalization of the primary electron flux to be undetermined and parametrized by the parameter of @xmath66 .
the value of @xmath66 is chosen to fit the data . before introducing the source of the primary positron ,
@xmath66 is set to be 0.8 . ) and
( [ eq : flux_tot ] ) with @xmath67 .
, width=5 ] for a comparison , we show the background with @xmath67 and current data as a function of the cosmic - ray energy in fig .
[ fig : data ] .
now we start to numerically analyze the influence of long - lived neutralino decays on the electron and positron fluxes in split susy . to fit the pamela and atic ( fermi ) data , we take _ = 210 ^ 26 ( 410 ^ 26 ) s , m_=2 tev , = 0.7 . since fermi and atic both
show the electron+positron flux , we will make detailed discussions in the light of the atic s measurement and extend the analysis to include fermi accordingly .
based on our earlier discussions , we also present our numerical results by diagrams shown in fig .
[ fig : dmdecay ] .
hence , in terms of eqs .
( [ eq : slepa ] ) , ( [ eq : bg ] ) and ( [ eq : flux_tot ] ) , the contributions of fig .
[ fig : dmdecay](a ) are displayed in fig .
[ fig : a - b](a ) , where the left - hand side is for the electron + positron flux while the right - hand side is the ratio of the positron flux to electron+positron flux . the thin line stands for the background with @xmath67 before including the split susy effects . the thick solid ( dashed ) line denotes the result with ( without ) the background .
since the pamela data show the ratio of fluxes , it should be a constant line with unity if we turn off the contributions of the background .
thus , for pamela , we only show the results associated with the background . from our results , it is clear that the energy spectrum of fig .
[ fig : dmdecay](a ) shown in eq .
( [ eq : slepa ] ) matches the pamela and atic data up to 1 tev well .
it is worthy of mentioning that since we only consider one channel of fig .
[ fig : dmdecay](a ) with the normalization of eq .
( [ eq : slepa ] ) , the energy spectra of cosmic rays have no dependences on the slepton mass and the couplings given in eqs .
( [ eq : rp ] ) and ( [ eq : int_neutralino ] )
. the same situation will appear in the analysis on the other individual diagram .
however , the parameters will be involved when we consider different diagrams together .
we will see this point more clear later .
the effects of fig .
[ fig : dmdecay](a ) [ ( b ) ] on the atic and pamela anomalies , where the thin solid line is the background with @xmath67 , the thick solid ( dashed , dash - dotted ) line for atic denotes the effect with ( without ) the background , and the dashed and dash - dotted lines in ( b ) represent the contribution of @xmath68 and @xmath69 , respectively . , width=5 ] for the effects of fig .
[ fig : dmdecay](b ) , where electron and positron can be emitted directly from neutralino and muon decays , we show their contributions in fig .
[ fig : a - b](b ) . here ,
the dashed and dash - dotted lines correspond to eq .
( [ eq : gammab1 ] ) and eq .
( [ eq : double_diff ] ) without the background , respectively .
the thick solid lines on left and right - hand sides are the combined effects with the background . from the figure
, we see clearly that below 300 gev , the spectrum from one electron and one muon ( @xmath70 ) in the neutralino decay fits the atic data well , but it is lower at the higher energy .
nevertheless , the @xmath70 mechanism still matches well with the pamela data .
( a ) and ( b ) with different partitions described by @xmath71 , where the solid , dashed and dash - dotted lines stand for @xmath72 , @xmath73 and @xmath74 , respectively . ] after the discussions of the individual contributions from figs .
[ fig : dmdecay](a ) and ( b ) , it is interesting to consider their combined effects . due to both mechanisms being mediated by the selectron ,
the energy spectrum now will depend on the free parameters @xmath311}$ ] and @xmath312}$ ] . in order to simplify the analysis
, we use a new parameter of @xmath75 that satisfies @xmath76 instead of the original ones of @xmath771(2)}|^2 $ ] in eqs .
( [ eq : dif_a ] ) and ( [ eq : diff_slepmu ] ) .
therefore , the new normalized energy spectrum for electron ( positron ) is given by & = & [ eq : dnab ] with & = & , + & = & ( + ^e^max__m _ de _ ) .
consequently , by combining eq .
( [ eq : dnab ] ) with eqs .
( [ eq : bg ] ) and ( [ eq : flux_tot ] ) , the results are shown in fig . [
fig : ab ] where the solid , dashed and dash - dotted lines correspond to the cases of @xmath78 , @xmath73 and @xmath79 , respectively . by the figure
, we see that the spectrum mediated by the selectron through the mechanisms of figs .
[ fig : dmdecay](a ) and ( b ) is not sensitive to the detailed partition for the percentage of @xmath311}$ ] and @xmath802}$ ] .
intriguingly , the resultant combination is consistent with the pamela and atic data . since the flux measured by fermi is more precise and smaller than that by atic in the high energy region , while
all involved parameters for atic and fermi are the same in the model , to singly fit the fermi data , we only need to tune the lifetime of the neutralino .
therefore , with @xmath81 s , the combination effects of figs . [
fig : dmdecay](a ) and ( b ) are presented in fig .
[ fig : fermiab ] .
but with @xmath81 s. ] it is clear to see that we can fit the fermi data in split susy .
however , the result for the positron flux ratio deviates from the pamela data .
next , we study the sneutrino - mediated effects shown in figs . [
fig : dmdecay](c ) and ( d ) . since electron or positron
can not be emitted from the vertex of the decaying neutralino in this case , the electron - positron pair or electron - antimuon ( positron - muon ) final state is only produced via the r - parity violating interactions .
therefore , the resultant cosmic - ray fluxes may differ from the processes mediated by the selectron . with eq .
( [ eq : snu_c ] ) and the normalization defined in eq .
( [ eq : slepa ] ) , we show the electron+positron flux and the ratio of the positron flux in fig .
[ fig : c - d](c ) , where the solid ( dashed ) line is associated with ( without ) the background .
clearly , the result of the mechanism not only roughly fit the atic s data but also explain the pamela s measurement well . by the analysis , we see that the energy spectra governed by @xmath82 of eq .
( [ eq : dif_a ] ) and @xmath83 of eq .
( [ eq : snu_c ] ) have similar results .
results explain the atic and pamela anomalies by fig .
[ fig : dmdecay](c ) [ ( d ) ] , where the solid ( dashed ) line for atic denotes the effect with ( without ) the background . ] as to the contributions of one electron ( positron ) and one anti - muon ( muon ) formulated by eq .
( [ eq : slepd ] ) , we present the results in fig .
[ fig : c - d](d ) .
we note that eq .
( [ eq : slepd ] ) involves two free parameters of @xmath312}$ ] and @xmath321}$ ] . for simplicity ,
we have set @xmath312}=\lambda_{[i2]1}$ ] .
similar to the @xmath70 mechanism of fig .
[ fig : dmdecay](b ) , it could also match the pamela data well .
although the shapes of fluxes in figs . [ fig : a - b](b ) and [ fig : c - d](d ) look similar , the decreasing tendency at the high energy in the latter is faster than that in the former .
moreover , in fig . [
fig : cd ] we display the combination of figs . [ fig : c - d](c ) and ( d ) , where we have followed the approach used in eq . ( [ eq : dnab ] ) to introduce a new parameter @xmath84 ( @xmath85 ) instead of the old parameters of @xmath311}$ ] and @xmath862(1)}$ ] .
the solid , dashed and dash - dotted lines in the figure denote @xmath87=(0.8 , 0.2 ) , ( 0.6 , 0.4 ) and ( 0.3 , 0.7 ) , respectively . , but for fig .
[ fig : dmdecay](c ) and ( d ) . ]
it is clear that the resultant fluxes mediated by the sneutrino at the energy above 200 gev are sensitive to the percentages of @xmath88 and @xmath89 .
when the percentage of @xmath88 @xmath90 decreases ( increases ) , the positron+electron flux is also decreased accordingly . intuitively , the resultant shape of the flux is very close to what fermi observes .
if we directly use the result in fig .
[ fig : cd ] to compare with the fermi data shown in fig .
[ fig : fermi_d ] , ( c ) and ( d ) due to the sneutrino . ]
we find that the result with @xmath91 could fit the fermi data very well .
importantly , unlike the slepton - mediated case , with the same parameters , fermi and pamela anomalies can be understood simultaneously . in summary
, we have investigated the fermi , pamela and atic anomalies in the framework of split susy without r - parity based on the neutralino decays .
we have shown that the energy spectra of electron and positron are from not only the direct neutralino decays denoted by @xmath0 , but also the decaying chains such as @xmath92 . we have studied both types of channels separately and unitedly in detail .
we have found that with @xmath93 s , slepton - mediated effects could explain the atic and pamela data well but not for the fermi and pamela data even with a different neutralino lifetime .
however , by the proper combination of @xmath0 and @xmath94 , the sneutrino - mediated effects could fit the fermi and pamela data simultaneously .
|
the hohenberg - kohn ( hk ) formulation of dft and the subsequent development of the kohn - sham ( ks ) scheme made dft the most popular tool for electronic structure calculations ( see , e.g. @xcite ) .
the profound meaning of the hk theorems for many - electron systems was further strengthened by the levy - lieb formulation of the problem @xcite . in the next section the relevant aspects of the levy - lieb formulation
will be reported .
such a formulation removes limitations of the hk theorem , such as the degeneracy of the ground state , and sets the exact correspondence between the 3-dimensional electron density and the 3n - dimensional wavefunction of the ground state of a system of @xmath0 electrons . in practical calculations ,
this formulation is never explicitly used and it is usually considered only a conceptual proof of validity of dft . in this paper
i present the levy - lieb formulation in a different perspective ; i propose such a formulation as the central algorithm which consistently merge two complementary but distinguished , approaches , namely ks dft and qmc .
the merging occurs in such a way that the numerical efficency of dft contributes to enhance the numerical efficency of qmc and that the poor accuracy of dft is enhanced by the accuracy of qmc .
a similar idea , restricted to orbital - free dft and to ground state path integral qmc was presented in a previous paper @xcite ; here the idea is extended to all qmc methods based on the explicit calculation of the ground state wavefunction ( above all the most recent characterized high accuracy of the wavefunction e.g. @xcite ) , and to the ks dft scheme .
another ( suggestive , but yet not fully proved ) implication of the procedure proposed is that the levy - lieb principle can be interpreted as a particular case of a decoding key of a set of 3-dimensional data into a set of @xmath2-dimensional data within the more general framework of information theory @xcite ; this interpretation may have important implications for problems of `` inverse chemistry '' ( see e.g. ref.@xcite ) .
however in this paper i wish only to mention the possible `` decoding '' interpretation and a detailed discussion can be found in ref.@xcite .
the levy - lieb constrained - search principle is formulated as : @xmath3 .
\label{eq1}\ ] ] with @xmath4 the ground state energy , @xmath5 an antisymmetric wavefunction of an @xmath0-electron system , @xmath6 the kinetic operator , @xmath7 the electron - electron coulomb potential operator and @xmath1 the external potential ( usually electron - nucleus coulomb potential ) .
the inner minimization is restricted to all antisymmetric wavefunctions @xmath8 leading to @xmath9 , i.e. @xmath10 ( where @xmath11 is the @xmath12 spatial domain ) , while the outer minimization is then independent of @xmath8 and searches over all the @xmath13 s which integrate to @xmath0 . within this framework , the universal functional of the hohenberg - kohn formulation of dft
is written as : @xmath14=min_{\psi\to \rho}\left<\psi\left|\widehat{t}+\widehat{v}_{ee}\right|\psi\right>. \label{eq2}\ ] ] the universal functional @xmath15 $ ] is unknown , however the accuracy of its approximations is the key to the accuracy of dft calculations .
more specifically , within the ks formulation of dft , that is the most widely used approach of dft for electronic structure calculations , it is not the entire @xmath15 $ ] that needs to be approximated , but only the so called exchange and correlation part of it , @xmath16 $ ] . consistently with the definition of @xmath15 $ ] , @xmath16 $ ]
can be defined as follows : @xmath17=f[\rho]-t_{nint}[\rho]-e_{h}[\rho ] \label{eq3}\ ] ] where @xmath18 $ ] is the kinetic functional for non - interacting particles ( whose explicit expression is also unknown , see e.g. @xcite)and @xmath19=\int\frac{\rho({\bf r})\rho({\bf r}^{'})}{|{\bf r}-{\bf r}^{'}|}d{\bf r}d{\bf r}^{'}$ ] is the hartree energy .
let us suppose we employ a given expression of @xmath16 $ ] ( let us refer to it as : @xmath20=\int\rho({\bf r})\epsilon^{0}_{xc}({\bf r})d{\bf r}$ ] ) in a dft - ks calculation with external potential @xmath1 .
the ks calculation done with @xmath20 $ ] will deliver a density of ground state @xmath21 . in parallel , let us consider the problem : @xmath22 within a qmc approach .
the search for @xmath23 , in this case , would correspond to find the ground state wavefunction of a gas of @xmath0 electrons , that is we sample the space of antisymmetric @xmath8 s , and a wavefunction is preferred to another if its energy is lower in absence of any other external constraints . however , if , within the qmc procedure , we add the constraint that @xmath8 must integrate to a given @xmath9 , ( e.g. to @xmath21 ) , then we numerically realize the levy - lieb principle , that is we practically realize the operation of @xmath24 of eq.[eq1 ] . + the relevant , and non trivial , consequence of the argument above is that , if @xmath20 $ ] was exact , then @xmath21 would be exact and the @xmath8 delivered by qmc as a solution of @xmath22 with the constraints that @xmath8 integrates to @xmath21 , would correspond to the exact ( within the accuracy of qmc ) wavefunction of the ground state for a system of @xmath0 electrons with external potential @xmath1 .
based on the considerations above , the proposed procedure to speed up qmc , consists of the following steps : * employ @xmath20 $ ] for a ks calculation and obtain @xmath21 . *
perform a qmc calculation where @xmath8 is sampled in such a way that it integrates to @xmath21 and minimizes @xmath25 .
let us call the corresponding wavefunction @xmath26 . *
if the dft functional employed is accurate enough , then @xmath26 would represent a very good * initial guess * for solving the full problem in qmc .
the essence of the scheme is that one , computationally trivial , dft calculation can * always * be used to simplify a complex qmc calculation , at least in its initial stage .
the practical question concerns the construction of an efficient numerical algorithm for sampling @xmath8 * at given * @xmath9 ( suggestions for a specific case are given in ref.@xcite ) . here for efficiency is meant that the computational cost for the convergence of the sampling at given @xmath9 must be negligible compared to the computational cost of convergence of the standard qmc procedure for searching @xmath8 .
the suggested procedure is well defined in each of its step and its success / failure depends only on the numerical efficiency in sampling @xmath8 * at given * @xmath9 .
however , one may aim to extend the procedure beyond its utility for the `` initial stage '' of a qmc calculation and ask whether it may be possible to design a combined ks dft - qmc scheme to iteratively solve the ground state problem ; in the next section i propose an iterative scheme .
the key issue for the efficiency of the scheme suggested in the previous section is the accuracy of @xmath20 $ ] , the more accurate is @xmath20 $ ] the closer @xmath26 from qmc is to the real ground state wavefunction . in general @xmath20 $ ] may not be enough accurate and in this section i suggest a procedure that updates @xmath16 $ ] and @xmath8 in iterative manner .
interestingly , upon convergence , the procedure goes beyond the `` initial stage '' of a qmc calculation and leads , in principle , to the ground state wavefunction . for this purpose
i define : @xmath14=\int \rho({\bf r})f({\bf r})d{\bf r}. \label{frho}\ ] ] next i need to introduce a `` universal '' kinetic and hartree energy reference for non interacting electrons .
when this reference energy is subtracted to @xmath15 $ ] one obtains the exchange and correlation energy as in eq.[eq3 ] .
@xmath27=\int \rho({\bf r})e_{h}({\bf r})d{\bf r};e_{h}({\bf r})=\int\frac{\rho({\bf r}^{'})}{|{\bf r}-{\bf r}^{'}|}d{\bf r}^ { ' } \label{ehrho}\ ] ] @xmath28=\int \rho({\bf r})t_{nint}({\bf r})d{\bf r } \label{tninit}\ ] ] where @xmath29 @xmath30 stays for hartree - fock , @xmath31 , and @xmath32 is such that the total kinetic energy writes , @xmath33 .
this means that the kinetic energy density for non interacting electrons is determined by the hartree - fock calculation of the @xmath0-electron system considered . for consistency with the definition of @xmath34
we define : @xmath27=\int \rho({\bf r})e^{hf}_{h}({\bf r})d{\bf r } \label{ehrho1}\ ] ] that is the hartree energy density is calculated by the hartree - fock method for the specific @xmath0-electron system considered . in this way we consider @xmath34 and @xmath35 as reference quantities in the limit in which electron correlations , due to the explicit electron - electron interaction , are absent .
the reason for calculating @xmath36 and @xmath34 with hf is due to the need of having a general reference energy for non interacting particles
. such a quantity shall be universal and independent of a specific @xmath9 generated by a specific choice of a @xmath16 $ ] , otherwise the iterative procedure shown below would be biased by the specific choice of the initial @xmath16 $ ] .
it must be noticed that the exchange term of the hf calculation is not considered and used since this latter is automatically calculated in @xmath37 within the qmc approach .
the hf calculation is not needed for the first step of the procedure ( `` initial stage '' of the qmc calculation ) outlined in the previous section , but , as it is reported next , it is required to update / improve @xmath20 $ ] at the successive steps . in any case
the hf calculation is needed only once and is numerically not expensive .
the iterative procedure consists of the following steps ( pictorially illustrated in fig.[fig1 ] ) : * employ a chosen @xmath20 $ ] for a ks calculation and obtain @xmath21 . *
perform a qmc calculation where @xmath8 is sampled via minimization of @xmath25 with the additional request that @xmath8 integrates to @xmath21 , let us indicate the resulting wavefunction as @xmath26 . * from @xmath26 we calculate @xmath38 : @xmath39 * at this point we can numerically update the exchange and correlation energy density : @xmath40 however rather than @xmath41 what is needed is the exchange and correlation potential : @xmath42}{\delta\rho({\bf r})}$ ] . in order to have @xmath43
, one possibility is that of fitting numerically @xmath41 in terms of @xmath44 .
the numerical fit delivers an analytical expression for @xmath45 .
the analytic expression is taken as an approximation of @xmath46 and can be employed in the ks procedure to derive ( analytically ) @xmath43 .
the fitting procedure could be performed , for example , with approaches similar to the `` kernel ridge regression '' ( krr ) a method for non - linear regression that prevents over - fitting @xcite and that is successfully employed in machine learning approaches ( see e.g.@xcite ) .
* we can now employ @xmath46 ( or better the corresponding @xmath43 ) for a ks calculation which will deliver an updated electron density @xmath47 .
+ next , starting from @xmath47 the scheme above can be repeated and continued until certain criterion of convergence on @xmath9 , chosen a priori , is satisfied .
* once the procedure converges , the qmc wavefunction obtained at the final @xmath48 iteration , @xmath49 , corresponds to the `` true '' ground state wavefunction of the system .
, scaledwidth=95.0% ] finally it must be reported that here the convergence of the iterative procedure , has not been proved mathematically .
however the idea remains valid as long as a check that the total energy decreases at each iterative step is performed .
as long as the total energy decreases , the dft - qmc procedure is useful and at any successive step brings us closer to the exact ground state . if the energy increases at the @xmath50-th iteration step , then , the wavefunction , @xmath51 , of the @xmath52-th iteration step shall be used as initial guess for a standard full qmc calculation .
we have proposed a scheme to link in a consistent way ks - dft and qmc via the levy - lieb principle .
the electron density of ks - dft , that is a quantity solution of a @xmath53-dimensional problem can be used to significantly restrict the @xmath2-dimensional space of wavefunctions sampled by the qmc procedure .
the procedure can be divided in two parts : first , given any exchange and correlation functional , the corresponding electron density , solution of the ks equation , can be used as a pre - selection criterion for sampling a wavefunction that can be employed as initial guess in a qmc calculation .
one could stop here and by employing such initial guess proceed with the standard full qmc procedure .
the second part of the scheme proposed is a continuation of the first part , that is a procedure to iteratively solve the ground state problem by combining at each step ks calculations and subsequent qmc sampling .
this second part requires an additional hf calculation , in order to define a general energy of reference for non interacting electrons , and a numerical fitting procedure .
it must be underlined that the whole procedure is based on the assumption that there exists an efficient numerical procedure of sampling @xmath8 at a given electron density @xmath13 ; despite some proposals were put forward in previous work , up to now there is no numerical implementation of such a scheme .
moreover the application of the krr method for the fitting procedure may require a considerable amount of computational resources , thus its application may also need to be optimized so that the total cost of the proposed scheme is convenient compared to that of a full qmc calculation .
thus this work must be considered as a protocol that may be worth to explore further and eventually implement in a computational code .
i would like to thank luca ghiringhelli for a critical reading of the manuscript , for the clarifying lecture about the krr method and for several valuable suggestions .
i would also like to thank anatole von lilienfeld for providing references for the krr method and its applications .
this work was supported by the deutsche forschungsgemeinschaft ( dfg ) , via the heisenberg grant , no .
de 1140/5 - 2 . 24
natexlab#1#1url # 1`#1`urlprefix w.yang and r.g.parr , _ density functional theory of atoms and molecules _ , oxford university press , new york , 1989 m.levy , proc.natl.acad.sci.u.s.a . * 76 * , 6062 ( 1979 ) ; see also : m.levy , phys.rev.a * 26 * , 1200 ( 1982 ) e.lieb , int .
quant . chem .
* 24 * , 243 - 277 ( 1983 ) .
an expanded version appears in density functional methods in physics , r. dreizler and j. da providencia eds . , plenum nato asi series * 123 * , 31 - 80 ( 1985 ) l.delle site , l.m.ghiringhelli and d.m.ceperley , int.j.quant.chem . * 113 * , 155 ( 2013 ) m.holzmann , d.m.ceperley , c.pierleoni and k.esler , phys.rev.e * 68 * , 046707 ( 2003 ) g.h . booth , a.j.w .
thom and a.alavi , j. chem .
phys . * 131 * , 054106 , ( 2009 ) l.delle site , int.j.quant.chem . doi:10.1002/qua.24823 ( 2015 ) .
t.weymuth and m.reiher , int.j.quant.chem .
* 114 * , 823 - 837 ( 2014 ) .
trickey , v.v .
karasiev , and a. vela , phys .
b , * 84 * , 075146 ( 2011 ) l.delle site , _ levy - lieb principle meets quantum monte carlo _ , pg 361 - 375 in * many - electron approaches in physics , chemistry and mathematics : a multidisciplinary view * , v.bach and l.delle site eds . studies in mathematical physics ,
springer international publishing switzerland , 2014 t.hastie , r.tibshirani and j.friedman , _ elements of statistical learning , data mining , inference and prediction _ ,
springer , new york , 2009 .
snyder , m.rupp , k.hansen , k - r .
mller , and k.burke , phys .
lett . * 108 * , 253002 ( 2012 ) .
snyder , m.rupp , k.hansen , l.blooston , k - r .
mller , and k.burke,j .
phys . * 139 * , 224104 ( 2013 ) m.rupp , a.tkatchenko , k - r .
mller , and o. anatole von lilienfeld , phys .
lett . * 108 * , 058301 ( 2012 ) .
a.lopez-bezanilla and o. anatole von lilienfeld , phys .
b * 89 * , 235411 ( 2014 ) .
|
the constrained - search principle introduced by levy and lieb , is proposed as a practical , though conceptually rigorous , link between density functional theory ( dft ) and quantum monte carlo ( qmc ) .
the resulting numerical protocol realizes in practice the implicit key statement of dft : + `` _ given the three dimensional electron density of the ground state of a system of @xmath0 electrons with external potential @xmath1 it is possible to find the corresponding @xmath2-dimensional wavefunction of ground state .
_ '' + from a numerical point of view , the proposed protocol can be employed to speed up the qmc procedure by employing dft densities as a * pre - selection * criterion for the sampling of wavefunctions .
|
in the beautiful book @xcite the title of chapter 12 reads : is nature , underneath it all , a ca ? .
such ideas have in fact been around for quite some time ( cf .
e.g.@xcite,@xcite,@xcite or @xcite , to mention a few references ) .
a little bit later t hooft analysed the possibility of deterministic ca underlying models of quantum field theory or quantum gravity ( @xcite and @xcite are two examples from a long list of papers ) . for more detailed historical information
see @xcite or @xcite . a nice collection of references can also be found in @xcite .
however , we would like to issue a warning against an overly optimistic attitude . while we share the general philosophy uttered in these works , there are some subtle points as t hooft remarks correctly(@xcite ) .
it is no easy task to incorporate something as complex as the typical entanglement structure of quantum theory into the , at first glance , quite simple and local ca - models .
we would like to emphasize that it is not sufficient to somehow simulate or reproduce these quantum phenomena numerically on a computer or ca .
what is actually called for is a structural isomorphism between those phenomena and corresponding emergent phenomena on ca .
this problem has been one of the reasons underlying our interest in ca having a fluctuating time - dependent geometry ( see below ) .
we recently observed that ideas about the discrete fine structure of space - time similar to our own working philosophy have been uttered in chapt .
9 of @xcite , in particular concerning the existence of what we like to call shortcuts or whormhole structure .
still another interesting point is discussed by svozil ( @xcite ) , i.e. the well - known problem of species doubling of fermionic degrees on regular lattices , which , as he argues , carries over to ca . among the various possibilities to resolve this problem he suggests a kind of dimensional reduction ( `` dimensional shadowing '' ) , which leads in the ca one is actually interested in , to non - local behavior ( see also @xcite p.649ff ) .
it is perhaps remarkable that , motivated by completely different ideas , we came to a similar conclusion concerning the importance of non - local behavior ( cf .
@xcite , see also @xcite ) . while presently the discussion in the physics community , when it comes to the high - energy end of fundamental physics , is dominated by string theory and/or loop quantum gravity , frameworks which are in a conceptual sense certainly more conservative , we nevertheless regard an approach to these primordial questions via networks and/or ca as quite promising .
in contrast to the above - mentioned ( more conservative ) approaches which start from continuum physics and hope to detect discrete space - time behavior at the end of the analysis ( for example by imposing quantum theory as a quasi god - given absolute framework on the underlying structures over the full range of scales ) , we prefer a more bottom - up - approach .
one of our reasons for this preference is that we do not believe that quantum theory holds sway unaltered over the many scales addressed by modern physics down to the pristine planckian regime .
like t hooft , we regard quantum theory rather as a kind of effective intermediate framework , which emerges from some more primordial structure of potentially very different nature .
we start from some underlying dynamic , discrete and highly erratic network substratum consisting of ( on a given scale ) irreducible agents interacting ( or interchanging pieces of information ) via elementary channels . on a more macroscopic ( or , rather , mesoscopic ) scale
, we then try to reconstruct the known continuum structures as emergent phenomena via a sequence of coarse graining and/or renormalisation steps ( see @xcite and @xcite ) .
while ca have been widely used in modeling complex behavior of molecular agents and the like ( a catchword being artificial life or conway s game of life ; for a random selection see e.g. @xcite , @xcite , @xcite , @xcite , @xcite or @xcite ) , papers on the more pristine and remote regions of planck - scale physics are understandably less numerous .
when we embarked on such a programme in the early nineties of the last century , we soon realized that the ordinary framework of ca , typically living on fixed and quite regular geometric arrays , appears to be far to rigid and regular in this particular context . in order to implement the lessons of general relativity we have to make their structure dynamical , that is , not only the local states on the vertices of the lattice but also the local states attached to the links need to be dynamic .
a fortiori , we would like the whole wiring diagram of links to be `` clock - time dependent '' . to put it briefly :
matter shall act on geometry and vice versa , where we , tentatively , associate the pattern of local vertex states with the matter distribution and the geometric structure of the network with geometry .
our first task therefore is to turn both the site and the link states into fully dynamical degrees of freedom , which mutually depend on each other in a dynamical and local way .
furthermore , all this is assumed to happen on very irregular arrays of nodes and links which dynamically arrange themselves according to some given evolution law
. then the hope is , that under certain favorable conditions , the system will undergo a ( series of ) phase transition(s ) from , for example , a disordered chaotic initial state into a kind of macroscopically ordered , extended pattern , which may be associated with a classical continuum space - time with some matter living in it .
one of our first ( published ) papers , in which we implemented such a programme was @xcite , see also @xcite ; for the notion of dimension of such irregular structures see @xcite .
we carefully inspected the literature known to us on ca at the time of writing those papers , but only several years later , when one of the authors ( m. requardt ) had the pleasure to review the book by ilachinski , we became aware of slightly earlier related ideas developed by ilachinski and halpern ( see e.g. @xcite or @xcite for reviews and further references ) . in the following sections we are going to relate these two originally independent approaches to each other and discuss the behavior of two interesting dynamical network models we employed and studied in greater detail .
furthermore , we give an overview of an extensive numerical and computational analysis of these model systems .
sdca have been introduced by ilachinski and halpern and are straightforward generalisations of ca ( for a more recent application see e.g. @xcite ) . in the simplest cases they are placed ( as most of ca ) on a finite or infinite regular grid , e.g. @xmath0 .
the generalisation consists of the assumption that also the links , connecting the sites of the lattice , can be created and deleted according to a local law .
more properly , we have link variables , @xmath1 , attaining the value 1 if site @xmath2 is linked to site @xmath3 and being zero otherwise . in this context
it is of course of great relevance which sites can be linked at all . in @xcite or @xcite , for example , links of the original background lattice @xmath4 belong to this pool together with diagonal links to the next - nearest neighbors . in the respective examples some start configuration
is chosen on the euclidean background lattice and one can observe , in the course of clock time , the emergence of additional diagonal links and the subsequent deletion of some of them , as well as deletion and reinsertion of the original horizontal and vertical links .
the local dynamical rule guarantees that only links connecting nearest or next - nearest neighbors participate in the process ( cf .
e.g. section 8.3 in @xcite ) .
this entails that the change of the wiring diagram proceeds still in a rather local and orderly way with respect to the initial euclidean lattice .
these restrictions are of course not necessary . in general , ca can be defined on an undirected graph . at each site
@xmath2 we have attached a site state @xmath5 , being capable of attaining some discrete values ( typically @xmath6 ) while link states @xmath1 can have the values @xmath7 or @xmath8 .
in sdca the wiring diagram , i.e. , the distribution of links , is now also a dynamical evolving structure .
a local law , being independent of the constantly varying wiring diagram can be formulated by employing the natural graph distance metric given by @xmath9 with @xmath10 a path , connecting the sites @xmath2 and @xmath3 and @xmath11 its length , i.e. , the number of links along the path ( this distance being infinite if the sites are in disconnected pieces of the network ) . with this metric
, the graph becomes a discrete metric space .
ball neighborhoods around a site @xmath2 are then defined by @xmath12 ( cf .
e.g. @xcite or @xcite ) . for convenience
we introduce some notation . the underlying time dependent lattice ( the wiring diagram )
is denoted by @xmath13 .
@xmath14 ( or @xmath15 ) designate the local site or link states ( in the simplest case @xmath16 , @xmath17 ) .
@xmath18 is a certain neighborhood of sites and links about the site @xmath2 .
a classical ca is given by a local dynamical law or rule , i.e. , a map from some @xmath18 to @xmath19 , the state space at site @xmath2 .
typically the type of neighborhood and the local rule are chosen to be the same over the full lattice .
things become a little bit more complicated if the wiring diagram is chosen to also become ( clock ) time dependent . in that case
it is more reasonable to define the neighborhoods by the distance metric , i.e. , choose some @xmath20 .
note that now the actual site and link content of @xmath20 is time dependent , while the definition of the neighborhood can be given in a time independent form . from a mathematical point of view
we could formulate rather arbitrary local rules , but physics has taught us to avoid too artificial or cumbersome rules , which depend on rather _ ad hoc _ assumptions .
so , quite reasonable laws appear to be totalistic or outer - totalistic rules , which act on the sum of site states and/or link states in the ball @xmath20 with a possible particular role played by @xmath5 itself . in the general case , the dynamics is given by a pair of local laws : @xmath21 in which we have been a bit sloppy in order not to overburden the formulas with too many indices .
to get an idea how this scheme works in concrete examples , see sect .
8.8 of @xcite or @xcite or the sections below .
+ remark : we would like to emphasize again , that , in the typical examples given above , link deletion or creation is restricted to nearest or next - nearest neighbors with respect to the background lattice ( e.g. @xmath4 ) .
the lattice evolution is hence still quite regular .
this is of some relevance in comparison to our cellular networks , which are capable of developing both local and translocal connections with respect to some reference space .
our networks are defined on general graphs , @xmath22 , with @xmath23 the set of vertices ( sites or nodes ) and @xmath24 the set of edges ( links or bonds ) .
the local site states can assume values in a certain discrete set .
in the examples we have studied , we follow the philosophy that the network should be allowed to find its typical range of states via the imposed dynamics .
that is , we allow the @xmath25 to vary in principle over the set @xmath26 , with @xmath27 a certain discrete quantum of information , energy or whatever
. the link states can assume the values @xmath28 ( we are assuming here the notation @xmath29 instead of @xmath30 as we regard the links as representing a kind of elementary coupling ) . viewed geometrically we associate the states @xmath31 with directed edges pointing from site @xmath32 to @xmath33 , or the other way around , or , in the last case , with a non - existing edge .
that is , at each clock time step , @xmath34 ( @xmath35 an elementary quantum of time ) , we have as underlying substratum a time dependent directed graph , @xmath36 .
our physical idea is that at each clock time step an elementary quantum @xmath27 is transported along each existing directed edge in the indicated direction .
to implement our general working philosophy of mutual interaction of overall site states and network geometry , we now describe some particular network laws , which we investigated in greater detail ( see the following section ) .
we mainly considered two different classes of evolution laws for vertex and edge states : * network type i @xmath37 * network type ii @xmath38 where @xmath39 and @xmath40 .
we see that in the first case , vertices are connected that have very different internal states , leading to large local fluctuations , while for the second class , vertices with similar internal states are connected .
we proceed by making some remarks in order to put our approach into the appropriate context .
+ remarks : + 1 .
it is important that , generically , laws , as introduced above , do not lead to a reversible time evolution , i.e. , there will typically be attractors or state - cycles in total phase space ( the overall configuration space of the node and bond states ) .
on the other hand , there exist strategies ( in the context of cellular automata ! ) to design particular reversible network laws ( cf . e.g. @xcite ) which are , however , typically of second order .
usually the existence of attractors is considered to be important for pattern formation . on the other hand
, it may suffice that the phase space , occupied by the system , shrinks in the course of evolution , that is , that one has a flow into smaller subvolumes of phase space .
2 . in the above class of laws a direct bond - bond interaction
is not yet implemented .
we are prepared to incorporate such a ( possibly important ) contribution in a next step if it turns out to be necessary . in any case
there are not so many ways to do this in a sensible way . stated differently , the class of possible , physically sensible interactions ,
is perhaps not so large .
3 . we would like to emphasize that the ( undynamical ) clock - time , @xmath41 , should not be confused with the `` true '' physical time , i.e. , the time operationally employed on much coarser scales .
the latter is rather supposed to be a collective variable and is expected ( or hoped ! ) to emerge via a cooperative effect .
clock - time may be an ideal element , i.e. , a notion which comes from outside , so to speak , but at least for the time being we have to introduce some mechanism , which allows us to label consecutive events or describe change . the following observation we make because it is relevant if one follows the general spirit of modern high energy physics .
the above dynamical law depends nowhere on the absolute values of the node charges but only on their relative differences . by the same token ,
charge is nowhere created or destroyed .
we have @xmath42 where , for simplicity , we represent the set of sites by their set of indices , @xmath43 , and @xmath44 denotes the difference between consecutive clock - time steps .
put differently , we have conservation of the global node charge . to avoid artificial ambiguities we can , e.g. ,
choose a fixed reference level and take as initial condition the constraint @xmath45 we conclude this subsection by summarizing the main steps of our working philosophy .
irrespective of the technical details of the dynamical evolution law under discussion , the following , in our view crucial , principles should be emulated in order to match fundamental requirements concerning the capability of _ emergent _ and _ complex _ behavior . 1 . as is the case with , say , gauge theory or general relativity , our evolution law on the surmised primordial level should implement the mutual interaction of two fundamental substructures , put sloppily : `` _ geometry _ '' acting on `` _ matter _ '' and vice versa , where in our context `` _ geometry _ ''
is assumed to correspond in a loose sense with the local and/or global bond states and `` _ matter _ '' with the structure of the node states .
2 . by the same token
, the alluded _ self - referential _ dynamical circuitry of mutual interactions is expected to favor a kind of _ undulating behavior _ or _ self - excitation _ above a return to some uninteresting ` _ equilibrium state _ ' as is frequently the case in systems consisting of a single component which directly feeds back on itself .
this propensity for the ` _ autonomous _ ' generation of undulation patterns is in our view an essential prerequisite for some form of `` _ protoquantum behavior _ ''
we hope to recover on some coarse grained and less primordial level of the network dynamics .
in the same sense we expect the overall pattern of switched - on and -off bonds to generate a kind of `` _ protogravity _ '' .
we now put our two cellular network models on a simplex graph with @xmath46 vertices @xmath32 and edges @xmath47 , @xmath48 .
more specifically , the maximally possible number of edges is @xmath49 .
we choose such a simplex graph as initial geometry . as an initial distribution for vertex states ( seed )
we choose a uniform ( random ) distribution scattered over the interval @xmath50 . in an early state of the work we also used other ( initial ) distributions as well but we did not find any significantly different results .
the initial values for edge states @xmath29 were chosen from @xmath51 with equal probability @xmath52 .
in other words , our initial state is a maximally entangled nucleus of vertices and edges and the idea is to follow its unfolding under the imposed evolution laws . in a sense
, this is a scenario which tries to imitate the big bang scenario .
the hope is , that from this nucleus some large - scale patterns may ultimately emerge for large clock - time . for numerical investigations ,
the size of the ca is by necessity rather limited . to obtain an estimate for properties of the large networks , which we are ultimately interested in , we simulated networks of increasing size and tried to extrapolate the expected properties for larger networks .
the average of all vertex states is approximately @xmath7 by construction and the the sum of all temporal changes of vertex states is exactly @xmath7 . for most other properties we found that the average over the width of the initial vertex state distribution , over @xmath53 and @xmath54 , and over specific realizations of initial conditions as well as time , has a linear dependence of the given property on the network size @xmath46 .
figure [ lineardep](a ) shows an example and table [ propertytable ] summarizes the observed dependencies .
a few of the quantities did not show linear dependencies , see figure [ lineardep](b ) and ( c ) . while the standard deviation of spatial fluctuations has an unknown dependency on network size ( [ lineardep](b ) ) , the number of `` off '' bonds clearly scales with the number of total edges , i.e. , with @xmath55 .
this suggests that a random graph approximation with constant probability for `` off '' edges might be possible . .scaling laws for averaged properties of the two network types .
most quantities scale linearly with the number of nodes @xmath46 . some scale with the square of @xmath46 , while others have an unknown scaling behaviour , denoted by a question mark in the table .
[ propertytable ] [ cols="<,<,^,^,^,^,^ " , ] in most results on a single size network we used @xmath56 and @xmath57 .
because of the finite phase space of the ca ( technically it is infinite , but the vertex states only fill a finite interval of @xmath58 due to the nature of the network laws ) , network states will eventually repeat , which leads to a limit cycle because of the memory - less dynamics .
we tested for the appearance of such limit cycles for different network size @xmath46 and to our surprise , networks of type i had with very few exceptions extremely short limit cycles of period @xmath59 .
the exceptions we were able to find , had periods of a multiple of @xmath59 , the longest found ( in a network with @xmath60 ) was @xmath61 .
the prevalence of such short limit cycles is still an open question and beyond this work .
we note in this context that already s. kauffmann observed such short cycles in his investigation of switching nets ( @xcite , @xcite ) and found it very amazing .
such short cycles were also found in random networks ( @xcite ) in a quite different context .
this phenomenon is remarkable in the face of the huge accessible phase spaces of typical models and points to some hidden ordering tendencies in these model classes .
what is even more startling is that this phenomenon prevails also in our case for model class @xmath8 when we introduce a further element of possible disorder by allowing edges to be dynamically created and deleted .
we formulate the following hypothesis .
we conjecture that this important phenomenon has its roots in the self - referential structure ( feed - back mechanisms ) of many of the used model systems .
it is instructive to observe the emergence of such short cycles in very small models on paper , setting for example @xmath62 , i.e. , no switching - off of edges and taking @xmath63 or @xmath64 .
taking , e.g. , @xmath65 and starting from @xmath66 , the network will eventually reach a state @xmath67 . without loss of generality we can assume @xmath68 and @xmath69 .
this state develops into a cycle of length 6 as illustrated in table [ table2]a . for @xmath70
the state eventually becomes @xmath71 , without loss of generality @xmath72 , @xmath73 , @xmath74 , resulting in the dynamics in table [ table2]b .
again , the length of the cycle is @xmath59 .
hence , @xmath59 is a good candidate for a short cycle length , which of course does not explain why such a short length should appear at all .
the transients in networks of type i are also rather short and grow slowly with the network size ( data not shown ) .
\a ) b ) + networks of type ii have much longer limit cycles and transients . because of numerical limitations we were only able to determine cycle lengths for small networks . as shown in figure [ table2]b ) the typical transient and cycle lengths both grow approximately exponentially . to characterize the networks resulting from the two different evolution laws , we measured several key quantities , including the distribution of node degrees ( also called the vertex degree sequence ) , the distribution of node states @xmath32 , the distribution of bond states @xmath75 , as well as temporal and spatial fluctuations of these quantities . the vertex degree distribution in dependence on @xmath53 for a network of @xmath76 vertices
is shown in figure [ figurex]a ) and b ) for network law i and ii respectively .
the second parameter was fixed as @xmath77 to implement a reasonable hysteresis in the dynamical addition and removal of edges and the degrees were observed after a transient of @xmath78 time steps , i.e. , prevalently still in a transient dynamics regime .
the network structure undergoes a series of changes for increasing @xmath79 .
networks of type i evolve from almost fully connected simplex networks to more sparse connectivities with increasing @xmath79 .
there is a regime , where few vertices with very high degree coexist with many vertices with a low degree ( around @xmath80 ) , which is reminiscent of the situation in small world networks .
we , however , observe a bimodal distribution ( with very sharp peaks in each given network , see figure [ figurex ] ) rather than a power law of abundance of node degrees . for large @xmath79
the network eventually breaks apart and all nodes have vertex degree @xmath7 . for networks of type ii
the situation is as expected inverse with respect to @xmath79 .
the networks are trivial with vertex degree @xmath7 for all nodes for small @xmath79 and connect increasingly dense for increasing @xmath79 . in this family of networks ,
the distribution of vertex degrees is always fairly broad and remains such up to large @xmath79 .
we observe an intriguing structure of multiple maxima of the distributions in a wide range of @xmath79 values .
the observed structure of groups of highly connected and less connected vertices in the averaged distributions may arise from each given network realization having these two groups of vertices or appear due to the existence of two different types of networks . to probe these possibilities , we examined the abundance distribution of node degrees for individual initial conditions ( fig .
[ figurey ] ) . in all examples
we observe the same structure as in the averaged picture ( fig .
[ figurex]a ) , such that we have to conclude that there is only one type of network for a given @xmath81 pair that has a structured vertex degree distribution .
furthermore , careful examination shows that this distribution can unlike in the averaged picture be fairly sharp , with often only one or two prevalent values for the vertex degree .
the same is true for networks of type ii : distributions resemble the averaged picture but often with sharp peaks for a single value for the vertex degree ( data not shown ) .
the temporal fluctuations , @xmath82 , of vertex degrees give us insight into the stability of the network structure .
for network law i we observe at @xmath83 an abrupt phase - transition from basically no temporal fluctuations in the node degrees ( `` frozen network '' ) to fairly high fluctuations ( `` liquid network '' ) . for even larger @xmath84
, the fluctuations slowly abate in agreement with the smaller overall vertex degrees .
it is surprising that the transition is so abrupt especially in the face of the much smoother development of the distribution of vertex degrees ( fig .
[ figurex ] ) .
while the node degrees instruct us on the geometry of the ca graph , i.e. , presumably on the structure of space , the state variables @xmath25 tell us something about an `` energy '' or `` mass '' density in some appropriate sense .
as explained above the sum over local states , @xmath85 , is conserved , and hence is the sum of changes @xmath86 .
this does , however , not imply that the distribution of @xmath25 or changes in @xmath25 are trivial .
figure [ figurendval ] shows the maps for different node value distributions depending on @xmath87 and @xmath88 .
the results are , as expected , very different for the two network types . networks of type i have a clear region of intermediate @xmath84 values where the distribution of node states is strongly bimodal for @xmath89 . for smaller @xmath84 ,
i.e. , for less bonds being switched off , there is a broad distribution of node states , and for larger @xmath84 values , we observe a sharp unimodal distribution around @xmath7 corresponding to disconnected graphs , in which the node states are basically frozen in their initial values .
intriguingly , the transitions between the different types of state distributions occur at sharp values of @xmath84 reminiscent of phase transitions .
networks of type ii have no bimodal distributions of node values , but there is a visible modulation in the width of node state distributions with different values of @xmath53 .
for some @xmath84 values , the distribution of observed states is rather sharp , for others rather wide .
it is also interesting to note that , contrary to the naive expectation , the total width of the distributions is slightly larger than for networks of type i ( note the light blue left and right tails for @xmath90 ) .
the fact that nodes with similar node state are connected and can reach some equilibrium and nodes with very different states are separate and can not interact ( equilibrate ) directly seems to allow such `` outlier '' states .
when studying such models as described above , it is of tantamount importance to find certain effective geometric characteristics of the large scale behavior of such irregular ( in the small ) networks .
this holds the more so if one plans to perform some coarse - graining or continuum limit .
it is evident that , in particular in the latter situation , only global features can be significant while , on the other hand , the finer details on small scales should be ignored .
the notion of dimension is one of these fundamental organizing concepts .
there exist of course quite a few different versions of this concept in the mathematical literature , most of which , while looking useful at first glance , have to be dismissed on second thoughts
. we will not give the pros and cons of all the different notions ( a more detailed discussion can e.g. be found in @xcite and @xcite ) , let us only make the following point clear .
the discrete models we are dealing with are not to be understood as some sort of tessellation of a preexisting continuous background manifold like in algebraic topology . in that case
it would for example be reasonable to use the dimensional concept employed in simplicial complexes . in our case ,
exactly the opposite is true .
we view the continuum as an emergent limit structure , being the result of a ( complicated ) renormalisation group like process of coarse - graining . in this process , due to the rescaling of geometric scales ,
more and more vertices and links are absorbed in the infinitesimal neighborhoods of the emerging points of the continuum . as a consequence ,
what amounts to a local definition of dimension in the continuum is actually a large scale concept on the network , involving practically infinitely many vertices and their wiring .
halpern and ilachinski make a certain suggestion in their work ( see e.g. @xcite ) , which also happens to be a distinctly local concept .
they define the effective dimensionality as the ratio of the number of next - nearest neighbors to the number of nearest neighbors averaged over the set of sites of the network .
it is a typical property of most of the different dimensional concepts that they usually coincide on very regular spaces .
this is for example well - known for the various notions of fractal dimension ( cf .
e.g. @xcite or @xcite ) .
for the notion introduced above we have for instance in the case of @xmath4 : @xmath91 , i.e. @xmath92 and correspondingly for higher dimensions .
however , for other lattices which are not so regular or translation invariant , i.e. having a more complicated local neighborhood structure , this is no longer true . while one would still like to associate on physical grounds in many cases their dimension with the corresponding embedding dimension of the ambient space , the above quotient may yield a different value .
the deeper reason is that the accidental near - order of the lattice may differ from its more important far - order .
this phenomenon and the following physical argument motivated us to choose a different notion of intrinsic dimension which has a lot of very nice and desirable properties as has been shown in the papers we cited above .
originally we were primarily motivated by the following reasoning .
what kind of intrinsic global property ( i.e. being independent of some embedding dimension or accidental near - order ) is relevant for the occurrence of phase transitions , critical behavior and the like ?
we wound up with the following answer : it is the increase of number of new agents or degrees of freedom one sees when one starts from a given lattice site and moves outward .
this led to the following definition ( note that we introduce two slightly different notions which again coincide in many cases ) .
the ( upper , lower ) internal scaling dimension with respect to the vertex @xmath2 is given by @xmath93 the ( upper , lower ) connectivity dimension is defined correspondingly as @xmath94 if upper and lower limit coincide , we call it the internal scaling dimension , the connectivity dimension , respectively .
in @xcite we exhibited the close connection of this concept with important properties in various fields of pure mathematics ( growth properties of metric spaces ) .
furthermore , when performing some continuum limit we could show that our notion of dimension makes contact with the various notions of fractal or haussdorff - dimension .
in the preceding sections we introduced and studied networks which are in various respects generalisations of structural dynamic ca , i.e. , networks with both the site states and the wiring ( link - distribution ) being dynamic and ( clock ) time dependent .
we thus realize an entangled dynamics of geometry and matter degrees of freedom as in , say , general relativity or quantum gravity .
in contrast to more ordinary ca we admit a very high degree of disorder in the small while we hope that our network models find an attracting set in phase space after some transient time , thus displaying some patterns of global order .
we are particularly interested in ordering phenomena on the geometric side .
that is , we look for collective geometric properties like , e.g. , dimension , which suggest that , after some coarse graining and/or rescaling , our networks display global smoothness properties which may indicate the transition into a continuum - like macro state . we underpin our investigation by a quite detailed quantitative computational analysis of various ( large scale ) characteristics of our model networks as , e.g. , vertex degree distribution , fluctuation patterns in site and/or link states , etc . there are indications that for certain choices of the parameters , labelling our model networks , we witness something akin to structural phase transitions .
it is particularly noteworthy that one of our model networks ( after a very short transient time ) enters a periodic state of period only six , and this being practically independent of the chosen initial state .
given the huge possible local fluctuations in both site states and link distribution the extremely short period is remarkable . this phenomenon has also been observed in other kinds of networks but nevertheless remains somewhat mysterious .
a.ilachinski : `` cellular automata , a discrete universe '' , world scient . publ . , singapore 2001 k.zuse:``the computing universe '' , int.j.theor.phys .
21(1982)589 r.p.feynman : `` simulating physics with computers '' , int.j.theor.phys .
21(1982)467 e.fredkin : digital mechanics , physica d45(1990)254 or the ( programmatic ) material on his homepage http://www.digitalphilosophy.org d.finkelstein : `` the space - time code '' , phys.rev .
184(1969)1261 g.t hooft : `` quantization of discrete deterministic theories '' , nucl.phys .
b 342(1990)471 g.t hooft : `` quantum gravity as a dissipative deterministic system '' , class.quant.grav .
16(1999)3263 , gr - qc/9903084 s.wolfram : a new kind of science ,
wolfram media 2002 k.svozil : computational universe , chaos , solitons , fractals 25(2005)845 or physics/0305048 g.t hooft : can quantum mechanics be reconciled with ca ? , int.j.theor.phys .
42(2003)349 k.svozil : `` are quantum fields cellular automata ? '' , phys.lett .
119a(1987)153 m.requardt : `` scale free small - world networks and the structure of quantum space - time '' , gr - qc/0308089 a.lochmann,m.requardt : `` an analysis of the transition zone between the various scaling regimes in the small - world model '' , j.stat.phys .
122(2006)255 or cond - mat/0409710 m.requardt : a geometric renormalisation group and fixed - point behavior in discrete quantum space - time
, j.math.phys .
44(2003)5588 or gr - qc/0110077 s.kauffmann : `` origins of order : self - organisation and selection in evolution '' , oxford univ.pr . ,
oxford 1993 s.kauffmann : `` at home in the universe , the search for laws of self - organisation and complexity '' , oxford univ.pr . ,
oxford 1995 s.kauffmann : `` antichaos and adaption '' , sci.am .
265(1991)76 c.g.langton : ed .
`` artificial life : proc . of an interdisciplinary workshop on the synthesis and simulation of living systems , sept .
1987 , los alamos , addison - wesley publ .
1989 m.gardner : `` on ca , self - reproduction . the garden of eden and the game of life '' , sci.am .
224(1971)112 m.requardt : ' ' cellular networks as models for planck - scale physics , j.phys .
a : math.gen .
31(1998)7997 or hep - th/9806135 t.nowotny,m.requardt : pregeometric concepts on graphs and cellular networks , chaos , solitons and fractals 10(1999)469 ( invited paper ) or hep - th/9801199 t.nowotny,m.requardt : dimension theory on graphs and networks , j.phys .
a : math.gen .
31(1998)2447 or hep - th/9707082 p.halpern : sticks and stones : a guide to sdca
, am.j.phys .
57(1989)405 r.alonso-sanz : `` reversible structurally ca with memory '' , preprint 2006 , submitted to j.cell.aut .
m.requardt : the continuum limit of discrete geometries , int.j.geom.meth.mod.phys .
3(2006)285 or math - ph/0507017 t.toffoli , n.margolus : `` cellular automaton machines '' , mit pr . ,
cambridge mass .
1987 r. huerta and m. rabinovich , `` reproducible sequence generation in random neural ensembles '' , phys rev lett 93 : 238104 ( 2004 ) g.a.edgar : `` measure , topology and fractal geometry '' , springer , n.y .
1990 k.j.falconer : `` fractal geometry '' , wiley , chichester 1990
|
we relate structurally dynamic cellular networks , a class of models we developed in fundamental space - time physics , to sdca , introduced some time ago by ilachinski and halpern .
we emphasize the crucial property of a non - linear interaction of network geometry with the matter degrees of freedom in order to emulate the supposedly highly erratic and strongly fluctuating space - time structure on the planck scale .
we then embark on a detailed numerical analysis of various large scale characteristics of several classes of models in order to understand what will happen if some sort of macroscopic or continuum limit is performed .
of particular relevance in this context is a notion of network dimension and its behavior in this limit .
furthermore , the possibility of phase transitions is discussed . 1.0 cm
|
in recent years , the halo occupation distribution and conditional luminosity function have become powerful statistical measures to probe the link between galaxies and their hosting dark matter halos .
although these statistical measures themselves do not give physical explanations of how galaxies form and evolve , they provide important constraints on various physical processes that govern the formation and evolution of galaxies , such as gravitational instability , gas cooling , star formation , merging , tidal stripping and heating , and a variety of feedback processes .
in particular , they constrain how their efficiencies scale with halo mass .
the halo occupation distribution ( hereafter hod ) , @xmath18 , which gives the probability of finding @xmath19 galaxies ( with some specified properties ) in a halo of mass @xmath20 , has been extensively used to study the galaxy distribution in dark matter halos and galaxy clustering on large scales ( e.g. jing , mo & brner 1998 ; peacock & smith 2000 ; seljak 2000 ; scoccimarro et al . 2001 ;
jing , brner & suto 2002 ; berlind & weinberg 2002 ; bullock , wechsler & somerville 2002 ; scranton 2002 ; kang et al .
2002 ; marinoni & hudson 2002 ; zheng et al . 2002 ; magliocchetti & porciani 2003 ; berlind et al . 2003 ; zehavi et al . 2004 , 2005 ; zheng et al . 2005 ; tinker et al .
the conditional luminosity function ( clf ) , @xmath21 , which refines the hod statistic by considering the average number of galaxies with luminosity @xmath22 that reside in a halo of mass @xmath20 , has also been extensively investigated ( yang , mo & van den bosch 2003 ; van den bosch , yang & mo 2003 ; vale & ostriker 2004 , 2006 ; cooray 2006 ; van den bosch et al .
2007a ) and has been applied to various redshift surveys , such as the 2dfgrs , the sloan digital sky survey ( sdss ) and deep2 ( e.g. yan ,
madgwick & white 2003 ; yang et al . 2004 ;
mo et al . 2004 ; wang et al . 2004 ; zehavi et al .
2005 ; yan , white & coil 2004 ) .
these investigations demonstrate that the halo occupation statistics are very useful in establishing and describing the connection between galaxies and dark matter halos .
furthermore , they also indicate that the galaxy / dark halo connection can provide important constraints on cosmology ( e.g.,van den bosch , mo & yang 2003 ; zheng & weinberg 2007 ) .
finally , the hod / clf framework also allows one to split the galaxy population in centrals and satellites , and to describe their properties separately ( e.g. cooray 2005 ; white et al . 2007 ;
zheng et al . 2007 ) .
as has been pointed out in yang et al .
( 2005c ; hereafter y05c ) , a shortcoming of the hod / clf models is that the results are not completely model independent .
typically , assumptions have to be made regarding the functional form of either @xmath18 or @xmath23 .
moreover , in all hod / clf studies to date , the occupation distributions have been determined in an indirect way : the free parameters of the assumed functional form are constrained using _ statistical _ data on the abundance and clustering properties of the galaxy population .
one may hope to circumvent this shortcoming by directly measure the dark matter distribution around galaxies .
such measurements can in principle be obtained through gravitational lensing and x - ray observations . however
, both methods are hampered by requirements on the data quality and uncertainties related to the interpretation of the data .
for instance , weak lensing measurements , which requires high - quality imaging , typically needs to resort to the stacking of many lens galaxies in order to get a detectable signal , but this stacking severely complicates the interpretation in terms of the halo masses of the lens galaxies . in the case of x - ray observations , robust constraints can only be obtained for massive clusters , but even here the interpretation of the data can be complicated due to the presence of substructure and deviations from hydrostatic equilibrium . an alternative method to directly probe
the galaxy - dark halo connection is to use galaxy groups as a representation of dark matter halos and to study how the galaxy population changes with the properties of the groups ( e.g. , y05c ; zandivarez et al .
2006 ; robotham et al .
2006 ; hansen et al .
2007 ) .
recently , we have constructed a large galaxy group catalogue based on the sloan digital sky survey data release 4 ( sdss dr4 ) , using an adaptive halo - based group finder ( yang et al .
2007 ; paper i ; y07 hereafter ) . detailed tests with mock galaxy catalogues have shown that this group finder is very successful in associating galaxies according to their common dark matter halos .
in particular , the group finder performs reliably not only for rich systems , but also for poor systems , including isolated central galaxies in low mass halos .
this makes it possible to study the galaxy - halo connection for systems covering a large dynamic range in masses .
various observational selection effects have been taken into account , especially the survey edge effects and fiber collisions .
the halo masses for the groups are estimated according to the abundance match , using the characteristic group luminosity and stellar masses ( see [ sec_data ] below ) . according to tests with mock galaxy catalogues ,
the halo masses are estimated with a standard deviation of about 0.3 dex . with these well - defined galaxy group catalogues
, one can not only study the properties of galaxies in different groups ( e.g. y05c ; yang et al .
2005d ; collister & lahav 2005 ; van den bosch et al . 2005
; robotham et al . 2006 ; zandivarez et al .
2006 ; weinmann et al . 2006a , b ; van den bosch et al .
2007b ; mcintosh et al .
2007 ) , but also probe how dark matter halos trace the large - scale structure of the universe ( e.g. yang et al . 2005b , 2006
; coil et al . 2006 ; berlind et al . 2007
; wang et al . 2007 in preparation ) . in this paper ,
which is the second in a series , we use the sdss dr4 group catalogue to probe various occupation statistics and measure the clfs for different populations of galaxies .
this paper is organized as follows : in section [ sec_data ] we describe the data ( galaxy and group catalogues ) used in this paper .
section [ sec_clfs ] presents our measurement of the clfs for all , red and blue galaxies
. sections [ sec_central ] , [ sec_hod ] and [ sec_satfrac ] describe the properties of central galaxies , the halo occupation statistics and the fraction of satellite galaxies , respectively . finally , we summarize our results in section [ sec_summary ] . throughout this paper ,
we use a @xmath24cdm ` concordance ' cosmology whose parameters are consistent with the three - year data release of the wmap mission : @xmath25 , @xmath26 , @xmath27 , @xmath28 and @xmath29 ( spergel et al . 2007 ) .
the analysis presented in this paper is based on the sdss dr4 galaxy group catalogue constructed by y07 using an adaptive halo - based group finder ( yang et al .
the related galaxy catalogue is the new york university value - added galaxy catalogue ( nyu - vagc ; blanton et al .
2005b ) , which is based on the sdss data release 4 ( adelman - mccarthy et al .
2006 ) , but with an independent set of significantly improved reductions .
from this catalogue we select all galaxies in the main galaxy sample with redshifts in the range @xmath30 and with a redshift completeness @xmath31 .
as described in y07 , three group samples were constructed from the corresponding galaxy samples : sample i , which only uses the @xmath32 galaxies with measured @xmath33-band magnitudes and redshifts from the sdss , sample ii which also includes 7091 galaxies with sdss @xmath33-band magnitudes but redshifts taken from alternative surveys , and sample iii which includes an additional 38672 galaxies that lack redshifts due to fiber collisions but that are assigned the redshifts of their nearest neighbors ( cf .
zehavi et al .
2002 ) . unless stated otherwise , our analysis is based on sample ii . for comparison
, we also present some results obtained from sample iii .
the magnitudes and colors of all galaxies are based on the standard sdss petrosian technique ( petrosian 1976 ; strauss et al .
2002 ) , have been corrected for galactic extinction ( schlegel , finkbeiner & davis 1998 ) , and have been @xmath34-corrected and evolution corrected to @xmath35 , using the method described in blanton et al .
( 2003a , b ) .
we use the notation @xmath36 to indicate the resulting absolute magnitude in the @xmath33-band .
stellar masses , indicated by @xmath37 , for all galaxies are computed using the relations between stellar mass - to - light ratio and @xmath38 color of bell et al .
( 2003 ; see y07 for details ) . in this study
we separate galaxies into red and blue subsamples according to their bi - normal distribution in the @xmath38 color ( baldry et al .
2004 ; blanton et al . 2005a
; li et al . 2006 ) .
[ fig : data_color ] shows the color - magnitude distribution of the galaxies in our sample ii ( dots ) together with the two peak values of the bi - normal distribution in each absolute magnitude bin ( open circles ) ( cheng li ; private communication ) .
the galaxies are separated into red and blue subsamples using the solid line , which is the best fit to the average of the two peak values in each absolute magnitude bin : @xmath39 where @xmath40 . for each group in our catalogue we have two estimates of its dark matter halo mass @xmath41 : ( 1 ) @xmath42 , which is based on the ranking of the characteristic group luminosity @xmath43 , and ( 2 ) @xmath44 , which is based on the ranking of the characteristic group stellar mass @xmath45 , respectively and @xmath46 are , respectively , the total luminosity and total stellar mass of all group members with @xmath47 . ] . as shown in y07 , these two halo masses agree reasonably well with each other , with a scatter that decreases from @xmath48 dex at the low - mass end to @xmath49 dex at the massive end .
detailed tests using mock galaxy redshift surveys have demonstrated that the group masses thus estimated can recover the true halo masses with a 1-@xmath50 deviation of @xmath51 dex , and are more reliable than those based on the velocity dispersion of group members ( y05c ; weinmann et al .
2006 ; berlind et al .
2006 ; y07 ) .
note also that survey edge effects have been taken into account in our group catalogue : groups that suffer severely from edge effects ( about 1.6% of the total ) have been removed from the catalogue . in most cases , we take the brightest galaxy in the group as the central galaxy ( bcg ) and all others as satellite galaxies . in addition , we also considered a case in which the most massive galaxy ( in terms of stellar mass ) in a group is considered as the central galaxy ( mcg ) .
tests have shown that for most of what follows , these two definitions yield indistinguishable results . whenever the two definitions lead to significant differences , we present results for both .
the conditional luminosity function ( clf ) of galaxies in dark halos , @xmath52 , which describes the average number of galaxies as a function of galaxy luminosity in a dark matter halo of a given mass , plays an important role in our understanding of how galaxies form in dark matter halos ( yang , mo & van den bosch 2003 ; van den bosch , yang & mo ; 2003 ; cooray 2006 ; vale & ostriker 2006 ; van den bosch et al .
we now use our group catalogue to directly determine @xmath23 .
the clf can be estimated by directly counting galaxies in groups . for a given galaxy luminosity @xmath53 ,
there is a limiting redshift , @xmath54 , beyond which galaxies with such a luminosity are not included in the sample .
as discussed in y07 , the group catalogue is complete to a certain limiting redshift for a given halo mass . in order to estimate the clf ,
@xmath55 at a given @xmath53 , we only use groups that are complete to the redshift limit @xmath54 .
the clf is obtained by simply counting the average number of galaxies ( in luminosity bins ) in groups of a given @xmath20 .
we show in fig .
[ fig : clf ] the resulting clfs for groups of different masses .
the contributions of central and satellite galaxies are plotted separately . for comparison ,
results obtained using both @xmath42 and @xmath44 are shown as symbols and dashed lines , respectively .
overall , these two halo masses give consistent results , except that the @xmath42-based clf of the central galaxies in low mass halos is more peaked than the @xmath44-based clf ( see the lower right - hand panel ) .
the errorbars shown in each panel correspond to 1-@xmath50 scatter obtained from 200 bootstrap samples of our group catalogue .
the clf for the total population ( centrals plus satellites ) matches reasonably well the schechter form down to halo masses of @xmath56 . for less massive halos , however , there is a prominent peak in the clf at the bright end due to the contribution of central galaxies , which makes the total clf deviate significantly from the schechter form .
as discussed in y05c , in low - mass halos where the group characteristic luminosity ( @xmath43 ) is dominated by the central galaxy , the @xmath43-@xmath42 conversion can produce an artificial peak in the clf at the bright end .
however , comparing the results obtained here with the test results shown in fig .
9 of y05c indicates that the strong peak in the lower right panel of fig .
[ fig : clf ] can not be entirely due to the @xmath43-@xmath42 conversion .
therefore , in what follows , we will model the clf for central and satellite galaxies separately . fig . [
fig : clf_color ] shows the clfs separately for red ( dashed lines ) and blue ( dotted lines ) galaxies .
note that massive halos clearly contain more red galaxies than blue galaxies ( both centrals and satellites ) , while the opposite applies to low mass halos .
the overall clf shapes for red and blue galaxies , however , are remarkably similar .
to quantify the clfs , we fit each of them with the following model .
we write the total clf as the sum of the clfs of central and satellite galaxies : @xmath57 we assume the contribution from the central galaxies to be a lognormal : @xmath58\,,\ ] ] where @xmath59 is the number of central galaxies per halo .
thus , @xmath60 for all galaxies , @xmath61 ( with @xmath62 being the fraction of red central galaxies ) for red galaxies , and @xmath63 for blue galaxies .
note that @xmath64 is , by definition , the expectation value for the ( 10-based ) logarithm of the luminosity of the central galaxy ; @xmath65 and that @xmath66 .
for the contribution from the satellite galaxies we adopt a modified schechter function : @xmath67.\ ] ] note that this function decreases faster at the bright end than a schechter function .
this clf parameterization has a total of five free parameters : @xmath2 , @xmath68 , @xmath69 , @xmath70 and @xmath71 . in practice , we find that @xmath72 to good approximation .
we therefore adopt @xmath73 , throughout , which reduces the number of free parameters to four .
for all the clfs measured above , the best fit parameters , using the model described by equation [ eq : clf_fit ] , are shown in fig [ fig : fit_clf ] .
results are shown separately for all ( left column panels ) , red ( middle column panels ) and blue ( right column panels ) galaxies .
the error - bars in the upper two rows reflect the 1-@xmath50 scatter obtained from our 200 bootstrap samples . here
we also compare results obtained using samples ii and iii .
note that sample ii does not include any galaxies missed due to fiber collisions , while sample iii includes all such galaxies by assigning each of them the redshift of its nearest neighbor .
although this fiber collision correction works well in roughly 60 percent of all cases , the remaining 40 percent are assigned redshifts that can be very different from their true values ( zehavi et al .
samples ii and iii should therefore be considered as two extremes as far as a treatment of fiber - collisions is concerned .
given that we also have two kinds of halo masses , @xmath42 and @xmath44 , we have a total of four different combinations for which we have determined the clf . the results for all four cases are shown as different symbols in each of the panels of fig [ fig : fit_clf ] . as an illustration of
how well the model fits the data , the solid lines in figs .
[ fig : clf ] and [ fig : clf_color ] indicate the corresponding best - fit models .
the upper row of fig . [ fig : fit_clf ] shows the best fit normalization of the clf for satellite galaxies , which describes the average number of satellite galaxies with luminosity @xmath74 in a group of a given halo mass .
as expected , sample iii gives a higher @xmath69 , especially for low - mass groups .
comparing @xmath69 for red ( upper middle panel ) and blue ( upper right panel ) galaxies , one sees that the fraction of red satellites increases with halo mass .
the second row shows the faint end slopes of the clfs , @xmath70 . in massive halos with @xmath75 ,
@xmath70 decreases ( i.e. , becomes more negative ) with increasing halo mass , both for red and for blue galaxies . in halos with @xmath76
, however , @xmath70 decreases with increasing halo mass for red satellites , but increases with @xmath77 for blue satellites , while the faint - end slope for the entire satellite population ( red and blue combined ) is roughly constant at @xmath78 .
the third row of panels shows that @xmath79 increases with halo mass , for both red and blue centrals .
a more detailed discussion regarding this and other properties of central galaxies will be presented in section [ sec_central ] ) .
finally , the last row of fig [ fig : fit_clf ] shows the width of the log - normal clf of central galaxies .
for the combined sample of red and blue galaxies we find an average value of @xmath80 this is in good agreement with constraints obtained by yang et al .
( 2003 ) and cooray ( 2006 ) from clustering and abundances of galaxies in the 2dfgrs and sdss , respectively .
however , zheng et al . ( 2007 ) found , based on hods models for the sdss and deep2 , that the log - normal width increases from @xmath81 for massive halos with @xmath82 to @xmath51 for low mass halos with @xmath83 .
this is quite different from our results , which suggest that @xmath84 _ decreases _ with decreasing halo mass . however , this is most likely an artefact of the method that we used to assign halo masses to our groups . as mentioned above , our halo masses are based on the ranking of either @xmath43 or @xmath85 , which implies that we have assumed a one - to - one relation between halo mass and these two indicators . at the low mass end , both @xmath43 and @xmath46
are expected to be strongly correlated with the luminosity of corresponding central galaxies .
therefore the values of @xmath68 obtained here should be considered as lower limits , in particular for low mass halos .
ccccccc & [ 14.40 , 15.00 ) & @xmath86 & @xmath87 & @xmath88 & @xmath89 & @xmath90 + & [ 14.10 , 14.40 ) & @xmath91 & @xmath92 & @xmath93 & @xmath94 & @xmath95 + & [ 13.80 , 14.10 ) & @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 + & [ 13.50 , 13.80 ) & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 + all & [ 13.20 , 13.50 ) & @xmath106 & @xmath107 & @xmath108 & @xmath109 & @xmath110 + & [ 12.90 , 13.20 ) & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 + & [ 12.60 , 12.90 ) & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 + & [ 12.30 , 12.60 ) & @xmath121 & @xmath122 & @xmath123 & @xmath124 & @xmath125 + & [ 12.00 , 12.30 ) & @xmath126 & @xmath127 & @xmath128 & @xmath129 & @xmath130 + + & [ 14.40 , 15.00 ) & @xmath86 & @xmath131 & @xmath132 & @xmath133 & @xmath134 + & [ 14.10 , 14.40 ) & @xmath91 & @xmath135 & @xmath136 & @xmath137 & @xmath138 + & [ 13.80 , 14.10 ) & @xmath96 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + & [ 13.50 , 13.80 ) & @xmath101 & @xmath143 & @xmath144 & @xmath145 & @xmath146 + red & [ 13.20 , 13.50 ) & @xmath106 & @xmath147 & @xmath148 & @xmath149 & @xmath150 + & [ 12.90 , 13.20 ) & @xmath111 & @xmath151 & @xmath152 & @xmath153 & @xmath154 + & [ 12.60 , 12.90 ) & @xmath116 & @xmath155 & @xmath156 & @xmath157 & @xmath158 + & [ 12.30 , 12.60 ) & @xmath121 & @xmath159 & @xmath160 & @xmath161 & @xmath162 + & [ 12.00 , 12.30 ) & @xmath126 & @xmath163 & @xmath164 & @xmath165 & @xmath166 + + & [ 14.40 , 15.00 ) & @xmath86 & @xmath167 & @xmath168 & @xmath169 & @xmath170 + & [ 14.10 , 14.40 ) & @xmath91 & @xmath171 & @xmath172 & @xmath173 & @xmath174 + & [ 13.80 , 14.10 ) & @xmath96 & @xmath175 & @xmath176 & @xmath177 & @xmath178 + & [ 13.50 , 13.80 ) & @xmath101 & @xmath179 & @xmath180 & @xmath181 & @xmath182 + blue & [ 13.20 , 13.50 ) & @xmath106 & @xmath183 & @xmath184 & @xmath185 & @xmath186 + & [ 12.90 , 13.20 ) & @xmath111 & @xmath187 & @xmath188 & @xmath189 & @xmath190 + & [ 12.60 , 12.90 ) & @xmath116 & @xmath191 & @xmath192 & @xmath193 & @xmath194 + & [ 12.30 , 12.60 ) & @xmath121 & @xmath195 & @xmath196 & @xmath197 & @xmath198 + & [ 12.00 , 12.30 ) & @xmath126 & @xmath199 & @xmath200 & @xmath201 & @xmath202 + [ tab : clf ] for reference , table [ tab : clf ] lists the _
average _ values of the clf fitting parameters obtained from the four combinations of samples ii and iii and group masses @xmath42 and @xmath44 . the errorbars indicate the scatter among these four combinations or the scatter obtained from the 200 bootstrap samples , whichever is larger .
as shown in weinmann et al . ( 2006b ) and baldry et al . ( 2006 ) , the red and blue fractions of galaxies as function of halo mass provide important constraints for models of galaxy formation . fig .
[ fig : f_red ] shows the red fractions of centrals ( left - hand panel ) and satellites ( right - hand panel ) as function of halo mass obtained from our group catalogue ( solid and dashed lines ) .
these fractions are obtained by simply dividing the number of red centrals ( satellites ) by the number of all centrals ( satellites ) in a given halo mass bin .
for both centrals and satellites the red fraction increases with increasing halo mass .
for comparison , results are shown for both @xmath42 ( solid lines ) and @xmath44 ( dashed lines ) .
these two different halo mass estimates give quite different results for central galaxies , especially for halos with @xmath203 .
the origin of this discrepancy can be understood in terms of fig .
[ fig : data_color ] . in this color - magnitude diagram
, the galaxies are separated into red and blue populations .
if we would convert this color - magnitude diagram into a color - stellar mass diagram , the red and blue populations would shift slightly towards higher and lower stellar masses , respectively ( according to the eq . 2 in y07 ) .
therefore , a certain upper percentile of galaxies that is ranked according to stellar mass will have a larger red fraction than the same percentile ranked according to luminosity .
because of the tight correlation between the luminosity of the central galaxy and @xmath42 , and between the stellar mass of the central galaxy and @xmath44 , it is clear that centrals in a given bin of @xmath44 have a larger red fraction than those in the same bin of @xmath42 .
we can also determine the red fractions of centrals from our best - fit clf parameterizations ( equation [ [ eq : phi_c ] ] ) . as shown above
, the parameterization of @xmath204 involves the parameter @xmath205 , which describes the fraction of central galaxies that are red .
for comparison , the symbols in the left - hand panel of fig .
[ fig : f_red ] show the best - fit values of @xmath206 obtained from the clfs for the @xmath42 masses ( open circles ) and @xmath44 masses ( solid squares ) .
it is reassuring that these best fit results match the direct determination of the red fractions reasonably well .
the small differences between the direct measurement ( lines ) and clf measurement ( symbols ) owe to the fact that in the latter case we have normalized the galaxies using the number of groups within the redshift limit @xmath54 ( see section [ sec_clfs ] ) .
the results shown in the left panel indicate that more than 80% of the central galaxies in halos more massive than @xmath207 are red galaxies , while in smaller halos with masses @xmath208 , less than 50% of the centrals
as discussed in y07 , @xmath44 may represent the true halo mass better than @xmath42 .
the result based on @xmath44 shows that @xmath206 decreases rapidly with decreasing halo mass at @xmath209 . the right panel of fig .
[ fig : f_red ] shows the fraction of red satellite galaxies as a function of halo mass .
as one can see , for satellite galaxies @xmath62 increase steadily from about 40% for halos with @xmath210 to about 80% in massive halos with @xmath211 . in van den bosch
et al . ( 2007b ) we have used this information to constrain the efficiency with which the star formation of galaxies is quenched once they become a satellite galaxy ( i.e. , after they are accreted into a larger halo ) .
we now turn to a more detailed investigation of the properties of central galaxies in our group catalogue .
we re - iterate that a central galaxy is either defined as the brightest group member ( bcg ) or the most massive group member ( mcg ) .
in fig . [ fig : l_c ] , we show the luminosity - halo mass ( left panel ) and stellar mass - halo mass ( right panel ) relations .
the solid circles and shaded areas indicate the median and 68% confidence levels of these relations obtained using @xmath42 as halo mass , while the open squares with error - bars are the results obtained using @xmath44 as halo mass .
clearly , and not surprisingly , the luminosity ( stellar mass ) of the bcg ( mcg ) increases with halo mass . in both cases
we find that the slope of the relation decreases significantly with increasing halo mass , in good agreement with previous results ( e.g. , vale & ostriker 2004,2006 ; cooray 2005 ; yang et al .
2003 ; y05c ; van den bosch et al .
the physical reason for this change in slope is probably a combination of agn feedback , and changes in the efficiencies of radiative cooling and dynamical friction ( e.g. lin et al .
2004 ; dekel 2004 ; cooray & milosavljevi 2005 ) . to quantify the @xmath212 relation shown in the left panel , we fit the data using the following function , @xmath213
this model contains four free parameters : a normalized luminosity , @xmath214 , a characteristic halo mass , @xmath215 , and two slopes , @xmath216 and @xmath217 .
the solid line shown in the left panel is the best fit to the average @xmath212 relation .
note that using @xmath42 or @xmath44 as the halo mass does not lead to any significant changes in the results .
the best fitting parameters are [ @xmath218 , @xmath219 , @xmath216 , @xmath217 ] = [ 10.45 , 12.54 , 0.17 , 0.51 ] .
thus , according to eq .
[ eq : lc_fit ] , @xmath2 scales with @xmath41 roughly as @xmath221 for halos with @xmath222 , and as @xmath223 for halos with @xmath224 .
unfortunately , since we do not have data for halos with @xmath225 , any change of behavior at the low - mass end is not constrained . for the @xmath226 relation shown in the right panel
, we use a similar function to fit the data : @xmath227 the solid line shown in the right panel is the best fit of this model to the data , and the best - fit parameters are [ @xmath228 , @xmath229 , @xmath216 , @xmath217 ] = [ 10.86 , 12.08 , 0.22 , 1.61 ] .
thus , @xmath230 scales with @xmath41 as @xmath231 for halos with @xmath232 , and as @xmath233 for halos with @xmath234 .
an useful quantity to describe the difference between the central and satellite galaxies is the so - called ` luminosity - gap ' between the bcg and the brightest satellite galaxy in a given halo .
this ` luminosity - gap ' holds important information regarding the formation and evolution of galaxies . for example , as discussed in donghia et al .
( 2005 ) and milosavljevi et al .
( 2006 ) , the luminosity - gap may be used to quantify the dynamical age of a system of galaxies : halos with @xmath235 close to unity must be relatively young , as dynamical friction causes multiple luminous galaxies in the same halo to merge on a relatively short time scale .
put differently , the distribution of @xmath235 holds important information regarding the importance of galactic cannibalism for bcgs ( e.g. , tremaine & richstone 1977 ; vale & ostriker 2007 ) using our group catalogue , it is straightforward to measure @xmath235 , as well as the stellar mass equivalent , @xmath236 , as function of group mass , as long as the group has at least two members . in the case of ` isolated galaxies ' , i.e. , groups with only one member ( the central ) , however , there are two possibilities : either the galaxy is truly isolated , in that its halo contains no satellite galaxies , or the halo contains one or more satellite galaxies that are fainter than the flux limit of the survey . to bracket the extremes we proceed as follows . for isolated galaxies we either assume that the brightest satellite has zero luminosity ( case 1 ) or that its luminosity is such that its apparent magnitude limit is equal to the magnitude limit of our sample ( case 2 ) .
[ fig : l1l2 ] shows the results thus obtained for both the luminosity gap ( left - hand panel ) and the stellar mass gap ( right - hand panel ) .
symbols and shaded areas indicate the medians and 68% confidence intervals , respectively , and results are shown for both case 1 ( open squares ) and case 2 ( solid dots ) . for groups with @xmath237 ,
case 1 and 2 yield identical results , simply because all these groups contain at least two member galaxies above the magnitude limit of the survey . in the most massive halos , with @xmath238 ,
the median gap is @xmath239 .
the median luminosity or stellar mass gap can be reliably measured down to a halo masses of @xmath240 , where the median values are @xmath241 and @xmath242 . for halos with @xmath243
, however , cases 1 and 2 yield very different results , indicating that the flux limit of the survey severely impedes our ability to accurately measure the gap statistics . finally we emphasize that the luminosity - gap statistics obtained from samples ii and iii separately are very similar , indicating that fiber collisions do not have a strong impact on the results presented here . as pointed out by tremaine & richstone ( 1977 ; and references therein ) , the ` luminosity - gap ' can be used to determine whether the bcgs in galaxy groups are statistically drawn from the same distribution as the satellite galaxies or whether they are ` special ' .
if the average magnitude difference between the bcgs and the brightest satellite galaxies is smaller than the standard deviation in the magnitudes of the bcgs , then they are consistent with being draw from the same distribution . to test this , the solid curve in the left - hand panel of fig .
[ fig : l1l2 ] indicates the best - fit values of @xmath244 obtained from the clfs .
a comparison with the median luminosity gaps suggests that the bcgs in groups , especially with @xmath245 , form a ` special ' subclass , in that their luminosities can not be considered the extreme values of the distribution of satellite luminosities . as a cautionary remark
, we emphasize that because of the method used to assign halo masses to the groups , the value of @xmath68 may be underestimated , especially for low mass groups ( see discussion in section [ sec_clfs ] ) .
however , even if @xmath68 were underestimated by a factor two at the low mass end , this would not change our conclusion that bcgs are special in low mass halos .
the ` luminosity - gap ' can also be described using its distribution for groups of a given mass . in the upper panels of fig .
[ fig : pl1l2 ] we show the distribution of @xmath246 for groups in the above mentioned case 1 ( unshaded histograms ) and case 2 ( shaded histograms ) , respectively .
results are shown for three different mass bins , as indicated in the panels . in the lower panels of fig .
[ fig : pl1l2 ] , results are shown for the corresponding stellar mass gap distributions .
one can see , the width of the distribution increases with decreasing halo mass .
systems with a relatively large luminosity gap , which most likely owes to the fact that the brightest galaxies in the halo have merged , are often termed `` fossil groups '' and have received a significant amount of attention in the recent literature ( see vikhlinin et al . 1999 ;
jones et al . 2003 ; donghia et al .
2005 ; milosavljevi et al . 2006 ; sommer - larsen 2006 ; van den bosch et al .
2007a ; sales et al .
2007 ; von benda - beckmann et al .
2007 ) . following jones et al . ( 2003 ) and milosavljevi et al .
( 2006 ) we define systems in which the brightest satellite galaxy is at least 2 magnitudes fainter than the bcg ( i.e. , @xmath247 , indicated as the dotted vertical lines in fig .
[ fig : pl1l2 ] ) , as `` fossil '' systems . from the sdss dr4 group catalogue
, we obtain that the fraction of fossil systems increases from @xmath248 percent for groups with @xmath249 , to @xmath250 percent for groups with @xmath251 , @xmath252 percent for groups with @xmath82 , and @xmath253 percent for groups with @xmath254 .
these results are in good agreement with a similar analysis of galaxy groups in the 2dfgrs by van den bosch et al .
( 2007a ) . on the other hand ,
jones et al .
( 2003 ) obtained an incidence rate of @xmath255 to @xmath256 percent for systems with an x - ray luminosity from diffuse , hot gas of @xmath257 .
however , since the groups in our sdss dr4 catalogue are not x - ray selected , a detailed comparison with our results is not possible . in a recent paper , donghia et al .
( 2005 ) used detailed hydrodynamical simulations to predict the fraction of halos with @xmath258 that have @xmath259 . from a total of twelve simulated groups ,
they obtain a fossil fraction of @xmath260 percent .
this value is much higher than the fraction of fossil systems we find in the sdss , which suggests a potential over - merging problem in their simulations .
more recently von benda - beckmann et al .
( 2007 ) used a combination of n - body simulations and empirical models for the connection between galaxy luminosity and halo mass ( taken from cooray & milosavljevi ) 2005 ) , and found that the fossil group fraction is about 24% among all systems with masses @xmath261 .
this is in good agreement with our direct measurement from the sdss data .
ccc @xmath262 & @xmath263 & @xmath264 + @xmath265 & @xmath266 & @xmath267 + @xmath268 & @xmath269 & @xmath270 + @xmath271 & @xmath272 & @xmath273 + @xmath274 & @xmath275 & @xmath276 + @xmath277 & @xmath278 & @xmath279 + @xmath280 & @xmath281 & @xmath282 + @xmath283 & @xmath284 & @xmath285 + [ tab : hos ] the upper panels of fig .
[ fig : n_mh ] show the mean halo occupation numbers , @xmath286 ( the mean number of all group members ) and @xmath287 ( the mean number of satellites ) , as functions of halo mass @xmath41 .
results are shown only for @xmath288 , but adopting @xmath289 gives very similar results .
the results shown in the left , middle and right panels correspond to galaxies in different luminosity ranges , as indicated in the panels .
similar to what is found in y05c , the sharp break at the low mass end at @xmath290 . ]
indicates an almost deterministic relation between the luminosity of a central galaxy and the mass of its dark matter halo .
the shoulder suggests that the brightest satellite galaxies is in general much fainter than the central galaxy ( e.g. zheng et al . 2005 ) . the mean number of satellite galaxies , which is shown as the dashed line , reveals a very good power law feature . to quantify this .
we model the halo occupation for satellite galaxies with @xmath291 , where @xmath292 is a characteristic halo mass above which there is on average at least one satellite galaxies and @xmath293 is the power law index .
the best fit parameters for satellite galaxies with different absolute magnitude cuts are listed in table [ tab : hos ] . as an illustration
, we show in the upper panels of fig .
[ fig : n_mh ] the corresponding best fit model predictions of the halo occupation of satellite galaxies as the dotted lines .
they all agree with the data extremely well .
in addition to the mean halo occupation number , we also investigate the second moment of the halo occupation distribution ( see y05c ) . here
we only consider satellite galaxies , because by definition a central galaxy is always assigned to a group in a deterministic way .
this quantity is crucial in modelling the two - point correlation function of galaxies on small scales ( e.g. , benson et al .
2000 ; berlind et al .
2003 ; y05c ; tinker et al .
2007 ) , and holds important information regarding the physical processes related to galaxy formation . in earlier investigations , a number of simple models were adopted to describe the second moment of the halo occupation distribution and its dependence on halo mass ( e.g. , berlind & weinberg 2002 ; yang et al .
in particular , using the group samples constructed from the 2dfgrs , y05c measured the second moment for _ all _ group members and found that the halo occupation distribution is close to poissonian in massive halos and significantly sub - poissonian in low mass halos . in the middle row of fig .
[ fig : n_mh ] , we show @xmath294 for _ satellite _ galaxies as a function of halo mass @xmath41 .
as one can see , @xmath295 is roughly proportional to @xmath296 . to see how the distribution deviates from a poisson distribution ,
we show , in the lower panels of fig [ fig : n_mh ] , @xmath297 as a function of @xmath298 .
a poisson distribution has @xmath299 ( doted lines ) , while a deterministic distribution has @xmath300 .
the results thus indicate that the number distribution of satellite galaxies follows roughly a poisson distribution .
[ fig : ns_mh ] shows the distribution of the number of satellite galaxies in groups , @xmath301 , for different halo mass bins .
the thick solid curves indicate poisson distributions with the same @xmath302 . as one can see , the observed @xmath301-distributions are well fitted by poisson distributions .
these properties , which have already been found in y05c , suggest a direct link between satellite galaxies and dark matter sub - halos . in a recent study ,
kravtsov et al . (
2004 ) , using large numerical simulations , have shown that the occupation distribution of dark matter sub - halos follows poisson statistics .
this suggests that there may be a tight link between satellite galaxies and dark matter sub - halos , which has been a standard assumption in various hod / clf models ( e.g. , vale & ostriker 2004 ; 2006 ) .
our results provide observational support for such a link .
cccc -22.34 & 0.021 @xmath303 0.001 & 0.023 @xmath303 0.001 & 0.017 @xmath303 0.003 + -21.95 & 0.053 @xmath303 0.010 & 0.058 @xmath303 0.008 & 0.040 @xmath303 0.018 + -21.56 & 0.100 @xmath303 0.012 & 0.116 @xmath303 0.013 & 0.069 @xmath303 0.011 + -21.18 & 0.141 @xmath303 0.025 & 0.167 @xmath303 0.029 &
0.104 @xmath303 0.021 + -20.80 & 0.186 @xmath303 0.040 & 0.230 @xmath303 0.045 & 0.135 @xmath303 0.033 + -20.41 & 0.230 @xmath303 0.047 & 0.289 @xmath303 0.055 & 0.164 @xmath303
0.036 + -20.01 & 0.264 @xmath303 0.052 & 0.340 @xmath303 0.062 & 0.185 @xmath303
0.039 + -19.61 & 0.306 @xmath303 0.050 & 0.400 @xmath303 0.060 & 0.211 @xmath303
0.037 + -19.22 & 0.336 @xmath303 0.056 & 0.453 @xmath303 0.067 & 0.233 @xmath303
0.040 + -18.82 & 0.358 @xmath303 0.062 & 0.505 @xmath303 0.078 & 0.253 @xmath303 0.042 + -18.41 & 0.372 @xmath303 0.054 & 0.548 @xmath303 0.069 & 0.269 @xmath303
0.033 + -18.01 & 0.382 @xmath303 0.057 & 0.584 @xmath303 0.080 & 0.275 @xmath303
0.031 + -17.61 & 0.396 @xmath303 0.064 & 0.624 @xmath303 0.083 & 0.280 @xmath303 0.035 + -17.21 & 0.400 @xmath303 0.070 & 0.650 @xmath303 0.086 & 0.288 @xmath303 0.044 + + 11.28 & 0.061 @xmath303 0.009 & 0.062 @xmath303 0.008 & 0.049 @xmath303 0.026 + 11.08 & 0.108 @xmath303 0.016 & 0.112 @xmath303 0.015 & 0.084 @xmath303 0.022 + 10.89 & 0.155 @xmath303 0.029 & 0.164 @xmath303 0.030 & 0.123 @xmath303 0.022 + 10.70 & 0.197 @xmath303 0.039 & 0.226 @xmath303 0.042 & 0.146 @xmath303
0.031 + 10.50 & 0.233 @xmath303 0.047 & 0.287 @xmath303 0.051 & 0.166 @xmath303 0.041 + 10.30 & 0.260 @xmath303 0.049 & 0.343 @xmath303 0.051 & 0.185 @xmath303 0.045 + 10.10 & 0.282 @xmath303 0.052 & 0.401 @xmath303 0.052 & 0.199 @xmath303 0.046 + 9.91 & 0.300 @xmath303 0.057 & 0.461 @xmath303 0.064 & 0.220 @xmath303 0.047 + 9.70 & 0.304 @xmath303 0.057 & 0.507 @xmath303 0.055 & 0.237 @xmath303 0.050 + 9.51 & 0.317 @xmath303 0.054 & 0.557 @xmath303 0.057 & 0.255 @xmath303 0.046 + 9.31 & 0.328 @xmath303 0.054 & 0.587 @xmath303 0.066 & 0.266 @xmath303 0.041 + 9.11 & 0.336 @xmath303 0.053 & 0.610 @xmath303 0.064 & 0.271 @xmath303 0.039 + [ tab : fsat ] the satellite fraction as function of luminosity , @xmath304 , is an important quantity for a proper interpretation of measurements of galaxy - galaxy lensing ( e.g. , guzik & seljak 2002 ; mandelbaum et al .
2006 ; yang et al .
2006 ) and pairwise velocity dispersion of galaxies ( e.g. , jing & brner 2004 ; slosar , seljak & tasitsiomi 2006 ; van den bosch et al .
2007a ; li et al .
in addition , since the halo bias depends on halo mass ( mo & white 1996 ) , and since a satellite galaxy resides in a more massive halo than a central galaxy of the same luminosity ( van den bosch et al .
2007b ) , the large scale clustering of galaxies of a given luminosity also depends strongly on the fraction of satellite galaxies .
here we estimate @xmath305 directly from our group catalogue . according to the test we carried out in the previous section ,
this fraction can be determined relatively accurately . in the left - hand panel of fig .
[ fig : f_sat ] we show @xmath304 as a function of galaxy luminosity .
the results are plotted separately for all ( solid lines ) , red ( dashed lines ) and blue ( dotted lines ) galaxies . since
fiber collisions are expected to significantly impact the number of close pairs , it can affect the satellite fractions @xmath306 . to assess the uncertainty induced by the fiber collisions , we show results for both samples ii and iii , as thick and thin lines , respectively . as mentioned earlier , samples ii and iii may respectively under- and over - estimate the number of group members because of their different treatments of fiber collisions .
most likely the satellite fractions of these two extreme cases bracket the true satellite fractions .
as one can see , the satellite fraction decreases with increasing luminosity , from @xmath307 at @xmath308 to @xmath309 at @xmath310 .
the satellite fraction of red galaxies at the faint end , @xmath311 at @xmath308 , is significantly higher than that of blue galaxies , @xmath312 at @xmath308 .
the satellite fraction as a function of galaxy stellar mass is shown in the right panel of fig .
[ fig : f_sat ] .
the overall behavior here is similar to that shown in the left panel . for reference
, we list in table [ tab : fsat ] the fraction of satellites as a function of galaxy luminosity or stellar mass , separately for all , red and blue galaxies .
the averages listed in this table are the averages between sample ii and iii , while the errors quoted represent the deviations of the samples from the average . in recent years
, the fraction of satellite galaxies has been studied extensively using hod / clf models ( cooray 2006 ; tinker et al .
2007 ; van den bosch et al .
2007a ) , and analyses of galaxy - galaxy lensing measurements ( e.g. mandelbaum et al .
2006 ) . for comparison , we overplot in fig .
[ fig : f_sat ] the results obtained by mandelbaum et al . for early - type galaxies ( solid circles ) and late - type galaxies ( open triangles ) , with error - bars indicating the 95% confidence level .
although their samples are defined differently from ours ( they separate galaxies into early- and late - types according to galaxy morphologies , while we separate galaxies according to colour ) , our results match theirs quite well .
using a large galaxy group catalogue constructed from the sdss data release 4 ( dr4 ) by y07 , we have investigated various halo occupation statistics of galaxies .
in particular , we have split the galaxy population in red and blue galaxies , and in centrals and satellites , and determined the conditional luminosity functions of these varies subsamples .
we have also presented luminosity gap statistics , satellite fractions , and halo occupation numbers for the galaxies in our group sample .
the main results are summarized as follows : 1 .
the conditional luminosity functions for central and satellite galaxies can be well modelled with a log - normal distribution and a modified schechter form , respectively .
the corresponding best fitting parameters are listed in table 1 . 2 .
the average scatter of the log - normal luminosity distribution of central galaxies decreases from @xmath313 dex at the massive end ( @xmath314 \ga 13.5 $ ] ) to @xmath315 dex at the low mass end ( @xmath316 \sim 12.0 $ ] ) .
however , due to the method used to assign halo masses to the groups , at the low mass end this should be considered a lower limit on the true amount of scatter .
the slope of the relation between the average luminosity of central galaxies ( in the @xmath317-band ) and halo mass , @xmath318 , decreases from @xmath319 for @xmath316 \ll 12.5 $ ] to @xmath320 for@xmath314 \gg 12.5 $ ] . for the stellar masses of the central galaxies we obtain that @xmath321 , decreases from @xmath322 for @xmath316 \ll 12.1 $ ] to @xmath323 for@xmath324 \gg 12.1 $ ] .
the halo ( group ) occupation numbers of satellite galaxies accurately follow poisson statistics .
since the same applies to dark matter sub - halos , this supports the standard picture that satellite galaxies are associated with dark matter sub - halos .
5 . in massive halos with masses
@xmath325 roughly 85% ( 80% ) of the central ( satellite ) galaxies are red .
these red fractions decrease to 50% ( 40% ) in halos with masses @xmath1 .
6 . by comparing the scatter in the luminosities of bcgs to the luminosity difference between the bcg and its brightest satellite , we find that the bcgs form a ` special ' subclass , in that their luminosities can not be considered the extreme values of the distribution of satellite luminosities , expecially in halos with masses @xmath326
the fractions of fossil groups , which are defined as groups a with luminosity gap @xmath327 , decreases with increasing of halo mass from 18%-60% in halos with @xmath328 to @xmath329 in halos with @xmath330 .
the satellite fractions obtained from our group catalogue as functions of both luminosity and stellar mass ( listed in table [ tab : fsat ] ) are in good agreement with independent constraints from analyses of galaxy clustering and galaxy - galaxy lensing .
these results can be used to constrain the various physical processes related to galaxy formation and to interpret the various statistics used to describe large scale structures ( e.g. , galaxy correlation functions , pairwise velocity dispersions , etc . ) .
most of our findings are in good agreement with previous studies ( e.g. y05c , zandivarez et al .
2006 ; robotham et al .
2006 ) and can be linked to the semi - analytical modelling of galaxy formations ( e.g. , kang et al . 2005 ; zheng et al . 2005 ; bower et al .
2006 ; croton et al .
2006 ; de lucia et al .
the luminosity and stellar mass gap can be used to probe the specific formation properties of central galaxies ( e.g. , vale & ostriker 2007 ) .
the fraction of the red and blue populations for central and satellite galaxies can be used to probe the color evolution of satellite galaxies ( ven den bosch et al .
we thank j.p . ostriker and zheng zheng for helpful comments , and cheng li for the galaxy color bi - normal fitting .
xy is supported by the _ one hundred talents _ project , shanghai pujiang program ( no .
07pj14102 ) , 973 program ( no .
2007cb815402 ) , the cas knowledge innovation program ( grant no .
kjcx2-yw - t05 ) and grants from nsfc ( nos .
10533030 , 10673023 ) .
hjm would like to acknowledge the support of nsf ast-0607535 , nasa aisr-126270 and nsf iis-0611948 .
funding for the sdss and sdss - ii has been provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss web site is http://www.sdss.org/. the sdss is managed by the astrophysical research consortium for the participating institutions .
the participating institutions are the american museum of natural history , astrophysical institute potsdam , university of basel , cambridge university , case western reserve university , university of chicago , drexel university , fermilab , the institute for advanced study , the japan participation group , johns hopkins university , the joint institute for nuclear astrophysics , the kavli institute for particle astrophysics and cosmology , the korean scientist group , the chinese academy of sciences ( lamost ) , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , ohio state university , university of pittsburgh , university of portsmouth , princeton university , the united states naval observatory , and the university of washington .
|
we investigate various galaxy occupation statistics of dark matter halos using a large galaxy group catalogue constructed from the sloan digital sky survey data release 4 ( sdss dr4 ) with an adaptive halo - based group finder .
the conditional luminosity function ( clf ) , which describes the luminosity distribution of galaxies in halos of a given mass , is measured separately for all , red and blue galaxies , as well as in terms of central and satellite galaxies .
the clfs for central and satellite galaxies can be well modelled with a log - normal distribution and a modified schechter form , respectively . about 85% of the central galaxies and about 80% of the satellite galaxies in halos with masses @xmath0 are red galaxies .
these numbers decrease to 50% and 40% , respectively , in halos with @xmath1 . for halos of a given mass
, the distribution of the luminosities of central galaxies , @xmath2 , has a dispersion of about @xmath3 dex .
the mean luminosity ( stellar mass ) of the central galaxies scales with halo mass as @xmath4 ( @xmath5 ) for halos with masses @xmath6 , and both relations are significantly steeper for less massive halos .
we also measure the luminosity ( stellar mass ) gap between the first and second brightest ( most massive ) member galaxies , @xmath7 ( @xmath8 ) .
these gap statistics , especially in halos with @xmath9 , indicate that the luminosities of central galaxies are clearly distinct from those of their satellites .
the fraction of fossil groups , defined as those groups with @xmath10 , ranges from @xmath11 for groups with @xmath12 to 18 - 60% for groups with @xmath13 .
the number distribution of satellite galaxies in groups of a given mass follows a poisson distribution , in agreement with the occupation statistics of dark matter sub - halos .
this provides strong support for the standard lore that satellite galaxies reside in sub - halos .
finally , we measure the fraction of satellites , which changes from @xmath14 for galaxies with @xmath15 to @xmath16 for galaxies with @xmath17 .
|
the infrared satellite iras found that some 15% of nearby main sequence stars exhibit infrared emission in excess of that expected from the stellar photosphere alone ( e.g. , aumann et al .
1984 ; backman & paresce 1993 ) .
this excess emission is thought to come from dust which is heated by the star and its temperature ( @xmath9 k ) implies that for the majority of stars this dust is in regions analogous to the kuiper belt in the solar system ( @xmath10 au ; wyatt et al .
this analogy is reinforced by detailed analysis of these systems which show that the dust must be continually replenished from a reservoir of larger ( km - sized ) planetesimals ( wyatt & dent 2002 ) .
very few stars have been seen to exhibit warm emission from dust at @xmath11 au from the star ; those that do tend to be young with age @xmath12 myr ( laureijs et al .
2002 ) , an observation which may indicate that these inner regions have been cleared by the formation of planets ( e.g. , williams et al . 2004 ) .
further indirect evidence for the presence of planets in these inner regions comes from the clumpy and asymmetric structures of the disks ( holland et al . 1998 ; greaves et al .
1998 ) ; modeling has shown that these structures may be caused by the effect of an unseen planet s gravity on the disk ( wyatt et al .
1999 ; ozernoy et al .
2000 ; wyatt 2003 ) .
thus these debris disks play a fundamental role in increasing our understanding of the outcome of planet formation .
while several hundred stars are now known to exhibit infrared excesses , the next fundamental step is to image this emission .
such images are important for a number of reasons : imaging confirms that the iras excess is associated with star , since the large iras beam size means that some excess sources are confused with background objects ( lisse et al .
2002 ; jayawardhana et al .
2002 ; sheret , dent , & wyatt 2004 ) .
it also gives a direct measure of the radial location of dust from which conclusions can be drawn about the size and emptiness of the inner regions , as well as about the size and composition of the dust ( wyatt & dent 2002 ; sheret et al .
the detection of asymmetric structures in the images can also be used to set constraints on an unseen planetary system in the inner region ( wyatt 2003 ) . once this type of detailed information has been accumulated for a sufficient number of disks it will be possible to consider their evolutionary sequence and the diversity of possible outcomes of planet formation around different types of star .
until recently only six disks had been imaged at wavelengths ranging from the optical ( clampin et al .
2003 ) through the mid - ir ( telesco et al . 2000 ) , to the sub - mm ( holland et al . 1998
; greaves et al . 1998 ; holland et al .
2003 ) , and the mm ( koerner et al . 2001 ; wilner et al . 2001 ) .
not surprisingly these were also the six brightest candidates found by iras .
five of these disks are found around bright a stars , and one around the nearby k star @xmath13 eri .
the challenge with imaging of debris disks is that they are faint ; sub - mm imaging of the brightest disks required several nights of telescope time . given that fluxes and size scales have large uncertainties based on the iras fluxes it is prohibitive to study all iras excess sources in the same detail .
recent observational campaigns have had success at resolving disks by focussing on young stars ( williams et al .
2004 ; kalas , liu & matthews 2004 ) ; the disks of stars younger than a few 100 myr might be expected to be brighter than those of older stars , since there is evidence that disk mass decreases with age ( habing et al .
2001 ; greaves & wyatt 2003 ) .
however , a disk was recently resolved in sub - millimeter imaging around a star which is older than the sun ( greaves et al .
since the three newly resolved disks are around g and m stars , these new images are filling in our understanding of how disk parameters are affected by spectral type and age . here
we report on the discovery of a resolved disk around the nearby star @xmath0 corvi ( hd109085 ) , located at @xmath14 pc from the sun .
the star is of spectral type f2v ( mora et al . 2001 ) and has an age of @xmath8 gyr ( see [ s : concl ] ) .
its disk is the first to be resolved around a main sequence f star .
previous surveys of the iras database found this star to exhibit an infrared excess ( stencel & backman 1991 ; mannings & barlow 1998 ) , and a recent sub - mm photometric survey concluded that this disk should be large enough to be resolved with sub - mm imaging ( sheret et al .
sheret et al .
also noted that there is a large mid - ir excess in this system which is expected to be at a different radial location to the sub - mm emission .
the structure of this paper is as follows .
sub - mm imaging observations of this system are described in [ s : obs ] and the results presented in [ s : res ] .
mid - ir observations of this system are also described in
[ s : obs ] and [ s : res ] .
modeling of the observations and of the disk s spectral energy distribution are presented in
[ s : disc ] and a discussion of the implications for the nature of this disk are given in
[ s : concl ] .
the sub - mm observations of @xmath0 corvi were made using the submillimetre common - user bolometer array , scuba at the james clerk maxwell telescope ( jcmt ) .
scuba images at wavelengths of 850 and 450 @xmath1 m simultaneously onto arrays containing 91 and 37 bolometers respectively covering a total field of view of @xmath15 diameter ( holland et al .
each bolometer has diffraction - limited resolution on the sky with beam sizes of @xmath16 and @xmath17 at 850 and 450 @xmath1 m respectively .
we observed @xmath0 corvi ( @xmath18 , @xmath19 ) for a total of 13.5 hours over 10 nights ( 6.4 hours in 2003 january 5 - 9 and 7.1 hours in 2003 march 13 - 16 and 21 ) at an average airmass of 1.4 .
all nights had excellent weather conditions with zenith opacities in the range 0.5 - 0.9 at 450 @xmath1 m and 0.12 - 0.20 at 850 @xmath1 m . since the arrays instantaneously undersample the sky
, a fully sampled image was achieved using the standard technique of jiggle - mapping . in this method
the final map is made up of a co - add of 2 second jiggle observations in which the arrays are pointed at different offsets from the star using the secondary mirror ( holland et al .
1999 ) .
given the small scale of the predicted structure , the pointing was checked thoroughly before and after each 20 - 40 minute observation on pointing sources 1213 - 179 or 3c279 .
the pointing accuracy was found to be @xmath20 , slightly better than the @xmath21 errors quoted for the jcmt . ) . in all analysis
the tracking error due to the f2 correction ( http://www.jach.hawaii.edu/jacpublic/jcmt/facility_description/pointing/tracking_fault.html ) was completely removed in the data reduction ; the maximum correction applied during any one observation was @xmath22 . ] the data for each night were calibrated using mars when available , or one of the standard secondary jcmt calibrators .
calibration uncertainties were estimated from the night to night gain variation as @xmath23 at 850 @xmath1 m and @xmath24 at 450 @xmath1 m .
point source calibration observations were modeled to show that the psf of the scuba observations could be well fit by gaussians with fwhm of @xmath16 and @xmath17 at 850 and 450 @xmath1 m respectively .
the first part of the data reduction was accomplished using the _ surf _ package ( jenness & lightfoot 1998 ) . the data were corrected for atmospheric extinction using 850 and 450 @xmath1 m sky opacities derived from the cso s 225ghz ( january ) and 350 @xmath1 m ( march ) dipper measurements ( weferling 2004 ) .
the cso dip measurements were taken at 10 minute intervals and their relation to sub - mm opacities is well understood ( archibald et al .
each 20 - 40 minute observation was reduced assuming an opacity derived from a polynomial fit to the long term trend of the opacity variation on each night .
the accuracy of this method was confirmed using observations of the 850 @xmath1 m opacity measured less frequently at the jcmt using the skydip method ( archibald et al .
2002 ) .
the standard observational technique of sky chopping ( at @xmath25 hz ) was used to remove the dominant sky background during each observation ; the chop throw employed was @xmath26 in january and @xmath27 in march .
the telescope primary was nodded ( every 16 seconds ) to place the source alternately in both chop beams and thus cancel slower varying sky gradients .
short timescale spatial and temporal variations in sky emissivity ( `` sky - noise '' ) were further removed using the _ surf _ routine remsky which subtracts the mean level of bolometers not on source during each 2 second jiggle observation .
however , even employing these sky subtraction methods a low - level residual background signal remained in each 20 - 40 minute observation at a level of @xmath28 and @xmath29 mjy / beam at 850 and 450 @xmath1 m respectively .
this dc - offset is believed to be caused by residual beam imbalance ( not cancelled by nodding ) due to a time - dependent spillover signal , most likely from beyond the edge of the primary as the secondary chops and jiggles during a 20 - 40 minute observation .
this residual sky level was removed by subtracting the mean surface brightness @xmath30 from the star from each observation .
since the dc level is constant across the array the subtraction does not result in higher resolution structure being introduced into the final map .
the final part of the reduction was accomplished using custom routines .
the data for observations on all nights were rebinned , with the measurements from different bolometers ( on different nights ) weighted according to their noise ; this noise level was estimated from the standard deviation of flux measurements made by each bolometer in an observation .
bolometers with above average noise on any given night were not included in the final map , and anomalous signals were clipped above the 3 @xmath31 level .
the resulting maps do not differ significantly to those produced using _ surf _
s rebin , however the routines also return a map of the uncertainty in the flux measurements in the map .
the mid - ir observations were performed using the thermal infrared multimode instrument timmi2 at the eso 3.6 m telescope at la silla .
we imaged @xmath0 corvi for 840 seconds in 2003 november 19 at an airmass of 1.7 .
the observations used the instrument s 11.9 @xmath1 m filter and a @xmath32 pixel scale was chosen giving a total field of view of 64x 48on the 320x240 pixel array .
the sky background was removed by chopping north - south with @xmath33 amplitude and the telescope emission and sky residuals were subtracted by nodding east - west , again with @xmath33 amplitude .
this observing strategy means that a straight co - add of the data results in an image which contains two positive and two negative images of any given source .
the image was calibrated using observations of the standard star hd71701 taken immediately before the science observation at an airmass of 1.5 ; the flux of the calibrator in this waveband , 6312 mjy , was derived from the template for the emission spectrum of this star given in cohen et al .
a conservative estimate of the calibration accuracy was determined from the gain variation throughout the night to be @xmath34% .
the psf of point sources on this night had fwhm of between @xmath35 and @xmath36 .
the rebinned 850 and 450 @xmath1 m images of a @xmath37x@xmath37 region centered on @xmath0 corvi are shown in fig . [
fig : scuba ] .
each @xmath32 pixel in the images represents the weighted mean of all bolometer measurements falling ( a ) within @xmath38 at 850 @xmath1 m , ( c ) within @xmath39 at 450 @xmath1 m , and ( e ) within @xmath40 at 450 @xmath1 m .
the error is not constant across the image because of the different noise levels in different bolometers , but its distribution in the region plotted can be approximated by gaussians with ( a ) @xmath41 , ( c ) @xmath42 , and ( e ) @xmath43 @xmath1jy arcsec@xmath44 .
this uncertainty is quantified in figs .
[ fig : scuba]b , [ fig : scuba]d , and [ fig : scuba]f which show the corresponding signal to noise maps .
the same rebin methods were applied to the calibration data which showed that point sources rebinned in this way are slightly smoothed and can be fitted by gaussians of fwhm ( a ) 158 , ( c ) 137 , and ( e ) 95 .
extended emission was detected centered on the star ( hereafter source a ) at each wavelength with greater than @xmath45 confidence .
another source ( hereafter source b ) was also detected with @xmath46 confidence at both wavelengths at @xmath47 nnw of @xmath0 corvi .
the @xmath48x@xmath48 field around the star was analyzed to determine the noise level .
structure at @xmath49 confidence is found randomly distributed across the field of view and has the same level and distribution as that expected from noise .. ] thus the zero levels in all maps were set to show only structure detected with @xmath50 confidence .
the remaining field is blank on this scale apart from a source ( hereafter source c ) detected with @xmath50 confidence @xmath51 sw of @xmath0 corvi at 850 @xmath1 m but not at 450 @xmath1 m ; the edge of source c can just be seen in the bottom left corner of figs .
[ fig : scuba]a and [ fig : scuba]b .
the fluxes , locations and morphologies of these sources are discussed in more detail below .
the extended emission found centred on the star ( source a ) we attribute to a circumstellar disk detected by iras in the far - ir ( e.g. , stencel & backman 1991 ) .
fitting a gaussian to the rebinned 850 @xmath1 m data indicates that the 850 @xmath1 m emission has a fwhm of @xmath52 and is centered on the star to within @xmath21 . subtracting the observed psf ( which has a fwhm of @xmath53 ) in quadrature from this measurement implies that the disk is significantly extended on a scale of @xmath54 , indicative of a disk of 100 au radius .
there is no evidence for any azimuthal asymmetry in the 850 @xmath1 m emission .
the structure of the disk is also evident in the 450 @xmath1 m observation at much higher resolution .
the 450 @xmath1 m data are presented in two ways : figs .
[ fig : scuba]c and [ fig : scuba]d show the data rebinned to a resolution of @xmath55 , while figs .
[ fig : scuba]e and [ fig : scuba]f were rebinned to a resolution of @xmath56 .
the former images are appropriate for studying the disk s larger scale structure , while the latter images show its high resolution structure . the highest contours in figs .
[ fig : scuba]c and [ fig : scuba]d show that the emission is elongated along a position angle of @xmath3 .
the origin of this axis of symmetry is seen more clearly in the higher resolution image of fig .
[ fig : scuba]e in which the emission is resolved into two peaks .
these peaks are at an ra and dec offset of @xmath57 , @xmath58 ( at 0.15 mjy arcsec@xmath44 ) and @xmath59 , @xmath60 ( at 0.13 mjy arcsec@xmath44 ) . given that the two peaks are detected at the @xmath61 and @xmath62 levels , the location of each peak is not known with better precision than @xmath63 ( beam size divided by signal - to - noise ) , and the peaks are equally bright to within the uncertainties . ) .
] thus the observed structure is entirely consistent with two equally bright peaks at 0.14 mjy arcsec@xmath44 that are equidistant from the star at an offset of @xmath64 ( 100 au ) and a position angle of @xmath65 ( counterclockwise from north ) . since this is the simplest interpretation of the data , it is the one we adopt in the rest of the paper . the total flux in a @xmath66 ( @xmath67 ) aperture centered on the star is @xmath68 ( @xmath69 ) mjy at 850 @xmath1 m and @xmath70 ( @xmath71 ) mjy at 450 @xmath1 m .
adding the calibration error in quadrature gives the uncertainty of these measurements as @xmath72 ( 2.4 ) mjy at 850 @xmath1 m and @xmath73 ( 12.9 ) mjy at 450 @xmath1 m .
these fluxes imply disk emission with a spectrum @xmath74 in the sub - mm , consistent with black body emission in the rayleigh - jeans regime .
the 850 @xmath1 m flux is almost a factor of two higher than that presented by sheret et al .
( 2004 ) who measured @xmath75 mjy for this source .
however , the two datasets are consistent because their observations were performed in photometry mode which is insensitive to emission extended on scales larger than the beamsize ; sheret et al .
noted that the fluxes they derived could underestimate the true flux by a factor of 2 if the source is extended .
source b is located @xmath76 w and @xmath77 n of @xmath0 corvi at 850 @xmath1 m .
its measured fwhm of 14at 850 @xmath1 m is not significantly different to that of a point source , and its total flux is @xmath78 mjy in an aperture of @xmath79 diameter .
this source thus provides an independent measure of the psf in the image for comparison with source a. at 450 @xmath1 m the emission peaks at @xmath80 w and @xmath81 n of @xmath0 corvi and has a total flux of @xmath82 mjy in an @xmath79 aperture .
source c was discovered in the 850 @xmath1 m map at an offset of @xmath83 e , @xmath84 s of @xmath0 corvi with a total flux of @xmath85 ( @xmath86 ) mjy in an @xmath79 ( @xmath66 ) aperture .
no emission was detected from the location of source c in the 450 @xmath1 m image at the @xmath87 upper limit level of @xmath88 ( 21 ) mjy in an @xmath79 ( @xmath66 ) aperture .
this implies a spectrum of emission @xmath89 .
several background sources are expected in any given sub - mm image .
since this star is not in the galactic plane ( it is at a galactic latitude of @xmath90 ) and is far from known regions of star formation , the density of background sources in this image would be expected to be the same as that measured in similar regions .
scott et al . ( 2002 ) estimate that there are about 620 sources per square degree above 5 mjy at 850 @xmath1 m , while there are as many as 2000 sources per square degree above 3 mjy at 850 @xmath1 m ( eales et al .
smail et al .
( 2002 ) estimate that there are some 2000 @xmath91 mjy sources at 450 @xmath1 m per square degree .
thus in the field of view analyzed in this paper , @xmath48x@xmath48 , we would expect to find : 1.5 sources above 3 mjy in our 850 @xmath1 m image and 1.5 sources above 10 mjy in our 450 @xmath1 m image .
these number counts are consistent with sources b and c being background objects and from now on we interpret them as such .
the probability of two background sources ( at either @xmath92 mjy or @xmath91 mjy at 850 and 450 @xmath1 m respectively ) falling within @xmath33 of @xmath0 corvi , and so providing confusion with source a , is less than 1:300 ; the chance of one such object falling within this distance is 1:16 , so background objects should be considered as a potential source of uncertainty in the low level morphology of this source . while source b is interpreted as a background object , we note that a 19 mjy source was observed at an offset of @xmath93 ( 660 au projected separation ) in 850 @xmath1 m images of @xmath94 pictoris ( holland et al .
thus it is possible that offset sub - mm clumps may be more commonly associated with debris disks .
the spectral slope of source b , emission @xmath95 , is similar to that of source a , meaning that it is not inconsistent with the sources being comprised of dust with similar properties .
however , this spectral slope is also similar to that of other background sub - mm objects ( @xmath96 , smail et al .
2002 ) and further study is required to determine the true nature of this object . if associated with the star the projected separation of source b ( 480 au ) would make it a highly unusual object .
the image resulting from the 11.9 @xmath1 m observations of @xmath0 corvi described in [ ss : midir ]
is presented in fig .
[ fig : midir ] .
the pattern of positive and negative sources shown in this image is that expected from a single source centered on the star .
each of the peaks was detected with a statistical signal to noise of @xmath97 , and is consistent with point source emission ( fwhm of @xmath98 ) .
the total measured flux was @xmath99 jy ( @xmath100 ) ; the dominant uncertainty in this flux estimate is the calibration uncertainty .
this is consistent , within the error bars of the measurement , with the @xmath101 jy at 12 @xmath1 m measured by iras ( table [ tab : fluxes ] ) .
further point - like sources within the field of view , including at the location of source b , were ruled out at @xmath102 mjy with @xmath87 certainty .
a disk structure characterized by two bright lobes equidistant from the star has previously been seen in the disks of fomalhaut and hr4796a and interpreted as evidence of an edge - on dust ring ( e.g. , holland et al .
1998 ; telesco et al .
2000 ; holland et al . 2003 ) .
however , two clumps straddling the star have also been observed and interpreted as azimuthal asymmetries in a face - on dust ring around vega ( holland et al .
1998 ; koerner et al . 2001 ; wilner et al . 2002 ) .
vega s disk was assumed to be face - on due to the symmetric nature of the lower level contours ( holland et al .
1998 ) . in this respect
the @xmath0 corvi disk bears more resemblance to vega than to fomalhaut : in fig .
[ fig : scuba]c the source is significantly resolved at all position angles , and is only slightly elongated with a fwhm of @xmath103 along a position angle of @xmath104 and @xmath105 at @xmath106 . subtracting these observed widths in quadrature from the fwhm of a point source ( @xmath55 )
, we infer that this emission comes from a ring at moderate inclination ( @xmath107 from edge - on ) to our line - of - sight .
we simulated the 450 @xmath1 m image by making a model of the predicted emission distribution , smoothing it by the psf ( @xmath17 gaussian ) , then using the resulting image to determine the expected flux measured by each bolometer in each jiggle observation in the scuba data and rebinning the model in the same way as the observation ( fig .
[ fig : scuba ] ) .
a noisy model was also made by adding gaussian noise at the same level as the deviation in the measurements for bolometers in individual observations to determine the extent to which noise affects the observed structure .
the simplest model which was considered was one in which the emission comes from an axisymmetric torus of width @xmath108 au which is inclined to our line - of - sight .
the radius and surface density of the ring were constrained by the fwhm and peak fluxes in the resulting images , leaving the inclination of the ring as the only free parameter .
simulated observations of models inclined by @xmath109 , @xmath110 , and @xmath111 from edge - on are shown in figs .
[ fig : mod]a , [ fig : mod]b , and [ fig : mod]c , respectively ; the radii of the rings in these models were determined to be 140 , 150 , and 180 au respectively , all with errors of @xmath112 au .
it is possible to rule out the edge - on model at the statistically significant level , because this model does not fit the run of flux vs distance from the peak along a position angle of @xmath106 ( the model is unresolved , fwhm of @xmath55 , in the left panel of fig .
[ fig : mod]a ) .
this means that while the peak fluxes in the images are well reproduced , the total flux within 15from this model is just 37 mjy , more than @xmath87 below that observed .
the face - on model is also ruled out at a statistically signficant level because it fails to reproduce an axis of symmetry .
this symmetry can be shown to be significant , since the total flux measured by bolometers falling within 10of the star and within @xmath113 of a position angle of @xmath114 is @xmath115 mjy , whereas the same measurements for position angles centered on @xmath109 and @xmath116 are @xmath117 and @xmath118 mjy respectively .
we conclude that if this disk is axisymmetric , an inclination of @xmath119 to our line - of - sight is most likely , and its radius is @xmath5 au .
an inclination of @xmath120 from edge - on is consistent with the disk being aligned with the stellar equator given that the star @xmath0 corvi has @xmath121 km s@xmath122 ( mora et al .
2001 ) whereas the mean @xmath123 for f0v stars is 106 km s@xmath122 ( abt & morrell 1995 ) .
an inclination of @xmath124 is still possible , however , if , like the vega disk , the structure of the @xmath0 corvi disk includes two clumps , one on either side of the star .
such an axis of symmetry is to be expected if the disk is being perturbed by planets . under a wide variety of conditions
two clumps approximately equidistant from the star are expected , either because p - r drag makes the dust migrate into the exterior mean motion resonances of a massive planet ( kuchner & holman 2003 ) or because the parent planetesimals feeding the dust disk were trapped into the resonances of a planet when that planet migrated outward ( wyatt 2003 ) . in fig .
[ fig : mod]d we show how the 450 @xmath1 m disk structure can be explained by a disk inclined by @xmath116 to our line - of - sight in which a neptune mass planet ( @xmath125 ) migrated out from 80 to 105 au over a period of 25 myr and so trapping planetesimals into its 3:2 and 4:3 resonances ( wyatt 2003 ) . in this model
the planet is at an offset of @xmath126 at a position angle of either 40 or @xmath127 .
the planet s mass ( and the time it took to complete its migration ) are poorly constrained in this model because a wide variety of parameters would cause two clumps ( wyatt 2003 ) .
we have presented two possible models which fit the 450 @xmath1 m observations to within the uncertainty due to noise ( figs .
[ fig : mod]b and [ fig : mod]d ) . clearly further observations are required to determine the true morphology and orientation of the disk .
high resolution observations , such as those provided by sub - mm or mm interferometry or optical coronography , would be able to distinguish between the two models presented here , or tell if an alternative model is required .
regardless of the axisymmetry of the dust ring and its inclination , the 450 @xmath1 m images show there is a lack of sub - mm emitting dust inside 100 au . to understand this , consider that the models presented in figs .
[ fig : mod]b and [ fig : mod]d fit the surface brightness distribution at both @xmath128 au from the star and at the stellar location , even though there is no dust at @xmath129 au in the models .
the reason is that emission at the stellar location is seen from the 150 au ring because of the resolution of the observation .
we set a conservative constraint on the emission from an unresolved sub - mm component from the @xmath130 uncertainty in the surface brightness at the center of figs .
[ fig : scuba]c and [ fig : scuba]e ; i.e. , an additional unresolved component is ruled out at @xmath131 mjy at 450 @xmath1 m , or less than one third of the total disk flux .
the sed of @xmath0 corvi from optical to sub - mm wavelengths is plotted in fig . [
fig : sed ] .
the iras fluxes were determined from scanpi ( iras scan processing and integration ) , available at the irsa website ( http://irsa.ipac.caltech.edu ) and the new values ( including a previously unpublished detection at 100 @xmath1 m ) are given in table 1 .
the stellar fluxes at mid - ir to sub - mm wavelengths were determined from a 7080 k kurucz atmosphere scaled to the near - ir fluxes in sylvester et al .
color corrected stellar fluxes were subtracted from the iras fluxes to give the excess fluxes ( which were also color corrected ) .
excess emission is found at all wavelengths from 12 - 850 @xmath1 m . as noted by previous authors ,
the spectrum of this excess emission is such that it can not all be accounted for by dust at a single temperature or distance from the star ( sheret et al .
2004 ) .
mid - ir spectroscopy presented in sylvester & mannings ( 2000 ) found that most of the 12 @xmath1 m excess emission must be centered on the star .
it is also important that our 11.9 @xmath1 m observations did not detect any additional nearby sources which could be contributing to the 12 @xmath1 m flux measured by iras ( which could originate anywhere within 30of the star ; lisse et al .
2002 ) ; in particular source b does not have a significant counterpart at 12 @xmath1 m .
combining these findings implies that most of the iras emission is centered on a region around the star that is smaller than our measured fwhm , i.e. , at @xmath132 au .
thus we modeled the disk as a single temperature cool component ( causing the excess at @xmath133 m ) with an additional single temperature hot component ( causing the excess at @xmath134 m ) .
the fluxes from the individual components were modeled assuming a black body emission distribution modified by a ratio @xmath135 at @xmath136 ( e.g. , dent et al .
2000 ) .
the cool component is well fit by emission at @xmath137 k , @xmath138 and
@xmath139 @xmath1 m .
the fractional luminosity of this component is @xmath140 and the inferred dust mass assuming an opacity of @xmath141 m@xmath142 kg@xmath122 ( e.g. , wyatt , dent & greaves 2003 and references therein ) is @xmath143 .
the black body temperature at a distance of 150 au is 34 k. thus the dust is only slightly warmer than black body , implying that the dust is comprised of grains larger than a few 10s of @xmath1 m .
this is consistent with the inferred @xmath94 which is comparable with that of debris disks of similar age ( dent et al .
2000 ) . in a further study we performed a more detailed sed model using realistic grain properties (
see , e.g. , wyatt & dent 2002 ; sheret et al .
2004 ) and found that this spectrum could be fitted with a dust ring at 150 au with the same dust size distribution as that expected from a collisional cascade , but one with an additional imposed cut - off for grains smaller than @xmath144 @xmath1 m rather than at @xmath145 m as expected from radiation pressure blow - out ( wyatt et al .
grains in the size range of 4 - 30 @xmath1 m must be absent because they are too hot ( @xmath146 k ) at this distance from the star to explain the shape of the sed , which would have stronger emission at 25 and 60 @xmath1 m if these grains were present .
possible reasons for the absence of small grains , also seen in the spectrum of @xmath13 eridani , were discussed in sheret et al .
these include the possibility that the @xmath147 @xmath1 m grains in the outer disk of @xmath0 corvi are destroyed in collisions with @xmath148 @xmath1 m grains which are in the process of being blown out of the inner regions by radiation pressure ( krivov et al .
we also note that a dearth of small grains is expected in the model in which clumpy structure is formed by planet migration ( wyatt 2003 ) .
the fit presented in fig .
[ fig : sed ] shows the hot component has a temperature of @xmath7 k ( assumed to be black body , i.e. , @xmath149 ) and has a fractional luminosity of @xmath150 .
while this fractional luminosity is an order of magnitude higher than that of the cool component , the mass of the hot component is inferred to be at least two orders of magnitude less than that of the cold component ( because of its higher temperature and lower inferred sub - mm flux ) .
this temperature implies that the dust lies 1 - 2 au from the star .
while this is the interpretation discussed in the rest of this paper , we note that this distance should be regarded as suggestive rather than definitive , since the two temperature model used to fit the sed may be an oversimplification of the true dust distribution .
it is also possible that the dust lies in a more extended distribution wherein the 12 @xmath1 m excess comes from dust which is hotter than 370 k ( and so is unresolved ) , and the 25 @xmath1 m excess comes from cooler dust located between 1 and 150 au .
high resolution mid - ir observations are required to determine the location of the 25 @xmath1 m emission .
the age of this star has been estimated in the literature to be 1.3 gyr from its position relative to evolutionary tracks in the hr diagram ( mallik , parthasarathy & pati 2003 ) .
such a high age is also consistent with the low x - ray luminosity of @xmath0 corvi .
hnsch et al .
( 1999 ) found this star to have @xmath151 erg s@xmath122 , a value which is significantly below that of members of the pleiades with similar @xmath152 ( micela et al .
1999 ) , implying an age of @xmath153 myr . on the other hand
, this x - ray luminosity lies very close to the mean value of a - f stars in the hyades cluster ( stern , schmitt & kahabka 1995 ) implying an age of 600 - 800 myr . here
we adopt an age of 1 gyr for @xmath0 corvi .
given such a large age , the dust in this system can not be primordial , since this is much longer than the p - r drag lifetime of the dust : dust with @xmath154 would migrate from 150 au to 0 au ( where it would evaporate ) on timescales of 20 myr ( burns et al .
this is unlikely to be the ultimate fate of the dust originating at 150 au , however , since the collisional lifetime of this dust is @xmath8 myr , so it is more likely to be destroyed in mutual collisions ( and then removed by radiation pressure ) before it reaches the inner regions . in any case , the dust must be replenished , presumably by the collisional destruction of a population of planetesimals orbiting at 150 au from @xmath0 corvi . using the same collisional modeling as that presented in wyatt & dent ( 2002 ) , we infer from the age of 1 gyr that the collisional cascade in this system starts with planetesimals a few km in size implying a total mass of @xmath155 in planetesimals in the ring at 150 au .
this is similar to the mass inferred for the collisional cascade of the disk around star fomalhaut which has an age of 200 myr ( wyatt & dent 2002 ) .
the relative lack of dust at @xmath129 au indicates that there is a significant lack of colliding planetesimals in this region .
this could be because planet formation has proceeded to such a stage that the planetesimals have grown into planets ( e.g. , kenyon & bromley 2002 ) ; the perturbations from the outermost planet of that system are potentially observable in the structure of the outer ring ( [ ss : ring ] ) .
the dust at a few au must also be replenished . just like the zodiacal cloud in the solar system
, this replenishment could come from either the collisional destruction of planetesimals in an asteroid belt at a few au or in the sublimation of comets , possibly comets which originated in the cooler outer ring .
mid - ir spectroscopy may be able to shed light on the composition of this dust , and so be able to determine whether this dust formed in a hot region close to the star or if it formed in a cold region far from the star .
the age of 1 gyr means that it is very surprising that this star has detectable hot emission .
the only other stars with mid - ir excesses are young ; e.g. , laureijs et al .
( 2002 ) found no stars older than 400 myr with excess 25 @xmath1 m emission and weinberger et al .
( 2004 ) found that even at 10 myr , very few stars have significant mid - ir excess .
while it is possible that this is a system which has always had an unusually massive hot dust population , another possibility is that the mid - ir excess is a transient phenomenon .
an increase in mid - ir excess might be expected after the break - up of a massive asteroid in an asteroid belt at a few au ( grogan , dermott , & durda 2001 ) , or after an event which perturbed comets from the ring at 150 au into the inner region .
such an event could have been the recent close passage of a nearby star ( larwood & kalas 2001 ) , or the recent scattering and subsequent migration of a neptune - sized planet from its formation at a few 10s of au out to @xmath2 au ( e.g. , thommes , duncan & levison 2002 ) .
the jcmt is operated by the joint astronomy centre on behalf of the uk particle physics and astronomy research council , the canadian national research council and the netherlands organization for scientific research .
this research has made use of the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this research was supported in part by pparc funding .
the careful analysis of the referee , c. m. lisse , has also greatly improved this paper .
abt , h. a. , & morrell , n. i. 1995 , , 99 , 135 archibald , e. n. , et al .
2002 , , 336 , 1 aumann , h. h. , et al .
1984 , , 278 , l23 backman , d. e. , & paresce , f. 1993 , in protostars and planets iii , ed . e. h. levy & j. i. lunine ( tucson : univ .
arizona press ) , 1253 burns , j. a. , lamy , p. l. , & soter , s. 1979 , icarus , 40 , 1 clampin , m. , et al .
2003 , , 126 , 385 cohen , m. , walker , r. g. , carter , b. , hammersley , p. , kidger , m. , & noguchi , k. 1999 , , 117 , 1864 dent , w. r. f. , walker , h. j. , holland , w. s. , & greaves , j. s. 2000 , , 314 , 702 eales , s. , lilly , s. , webb , t. , dunne , l. , gear , w. , clements , d. , & yun , m. 2000 , , 120 , 2244 greaves ,
j. s. , et al .
1998 , , 506 , l133 greaves , j. s. , & wyatt , m. c. 2003 , , 345 , 1212 greaves , j. s. , wyatt , m. c. , holland , w. s. , & dent , w. r. f. 2004 , , 351 , l54 grogan , k. , dermott , s. f. , & durda , d. d. 2001 , icarus , 152 , 251 habing , h. j. , et al .
2001 , a&a , 365 , 545 holland , w. s. , et al .
1998 , , 392 , 788 holland , w. s. , et al .
1999 , , 303 , 659 holland , w. s. , et al .
2003 , , 582 , 1141 hnsch , m. , schmitt , j. h. m. m. , sterzik , m. f. , & voges , w. 1999 , a&as , 135 , 319 jayawardhana , r. , holland , w. s. , kalas , p. , greaves , j. s. , dent , w. r. f. , wyatt , m. c. , & marcy , g. m. 2002 , , 570 , l93 jenness t. , lightfoot j. f. 1998 , in astronomical analysis software and systems vii , asp conf .
145 , eds .
r. albrecht , r. n. hook & h. a. bushouse , ( san francisco : astron .
soc . pac . ) , 216 kalas , p. , liu , m. c. , & matthews , b. c. 2004 , science , 303 , 1990 kenyon , s. j. , & bromley , b. c. 2002 , , 577 , l35 koerner , d. w. , sargent , a. i. , & ostroff , n. a. 2001 , , 560 , l181 krivov , a. v. , mann , i. , & krivova , n. a. 2000 , a&a , 362 , 1127 kuchner , m. j. , & holman , m. j. , , 588 , 1110 larwood , j. d. , & kalas , p. g. 2001 , , 323 , 402 laureijs , r. j. , jourdain de muizon , m. , leech , k. , siebenmorgen , r. , dominik , c. , habing , h. j. , trams , n. , & kessler , m. f. 2002 , a&a , 387 , 285 lisse , c. , et al .
2002 , , 570 , 779 micela , g. , et al . 1999 ,
a&a , 341 , 751 mallik , s. v. , parthasarathy , m. , & pati , a. k. 2003 , a&a , 409 , 251 mannings , v. , & barlow , m. j. 1998 , , 497 , 330 mora , a. , et al .
2001 , a&a , 378 , 116 ozernoy , l. m. , gorkavyi , n. n. , mather , j. c. , & taidakova , t. a. 2000 , , 537 , l147 scott , s. e. , et al .
2002 , , 331 , 817 sheret , i. , dent , w. r. f. , & wyatt , m. c. 2004 , , 348 , 1282 smail , i. , ivison , r. j. , blain , a. w. , & kneib , j .- p .
2002 , , 331 , 495 stencel , r. e. , & backman , d. e. 1991 , , 75 , 905 stern , r. a. , schmitt , j. h. m. m. , & kahabka , p. t. 1995 , , 448 , 683 sylvester , r. j. , & mannings , v. 2000 , , 313 , 73 sylvester , r. j. , skinner , c. j. , barlow , m. j. , & mannings , v. 1996 , , 279 , 915 telesco , c. m. , et al . 2000 , , 530 , 329 thommes , e. w. , duncan , m. j. , & levison , h. f. 2002 , , 123 , 2862 weferling , b. 2004 , a&a , in press weinberger , a. , becklin , e. e. , zuckerman , b. , & song , i. 2004 , , 127 , 2246 williams , j. p. , najita , j. , liu , m. c. , bottinelli , s. , carpenter , j. m. , hillenbrand , l. a. , meyer , m. r. , & soderblom , d. r. 2004 , , 604 , 414 wilner , d. j. , holman , m. j. , kuchner , m. j. , & ho , p. t. p. 2002
, , 569 , l115 wyatt , m. c. 2003 , , 598 , 1321 wyatt , m. c. , & dent , w. r. f. 2002 , , 334 , 589 wyatt , m. c. , dent , w. r. f. , & greaves , j. s. 2003 , , 342 , 876 wyatt , m. c. , dermott , s. f. , telesco , c. m. , fisher , r. s. , grogan , k. , holmes , e. k. , & pia , r. k. 1999 , , 527 , 918 wyatt , m. c. , holland , w. s. , greaves , j. s. , dent , w. r. f. 2004 , earth moon planets , 92 , 423
|
we present sub - millimeter and mid - infrared images of the circumstellar disk around the nearby f2v star @xmath0 corvi .
the disk is resolved at 850 @xmath1 m with a size of @xmath2 au . at 450 @xmath1 m
the emission is found to be extended at all position angles , with significant elongation along a position angle of @xmath3 ; at the highest resolution ( @xmath4 ) this emission is resolved into two peaks which are to within the uncertainties offset symmetrically from the star at 100 au projected separation . modeling
the appearance of emission from a narrow ring in the sub - mm images shows the observed structure can not be caused by an edge - on or face - on axisymmetric ring ; the observations are consistent with a ring of radius @xmath5 au seen at @xmath6 inclination .
more face - on orientations are possible if the dust distribution includes two clumps similar to vega ; we show how such a clumpy structure could arise from the migration over 25 myr of a neptune mass planet from 80 - 105 au .
the inner 100 au of the system appears relatively empty of sub - mm emitting dust , indicating that this region may have been cleared by the formation of planets , but the disk emission spectrum shows that iras detected an additional hot component with a characteristic temperature of @xmath7 k ( implying a distance of 1 - 2 au ) . at 11.9 @xmath1
m we found the emission to be unresolved with no background sources which could be contaminating the fluxes measured by iras .
the age of this star is estimated to be @xmath8 gyr .
it is very unusual for such an old main sequence star to exhibit significant mid - ir emission .
the proximity of this source makes it a perfect candidate for further study from optical to mm wavelengths to determine the distribution of its dust .
|
the science of networks is indeed a new kind of science @xcite for its interdisciplinary character and its explosive development ; for some recent monographs we refer to @xcite .
the list of applications of networks contains examples from physics , informatics , biology and social sciences .
basic characteristics of networks are the degree distribution @xmath0 and the clustering coefficient @xmath1 .
the degree of a given node is the number of other nodes connected to that node ; the clustering coefficient measures the probability that two neighbours of a given node are connected to each other .
as it was indicated only recently by mark newman @xcite , many real networks show a high clustering coefficient , usually some tens of percent . on the contrary to this
fact , model random networks show rather low @xmath1 , unless a special procedure is applied to enhance it .
examples of such procedures are described in @xcite ; we owe this list again to @xcite .
the idea is to enhance @xmath1 gradually by linking nodes which are neighbours of the same node . when a node has three neighbours , it is convenient to replace it by a triangle @xcite ; the trick is similar to the star - triangle or star - delta transformation .
the latter has been used also to construct apollonian networks with a high values of the clustering coefficient @xcite , and to prove some theorems in the theory of percolation @xcite .
+ the transformation from a graph @xmath4 to its line graph @xmath5 @xcite used here can be seen as a simple reformulation of the star - triangle transformation .
the latter converts a subgraph @xmath6 , i.e. a node with three neighbours , into a subgraph @xmath7 , where these three neighbours are linked to each other and the central node is deleted . in the transformation @xmath8 the same original subgraph @xmath6 is converted also into a graph of three nodes linked to each other .
the idea of the line graph , known also as edge graph @xcite , is to convert all links into nodes @xcite . in this way
a new network appears , where the number of nodes is equal to the number of links in the original network . in the new network ,
two nodes are connected if they are formed from links which shared the same node .
in particular , a node of degree @xmath9 is converted to a fully connected subgraph of @xmath9 nodes .
the definition of distance ensures in particular , that the small world effect in the original network persists also in the transformed network .
+ recently the line graph constructed from the scale - free network was discussed in @xcite analytically and numerically .
the results for dense networks ( @xmath2=10 , 20 and 30 ) supported the rule that the transformed network is also scale - free .
the exponent @xmath10 defined by the degree distribution @xmath11 of the transformed graph was shown to fulfil the relation @xmath12 , where @xmath13 was the same exponent for the initial network .
the line graph was also found to be useful in the problem of identification of communities in networks @xcite .
+ the aim of this work is to investigate the clustering coefficient @xmath1 in the line graphs transformed from random networks .
our motivation is twofold .
first , we share the point of view expressed by mark newman , as reported in our first paragraph .
our former simulations @xcite can actually be seen as the same transformation limited to local subgraphs and to nodes of degree three .
this technique allowed us to enhance the clustering coefficient in networks .
second issue can be briefly expressed as follows . in social networks ,
an active contact between two linked actors excludes at least to some extent the contact between each of these two actors and each of their other neigbours .
if we introduce two states of a link , termed for example open and closed , then it is clear that there is an anticorrelation between links which share the same node ; two such links can not be open simultaneously .
it can be convenient , then , to work on the network of links instead of the network of nodes .
however , mathematical formulation of many problems is expressed in terms of networks of nodes .
here is the area where the transformation can be useful .
+ in the next section we describe some examples of the transformed networks .
section 3 is devoted to our numerical results .
short discussion closes the text .
a chain of @xmath14 nodes is equivalent to a chain of @xmath14 links , then under action of the transformation @xmath8 it is converted into itself . in a fully connected graph of @xmath14 nodes each pair is connected ; there is @xmath15 links . after the transformation from @xmath4 to @xmath5 ,
this is the number of nodes .
the transformed graph is not fully connected ; this can be shown easily for @xmath16 . there
, link @xmath17 is connected to @xmath18 and @xmath19 , but not to @xmath20 , etc . for arbitrary @xmath14 ,
each node has @xmath21 neighbours , then each link shares its each node with @xmath22 other links . in the transformed network ,
each node has therefore @xmath23 neighbours .
the number of links in the transformed network is then @xmath24 .
in particular , for @xmath16 in the original graph we have 12 links in the transformed graph - an octahedron , if links are equally long
. + similar arguments apply to a regular graph , where each node is of the same degree , say @xmath9 .
each link joins two nodes with @xmath25 other neigbours . under the transformation @xmath8
, each node is converted into a @xmath9-clique .
obviously , each link in the original network joins two nodes ; then it contributes - as a node in the transformed network - to two cliques .
its degree in the transformed network is then @xmath26 .
the number of links in the transformed network is then @xmath27 . in the case of @xmath16 and @xmath28 again a tetrahedron is transformed into an octahedron .
+ let us consider a network with the degree distribution @xmath0 , where is no correlation between degrees of neighboring nodes .
the above arguments are now as follows .
a link in the original network joins two nodes with degrees @xmath29 and @xmath30 ; the considered link is not counted . in the transformed network , this link is a node of degree @xmath31 .
the degree distribution @xmath32 in the transformed network is then @xcite @xmath33 in the case when @xmath0 is the poisson distribution with @xmath34 the degree distribution @xmath32 for the transformed network is a new poisson distribution with @xmath35 . for the geometrical distribution @xmath36 we get
@xmath37 this distribution , when presented as @xmath38 against @xmath9 , gives only logarithmic deviation from the degree distribution @xmath0 of the original network .
for the power function @xmath39 we have no simple result .
some analytical considerations in terms of polygamma functions can be found in @xcite .
+ now we consider the clustering coefficient @xmath1 . in a fully connected graph of @xmath14 nodes each link @xmath17
is converted under @xmath8 to a node of degree @xmath23 .
maximal number of links between these @xmath40 neighbours is @xmath41 .
the actual number of links within each clique of @xmath42
. there is also @xmath22 links between nodes converted from links which met at the @xmath22 nodes different from nodes @xmath43 and @xmath44 .
then in total we have @xmath45 links , what gives @xmath46 the same for each node . for regular graphs of degree
each link is converted to a node of degree @xmath26 , with @xmath47 possible links between its neighbours . neglecting triangles built on the considered link
, we have only @xmath48 links within cliques .
then the clustering coefficient is @xmath49 for a network with degree distribution @xmath0 we have to find an average @xmath50 there , the contribution of pairs of nodes where @xmath51 is zero .
the connectivity matrix for the transformed network is constructed here as follows . in the conectivity matrix @xmath52 of the original network
each unit above the main diagonal means a link .
we substitute these units by their consecutive numbers : @xmath53 1 , 2 and so on .
let us call the obtained matrix @xmath54 .
the last number @xmath55 is equal to the number of links in the original network ; it is then equal also to the size of the transformed network . in the conectivity matrix @xmath56 of this network , elements @xmath57 and @xmath58
are connected if their numbers @xmath57 and @xmath58 appear in the matrix @xmath59 in the same row or in the same column . +
the original erds - rnyi network is generated from @xmath60 nodes .
then , the number of nodes in the transformed network is about @xmath61 , where @xmath62 is the density of links in the original network . with @xmath63 , as in fig . 1
, we expect @xmath2 close to @xmath64 in the original network
. the number of nodes in the transformed network should be about @xmath65 . in the example presented in fig .
1 we have 50147 nodes .
the result of the analytical calculation made above indicates that the mean degree of the transformed network should be equal to 20 , what confirms the simulation .
+ the original exponential network is grown from a fully connected cluster of @xmath66 nodes .
each next node is attached to randomly selected @xmath66 different nodes .
no preference of attachment is applied for the exponential networks .
the original network has @xmath60 nodes , and the transformed network has about @xmath67 nodes .
the original network is known to have the exponential distribution of node degree ; @xmath68 plotted against @xmath9 is a straight line .
the degree distributions of the transformed network , shown in fig .
2 , seem also to be close to the exponential function , as in the original network .
the curves shown are obtained from eq .
the size of the transformed network is 29994 , 49985 and 79964 for @xmath693 , 5 and 8 , respectively . + to generate the scale - free networks numerically we have only to add the preferential attachment ;
nodes are selected with the probability proportional to their degree.the original scale - free network is again @xmath70 nodes . in fig .
3 we show the degree distributions for the growing parameter @xmath71 and @xmath72 .
the plots are not far from straight lines in the log - log scale . on the contrary to the exponential network ,
the slope of the obtained curves does not depend on the mean degree @xmath73 .
the overall results agree with those of @xcite .
4 we show the comparison of the data on the clustering coefficient @xmath1 , as calculated from direct numerical simulations ( points ) and from eq . 5 ( lines ) .
it appears that the respective plots met when the mean degree @xmath2 is large enough .
this accordance indicates that our model assumption on the lack of correlations works well for dense networks . as it can be seen in fig .
4 , the largest departure of the clustering coefficient @xmath1 calculated numerically from the analytical values are found for the exponential networks .
this suggests that for these networks , the degree - degree correlations are the largest .
the rule that a given degree distribution is transformed under @xmath8 into the degree distribution from the same family has some justification in the transformation itself .
namely , a node of degree @xmath9 is transformed into a set of nodes of at least the same degree @xmath9 .
this means in particular that hubs are transformed into cliques of hubs , and chains are transformed into chains . for the scale - free networks
, our results on the degree distribution @xmath0 agree with those of @xcite .
+ accordance of the results calculated numerically with the analytical formulas means in our case that the corrections introduced by degree - degree correlations @xcite are relatively irrelevant .
we demonstrated that the transformation @xmath8 leads to clustered networks , where the clustering coefficient @xmath1 is not smaller than 0.5 .
this limit value can be lower , if @xmath8 is applied to selected local subnetworks and not to the whole system .
the density of the transformed nodes can be used to tune @xmath1 , similarly to @xcite .
+ an application of the transformation @xmath8 to the communication networks needs now a specification of interaction between links , i.e. between nodes of the transformed network .
the task is out of frames of this paper .
we hope that the idea of interaction between links can find applications in networks , where a relation between two nodes excludes at least partially the relations between one of these nodes and its neighbours .
we have in mind trade networks , sexual networks , decision trees and transport networks , where parallel links are activated along the principle this or this. links can also activate each other , according to this , then this. examples could be found in genetic networks , chains of catalytic reactions and social systems , where a process is simultaneously an active agent .
the authors are grateful to zdzisaw burda and to our anonymous referee for helpful remarks and useful references .
one of the authors ( a.m. ) is grateful to unesco for funds in frames of the fellowship programme unesco / poland . the research is partially supported within the fp7 project socionical , no .
the calculations were performed in the ack cyfronet , cracow , grants no .
mnisw / sgi3700 /agh /030/ 2007 and mnisw / sgi3700/agh /031/ 2007 .
|
we investigate the degree distribution @xmath0 and the clustering coefficient @xmath1 of the line graphs constructed on the erds - rnyi networks , the exponential and the scale - free growing networks .
we show that the character of the degree distribution in these graphs remains poissonian , exponential and power law , respectively , i.e. the same as in the original networks .
when the mean degree @xmath2 increases , the obtained clustering coefficient @xmath1 tends to 0.50 for the transformed erds - rnyi networks , to 0.53 for the transformed exponential networks and to 0.61 for the transformed scale - free networks .
these results are close to theoretical values , obtained with the model assumption that the degree - degree correlations in the initial networks are negligible . * clustering in random line graphs * + anna maka - kraso , advera mwijage@xmath3 and krzysztof kuakowski + _ _ faculty of physics and applied computer science , agh university of science and technology , al .
mickiewicza 30 , pl-30059 krakw , poland + @xmath3on leave from mbeya institute of science and technology , p.o.box 131 mbeya , tanzania [email protected] _ pacs numbers : _ 64.60.aq ; 02.10.ox ;
05.10.ln _ keywords : _ line graphs ; random networks ; clustering coefficient ; degree distribution
|
a clear signal of the new physics ( np ) that lies beyond the standard model ( sm ) would be the existence of heavy new particles or `` resonances '' . in tev mass range , the main arena of search for these objects
will soon be the cern lhc experiments as these particles can be produced or exchanged in high energy collisions .
the direct experimental evidence for a resonance is the peak in the energy dependence of the measured cross sections seen above the sm background .
once the resonance signal is observed , further analysis is needed to distinguish between scenarios that potentially may cause this effect .
first step in this analysis would be the determination of the spin of the resonance which provides an important selection among different classes of non - standard interactions . in the second step ,
then , one should test how likely this signal is accounted for by a certain np model based on the parameter space of the model , particle spectrum and production mechanism(s ) for the observed resonance . in this
regard , suppose lhc will observe a resonance in the @xmath2 1.5 - 2 tev mass range in the zz invariant mass distribution in the pp @xmath3 zz @xmath3 4 charged leptons mode .
now , what could be the underlying theory that is responsible for this resonance ? as a first possibility , recall @xcite that the first kaluza - klein ( kk ) mode of the spin-2 graviton from randall - sundrum ( rs ) extra - dimensional model @xcite may well account for this enhancement .
the reason is that for graviton with mass in the range @xmath2 1.5 - 2 tev , the graviton production via gluon fusion with subsequent decay to longitudinal z boson pair is significantly dominating over all other possible sources within the rs model . in particular , the vector boson fusion ( vbf ) production mode is found to be about an order of magnitude smaller for this mass range @xcite .
however , it is hard to reach the graviton mass bigger than @xmath2 2 tev as the graviton cross - section is small and the irreducible sm background to pp @xmath3 zz is starting to dominate for higher graviton masses . another spin-2 candidate for the enhancement in the @xmath2 1.5 - 2 tev mass range in this mode is the tensor bound state of some strongly - coupled technicolor ( tc ) theory @xcite produced via vbf and decaying to the longitudinal z boson pair .
such spin-2 resonance plays also an important role in these tc models since it helps to delay unitarity violation in the longitudinal @xmath4 and @xmath5 scatterings . in these scatterings , spin-1 isospin-1
( technirho ) vector resonance will appear only in the @xmath6 and @xmath7-channels in the @xmath4 scattering and will not appear in the @xmath8 scattering at all .
also , in the tc theories we will consider , the lightest spin-0 resonance is expected to be lighter than @xmath2 1 tev @xcite .
finally , due to the parity conservation , the axial resonance can not directly participate in these tree - level scatterings either .
this leaves the lightest spin-2 tc bound state as a strong candidate to account for the assumed - to - be - observed resonance .
thus , in this paper we start with the assumption that there is indeed a spin-2 resonance observed and refer to other studies for discrimination techniques between spin-0 , spin-1 and spin-2 resonances @xcite .
we will argue that due to the different production modes for the spin-2 states in rs and tc theories , it is possible to distinguish these models based on the angular analysis of the reconstructed longitudinal z bosons
. we will consider three different tc scenarios : minimal walking tc ( mwt ) @xcite , next to mwt ( nmwt ) @xcite and qcd - like tc theory .
first two models have received significant attention in the literature due to the fact that these models successfully pass the electroweak precision tests ( ewpt ) @xcite .
in addition , mwt leads to new candidates for cold dark matter @xcite and , with addition of modest amount of tc - neutral matter , also to a novel unification of the sm couplings @xcite . in all of these tc theories
, the technimatter will be singlet under qcd and , thus , the vbf will be the dominant production mechanism for the spin-2 bound state .
parameters of the tc theories will be chosen such as to preserve the unitarity up to at least ( m@xmath9 + 0.4 tev ) and to account for the experimental evidence of the resonance .
once this is accomplished , we will mimic the same resonance in the rs model and compare the angular distributions of reconstructed z bosons from both theories . due to the different production mechanisms
the angular distributions will be different and we will see that , in a long run , lhc will have a fair chance to conclusively distinguish between spin-2 states from tc and rs theories .
the paper is organized as follows . in section
ii we discuss how the different production mechanisms for the spin-2 state reveal themselves in the angular distributions of the longitudinal z bosons . in section
iii we present the amplitudes for the @xmath4 and @xmath8 scatterings and specify the parameters values for the three tc models we consider based on the unitarity and consideration of modified weinberg sum rules ( wsr ) . in parallel with section iii , section iv deals with the amplitudes and the parameters for the rs model . in section
v we present our numerical results relevant for lhc phenomenology .
final discussions and conclusions are presented in section vi .
in this section we show how the angular distribution of the reconstructed z bosons is connected to the production mechanism for the spin-2 resonance .
such resonance has the total of five possible polarization states , but gluons can produce only @xmath10 and @xmath11 since gluons do not have longitudinal polarizations and the total angular momentum has to be equal to @xmath12 ( where we have chosen beam axis to be in the z - direction ) . now , suppose that the two gauge bosons from the decay of the resonance are produced at the polar angle @xmath13 .
we rotate the gluon - produced resonance state specified by polarization tensor @xmath14 by this angle @xcite : @xmath15 where @xmath16 corresponds to the resonance state with the z - axis aligned with the direction of the decay products , and @xmath17 is the familiar wigner @xmath18-matrix .
independent wigner small @xmath19-matrix elements for the spin-2 state are presented in the appendix for completeness , see also @xcite .
now we may easily derive the angular dependence of the helicity amplitudes for the channels appearing in our analysis .
they follow from eq.([rotation ] ) for the @xmath20 resonance state which is produced either by two @xmath21 or by two @xmath22 gluon states : @xmath23 now just use clebsch - gordan decomposition of the @xmath24 and @xmath25 states in terms of 1@xmath261 final spin states to observe that , for example , helicity amplitude @xmath27\equiv { \mathcal{a}}_{\lambda_1 \lambda_2 \lambda_3 \lambda_4}$ ] for @xmath28 , @xmath29 , and @xmath30 .
notice that we have not included the @xmath31 and @xmath32 terms in eq.([decomp ] ) as the z bosons from the decay of the spin-2 resonance have longitudinal polarization and , thus , these terms can not contribute .
the @xmath33 state of the resonance does not contribute due to the fact that a gluon is massless . for longitudinal vbf , the only possible mode for the produced resonance is @xmath33 and the corresponding amplitude behaves as @xmath34 .
inherent to our analysis is the assumption that the spin-2 resonance is produced essentially at rest so that its decay products are mostly back to back .
we assume that the center of mass frame of the resonance can be determined experimentally in the clean pp @xmath3 zz @xmath3 4 charged leptons mode where the decay chain can be fully reconstructed .
the original idea of technicolor @xcite is the postulate of the existence of additional tc gauge interaction similar to qcd with the dynamical scale of the order of the electroweak scale and some new matter fields called technifermions taken to transform according to a suitable representation of the tc gauge group .
one of the explicit tc models we will consider in this paper will be the `` scaled - up '' version of qcd with technifermions in the fundamental representation of the tc gauge group .
in two other models , technifermions will be in the adjoint ( mwt ) and two - index symmetric ( nmwt ) representations of the tc gauge groups su(2 ) and su(3 ) , respectively .
furthermore , in all these models the technifermions are taken to be singlet under qcd and assigned anomaly free @xmath35 hypercharges .
the electroweak gauge group @xmath36 is a part of the new global symmetry group associated with chiral dynamics of the technifermions .
when strong dynamics of tc forces spontaneously breaks these new global symmetries the weak gauge bosons obtain their masses through the higgs mechanism in usual manner while the photon of electromagnetism remains massless .
we begin with the brief review of the formalism for the scattering of the ( techni)pions . at high energies , the scattering of longitudinally polarized vector bosons ( @xmath37 ) can be approximated by the scattering of the would - be - goldstone bosons @xmath38 .
if we think of these goldstone fields in analogy with the pions of qcd , we expect the @xmath39 scattering amplitudes to be unitarized by a spin - one , isospin - one vector resonance , like the techni - rho . as another alternative ,
if we think of the goldstone fields in terms of the linear sigma model , we expect the scattering amplitudes to be unitarized by a spin - zero , isospin - zero scalar field like the higgs boson .
both of these states , together with the spin-2 resonance will be included when we consider unitarity of the scattering amplitudes .
we are interested in the strongly interacting longitudinal @xmath37 bosons in the tev region .
we will ignore the up - down fermion mass splittings and , therefore , the su(2 ) `` isospin '' is conserved .
the @xmath39 scattering amplitudes can then be written in terms of isospin amplitudes with the following assignment of the isospin indices , @xmath40 where @xmath41 denotes either @xmath42 or @xmath43 , where @xmath44 and @xmath45 .
the scattering amplitude is given by @xmath46 where @xmath47 , and @xmath48 , @xmath6 , and @xmath7 are the usual mandelstam variables .
all the physics of @xmath49 scattering is contained in the amplitude function @xmath50 .
given the amplitude function , the physical amplitudes for boson - boson scattering are given as follows @xcite , @xmath51 we will need only @xmath52 and @xmath53 amplitudes .
contribution of the spin-1 ( technirho ) and spin-0 ( composite higgs ) resonances to the scattering is @xcite : @xmath54 \ , \label{eq : inv_2}\ ] ] here @xmath55 gev appropriate for the strong dynamics at the electroweak scale .
scaling up from qcd , note that the mass of the vector resonance should be as large as ( 246 gev/93 mev)@xmath56770 mev @xmath57 2 tev .
however in a theory with walking dynamics the resonances are expected to be lighter than in a running , qcd - like , setup .
similarly the coupling @xmath58 can be estimated by recalling that the qcd value that follows from @xmath59 mev would be @xmath60 .
the contribution of a spin - two meson @xmath61 to the invariant amplitude comes from the effective chiral interaction lagrangian between the traceless symmetric tensor nonet field @xmath62 and the pions @xcite , @xmath63= -\frac{g_2}{\sqrt{2 } } ( f_2)_{\mu\nu}\partial^{\mu } \vec{\pi}\cdot \partial^{\nu } \vec{\pi}+\cdots$ ] , and the resulting contribution to the amplitude is @xmath64-\frac{g_2 ^ 2 s^3}{12 m_{f_2}^4 } \ , \end{aligned}\ ] ] where @xmath65 and @xmath66 are the mass of the spin-2 meson and its coupling with the pions , respectively .
again , one can obtain a reference value for @xmath66 by relating to qcd where @xmath67 mev and @xmath68 mev gives @xmath69 gev@xmath70 so that @xmath71 . scaling the dimensionful numbers up to the eletroweak scale results in @xmath72 tev@xmath70 .
it is easy to check that in the @xmath73 limit the @xmath74 $ ] piece @xmath75 in accordance with the angular analysis section [ angular ] .
we note that the tensor interaction @xmath76 above can , on the gravity side , be compared to the gauge - invariant lagrangian for the kk graviton @xmath77 field coupled to the scalar bosons which is @xmath78 , where @xmath79 @xcite .
the appearance of the @xmath80 term in eq .
( [ tensor ] ) , which would not be present in the graviton case , is due to the absence of the @xmath81 term , _
i.e. _ the tracelessness condition for the @xmath62 . to constrain the parameter space by unitarity of the @xmath82 scattering
we apply the analysis of @xcite which we briefly review here for completeness . for the unitarity analysis the most general amplitude
should be expanded in its isospin @xmath83 and spin @xmath84 components , @xmath85 , but it turns out that the @xmath86 @xmath87 component , @xmath88 \ , \ ] ] has the worst high energy behavior , and is therefore taken as a basis for the analysis .
the possible resonances which contribute to the unitarity of @xmath82 scattering here are a light higgs , an axial ( which will be lighter than the vector in two of the three tc theories ) , vector and the tensor .
the axial can only be lighter than the vector in a walking technicolor theory ( wt ) , where the second weinberg sum rule is modified .
moreover , the chances of the axial being lighter than the vector are increased as the conformal window is approached , and the @xmath89 parameter decreases .
also , light composite higgs can naturally emerge in strongly coupled theories with matter in higher dimensional representations such as mwt and nmwt @xcite .
the axial resonance can not directly participate in the tree - level exchanges in the @xmath82 scattering due to parity invariance .
rather , the constraint due to @xmath90 field is indirect and appears in the @xmath82 scattering because the pion eaten by the @xmath91 boson contains a certain amount of the longitudinal component of @xmath90 . as a consequence ,
the @xmath58 and @xmath92 coupling are affected by the presence of @xmath90 , but since the dependence on @xmath93 comes together with other new parameters , it turns out that @xmath58 and @xmath92 remain completely free to take on any value . to see the effects of a light axial resonance in the @xmath82 scattering one imposes the wsr s and simultaneously requires that the @xmath89 parameter is small .
since the wsr s are affected by walking dynamics the resulting constraints on the allowed region in the @xmath94 space are different with walking dynamics than in a theory with a qcd - like dynamics and a heavy axial .
the wsr s , appropriately modified to account for the walking dynamics @xcite , read @xmath95 \ , \label{eq : wsr0 } \\ & & f_v^2-f_a^2=f_\pi^2 \label{eq : wsr1 } \ , \\ & & f_v^2 m_v^2 - f_a^2 m_a^2 = a\frac{8\pi^2}{d(r)}f_\pi^4 \label{eq : wsr2 } \ , \end{aligned}\ ] ] where @xmath96 ( @xmath97 ) and @xmath98 ( @xmath93 ) are decay constant and mass of the vector ( axial ) resonance , @xmath99 is the dimension of the fermion representation of the underlying gauge theory , and @xmath100 is an unknown number . in wt @xmath100
is expected to be positive and of order one while in a qcd - like theory @xmath101 . for the mwt with two flavors in the adjoint representation of su(2 )
we take @xmath102 , and the naive contribution to the @xmath89 parameter is @xmath103 . in the nmwt we have two flavors in the two - index symmetric representation of su(3 ) ,
@xmath104 , and the naive estimate of @xmath89 is @xmath105 . we will also consider the constraints for a running theory , _
i.e. _ @xmath101 .
furthermore , in addition to the wsr s of eqs .
( [ eq : wsr0 ] ) - ( [ eq : wsr2 ] ) one demands that vector and axial resonances should not be broad by excluding the regions in the parameter space corresponding to , @xmath106 finally , combining the above constraints with the unitarity constraints gives the allowed regions for the parameters @xmath107,@xmath108,@xmath109 and @xmath110 in the three tc theories we will consider ( see appendix of @xcite where these constraints were translated to the constraints on the values of @xmath108 . in our analysis
we added the tensor mass and coupling as extra parameters while in @xcite only @xmath107,@xmath108 were considered ) .
specific values for the parameters will be given in detail in the section [ numerics ] , where we present our numerical results . to conclude this section and to prepare for the numerical analysis of the total cross sections , we discuss the relevant production mechanism for spin-2 tc bound state , namely , vbf via @xmath111 or @xmath112 @xcite .
the probability for emission of ( an almost ) collinear longitudinal @xmath113 by a quark ( or anti - quark ) is suppressed by electroweak factor of @xmath114 .
however , the coupling of longitudinal @xmath113 to spin-2 tc tensor is enhanced .
moreover , vbf can proceed via valence quarks through the @xmath115 , @xmath116 or @xmath117 scatterings in addition to @xmath118 and @xmath119 annihilation ( which are suppressed due to the smaller sea quark content ) .
hence , the other amplitudes with transverse polarizations for initial or final state bosons can be neglected due to the smaller couplings to the spin-2 tc tensor . as was explained in sec .
[ angular ] , the expected angular dependence of the matrix element is @xmath120 as we will also see to arise from our numerical calculation .
the required parton - level cross - section is given by @xmath121 where the subscript @xmath122 on @xmath123 denotes longitudinal polarization .
the probability distribution for a quark of energy @xmath124 to emit a longitudinally polarized gauge boson of energy @xmath125 and transverse momentum @xmath126 ( relative to quark momentum ) is approximated by @xcite : @xmath127 ^ 2 } \nonumber \\ \label{lemission}\end{aligned}\ ] ] for an example of @xmath111 fusion from @xmath117 scattering , the proton - level cross - section can then be written as @xmath128 where in the second line , we have used the fact that from eq .
( [ lemission ] ) the average @xmath129 of the longitudinal @xmath123 is given by @xmath130 .
here , @xmath131 tev is roughly the energy of the longitudinal @xmath123 in order to produce an _ on_-shell composite tensor state .
hence , we can neglect transverse momenta in the parton - level cross - section , i.e. , set @xmath132 and integrate over the transverse momenta to obtain total probabilities , @xmath133 .
also , @xmath134 is the parton distribution function ( pdf ) corresponding to @xmath7 , @xmath19 quarks in the proton ; the @xmath7 quark ( or @xmath135 ) can come from the first proton and @xmath19 quark ( or @xmath136 ) from the second proton or vice versa .
expressions for contributions from @xmath137 emission from various other combinations of quarks and anti - quarks inside the protons can be similarly obtained . finally , we mention that the vbf offers the possibility to tag two additional highly energetic forward jets @xcite . in this case
we have to consider the @xmath112 + 2 jets sm background .
additional complication is the fact that in the present configuration of lhc experiments , pseudorapidity range @xmath2 6 - 10 is not covered .
we will not pursue the forward - jets - tagging option here and only require two ( almost ) back - to - back z bosons in the central region with `` central - jet - vetoing '' @xcite .
in this section , we briefly review the rs model and present the amplitudes needed for the numerical analysis . the rs framework @xcite features an extra - dimension which is taken to be a slice of ads@xmath138 space . at the endpoints of this five - dimensional space ( @xmath139 ) reside two branes which are usually labeled as an ultraviolet ( uv ) planck brane and an ir ( tev ) brane .
the large hierarchy of scales is resolved by a geometrical exponential factor . due to the warped geometry ,
the relationship between the @xmath140 mass scales ( taken to be of order @xmath141 where @xmath141 is the reduced planck mass in four dimensions ) and those in an effective @xmath142 description depend on the field - localization in the extra dimension .
the zero - mode of the graviton is localized near the uv brane which has a planckian fundamental scale , whereas the higgs sector is localized near the ir brane where the `` red - shifted '' fundamental scale is in the tev - range .
postulating modest - sized @xmath143 dimension with radius @xmath144 and curvature @xmath145 the ratio of the tev and planck scales is @xmath146 and can be numerically obtained by setting @xmath147 @xmath14811 .
the sm fields are allowed to propagate in the extra dimension @xcite . in this scenario
there are kk excitations of sm gauge and fermion fields in addition to those of the graviton .
these states have masses in the tev range and are localized near the tev brane .
the sm particles are the zero - modes of the 5d fields , and the profile of a sm fermion in the extra dimension depends on its 5d mass . by localizing light fermions near the planck brane and heavier ones near the tev brane ,
the contributions to the fcnc and ewpt are suppressed by factors tev/@xmath149 where @xmath150 . as a consequence ,
the kk graviton whose profile is peaked at the tev brane will couple mostly to the top quark , higgs ( or , by equivalence theorem , to the longitudinal @xmath91 and @xmath151 bosons ) , and kk excitations of the sm fields @xcite .
standard model gluons have a flat profile so that their coupling to kk graviton is suppressed only by a factor of the size of the extra dimension ( in units of radius of curvature ) , i.e. , @xmath152 .
now , we consider the couplings relevant for the production and decay . the production is dominated by gluon fusion and the coupling of gluons to kk gravitons is given by @xcite : @xmath153 } { k \pi r \left ( x_n^g \right)^2 | j_2 \left ( x_n^g \right ) | } \end{aligned}\ ] ] where @xmath154 denote bessel functions and the values @xmath155 give masses of the first four kk gravitons : @xmath156 .
also , in scenarios where the higgs field is localized at the tev brane , the coupling of the graviton to the longitudinal z bosons is equal to @xmath157 .
the relevant matrix elements for the process @xmath158 , via kk graviton are @xcite : @xmath159 } { \hat{s } - m_n^2 + i \gamma_g m_n } \label{mg1}\ ] ] where @xmath160 refer to initial and final state polarizations , c@xmath161 and @xmath162 are color factors , @xmath163 is the total decay width of kk graviton in our treatment and we have used @xmath164 . for c=1 ,
the width of the graviton is @xmath165 of the graviton mass which will be the input for our numerical analysis later .
this width value is higher than the one obtained in ref.@xcite but see @xcite , where additional graviton decay modes to the @xmath166 were considered .
we have @xmath167 \hat{s } } { 8 m_z^2 } \nonumber \nonumber \label{ggfusion}\end{aligned}\ ] ] where @xmath168 and the hatted variables are in the parton center of mass frame . the other amplitudes with transverse polarizations for @xmath151 bosons ( i.e. , @xmath169 )
can be neglected since these are suppressed relative to the above by @xmath170 . as we showed in section [ angular ] and as now can be seen also explicitly ,
@xmath171 as @xmath172 . similarly as in the case of technicolor , to reliably compare against the possible experimental signal we need the parton - level signal ( @xmath173 ) cross - section , averaged over initial state spins and colors
. this is given by @xcite : @xmath174 where a factor of @xmath175 has been included for identical bosons in the final state , initial helicity averaging has been accounted for by a factor of @xmath176 and a factor of @xmath177 accounts for color averaging .
note that @xmath178 is the only independent non - zero matrix element for the above process .
the total parton level cross section @xmath179 is related to the proton - level total signal cross - section as usual : @xmath180 where @xmath181 are the gluon pdf s and @xmath182 is the typical momentum transfer in the partonic process for resonant production of a kk graviton .
in this section we consider our predictions for the lhc . we evaluate numerically the total proton - level cross - sections as defined in previous sections . to begin with ,
we consider the sm background and then move on to present our numerical results .
we concentrate on the leptonic decay mode of the two @xmath151 bosons ( @xmath183 ) , based on considerations of the backgrounds .
the irreducible background to the @xmath112 final state , i.e. , sm contribution to @xmath184 is dominated by @xmath185 annihilation : gluon fusion is very small in the sm since it proceeds via a fermion loop .
hence , the interference of kk graviton signal ( dominated by @xmath186 ) with the sm background is negligible .
we also checked that @xmath112 + 2 jets sm background and its interference with the tc signal , with the requirements specified in the previous sections , are negligible .
the parton - level cross - section , averaged over quark colors and spins is given by @xcite @xmath187 , \nonumber \\\ ] ] where @xmath188 , @xmath189 , @xmath190 and @xmath191 this cross - section exhibits forward / backward peaking due to @xmath192 channel exchange , whereas the resonant kk graviton and tc signals do not have this feature .
hence , a cut on pseudo - rapidity @xmath193 is useful to reduce this background keeping the signal ( almost ) unchanged .
the total proton - level cross section is obtained as in the rs model case , namely , using eq .
( [ sigtot ] ) .
finally , we will exploit the fact that the sm background is dominated by transversely polarized z bosons .
thus , if we boost to the z rest frame and make an appropriate cut on the angle between the charged lepton momenta and the direction of the boost we select more data with longitudinally polarized z bosons @xcite . we can now present the comparison between tc and rs models . for numerical simulations
, we checked the sm background with calchep program @xcite and used mathematica program for graviton and tc signals . cteq5lo
pdf s were exploited throughout ( in their mathematica distribution package @xcite as well as intrinsically called by calchep ) .
since the main difference in angular distribution between rs and tc spin-2 signals is in the central rapidity region and also since sm background has forward / backward peaking we impose hard pseudorapidity cuts which we will specify below .
also , throughout the analysis we fix @xmath194=0.1m@xmath195 , @xmath196=0.1@xmath65 , c@xmath197=1 in rough agreement with the expectations from the rs and scaled - up qcd estimates .
we neglect the w and z boson masses which is a reasonable approximation to the accuracy of our final result .
given the above values , @xmath65 remains the only parameter in the rs theory .
our strategy is then , for any given mass of the spin two resonance , to find a set of allowed parameters for technicolor best mimicking the rs - case .
depending on the value of @xmath65 this corresponds to either qcd - like or walking technicolor as follows : * running theory * : @xmath65=2 tev , @xmath198 gev@xmath70 , @xmath199h=0.6 , @xmath200=130 gev , @xmath201 tev , @xmath202=2.26 .
now , we integrate over three - resonance - width region .
the results are shown in fig.[rs2 ] and in table.[tablezz ] .
we have @xmath2 20 total events in the @xmath203 region .
also , the angular dependence due to the presence of the spin-2 resonances shown in fig.[rs2 ] is given by @xmath204 and @xmath205 in rs and tc cases , respectively .
we included the @xmath206 channel into statistics of table.[tablezz ] due to the fact that the @xmath2500 gev energy @xmath207 s from @xmath151 decay will have a decay length of @xmath208 mm and therefore might leave visible tracks in the detector @xcite .
* mwt * : @xmath65=1.8 tev,@xmath198 gev@xmath70 , @xmath199h=0.57 , @xmath200=130 gev , @xmath209 tev , @xmath202=2.0 .
this parameter space corresponds to the region where axial is lighter than the vector .
we integrated over three - resonance - widhts region and made the cut @xmath2100.7 on the angle @xmath13 between the charged lepton momenta and the boost direction of the z boson .
the results are shown in fig.[rs3 ] and in table.[tablezz ] .
now , we have 29 and 19 total events for rs and tc cases respectively in the @xmath203 region . the spin-2 and spin-1 mass values for this case
were also intended to mimic the corresponding rs prediction @xmath2111.5 @xmath212 , where @xmath212 is the first spin-1 kk mass @xcite .
* nmwt * : @xmath65=1.6 tev,@xmath198 gev@xmath70 , @xmath199h=0.55 , @xmath200=130 gev , @xmath213 tev , @xmath202=1.8 .
this parameter space also corresponds to the region where axial is lighter than the vector .
we integrated over three - resonance - width region and also took @xmath214 .
in addition , the cut @xmath2100.7 on the angle @xmath13 between the charged lepton momenta and the boost direction of the z boson was again applied .
the results are shown in table.[tablezz ] and in fig.[rs4 ] . in this case , we have @xmath2 48 and 19 total events in the @xmath214 region for rs and tc cases , respectively .
these numerical values demonstrate the generic problem that lowering the spin-2 tensor mass , rs production dominates more and more over corresponding tc and sm background cases because gluon pdf s are larger in this region as compared to quark ones .
thus , it is very hard to mimic the production for both sides with low spin-2 mass .
thus , if the lhc happens to observe spin-2 state with such a low mass , rs would be the favored explanation out of the possibilities considered here .
one might think that it is possible to lower the rs production by decreasing the value of c@xmath161 parameter .
but then from eq .
( [ width ] ) the graviton width decreases significantly .
then the width of the rs graviton and spin-2 tc bound states are not expected to be the same anymore and therefore this should help to distinguish them in experiment .
we can not lower the spin-2 tc resonance width because for such a big value of the tensor coupling in our numerical analysis the width of the spin-2 tc bound state is expected to get bigger , not smaller , than in the guiding scaled - up - qcd case .
the fact that in our analysis we kept width of the spin-2 tc resonance close to the expected one from qcd but significantly increased the tensor coupling to the pions provides another tension in this problem .
we emphasize that the above dominance of rs vs tc spin two production is characteristic to the rs model described in section [ rs ] .
in particular , the ultraviolet scale in this model was taken to be @xmath215 but other possibilities can be envisioned . among these possibilities , in @xcite volume - truncated version , by a factor y @xmath148 6 , of the rs model was proposed with an ultraviolet cut off scale @xmath216 tev . in this `` little rs ( lrs ) '' model , the graviton cross - section considered in our paper would be lowered by a factor of @xmath217 and thus could agree in magnitue with the corresponding tc case with low spin-2 mass better @xcite .
however , the differences in the angular distributions between rs and tc remain robust .
zz mode integrated in three - resonance - width region ( a ) rs case ( b ) qcd - like tc theory case .
, title="fig:",width=288 ] zz mode integrated in three - resonance - width region ( a ) rs case ( b ) qcd - like tc theory case .
, title="fig : " ] + ( a)(b ) zz mode integrated in three - resonance - width region ( a ) rs case ( b ) mwt theory case.,title="fig:",width=288 ] zz mode integrated in three - resonance - width region ( a ) rs case ( b ) mwt theory case.,title="fig : " ] + ( a)(b ) zz mode integrated in three - resonance - width region ( a ) rs case ( b ) nmwt theory case.,title="fig:",width=288 ] zz mode integrated in three - resonance - width region ( a ) rs case ( b ) nmwt theory case.,title="fig : " ] + ( a)(b ) .signal @xmath218 4 charged leptons ( @xmath219,@xmath220,@xmath207 ) cross - section @xmath221(in fb ) with the corresponding leading sm background .
numbers correspond to @xmath203 cut case ( except the nmwt case where @xmath214 ) .
we assume 100 @xmath222 efficiency for our clean 4-lepton signal . [ cols="^,^,^,^,^",options="header " , ] we want to stress that once we assume that the 1.5 - 2 tev resonance is observed in experiment , it means that the tensor has to dominate in tc case under consideration which in turn implies that due to different production mechanism there will be a difference between rs and tc cases . of course , it is possible that in tc case parameters are different than the ones we considered but then it is not always possible to find resonant behavior because spin-2 is _ the only _ resonance in our treatment which can appear in the s - channel ( except the composite higgs which is expected to be light ) and so the tensor coupling should be large enough ( qcd reference value would be appropriate choice , for example ) .
the problem for such a qcd - like value of the tensor coupling is the dominance of the sm background , which is about twice bigger than the tc signal .
summarizing , we choose the tc parameters based on the a ) requirement of unitarity , b ) saturation of weinberg sum rules and c ) experimental observation of a spin-2 resonance . also , we would like to point out that in the typical rs model we are considering , the ratio between the graviton coupling to gluons and the graviton coupling to longitudinal gauge bosons is @xmath223 1/35 and , thus , one can not change the relative size of the gluon fusion vs. vbf graviton productions mechanisms .
therefore , the vbf mechanism remains subleading to the gluon fusion even when , similar to tc case , one tries to increase the graviton coupling to `` would - be '' goldstones ( this would , however , be possible by increasing the c@xmath197 value ) .
we analyzed three tc models and showed that it is possible to find parameter space for them which leads to spin-2 production similar to the rs case for the same mass and width of the tensor resonance .
we emphasize that the choices of the parameter values we made for the tc theory are rather uniquely fixed by the requirement of unitarity and the wsr .
the only unnatural value was the tensor coupling constant @xmath66 which was somewhat large compared to the qcd - like theory predictions .
the most natural region of the spin-2 mass where the two models can yield similar results was found to be about 2 tev . for smaller mass ,
rs production is expected to dominate due to high gluon densities and for higher mass the sm background becomes dominant .
we showed that still ( assuming the two models give similar signals ) it is possible to discriminate the models based on angular analysis after we know the mass , width , and s / b values from experiment . knowing these values will suggest the corresponding angular distributions for the two theories .
we note that our analysis can be also used as a prediction for the tc theories only ( without the reference to the rs ) . in this case , perhaps the tensor coupling value should be taken smaller ( close to the qcd value @xmath2244 tev@xmath70 ) .
still , in the resonant region , the tensor will dominate as long as the other parameters are taking values similar to the ones we used in our analysis .
angular distributions for this case will be dictated by the s / b ratio .
we thank francesco sannino and amarjit soni for a careful reading of the manuscript and many fruitful discussions .
@xmath225 k. agashe , h. davoudiasl , g. perez and a. soni , phys .
d * 76 * , 036006 ( 2007 ) [ arxiv : hep - ph/0701186 ] .
l. randall and r. sundrum , phys .
lett . * 83 * , 3370 ( 1999 ) [ arxiv : hep - ph/9905221 ] .
s. weinberg , phys .
d * 19 * , 1277 ( 1979 ) .
l. susskind , phys .
d * 20 * , 2619 ( 1979 ) .
d. k. hong , s. d. h. hsu and f. sannino , phys .
b * 597 * , 89 ( 2004 ) [ arxiv : hep - ph/0406200 ] .
b. c. allanach , k. odagiri , m. a. parker and b. r. webber , jhep * 0009 * , 019 ( 2000 ) [ arxiv : hep - ph/0006114 ] . b. c. allanach , k. odagiri , m. j. palmer , m. a. parker , a. sabetfakhri and b. r. webber , jhep * 0212 * , 039 ( 2002 ) [ arxiv : hep - ph/0211205 ] .
r. cousins , j. mumford , j. tucker and v. valuev , jhep * 0511 * , 046 ( 2005 ) .
p. osland , a. a. pankov , n. paver and a. v. tsytrinov , phys .
d * 78 * , 035008 ( 2008 ) [ arxiv:0805.2734 [ hep - ph ] ] .
f. sannino and k. tuominen , phys .
d * 71 * , 051901 ( 2005 ) d. d. dietrich , f. sannino and k. tuominen , phys .
d * 72 * , 055001 ( 2005 ) d. d. dietrich and f. sannino , phys .
d * 75 * , 085018 ( 2007 ) r. foadi , m. t. frandsen , t. a. ryttov and f. sannino , phys .
d * 76 * , 055005 ( 2007 ) f. sannino , arxiv:0804.0182 [ hep - ph ] .
s. b. gudnason , t. a. ryttov and f. sannino , phys . rev .
d * 76 * , 015005 ( 2007 ) [ arxiv : hep - ph/0612230 ] . s. b. gudnason , c. kouvaris and f. sannino , phys .
d * 74 * , 095008 ( 2006 ) [ arxiv : hep - ph/0608055 ] . c. kouvaris , phys .
d * 76 * , 015011 ( 2007 ) [ arxiv : hep - ph/0703266 ] . k. kainulainen , k. tuominen and j. virkajrvi , phys .
d * 75 * , 085003 ( 2007 ) [ arxiv : hep - ph/0612247 ] . m. y. khlopov and c. kouvaris , phys .
d * 78 * , 065040 ( 2008 ) [ arxiv:0806.1191 [ astro - ph ] ] . c. kouvaris , phys .
d * 78 * , 075024 ( 2008 ) [ arxiv:0807.3124 [ hep - ph ] ] .
s. u. chung , `` spin formalisms , '' cern-71 - 08 ; lectures given in the academic training program of cern 1969 - 1970 .
o. antipin and a. soni , jhep * 0810 * , 018 ( 2008 ) [ arxiv:0806.3427 [ hep - ph ] ] . s. m. berman and m. jacob , phys . rev .
* 139 * , b1023 ( 1965 ) .
m. golden , t. han and g. valencia , arxiv : hep - ph/9511206 .
r. foadi , m. jarvinen and f. sannino , arxiv:0811.3719 [ hep - ph ] .
f. sannino and j. schechter , phys .
d * 52 * , 96 ( 1995 ) [ arxiv : hep - ph/9501417 ] .
m. harada , f. sannino and j. schechter , phys .
d * 54 * , 1991 ( 1996 ) [ arxiv : hep - ph/9511335 ] .
t. han , j. d. lykken and r. j. zhang , phys .
d * 59 * , 105006 ( 1999 ) [ arxiv : hep - ph/9811350 ] .
h. davoudiasl , j. l. hewett and t. g. rizzo , phys .
b * 473 * , 43 ( 2000 ) [ arxiv : hep - ph/9911262 ] .
a. pomarol , phys .
b * 486 * , 153 ( 2000 ) [ arxiv : hep - ph/9911294 ] .
y. grossman and m. neubert , phys .
b * 474 * , 361 ( 2000 ) [ arxiv : hep - ph/9912408 ] .
s. j. huber and q. shafi , phys .
b * 498 * , 256 ( 2001 ) [ arxiv : hep - ph/0010195 ] .
t. gherghetta and a. pomarol , nucl .
b * 586 * , 141 ( 2000 ) [ arxiv : hep - ph/0003129 ] .
a. l. fitzpatrick , j. kaplan , l. randall and l. t. wang , arxiv : hep - ph/0701150 .
h. davoudiasl , j. l. hewett and t. g. rizzo , phys .
lett . * 84 * , 2080 ( 2000 ) [ arxiv : hep - ph/9909255 ] . h. davoudiasl , j. l. hewett and t. g. rizzo , phys .
d * 63 * , 075004 ( 2001 ) [ arxiv : hep - ph/0006041 ] .
t. han , arxiv : hep - ph/0508097 .
v. d. barger , k. m. cheung , t. han and r. j. n. phillips , phys .
d * 42 * , 3052 ( 1990 ) .
h. u. bengtsson , w. s. hou , a. soni and d. h. stork , phys .
* 55 * , 2762 ( 1985 ) .
o. antipin , d. atwood and a. soni , phys .
b * 666 * , 155 ( 2008 ) [ arxiv:0711.3175 [ hep - ph ] ] .
h. davoudiasl , g. perez and a. soni , phys .
b * 665 * , 67 ( 2008 ) [ arxiv:0802.0203 [ hep - ph ] ] .
|
we explore the possibility to discriminate between certain strongly - coupled technicolor ( tc ) models and warped extra - dimensional models where the standard model fields are propagating in the extra dimension .
we consider a generic qcd - like tc model with running coupling as well as two tc models with walking dynamics .
we argue that due to the different production mechanisms for the lowest - lying composite tensor state in these tc theories compared to the first kaluza - klein graviton mode of warped extra - dimensional case , it is possible to distinguish between these models based on the angular analysis of the reconstructed longitudinal z bosons in the @xmath0 four charged leptons channel .
|
in this paper we address the structure of self - conjugate core partitions .
a _ @xmath0-core partition _ ( more briefly _ @xmath0-core _ ) is a partition where no hook of size @xmath0 appears .
we let @xmath4 be the number of @xmath0-core partitions of @xmath2 and let @xmath5 be the number of self - conjugate @xmath0-core partitions of @xmath2 .
the study of self - conjugate partitions arises from the representation theory of the symmetric group @xmath6 and the alternating group @xmath7 . at the turn of the century
, young discovered that the irreducible characters of @xmath6 are labeled by partitions of @xmath2 , and in particular , the self - conjugate partitions label those that split into two conjugate irreducible representations of @xmath7 upon restriction . about the same time , frobenius discovered that the hook lengths on the diagonal of a self - conjugate partition determine the irrationalities that occur in the character table of @xmath7 .
the study of core partitions also arises in representation theory ; nakayama conjectured in the forties ( later proved by brauer and robinson ) that two irreducible characters of @xmath6 are in the same @xmath0-block if their labeling partitions have the same @xmath0-core . for this result and more on the development of the theory ,
see james and kerber @xcite .
more recently , core partitions have found to be related to mock theta functions , actions of the affine symmetric group , and ramanujan - type congruences .
self - conjugate partitions and core partitions intersect in several important ways .
hanusa and jones @xcite prove that for a fixed @xmath0 , self - conjugate @xmath0-core partitions are in bijection with minimal length coset representatives in the coxeter group quotient @xmath8 and they determine the action of the group generators on the set of self - conjugate @xmath0-cores .
self - conjugate core partitions are central to an ongoing investigation into the representation - theoretic navarro conjecture in the case of the alternating groups @xcite .
the last several decades have seen a growing interest in counting core partitions ; restricting to the case of self - conjugate partitions has opened new directions in research . here
we survey results on core partitions and their self - conjugate analogues and we propose a new conjecture that parallels one of stanton .
the @xmath0-core positivity conjecture asserts that every natural number has a @xmath0-core partition for every integer @xmath9 .
it was finally proved by granville and ono @xcite after initial results by ono and by erdmann and michler .
baldwin et al @xcite proved that every integer @xmath10 has a self - conjugate @xmath0-core partition for @xmath11 , with the exception of @xmath12 , for which infinitely many integers do not have such a partition .
olsson @xcite and garvan , kim , and stanton @xcite proved a generating function for @xmath5 , succeeding olsson s @xcite proof of the generating function for @xmath4 .
as an aside , conjecture [ conj:9core ] further highlights the peculiarity of self - conjugate @xmath13-core partitions .
recently , _ simultaneous _ core partitions have been investigated
partitions that are both @xmath14- and @xmath0-cores , where @xmath14 and @xmath0 are relatively prime .
anderson @xcite proved that there are @xmath15 many of such partitions , and olsson and stanton @xcite proved that the largest such partition is of of size @xmath16 .
ford , mai and sze @xcite have proved an analog of anderson s result in the case of self - conjugate simultaneous core partitions , showing that that there are @xmath17 such partitions when @xmath14 and @xmath0 are relatively prime . in 1999 , stanton @xcite posed the following monotonicity conjecture .
[ conj : stanton ] suppose that @xmath2 and @xmath0 are natural numbers and that @xmath18 .
then @xmath19 this was proved for values of @xmath0 that are large as a function of @xmath2 by craven @xcite and for large @xmath2 by anderson @xcite : suppose that @xmath2 is an integer , and let @xmath0 be an integer such that @xmath20 , and @xmath21
. then @xmath22 . if @xmath23 and @xmath24 are fixed integers satisfying @xmath25 then @xmath26 for sufficiently large @xmath2 .
more recently , stanton s conjecture was proved for many more values of @xmath0 and @xmath2 by kim and rouse @xcite , including when @xmath27 and @xmath28 . while the monotonicity criterion is conjectured for partitions in general , the set of self - conjugate partitions
do not satisfy a monotonicity criterion for any @xmath29 .
( this is corollary [ cor : monot ] ; see appendix [ sec : appendix ] for a table of values . )
however , we have found experimentally that @xmath30 for almost all values of @xmath9 and @xmath31 .
parallel to stanton s monotonicity conjecture , we propose the following monotonicity conjectures for self - conjugate core partitions .
@xmath32 [ conj : even ] @xmath33 [ conj : odd ] in this article , we discuss the given upper and lower bounds for these conjectures and prove the following partial results towards these conjectures .
[ thm : evenpc ] @xmath34 [ thm : oddpc ] @xmath35 along the way , we prove formulas for @xmath5 as a function of the number of self - conjugate partitions of @xmath36 for @xmath37 in theorems [ thm : sc2 t ] and [ thm : sc2t+1 ] . as a supplement to the positivity literature
, we discuss the positivity of @xmath3-core partitions of @xmath2 in conjecture [ conj:6core ] . for those readers familiar with the representation theory of the symmetric group @xmath6 and the alternating group @xmath7 , we provide a consequence of theorem [ thm : oddpc ] .
( for more information on the representation theory , see ( * ? ? ?
* chapter 4 ) or ( * ? ? ? * chapter 6 ) ) .
let @xmath0 be an odd prime . from (
* proposition 12.2 ) , we know that the defect zero @xmath0-blocks of @xmath6 restrict to defect zero @xmath0-blocks of @xmath7 in the following way .
when blocks @xmath38 and @xmath39 are labeled by distinct @xmath0-core partitions @xmath40 and @xmath41 of @xmath2 which satisfy @xmath42 , then they restrict to the same defect zero @xmath0-block of @xmath7 .
when a block @xmath43 is labeled by a self - conjugate partition of @xmath2 , it splits into two distinct defect zero @xmath0-blocks of @xmath7 upon restriction .
these are the _ splitting blocks _ of @xmath6 .
so , in particular , theorem [ thm : oddpc ] implies the following .
let @xmath44 be primes such that @xmath45 and @xmath46 .
for any prime @xmath0 , let @xmath47 be the set of defect zero @xmath0-blocks of @xmath7 that arise from splitting @xmath0-blocks of @xmath6
. then @xmath48 .
given a partition @xmath49 , let @xmath50 be the irreducible character of @xmath6 associated to @xmath51 and consider @xmath52 the product of all the hook lengths that appear in the young diagram of @xmath49 .
the frame thrall robinson hook length formula says that the _ character degree _ @xmath53 is @xmath54 @xcite .
for @xmath55 , define @xmath56 to be the highest power of @xmath0 dividing @xmath36 .
we have the following additional corollary .
let @xmath57 and @xmath58 be primes such that @xmath45 and @xmath46 . for any prime @xmath0 ,
let @xmath59 be the set of irreducible characters @xmath60 of @xmath6 which split upon restriction to @xmath7 such that @xmath61 .
then @xmath62 .
this paper is organized as follows . in section [ sec : facts ] , we recall basic facts about partitions , @xmath0-cores , and @xmath0-quotients , and prove new results on self - conjugate partitions . in section [ sec : mono ] , we discuss monotonicity and positivity results and conjectures depending on the parity of @xmath0 . our research in self - conjugate partitions branches out in many directions the last section of this paper brings attention to future research directions in representation theory , asymptotic analysis , unimodality , and numerical identities and inequalities .
we note that the results and perspective of craven in @xcite motivate much of our approach , and we obtain some similar results .
in order to state our results , we recall some basic definitions .
more details can be found in ( * ? ? ?
* sections 12 ) or ( * ? ? ?
* chapter 2 ) .
a _ partition _
@xmath49 of @xmath2 is a non - increasing sequence @xmath63 of positive integers such that @xmath64 .
each @xmath65 will be called a _ component _ of @xmath49 .
the _ young diagram _ associated to a partition @xmath49 is an up- and left - aligned series of rows of boxes , where the @xmath66-th row has @xmath65 boxes .
we label the positions of boxes in the young diagram using matrix notation ; the _ @xmath67-th position _ is the box in the @xmath68-th row and @xmath69-th column , so that the box in position @xmath70 is the upper - leftmost box . given a partition @xmath51 , its _ conjugate _ @xmath71 is a partition where the number of boxes in the @xmath66-th column of @xmath72 is the number of boxes in the @xmath66-th row of @xmath51 .
a partition is _ self - conjugate _ if @xmath73 . for a box @xmath43 in position @xmath67 , its _ hook _
@xmath74 is a set of boxes in the young diagram consisting of @xmath43 and the set of boxes in the @xmath68-th row to the right of @xmath43 and the boxes in the @xmath69-th column below @xmath43 ; its _ hook length _
@xmath75 is the number of boxes in @xmath74 .
a _ diagonal hook _ or _ diagonal hook length _ corresponds to a box on the ( main ) diagonal of the young diagram . because a self - conjugate partition @xmath49 is uniquely determined by its diagonal hook lengths
, we will use the notation @xmath76=(@xmath77 to refer to the decreasing sequence of diagonal hook lengths @xmath78 .
if @xmath49 contains a hook @xmath79 of length @xmath66 , we say that @xmath79 is an @xmath66-hook , and we can obtain an integer partition @xmath80 of @xmath81 from @xmath49 by _ removing _ @xmath79 in the following way : delete the boxes that constitute @xmath79 from the young diagram and migrate the detached partition ( if there is one ) up - and - to - the - left .
the following lemmas are related to hook lengths in self - conjugate partitions and are provided without proof .
let @xmath51 be a self - conjugate partition of @xmath2 defined by its diagonal hook lengths @xmath82 .
then for @xmath83 , the hook length @xmath75 equals @xmath84 . when @xmath85 , the hook length @xmath75 is strictly less than @xmath86 .
[ p : hij ] let @xmath51 be a self - conjugate partition of @xmath2 defined by its diagonal hook lengths @xmath82 .
then @xmath87 for all positions @xmath67 in the young diagram of @xmath51 , with the possible exception of @xmath88 .
[ p : bighooks ] we define @xmath89 to be the set of self - conjugate partitions of @xmath2 , @xmath90 to be the set of self - conjugate @xmath0-core partitions of @xmath2 and @xmath91 and @xmath92 .
clearly @xmath93 . the generating function for
the number of @xmath0-core partitions is due to olsson ( * ? ? ?
* proposition 3.3 ) , while the generating function for the number of self - conjugate @xmath0-core partitions is due to olsson ( * ? ? ?
* equation ( 2.40 ) ) and garvan , kim , and stanton ( * ? ? ?
* equation ( 7.1 ) ) : @xmath94 @xmath95 the _ @xmath0-core @xmath96 of @xmath49 _ is the partition obtained from @xmath49 by repeatedly removing @xmath0-hooks until none remain ; by theorem 2.7.16 in @xcite , @xmath97 is unique .
we introduce without definition the _ t - quotient _ of @xmath98 a sequence @xmath99 of partitions which record the hooks of @xmath49 which are divisible by @xmath0 .
we say that a @xmath0-quotient is _ self - conjugate _ when @xmath100 is the conjugate partition of @xmath101 for all @xmath102 .
the following results can be found in @xcite as propositions 3.6 and 3.5 .
given a partition @xmath49 of @xmath2 , its @xmath0-core @xmath103 and @xmath0-quotient@xmath99 satisfy @xmath104 .
further , there are exactly @xmath105 hooks in @xmath49 that are divisible by @xmath0 .
[ symquo ] a partition @xmath49 of @xmath2 is self - conjugate if and only if its @xmath0-core @xmath96 and @xmath0-quotient @xmath99 ( with the appropriate normalization ) are both self - conjugate . for the interested reader , the series of examples starting with 2.7.14 and ending with 2.7.28 in @xcite provide details on how calculate the @xmath0-core and @xmath0-quotient of a partition ( by way of its abacus diagram ) .
to show the symmetry inherent in the @xmath0-core and @xmath0-quotient of a self - conjugate partition , figure [ fig:5core ] shows the @xmath106-core and @xmath106-quotient of the partition defined by diagonal hooks @xmath107 .
@xmath108 and @xmath109{0.2 in } \epsfig{figure = figures / lam2915q0.eps , height=0.30 in } \end{minipage},\varnothing , \begin{minipage}[c]{0.35 in } \epsfig{figure = figures / lam2915q2.eps , height=0.30 in } \end{minipage } , \varnothing , \begin{minipage}[t]{0.3 in } \raisebox{0.04in}{\epsfig{figure = figures / lam2915q4.eps , height=0.15 in } } \end{minipage}~ \right)$ ] the following result describes the possible ways to remove a minimal amount of @xmath0-hooks from a self - conjugate partition to obtain a self - conjugate partition .
this is discussed further in section 4 of @xcite .
[ 2tremove ] let @xmath49 be a self - conjugate partition of @xmath2 that is not a @xmath0-core . 1 .
when @xmath0 is even , there there exists a pair of off - diagonal @xmath0-hooks such that upon their removal , the resultant partition is a self - conjugate partition of @xmath110 .
2 . when @xmath0 is odd , then one of the following must exist : a pair of off - diagonal @xmath0-hooks as in or a diagonal @xmath0-hook such that upon its removal , the resultant partition is a self - conjugate partition of @xmath111 .
the following result is key in proving our main results .
let @xmath2 and @xmath0 be positive integers .
then @xmath112 and @xmath113 where @xmath114 is the number of sequences of length @xmath0 of ( possibly empty ) partitions @xmath100 such that @xmath115 .
consider the set @xmath116 of self - conjugate partitions of @xmath2 that are not @xmath117-cores and let @xmath118 , whereby @xmath119 . by lemma
[ 2tremove ] , the @xmath117-core of any non-@xmath117-core must be obtained by the removing an even number of @xmath117-hooks .
furthermore , by proposition [ symquo ] , its @xmath117-core and ( non - empty ) @xmath117-quotient are both self - conjugate .
when one removes @xmath120 @xmath117-hooks , the @xmath117-core is a partition of @xmath121 and there are @xmath122 possible @xmath117-quotients . summing over valid values of @xmath68
gives equation .
consider the set @xmath123 of self - conjugate partitions of @xmath2 that are not @xmath124-cores and let @xmath125 .
the argument proceeds similarly as above , with the additional condition that the core of a non-@xmath124-core can be obtained by removing @xmath120 off - diagonal @xmath124-hooks and/or @xmath69 diagonal @xmath124-hooks , in which case the @xmath124-quotient has a non - empty partition @xmath126 of @xmath69 that is itself self - conjugate .
( note that this means @xmath69 will never be @xmath127 . )
there are a total of @xmath128 possible @xmath124-quotients which remove a total of @xmath129 @xmath124-hooks , and their @xmath124-cores are partitions of @xmath130 .
summing over valid values of @xmath68 and @xmath69 gives equation .
we establish bounds on @xmath132 and @xmath133 , which will be used in the next section to prove theorems [ thm : evenpc ] and [ thm : oddpc ] .
the technique used here is an adaptation of section 3 in @xcite .
let @xmath2 be an integer greater than or equal to @xmath134 .
then @xmath135 .
[ lem : scn-2 ] for a given @xmath136 , define two sets of self - conjugate partitions : * the set of self - conjugate partitions of @xmath2 whose diagonal hooks satisfy @xmath137 .
if @xmath2 is odd , also include @xmath138 .
* the set of self - conjugate partitions of @xmath2 whose diagonal hooks satisfy @xmath139 and whose parts are not all the same ( when @xmath2 is a square number ) . *
the set of self - conjugate partitions of @xmath2 in neither @xmath7 nor @xmath140 .
there is a bijection @xmath141 which takes a self - conjugate partition of @xmath142 and adds one box to the first row and to the first column .
we conclude that @xmath143 .
when @xmath140 is nonempty , there is also an surjection @xmath144 .
( @xmath140 is nonempty for all values of @xmath145 . ) for @xmath146 , define @xmath147 by the following steps .
first , if @xmath51 has one diagonal hook , define @xmath148 to have diagonal hooks @xmath149 if @xmath150 or @xmath151 if @xmath152 .
otherwise , suppose that the diagonal hooks of @xmath51 are @xmath153 ; create a self - conjugate partition @xmath154 with diagonal hooks @xmath155 , where @xmath156 and which keeps all other diagonal hooks the same ( @xmath157 for all @xmath158 ) .
next , determine ( if it exists ) the first @xmath68 such that @xmath159 . define @xmath148 to be the partition which adds one box to the @xmath160-st row and to the @xmath160-st column of @xmath154 . if no such @xmath68 exists , then @xmath161 is of the form @xmath162 for @xmath163 .
if @xmath51 has two diagonal hooks , then define @xmath148 to have diagonal hooks @xmath164 .
otherwise , @xmath154 has three or more diagonal hooks and @xmath165 ; define @xmath148 to have diagonal hooks @xmath166 .
the function @xmath167 is well defined because the image of every self - conjugate partition satisfies @xmath139 and is a surjection because for the function @xmath168 that removes the last box in the last row and the last box in the last column , then for any @xmath169 , it is true that @xmath170 . for @xmath169 , define the set @xmath171 to be the preimages of @xmath169 under @xmath167 .
the largest that this set can be is for the following @xmath172 , with diagonal hooks @xmath173 in each of these cases , @xmath174 . from the definitions of @xmath175 and @xmath167 ,
we can now bound @xmath131 as a function of @xmath176 when @xmath136 : @xmath177 the equation @xmath135 also holds for @xmath178 .
let @xmath2 be an integer greater than or equal to @xmath179 .
then @xmath180 .
[ lem : scn-4 ] lemma [ lem : scn-2 ] implies @xmath181 for @xmath182 . the equation @xmath180 also holds for @xmath183 .
the sequence @xmath184 ( a000700 in the on - line encyclopedia of integer sequences @xcite ) starts @xmath185 this , and lemma [ lem : scn-2 ] implies that @xmath186 for integers @xmath187 .
it also follows that @xmath188 for @xmath189 .
in this section , we prove formulas for @xmath190 for certain values of @xmath0 and @xmath2 , discussing their consequences for our monotonicity conjectures and the positivity of self - conjugate @xmath0-cores .
we first discuss formulas for @xmath191 for large values of @xmath117 . because the largest diagonal hook @xmath192 is odd in every self - conjugate core partition
, we have the following corollary of lemma [ p : bighooks ] .
[ cor : n/2 ] every self - conjugate partition of @xmath2 is a @xmath117-core for all integers @xmath0 satisfying @xmath193 . in particular , @xmath194 for integers @xmath0 satisfying @xmath193 .
proposition [ prop : evenlarge ] establishes a simple formula for @xmath191 for values of @xmath117 between @xmath195 and @xmath196 , which will be useful for proving theorem [ thm : evenpc ] .
[ prop : evenlarge ] let @xmath2 be a positive integer and suppose @xmath0 is an integer satisfying @xmath197
. then @xmath198 when @xmath197 , the sum in equation consists only of its first term , @xmath199 .
equation follows because @xmath200 and from corollary [ cor : n/2 ] because @xmath201
. we must be careful for values of @xmath117 near @xmath196 . substituting @xmath202 and @xmath203 into equation
establishes that when @xmath204 and @xmath205 , @xmath206 , which explains the upper bound we give for the even monotonicity conjecture .
explicit formulas are given as corollary [ cor:2n/4 ] .
[ cor:2n/4 ] let @xmath2 be an integer greater than or equal to @xmath207 .
then @xmath208 furthermore , let @xmath2 be an integer greater than or equal to @xmath209
. then @xmath210 in the formula for @xmath191 , the coefficients of @xmath0 are simply @xmath211 , which depends on @xmath2 modulo @xmath207 .
the range for which the formulas are valid comes from solving @xmath212 or @xmath213 . for successively smaller values of @xmath117 , formulas similar to those in corollary [ cor:2n/4 ] can be found .
for example , when @xmath2 is an integer greater than or equal to @xmath214 , then @xmath215 in general for self - conjugate @xmath216-cores , the range of validity of the formula for @xmath217 is for @xmath218 .
while equation does encompass all formulas of this type , these formulas are interesting in their own right . in general , we can apply equation repeatedly to find a formula for @xmath191 for all values of @xmath117 ; the formula only involves polynomials of @xmath0 and values of @xmath219 for @xmath37 .
[ thm : sc2 t ] we have the following formula for @xmath191 .
@xmath220 where the sum is over all sequences of positive integers @xmath221 such that its sum @xmath222 we now prove theorem [ thm : evenpc ] . by equation
, it suffices to prove @xmath223 , which is equivalent to @xmath224 .
lemma [ lem : scn-4 ] implies that @xmath225 when @xmath226 ; this condition is satisfied because the upper bound for @xmath117 under consideration implies @xmath227 .
last , because @xmath228 then @xmath229 , from which we have @xmath230 .
this completes the proof . before discussing monotonicity for small values of @xmath117
, we first discuss what is known about positivity in self - conjugate @xmath117-core partitions .
the only partitions which are @xmath127-cores are the staircase partitions @xmath231 , which are all self - conjugate . as a consequence , @xmath232 is non - zero exactly when @xmath2 is a triangular number .
ono and sze ( * ? ? ?
* theorem 3 ) characterize the integers having no self - conjugate @xmath207-core : @xmath233 if and only if the prime factorization of @xmath234 contains a prime of the form @xmath235 to an odd power .
baldwin et al .
@xcite prove that @xmath5 is positive for @xmath236 and @xmath237 , and give the example of @xmath238 to show that @xmath239 is not always positive .
however , they do not characterize when @xmath239 is zero . by using its generating function
, we generated the values of @xmath239 for @xmath240 , from which we conjecture the following .
let @xmath2 be a positive integer .
then @xmath241 except when @xmath242 .
[ conj:6core ] in the even monotonicity conjecture , we give the lower bound @xmath117 equals @xmath3 . indeed , there are integers @xmath2 such that @xmath243 , even for values of @xmath2 larger than @xmath244 .
( corollary [ cor:2n/4 ] establishes that @xmath245 . )
we conjecture that the set of such integers is finite , again aided by a computer search of non - negative integers @xmath2 up to @xmath246 .
let @xmath2 be an integer larger than @xmath244 .
then @xmath247 when @xmath248 and @xmath249 when @xmath250 .
there are no values of @xmath251 such that @xmath252 . for @xmath253 ,
there are no partitions of @xmath2 containing a hook length of @xmath254 . by lemma
[ p : bighooks ] , we know that the values of @xmath255 for @xmath256 are determined by the number of self - conjugate core partitions that have @xmath254 as its first diagonal hook . in other words , let @xmath2 be a positive integer and suppose that @xmath0 satisfies @xmath257 .
then @xmath258 [ cor : n/2odd ] corollaries [ cor : n/2 ] and [ cor : n/2odd ] imply : for fixed @xmath29 , the sequence @xmath259 is not monotonic .
[ cor : monot ] corollary [ cor : n/2odd ] also implies that for @xmath0 satisfying @xmath260 , @xmath261 . because @xmath186 for integers @xmath187 , we have the following corollary .
let @xmath2 be a positive integer and suppose that @xmath0 satisfies @xmath262 .
then @xmath263 [ cor : largeodd ] we now establish a formula for @xmath255 for values of @xmath254 between @xmath264 and @xmath196 . [
prop : oddlarge ] let @xmath2 be a positive integer and suppose @xmath0 is an integer satisfying @xmath265 . then @xmath266 when @xmath265 , the sum in equation is the sum of only two non - zero terms , @xmath267 which simplifies to @xmath268 .
we remark that @xmath269 and @xmath270 , so corollary [ cor : n/2odd ] implies that @xmath271+t\,sc(n-4t-2),\ ] ] from which equation follows . formulas like those in corollary [ cor:2n/4 ] can be found now for odd cores .
for example , when @xmath2 be an integer greater than or equal to @xmath272 , then @xmath273 as in the even core case , we can use equation to find a formula for @xmath255 involving polynomials of @xmath0 and values of @xmath219 for @xmath37 .
[ thm : sc2t+1 ] we have the following formula for @xmath255 .
@xmath274 where the sum is over all pairs of sequences of non - negative integers @xmath221 and @xmath275 such that @xmath276 for all @xmath277 and their sums satisfy @xmath278 .
we now prove theorem [ thm : oddpc ] . alongside corollary
[ cor : largeodd ] it remains to establish that @xmath279 for @xmath265 .
when @xmath280 and @xmath281 , then @xmath282 , so @xmath283 and we have @xmath284 when @xmath285 , proposition [ prop : oddlarge ] implies we need to prove @xmath286 when @xmath287 , then @xmath288 and @xmath289 .
if @xmath290 , then equation is @xmath291 , which is certainly true when @xmath283 .
otherwise , @xmath292 , so equation becomes @xmath293 , which is true when @xmath294 ; for the given range of @xmath254 , this requires @xmath295 .
( this result also holds for @xmath296 and @xmath297 . ) when @xmath298 , we will prove equation by proving that @xmath299 and relying on the fact that @xmath300 for @xmath2 and @xmath0 in our range . since @xmath301 ,
then @xmath302 , so lemma [ lem : scn-4 ] applies to give @xmath303 .
when @xmath304 , then @xmath305 , which in turn is less than @xmath254 .
therefore @xmath306 , so @xmath307 , implying @xmath308 , from which @xmath299 , as desired
. the lower bound of @xmath296 in theorem [ thm : oddpc ] is necessary from proposition [ prop : oddlarge ] , we have @xmath309 , @xmath310 , @xmath311 , @xmath312 , and @xmath313 .
there are other anomalies in for other values of @xmath314 and @xmath315 : we have @xmath316 , @xmath317 , @xmath318 . also of note are the two cases @xmath319 and @xmath320 .
robbins ( * ? ? ?
* theorem 7 ) and baruah and berndt ( * ? ? ?
* theorem 5.2 ) prove that the only integers having at least one self - conjugate @xmath321-core ( in fact , there is exactly one ) are of the form @xmath322 or @xmath323 for some non - negative integer @xmath324 .
garvan , kim , and stanton @xcite characterize the integers having no self - conjugate @xmath106-core : @xmath325 if and only if the prime factorization of @xmath2 contains a prime of the form @xmath235 to an odd power . in addition , they cite an observation of doug mcdoniel involving representations of integers as sums of three squares that proves that @xmath326 if and only if @xmath327 for integers @xmath36 and @xmath66 .
baldwin et al .
@xcite prove that @xmath328 for all @xmath2 of the form @xmath329 and cite a communication with peter montgomery which proves that this is a complete characterization of integers having no self - conjugate @xmath13-core partitions .
in the odd monotonicity conjecture , we give the lower bound @xmath254 equals @xmath13 . unlike in the even case
, it appears that @xmath330 for infinitely many values of @xmath2 ; the non - negative values of @xmath2 up to @xmath246 for which @xmath330 are @xmath331 note that these include many ( but not all ) values of @xmath332 ; this condition is neither necessary nor sufficient .
there are infinitely many positive integers @xmath2 such that @xmath330 .
[ conj:9core ] the choice of the lower bound @xmath333 in the odd monotonicity conjecture was chosen because @xmath334 when @xmath335 and @xmath336 when @xmath2 equals 11 or 23 .
that these are the only values satisfying @xmath337 has been verified for all non - negative @xmath338 .
in addition to the conjectures stated above , we have assembled multiple avenues for future exploration . to find a stronger monotonicity result for defect zero blocks of the alternating group , one would need to understand non - self - conjugate @xmath0-core partitions @xmath339 as well .
defect zero @xmath0-blocks arise in two ways .
the ones from @xmath6 that split upon restriction are counted by @xmath340 and those from @xmath6 that do not split upon restriction are counted by @xmath341 .
experimentally , @xmath342 for @xmath343 , so we conjecture the following .
suppose @xmath44 are odd primes such that @xmath344 .
the number of defect zero @xmath57-blocks of @xmath7 is strictly less than the number of defect zero @xmath58-blocks of @xmath7 . a deeper question than the monotonicity of @xmath345 has to do with the distribution of @xmath346 for a fixed @xmath2 , and as @xmath2 goes to infinity .
define the functions @xmath347 and @xmath348 which are the normalized net increase in the number of partitions of @xmath2 that are @xmath349-cores and not @xmath0-cores and the normalized net increase in the number of self - conjugate core partitions of @xmath2 that are @xmath1-cores and not @xmath0-cores . for fixed @xmath2 , we can see that @xmath350 plotting the functions ( of @xmath0 ) , @xmath351 , @xmath352 , and @xmath353 , for fixed values of @xmath2 between @xmath354 and @xmath355 gives the graphs in figure [ fig : distr ] . in @xcite ,
craven proves the following theorem .
suppose that @xmath356 is a real number .
then as @xmath2 tends to infinity , @xmath357 . as a consequence , as @xmath2 goes to infinity , @xmath351 approaches the function that is identically zero .
this is seen in figure [ fig : distr](a ) by noticing that the function values in the sequence of curves at a fixed value on the @xmath358-axis eventually decreases to zero .
this appears to be true for self - conjugate partitions as well .
suppose that @xmath356 is a real number .
then as @xmath2 tends to infinity , @xmath359 .
it appears that much more is true .
recall that a sequence @xmath360 is _ unimodal _ if there exists a number @xmath361 such that @xmath362 unimodality is a property that arises naturally in many areas , including combinatorics , geometry , and algebra ; brenti s survey article @xcite gives examples and references . in @xcite
, stanton discusses the unimodality of the coefficients of the generating function for partitions and self - conjugate partitions whose young diagrams fit inside a given shape .
it appears that for @xmath2 fixed and large enough , the sequences @xmath351 , @xmath352 , and @xmath353 are unimodal .
we state these as conjectures . for fixed @xmath363 ,
the sequence @xmath364 is unimodal .
[ conj : unimodal ] for fixed @xmath365 , the sequence @xmath366 is unimodal .
further , for fixed @xmath367 , the sequence @xmath368 is unimodal .
the formulas given in propositions [ prop : evenlarge ] and [ prop : oddlarge ] allow for partial results toward conjecture [ conj : unimodal ] , but the hard work is yet to be done .
more pointedly , we can ask for the shape of the distribution perhaps it is approaching a normal distribution , but after its peak it appears to decrease with a tail that is fatter than normal . because the pointwise limit of the distribution is the zero distribution ( by craven s theorem ) , the `` right question '' is more along the lines of finding the shape of the distribution as @xmath2 goes to infinity .
we state this as an open question
. for @xmath2 sufficiently large , is there a limiting shape of the distributions of @xmath351 , @xmath352 , and @xmath353 ? ideally , one would be able to find a combinatorial interpretation for @xmath346 to prove its positivity and understand its asymptotics .
is there a simple combinatorial description of @xmath369 ? of @xmath346 ?
another direction is related to numerical identities involving core partitions .
garvan , kim , and stanton prove that @xmath370 , @xmath371 , and @xmath372 using ( * ? ? ?
* equation ( 7.4 ) ) .
using ramanujan s theta functions , baruah and berndt @xcite prove @xmath373 and sarmah @xcite proves @xmath374 .
further , berkovich and yesilyurt @xcite prove inequalities such as @xmath375 and @xmath376 .
we aimed to find similar identities and inequalities .
experimental data suggests the following conjectures .
[ conj:1.9 ] let @xmath2 be a non - negative integer . 1 .
suppose @xmath377 .
then @xmath378 .
2 . suppose @xmath379
. then @xmath380 .
3 . suppose @xmath187
. then @xmath381 .
4 . suppose @xmath379
. then @xmath382 .
conjecture [ conj:1.9 ] gives some conjectures in a family of inequalities of the form @xmath383 .
it appears that for @xmath12 and @xmath384 , then there exists a constant @xmath385 where this is true for all @xmath386 not equal to @xmath127 modulo @xmath207 .
it would be of interest to determine the value and interpretation of these constants .
we do not expect identities of the form @xmath387 for integers @xmath388 and @xmath386 for odd @xmath315 and even @xmath236 , nor do we expect inequalities of the form @xmath383 for odd @xmath389 and even @xmath390 .
here we present tables of values of @xmath5 and @xmath346 , generated by extracting coefficients from the generating function in equation .
both authors would like to thank matt fayers , ben ford , and ken ono for conversations on this topic .
we thank noam elkies for his conversations about the positivity of 7-core partitions and the correspondence with representations of integers as the sum of three squares .
we also would like to thank nayandeep deka baruah , jrn olsson , and an anonymous referee for correcting and suggesting references related to our work .
the second author thanks jean - baptiste gramain for his hospitality while visiting the university of paris and for early discussions on a stanton - type conjecture for self - conjugate partitions while there .
francesco brenti .
log - concave and unimodal sequences in algebra , combinatorics , and geometry : an update . in _
jerusalem combinatorics 93 _ , vol .
178 of _ contemp .
_ amer . math .
soc . , providence , ri , 1994 .
|
a conjecture on the monotonicity of @xmath0-core partitions in an article of stanton [ open positivity conjectures for integer partitions , _
trends math_. , 2:19 - 25 , 1999 ] has been the catalyst for much recent research on @xmath0-core partitions .
we conjecture stanton - like monotonicity results comparing self - conjugate @xmath1- and @xmath0-core partitions of @xmath2 .
we obtain partial results toward these conjectures for values of @xmath0 that are large with respect to @xmath2 , and an application to the block theory of the symmetric and alternating groups . to this end
we prove formulas for the number of self - conjugate @xmath0-core partitions of @xmath2 as a function of the number of self - conjugate partitions of smaller @xmath2 .
additionally , we discuss the positivity of self - conjugate @xmath3-core partitions and introduce areas for future research in representation theory , asymptotic analysis , unimodality , and numerical identities and inequalities .
|
time - differential perturbed angular distribution ( tdpad ) methods : the @xmath0al beam was stopped in a cu crystal stopper mounted on the experimental apparatus for tdpad measurements , which was placed in a focal plane after the achromatic focal plane f7 .
the tdpad apparatus consists of a cu crystal stopper , a dipole magnet , ge detectors , a plastic scintillator and a collimator , as shown in the inset of fig .
[ fig_exp ] .
the cu stopper is 3.0 mm in thickness and 30@xmath7030 mm@xmath71 in area , and the dipole magnet provides a static magnetic field @xmath72 t. @xmath21al are implanted into the cu crystal , and de - excitation @xmath4 rays are detected with four ge detectors located at a distance of 7.0 cm from the stopper and at angles of @xmath73 and @xmath74 with respect to the beam axis .
the relative detection efficiency was 35% for one and 15@xmath7520% for the other three .
a plastic scintillator of 0.1 mm in thickness was placed upstream of the stopper , the signal from which provided the time - zero trigger for the tdpad measurement .
the tdpad apparatus enabled us to determine the spin alignment as well as the @xmath46-factor of @xmath21al by observing the changes in anisotropy of the de - excitation @xmath4 rays emitted from spin - aligned @xmath21al in synchronization with the spin precession in the presence of an external magnetic field .
experiments were performed under program no .
np0702-ribf018 at ribf , operated by riken nishina center and cns , the university of tokyo .
we thank the riken ring cyclotron staff for their cooperation during experiments .
y.i . is grateful to the special postdoctoral researchers program , riken .
this work was partly supported by the jsps kakenhi ( 22340071 and 20532089 ) , by the jsps and maee under the japan - france integrated action program ( sakura ) , and by the bulgarian national science fund ( grant did-02/16 ) .
|
the degree of freedom of spin in quantum systems serves as an unparalleled laboratory where intriguing quantum physical properties can be observed , and the ability to control spin is a powerful tool in physics research .
we propose a novel method for controlling spin in a system of rare isotopes which takes advantage of the mechanism of the projectile fragmentation reaction combined with the momentum - dispersion matching technique .
the present method was verified in an experiment at the riken ri beam factory , in which a degree of alignment of 8% was achieved for the spin of a rare isotope @xmath0al .
the figure of merit for the present method was found to be greater than that of the conventional method by a factor of more than 50 .
the immense efforts expended to fully comprehend and control _ quantum systems _ since their discovery are now entering an intriguing stage , namely the controlling of the degree of freedom of spin @xcite .
the case of nuclear systems is not an exception . in recent years
, nuclear physicists have been focusing their efforts on expanding the domain of known species in the nuclear chart , which is a two - dimensional map spanned by the axes of @xmath1 ( number of neutrons ) in the east direction and @xmath2 ( number of protons ) in the north direction . the key technique used to explore the south eastern ( neutron - rich , or negative in isospin @xmath3 ) and north western ( proton - rich , or positive in @xmath3 ) fronts of
the map has been the projectile fragmentation ( pf ) reaction , in which an accelerated stable nucleus is transmuted into an unstable one through abrasion upon collision with a target .
several new facilities for providing rare - isotope ( ri ) beams by this technique , such as ribf @xcite in japan , frib @xcite in the united states , and fair @xcite in europe , have been completed or designed for exploration of the frontiers of the nuclear chart . beyond such efforts toward exploring the @xmath1 and @xmath2 axes
, nuclear spin may be a `` third axis '' to be pursued .
the study reported in the present article concerns the control of the spin orientation of an unstable nucleus produced in a ri beam at such fragmentation - based ri beam facilities . the ability to control spin , when applied to state - of - the - art ri beams ,
is expected to provide unprecedented opportunities for research on nuclear structure of species situated outside the traditional region of the nuclear chart , as well as for application in materials sciences , where spin - controlled radioactive nuclei implanted in a sample could serve as probes for investigating the structure and dynamics of condensed matter @xcite .
the fragmentation of a projectile nucleus in high - energy nucleus - nucleus collisions is described remarkably well by a simple model that assumes the projectile fragment produced in the pf reaction to be a mere `` spectator '' of the projectile nucleus ; as a spectator , this fragment survives frequent nucleon - nucleon interactions , and the other nucleons ( `` participants '' ) are abraded off through the reaction @xcite , as illustrated in fig .
[ fig_pf ] . in the model
, the projectile fragment acquires an angular momentum ( in other words , a nuclear spin ) , whose orientation is determined simply as a function of the momentum of the outgoing fragment . although the spin orientation may practically be reduced due to cascade @xmath4 decays following the fragmentation , we assume that a significant amount of spin orientation survives in the fragment . here
the degrees of spin orientation of rank one and two , in particular , are referred to as _ spin polarization _ and _ spin alignment _ , respectively .
this implies a unique relation between the spin orientation and the direction of the removed momentum @xmath5 , as illustrated in fig .
[ fig_pf ] , which can be utilized as an obvious means for producing spin - oriented ri beams .
one advantage of this method of orienting the fragment spin is that the resulting spin orientation does not depend on the chemical or atomic properties of the ri .
however , the method also shows a drawback in the sense that the spin orientation thus produced in the pf reaction tends to be partially or completely attenuated since the fragmentation generally involves the removal of a large number of nucleons from the projectile .
this is quite a non - negligible flaw with respect to the yields attainable for spin - oriented beams since high - intensity primary beams are only available for a limited set of nuclear species , and consequently in most cases ris of interest must be produced through the removal of a large number of nucleons from the projectile .
accordingly , there has been high demand for a new technique for preventing the attenuation in spin orientation caused by large differences in mass between the projectile and the fragment . in this paper , we present a method for producing highly spin - aligned ri beams by employing a two - step pf process in combination with the momentum - dispersion matching technique .
is incident to the target . in a `` participant - spectator '' model ,
the nuclear spin @xmath6 of the fragment arises from the angular momentum @xmath7 with respect to the center of mass of the projectile nucleus , owing to the removal of nucleons that , before collision , were in internal motion in the projectile nucleus ( fermi motion ) . here
, @xmath5 is the sum of momenta of the removed nucleons ( participants ) , and @xmath8 is the position vector of the participants in the projectile rest frame .
furthermore , the linear momentum @xmath9 of the fragment is given as @xmath10 .
thus the orientation of the spin @xmath11 is determined plainly as a function of the momentum @xmath9 of outgoing fragment . in the figure , the axis of polarization is perpendicular to the reaction plane , while the axis of alignment is parallel to the beam axis .
insets ( a ) , ( b ) and ( c ) illustrate cases which produce spin orientation for the fragments in the left wing , center and right wing of the momentum distribution , respectively . ]
[ fig_pf ] al from a primary beam of @xmath12ca .
the graphs below each scheme represent the typical momentum distribution and the corresponding alignment , with abscissas representing the momentum @xmath13 of @xmath0al .
( a ) single - step pf method .
the @xmath0al beam is directly produced from @xmath12ca .
since pf involves a large number of nucleons , the expected spin alignment is small .
( b ) two - step pf method .
@xmath0al is produced via an intermediate nucleus @xmath14al .
the expected spin alignment is high , whereas the production yield is low due to the two fold selection with momentum slits .
( c ) two - step pf method with dispersion matching .
direct selection of the change in momentum @xmath15 in the second pf can be achieved by placing a secondary target in the momentum - dispersive focal plane and a slit in the double - achromatic focal plane , because the momentum spread @xmath16 of the incident beam is compensated for by fulfilling the condition of momentum - dispersion matching .
this method yields an intensive spin - aligned ri beam while avoiding cancellation between the opposite signs of spin alignment caused by the momentum spread @xmath16 .
, width=604 ] [ fig_2step ] ca ions , preaccelerated to 114 mev / nucleon , is introduced into src ( superconducting ring cyclotron ) , where it is further accelerated to 345 mev / nucleon .
the beam is subsequently adjusted to be incident to a primary target located in the focal plane f0 .
the projectile fragments are analyzed and selected in the bigrips beamline , where f1 , f4 , f5 and f6 are the momentum - dispersive focal planes and f2 , f3 and f7 are the double - achromatic focal planes . in the present experiment , a second pf
producing @xmath0al from @xmath14al takes place in f5 .
the inset shows the apparatus used for the tdpad experiment .
, width=604 ] [ fig_exp ] figure [ fig_2step ] illustrates three different schemes for producing spin - aligned ri beams , where each scheme uses a different configuration of elements , namely primary and secondary targets and slits for selection .
the most basic scheme employs the configuration in ( a ) , in which a nucleus of interest is directly produced from a primary beam through a single occurrence of the pf ( a single - step pf reaction ) . as stated earlier
, this scheme suffers from the drawback that the degree of spin alignment tends to be attenuated when the pf involves the removal of a large number of nucleons from the projectile . with the aim to overcome this problem , configuration ( b ) adopts a two - step pf reaction , where a beam of nuclei produced in the first pf reaction ( secondary beam ) is used to obtain a beam of the nuclide of interest through a second pf reaction .
in particular , using a slit installed at a momentum - dispersive focal plane , the particles forming the secondary beam are chosen to be of a nuclide containing one proton or neutron more than the nuclide of interest .
thus , the target ri beam is produced via a pf reaction in which only one proton or neutron is removed . for the ri beam obtained with this scheme
, the spin alignment is expected to be high due to the simplicity of the reaction @xcite .
we also note that a significant increase in the total production yield is suggested by experiments @xcite in in - beam @xmath4-ray spectroscopy . in the scheme presented above ,
however , the production yield is typically reduced by a factor of @xmath171/1000 because the production of target nuclei requires the successive occurrence of two highly particular reactions . a hint on how to eliminate the disadvantage of the scheme in ( b )
emerged from the recognition that the quantity that determines the spin alignment is solely the momentum change @xmath15 in the process of fragmentation that produces the nuclei of interest , and that the spin alignment is not sensitive to the momentum of the incident nucleus . in scheme ( b )
, @xmath15 is selected with the aid of two momentum slits , the first of which is used to select the momentum of the secondary beam and the second slit determines the outgoing momentum of the secondary pf . the tremendous and unnecessary drop in yield
is avoided by discarding the two - fold selection and introducing a single direct selection of the target @xmath15 itself .
this is realized by placing a secondary target in the momentum - dispersive focal plane and a slit in the double - achromatic focal plane , as illustrated in fig .
[ fig_2step ] ( c ) .
the concept of realizing maximum spectral resolution in momentum loss by compensating for the beam momentum spread of the incident beam , as executed here , is known as _ dispersion matching _ in ion optics @xcite .
the important point of this technique is that the reaction products that acquire equal amounts of momentum change upon the second fragmentation are focused onto a single physical location .
the application of this technique to pf - induced spin alignment can prevent the cancellation of opposite signs of spin alignment caused by momentum spread , which secondary beams unavoidably undergo .
the validity of scheme ( c ) was first tested with the in - flight superconducting ri separator bigrips @xcite at the riken ribf facility @xcite .
the arrangement for the production of spin - aligned ri beams with the present method is shown in fig .
[ fig_exp ] . in the reaction at the primary target position f0 ,
@xmath14al was produced by a pf reaction of a 345-mev / nucleon @xmath12ca beam on a @xmath18be target with a thickness of 1.85 g/@xmath19 , chosen to provide a maximum production yield for the secondary @xmath14al beam .
a wedge - shaped aluminium degrader with a mean thickness of 4.05 g/@xmath19 was placed at the first momentum - dispersive focal plane f1 , where the momentum acceptance at f1 was @xmath203% .
the secondary @xmath14al beam was introduced to a second wedge - shaped aluminium target with a mean thickness of 2.70 g/@xmath19 , placed at the second momentum - dispersive focal plane f5 . the @xmath0al nuclei ( including those in isomeric state @xmath21al )
were produced through a pf reaction involving the removal of one neutron from @xmath14al .
the thickness of the secondary target was chosen such that the energy loss from the target was comparable with the theoretical estimate for the width of the momentum distribution @xcite for single - nucleon removal . in the present case , @xmath22 mev/@xmath23 , and the momentum width for @xmath0al was measured to be @xmath24 mev/@xmath23 or 0.4% .
the @xmath0al beam was subsequently transported to focal plane f7 whereby the momentum dispersion between f5 to f7 was tuned to be with the same magnitude and opposite sign as that from f0 to f5 ( momentum matching ) , effectively canceling out the momentum dispersion from the site of the first pf reaction to f7 .
we note that an admixture of @xmath0al particles in the @xmath14al secondary beam was found to be negligibly small , due to the difference in the magnetic rigidity .
thus , the @xmath0al particles that were solely produced at f5 were transported to f7 .
the slit at f7 was used to select a region of momentum change at the second pf as @xmath25% about the center of relative momentum distribution .
the @xmath0al beam was then introduced to an experimental apparatus , shown in the inset of fig .
[ fig_exp ] , for time - differential perturbed angular distribution ( tdpad ) measurements .
( see `` methods '' for details . )
rays emitted from @xmath21al at 222 kev and 735 kev were observed .
( b ) time variations of @xmath26 and @xmath27(t ) for 222-kev @xmath4 rays .
events around @xmath28 originate from prompt @xmath4 rays .
( c ) @xmath29 ratio deduced from @xmath26 and @xmath27(t ) , according to eq .
[ eq_rfunc1 ] .
the solid line represents the theoretical @xmath29 function in eq .
[ eq_rfunc2 ] after fitting to the experimental @xmath29 . ] [ fig_rfunc ] the degree of spin alignment @xmath30 was determined from a ratio @xmath29 defined as @xmath31 where @xmath32 ( @xmath27 ) is the sum of the photo - peak count rates at ge 1 and ge 3 ( ge 2 and ge 4 ) , which are two pairs of ge detectors placed diagonally to each other , as depicted in the inset of fig .
[ fig_exp ] , and @xmath33 denotes a correction factor for the detection efficiency . theoretically , the @xmath29 ratio is expressed as a function of @xmath34 as @xmath35 in terms of the rank - two anisotropy parameter @xmath36 , which is defined as @xmath37 .
terms with higher ranks were evaluated to be negligible in the present case of @xmath21al . here , @xmath30 denotes the degree of spin alignment @xmath38 where @xmath39 is the occupation probability for magnetic sublevel @xmath40 , and @xmath41 the nuclear spin .
@xmath42 is the statistical tensor for complete alignment , and @xmath43 is the radiation parameter @xcite .
the parameter @xmath44 ( larmor frequency ) is given by @xmath45 , where @xmath46 is the @xmath46-factor of @xmath0al in units of the nuclear magneton @xmath47 , and @xmath48 is the initial phase of @xmath29 .
the @xmath0al nucleus is known to exhibit an isomeric state @xmath21al @xcite at 957 kev with a half - life of 200(20 ) ns . the spin and parity of @xmath21al
have not been fixed among the @xmath49 and @xmath50 candidates .
it is known that @xmath21al undergoes de - excitation by e2 transition @xcite with emission of @xmath4 rays with an energy of 222 kev and subsequently decays in cascade to the ground state by emitting 735-kev @xmath4 rays .
figure [ fig_rfunc ] ( a ) shows a @xmath4-ray energy spectrum measured with the ge detectors , where 222-kev de - excitation @xmath4 rays are clearly observed as a peak . the time variations @xmath26 and @xmath51 of the intensities for this peak obtained with detectors pairs ge 1 - 3 and ge 2 - 4 , respectively , are presented in fig . [ fig_rfunc ] ( b ) , in which the corresponding abscissas represent the time difference of the signals at either of the ge detector pairs relative to the beam particle signal at a plastic scintillator placed in front of the stopper crystal .
the @xmath29 ratio evaluated according to eq .
[ eq_rfunc1 ] is shown in fig .
[ fig_rfunc ] ( c ) . from the least @xmath52
fitting of the theoretical function of eq .
[ eq_rfunc2 ] to the experimental @xmath29 ratio of eq .
[ eq_rfunc1 ] , we obtained the degree of spin alignment as @xmath53% , and the @xmath46-factor of @xmath21al was determined for the first time to be @xmath54 .
also , the spin and parity were assigned to be @xmath55 through comparison of the @xmath46-factor with theoretical calculations . detailed analysis and extended discussion regarding the @xmath0al nuclear structure based on the obtained @xmath46-factor and spin - parity will be presented elsewhere .
a remeasurement of the degree of spin alignment was also performed during the experiment , in which the momentum acceptance in the f5 focal plane was narrowed to be @xmath200.5% , while maintaining other conditions unchanged .
this measurement corresponded to the two - step pf reaction without dispersion matching ( case ( b ) in fig .
[ fig_2step ] ) .
the degree of spin alignment derived from this measurement , 9(2)% , is consistent with the above value obtained with the proposed method , 8(1)% , thus confirming that the present method of producing spin - aligned ri beams is valid and performs well .
.comparison of the two - step and the single - step pf methods . @xmath56 and @xmath57 denote the width of the momentum distribution and the relative width of selected region around the center of the fragment momentum distribution , respectively .
@xmath58 , @xmath59 and @xmath60 are the yield of @xmath0al beam particle , the yield of its isomeric state @xmath21al , and the purity of @xmath0al particles in the ri beam , respectively , at the final focal plane for each method .
the isomer to ground - state ratio @xmath61 for the production of @xmath0al was deduced as @xmath62 , where @xmath63 is the correction factor for the in - flight decays of @xmath21al from the production target to the final focal plane , which are 0.47 and 0.17 for the two - step and single - step methods , respectively .
@xmath30 is the degree of spin alignment .
the values of fom are defined as proportional to @xmath64 and normalized so that the fom for the two - step method is unity . [ cols="<,^,^",options="header " , ] a supplementary experiment was carried out in order to compare the performance of the present method with that of the single - step method .
@xmath0al was directly produced in a pf reaction of a @xmath12ca beam on a 4-mm thick be target .
the thickness of the production target was chosen such that the energy loss in the target was comparable with the goldhaber width @xcite ( expected to be 4% in this case ) . in order to compare with the case of the two - step method under the equivalent condition
, this measurement was carried out by selecting a momentum region of @xmath200.5% around the center of fragment momentum distribution .
for this momentum region a maximum prolate alignment is expected . as a result
, the spin alignment was measured to be less than 0.8% ( 2@xmath56 confidence level ) .
a comparison of the two methods is summarized in table [ tab_fom ] .
the figure of merit ( fom ) for the production of such spin - aligned ri beams should be defined to be proportional to the yield and the square of the degree of alignment . in the measurement with the single - step pf reaction , a primary beam
whose intensity was deliberately attenuated by a factor of 1/100 was used in order to avoid saturation in the counting rate at the data acquisition system . here ,
the fom was compared on the basis of actual effectiveness without correction for the attenuation , in which the resulting fom for the new method was found to be improved by a factor of more than 50 .
note that the degree of spin alignment in the single - step pf reaction could not be determined within a measurement time comparable with that of the two - step pf reaction .
the superiority in fom of the new method over the single - step pf reaction method should be even more pronounced for nuclei located farther from the primary beam .
theoretically , the maximum of the spin alignment for the case of single - nucleon removal from @xmath14al with a momentum acceptance of @xmath65% is estimated to be 30% in a way similar to that described in @xcite .
the estimation is based on a model proposed by hfner and nemes @xcite , where the cross - section for the abrasion of one nucleon leading to a fragment of substate @xmath40 with momentum @xmath13 is proportional to the probability of finding a particle of substate @xmath66 with momentum @xmath67 at the surface of the target nucleus .
the maximum evaluated in this way is in fact four times greater than that obtained experimentally , which may result from de - excitation from higher states populated through the pf reaction , such as the ( @xmath68 ) @xcite and @xmath69 @xcite states .
this suggests that the ability to select the reaction path in populating the state of interest is key to achieving augmented spin alignment .
thus , spin alignment via pf reactions depends strongly on both the reaction mechanism and the nuclear structure . under these circumstances ,
the achieved degree of alignment , 1/4 of the theoretical maximum which was obtained despite the situation that the reaction path to the isomeric state was not unique , is rather satisfactory . if we choose a nucleus produced by a unique reaction path , a degree of spin alignment closer to the theoretical maximum might be possible to achieve .
-factor of its isomeric state with a 5 @xmath56 confidence level in a one - day beam time . in the plot ,
primary beams are restricted to the typical beam particles which are available at high intensities at ribf .
also , the following conditions are assumed : the degree of spin alignment is 10% for single - nucleon removal from the beam particle , and reduces exponentially to 1% down to 10-nucleon removal , as has been determined empirically ; the values of intensity actually available at ribf is assumed for each species of the primary beam ; the cross - sections for the pf reactions are estimated based on parameter sets known as epax2 @xcite , and the cross - section for the secondary pf reaction is assumed to be 1/1000 , as usual ; the isomeric to ground state population ratio for the nucleus of interest in the pf reaction is 50% ; and the external magnetic field up to 1 tesla is available for the tdpad measurement .
, width=604 ] [ fig_chart ] figure [ fig_chart ] shows the result of simulating the accessibility of unstable nuclei via the two - step pf method ( red region ) and the conventional method ( blue region ) . in the simulation , the primary beam
is assumed to be restricted to within a class of beam species which are available with high intensities at ribf @xcite .
clearly , the adoption of the two - step method drastically expands the set of accessible nuclei in the nuclear chart .
in addition to a simplest case that the nucleus of interest was produced through the one - nucleon removal reaction as presented in this article , the two - step scheme also allows to utilize few - nucleon removal reactions as well as few - nucleon pickup reactions which are known to produce significant spin orientation @xcite .
the fom of our proposed method was found to be more than 50 times greater than that of the conventional single - step pf reaction in this particular case and numerical simulations indicate that the present method dramatically broadens the domain of accessible nuclei in the nuclear chart . such an ability to control spin , when applied to state - of - the - art ri beams ,
is expected to provide unprecedented opportunities for research on the nuclear structure of species situated outside the traditional region of the nuclear chart , as well as for applications in material research where spin - controlled radioactive nuclei implanted in a sample serve as probes into the structure and dynamics of condensed matter .
|
let @xmath0 be a closed orientable surface of genus @xmath2 , and consider the deformation space of convex @xmath3 structures @xmath1 on @xmath0 .
topologically , we know that @xmath1 is a cell of dimension @xmath4 by goldman @xcite . moreover , in @xcite , goldman proved that , just as in the case of hyperbolic structures on @xmath0 , convex @xmath3 structures on @xmath0 are given by specifying convex @xmath3 structures on the pairs of pants in a pants decomposition of @xmath0 and then assembling them together .
this allowed him to parameterize @xmath1 in a way similar to the fenchel - nielsen parameterization of teichmller space @xmath5 . to define the goldman parameterization of @xmath1 , we first need to choose a pants decomposition @xmath6 for @xmath0 .
there are then three kinds of parameters in the goldman parameterization ; the twist - bulge parameters , the boundary invariants and the internal parameters .
the twist - bulge parameters describe how to glue pairs of pants together , and are the analog of the fenchel - nielsen twist coordinates .
the boundary invariants contain the eigenvalue information of the holonomy about each simple closed curve in @xmath6 , and correspond to the fenchel - nielsen length coordinates .
however , unlike the hyperbolic case , specifying only the boundary invariants and the twist - bulge parameters is insufficient to nail down a point in @xmath1 .
the internal parameters are thus designed to describe this residual deformation " . for every pair of pants in the pants decomposition of @xmath0
, we have two internal parameters , which take values in a @xmath7-cell @xmath8 in @xmath9 .
choose a goldman parameterization of @xmath1 , and consider a sequence in @xmath1 so that all the boundary invariants vary within a compact set , but every pair of internal parameters eventually leave every compact set in @xmath8 .
we call such a sequence a _ goldman sequence_. one should think of a goldman sequence as a deformation of the convex @xmath3 structure on @xmath0 away " from the image of the natural embedding @xmath10 , also known as the _ fuchsian locus_. the following is the main theorem of this paper .
let @xmath6 be a pants decomposition on @xmath0 and choose a goldman parameterization of @xmath1 that is compatible with @xmath6 .
let @xmath11 be a goldman sequence in @xmath1 . then the following hold : 1 .
the length of the shortest ( in the hilbert metric ) homotopically non - trivial closed curve in @xmath12 that is not homotopic to a multiple of a curve in @xmath6 grows arbitrarily large as @xmath13 approaches @xmath14 .
the topological entropy of @xmath12 converges to @xmath15 as @xmath13 approaches @xmath14 . here , the _
hilbert metric _ is a canonical finsler metric that one can define on any convex @xmath3 surface , and the _ topological entropy _ is a quantification of the complexity " of the geodesic flow on the unit tangent bundle of the convex @xmath3 surface .
these are discussed more carefully in section [ thehilbertmetric ] and section [ entropygeodesicflow ] respectively . a slightly more general version of the above theorem ( allowing for surfaces with boundary ) is stated as theorem [ mainthm1 ] .
this theorem has several interesting implications , which we list here .
the general theme of these corollaries is to highlight the differences between @xmath5 and @xmath1 .
there is a natural action of the mapping class group of @xmath0 , denoted @xmath16 , on any deformation space of @xmath17-structures on @xmath0 by pre - composition with the marking . in the case when this deformation space is @xmath5 , there are two well - known results . the first is known as _ mumford compactness _
@xcite , which states that for any @xmath18 , the quotient of @xmath19 by @xmath16 is compact .
a corollary of ( 1 ) of the main theorem ( via an easy diagonalization argument ) is that there exists a sequence @xmath11 in @xmath1 such that the length of the shortest closed curve in @xmath12 grows arbitrarily large .
since the length of the shortest closed curve is a continuous function on @xmath1 that is invariant under the action of @xmath16 , this implies that in contrast to the case of @xmath5 , mumford compactness fails when the deformation space is @xmath1 . for any @xmath18 ,
the quotient of @xmath20 by @xmath16 is not compact .
this corollary can also be obtained by applying proposition 3.4 of benoist - hulin @xcite to loftin s work in @xcite .
the second is a result by abikoff @xcite , who proved that the action of @xmath16 on @xmath5 extends continuously to the augmented teichmller space @xmath21 , and the quotient @xmath22 is a compactification of @xmath23 . by considering an algebraic construction of @xmath21
, one can define a natural analog of this augmentation for @xmath1 , on which the natural action of @xmath16 on @xmath1 extends continuously .
( canary - storm @xcite constructed such an analog for the deformation space of kleinian surface groups , which can also be done in the convex @xmath3 structures setting . ) a consequence of ( 1 ) of our theorem is that the quotient of this analogous augmentation by @xmath16 is not compact .
also , crampon @xcite proved that the topological entropy of the convex @xmath3 structures in @xmath1 are bounded above by @xmath24 , and that the value @xmath24 is achieved if and only if the structure lies in the fuchsian locus .
later , nie @xcite showed that one can find a sequence in @xmath1 so that the topological entropy converges to @xmath15 .
( in fact , their results hold for convex real projective structures on some manifolds of higher dimensions as well . )
the next corollary of our theorem is then the natural next step in this progression of questions , and gives a positive answer to a question asked by crampon and marquis .
( questions 13 of @xcite . ) for any @xmath25 $ ] , there is a diverging sequence of convex projective structures so that the topological entropy of the structures along this sequence converges to @xmath26 .
finally , our theorem also implies how several other geometric properties , such as the area of the convex @xmath3 surface , degenerate along goldman sequences .
let @xmath11 be a goldman sequence in @xmath1 , then the area of @xmath12 grows to @xmath14 as @xmath13 approaches @xmath14 .
now , we will give a brief description of the strategy to prove the main theorem .
let @xmath27 be any convex @xmath3 structure in @xmath1 , i.e. a marked convex @xmath3 surface .
the key behind the proof is the observation that we can write the hilbert distance between certain pairs of points in @xmath27 in terms of the goldman parameters for @xmath27 .
moreover , the formulas for these distances are relatively simple , so we can understand how they change as we vary the goldman parameters . with this in mind , we find a way to decompose each closed curve in @xmath27 into geodesic segments whose lengths we can bound from below by the distances between these special pairs of points . fortuitously ,
the lower bound that we obtain this way goes to infinity as we deform the convex projective structure along goldman sequences .
this gives us ( 1 ) . to obtain ( 2 )
, we use the fact that the geodesic flow corresponding to the convex @xmath3 structures on @xmath0 is anosov , and that the non - wandering set for this flow is all of @xmath28 .
the work of bowen @xcite then allows us to compute the topological entropy of the geodesic flow by the asymptotic exponential growth rate of orbits .
it turns out that the lower bound in ( 1 ) is strong enough to yield an upper bound on the topological entropy , which vanishes as we deform along goldman sequences .
the rest of this paper is divided into two sections . in section [ preliminaries ]
, we give a brief introduction to the main objects of study in this paper and explain the tools used to prove the main theorem . in section [ maincontent ] , we will give the proof of the main theorem , as well as state and prove a couple of corollaries .
this work has benefitted from conversations with gye - seon lee while he and the author were at the center for quantum geometry of moduli spaces during august of 2013 .
the author also especially wishes to thank richard canary for introducing him to this subject , and the many fruitful discussions they have had in the course of writing this paper .
finally , the author is very grateful for the referee s careful reading of this paper and his / her many helpful comments .
in this subsection , we will define the main objects of study in this paper , and introduce some of their well - known properties . before we do so
, we will make some comments on terminology and notation .
for the rest of this paper , we will only consider orientable surfaces that admit a pants decomposition .
hence , we will use the word surface " to mean compact smooth orientable surface with negative euler characteristic " .
also , we will always use @xmath0 to denote a surface ( without any projective structure ) and @xmath29 to denote a smooth pair of pants . by an oriented curve in @xmath0 ,
we will mean an equivalence class of continuous injective maps @xmath30\to m$ ] ( or @xmath31 if @xmath32 is closed ) , where @xmath32 is equivalent to @xmath33 if they have the same image and @xmath32 is homotopic to @xmath33 .
a ( unoriented ) curve in @xmath0 is then an equivalence class of continuous injective maps @xmath30\to m$ ] ( or @xmath31 if @xmath32 is closed ) , where @xmath32 is equivalent to @xmath33 if they have the same image and @xmath32 is homotopic to either @xmath33 or @xmath33 with its parameterization reversed .
we will denote both oriented and unoriented curves by a choice of a representative .
however , we will also abuse notation by denoting the image of @xmath32 by @xmath32 .
this ambiguity is introduced so as to simplify notation ; moreover , it should be clear from context which @xmath32 we are referring to .
the deformation spaces that we will be considering in this paper are of surfaces modeled locally on @xmath3 , with an additional convexity condition .
first , recall that the group of projective transformations on @xmath3 is the group @xmath34 , which can be identified with @xmath35 by choosing the unique representative in each equivalence class in @xmath36 which has determinant @xmath24 .
[ projectivestructure ] let @xmath0 be a surface ( possible with boundary ) .
an _ @xmath3 structure _
@xmath27 on @xmath0 is a maximal atlas of smooth charts @xmath37 such that 1 .
if @xmath38 is nonempty , then the transition maps @xmath39 are restrictions of projective transformations on @xmath3 to each component of @xmath38 .
2 . @xmath40 maps each connected component of @xmath41 to a line segment in @xmath3 .
the smooth surface @xmath0 equipped with an @xmath3 structure is called an _
@xmath3 surface_. two @xmath3 structures @xmath27 and @xmath42 are _ isotopic _ if there is a diffeomorphism @xmath43 that is isotopic to the identity , so that @xmath44 .
choose an @xmath3 structure @xmath27 on @xmath0 , fix any point @xmath45 and choose @xmath46 so that @xmath47 , where @xmath48 is a universal covering .
given any initial germ @xmath49 of a chart in @xmath27 at the point @xmath50 , we can construct a unique local diffeomorphism @xmath51 and a unique group homomorphism @xmath52 such that the germ of @xmath53 at @xmath54 is @xmath55 and @xmath53 is @xmath56-equivariant . here
, @xmath53 is called the _ developing map _ , @xmath56 is called the _ holonomy representation _ , and the pair @xmath57 is known as the _ developing pair_. if we do not make a choice of the initial germ @xmath49 , then the developing pair @xmath57 is only well - defined up to the action of @xmath35 , where @xmath35 acts on the developing map by post composition and on the holonomy representation by conjugation . furthermore ,
if @xmath27 and @xmath42 are isotopic , then there is some @xmath58 so that @xmath59 , where @xmath60 is conjugation by @xmath61 , and @xmath53 is isotopic to @xmath62 via @xmath56-equivariant diffeomorphisms . for more details ,
refer to sections 2 and 3 of goldman @xcite or section 4 of goldman @xcite . since we will only be considering @xmath3 structures up to isotopy
, our developing pairs will only be well - defined up to @xmath35 action and isotopy of the developing map
. we will henceforth simplify terminology by referring to an isotopy class of @xmath3 structures as an @xmath3 structure .
next , we want to define what it means for an @xmath3 structure on a surface to be convex . before we do so , we need to introduce a couple more definitions .
a subset @xmath63 in @xmath3 is _ properly convex _ if 1 . for any pair of distinct points @xmath64 in @xmath63
, there is a line segment between @xmath50 and @xmath65 that is entirely contained in @xmath63 .
2 . there is a line @xmath66 in @xmath3 that does not intersect the closure @xmath67 of @xmath63 .
a properly convex subset @xmath63 is _ strictly convex _ if @xmath68 does not contain any line segments .
we say that @xmath58 is _ positive hyperbolic _ if @xmath61 is diagonalizable with positive , pairwise distinct eigenvalues .
we will denote by @xmath69 the set of positive hyperbolic elements in @xmath35 . for any @xmath70
, we define its _ attracting fixed point _ and _ repelling fixed point _ to be the points in @xmath3 corresponding to the eigenvectors in @xmath71 of @xmath61 with the largest and smallest eigenvalues respectively .
any line segment in @xmath3 connecting the attracting and repelling fixed points of @xmath61 is called an _ axis _ of @xmath61 . for any positive hyperbolic @xmath58 ,
denote the smallest eigenvalue of @xmath61 by @xmath72 and the sum of the other two eigenvalues by @xmath73 .
then define @xmath74 and observe that this set can also be written as the @xmath7-cell @xmath75 with these definitions , we can state the convexity condition mentioned above . 1 .
an @xmath3 structure @xmath27 is called _ convex _ if @xmath53 is a diffeomorphism onto a properly convex subset of @xmath3 , and every non - identity element in @xmath76 is mapped by @xmath56 into @xmath69 .
the _ deformation space of convex @xmath3 structures _ on @xmath0 , denoted @xmath1 , is the set of ( isotopy classes ) of convex @xmath3 structures on @xmath0 .
3 . a _ convex @xmath3 surface _ is the quotient of a properly convex subset @xmath63 by a group of projective transformations that acts freely , properly discontinuously and cocompactly on @xmath63 .
intuitively , one can think of @xmath1 as the set of essentially different " ways to put a convex @xmath3 structure on @xmath0 , subject to some hyperbolicity " conditions on the holonomy about the boundary components of @xmath0 . in the case when @xmath0 is a closed surface and
@xmath53 is a diffeomorphism onto a properly convex subset of @xmath3 , then the image of @xmath53 is strictly convex with @xmath77 boundary ( kuiper @xcite and benzecri @xcite ; also see theorem 1.1 of benoist @xcite for a more general statement ) , and every non - identity element in @xmath76 is automatically mapped by @xmath56 into @xmath69 .
( theorem 3.2 of goldman @xcite . )
the klein model of hyperbolic space allows one to realize every hyperbolic structure on a surface as a convex projective structure .
this thus induces a natural embedding of teichmller space @xmath5 into @xmath1 , whose image is called the _ fuchsian locus_. for any convex @xmath3 structure @xmath27 , we will denote the image of @xmath53 by @xmath78 . in this case
, @xmath56 is also injective , and @xmath53 descends to a diffeomorphism @xmath79 , which is also known as a _
marking_. since @xmath80 is a convex @xmath3 surface , we can think of a convex @xmath3 structure @xmath27 on @xmath0 as a _ marked convex @xmath3 surface _
, i.e. @xmath81 $ ] , where @xmath80 is a convex @xmath3 surface and @xmath82 is the marking ( which is well - defined up to isotopy ) .
we will switch freely between these two different ways of thinking about @xmath27 .
let @xmath80 be a convex @xmath3 surface and @xmath83 a covering map .
a _ closed line _
@xmath32 in @xmath80 is a closed curve that lifts to a straight line in @xmath84 .
it is easy to check that the following proposition holds , so we omit the proof .
[ properties ] let @xmath81 $ ] be a convex @xmath3 structure on @xmath0 .
then : 1 . for any @xmath85 ,
the attracting and repelling fixed points of @xmath86 lie on @xmath87 .
every homotopically non - trivial closed curve in @xmath80 is freely homotopic to a unique closed line in @xmath80 .
3 . every path in @xmath80 is homotopic ( relative to its end points ) to a unique line segment in @xmath80 . in this subsection
, we will set up notation and establish some properties of a projective invariant which is commonly known as the cross ratio .
[ cross ratio def ] let @xmath88 be four lines in @xmath3 that intersect at a common point @xmath89 , so that no three of the four @xmath90 agree .
choose vectors @xmath91 so that @xmath92 is the projectivization of @xmath93 , and for all @xmath94 , @xmath90 is the projectivization of @xmath95 . then define the _ cross ratio _
@xmath96 where the right hand side is evaluated as a number in the one point compactification @xmath97 of @xmath98 via a choice of linear identification between @xmath99 and @xmath98 .
one can easily verify that the cross ratio depends neither on the choice of @xmath100 , nor on the linear identification between @xmath99 and @xmath98 .
also , the condition that no three of the four @xmath90 agree ensures that if any of the two terms in the numerator on the right hand side evaluate to @xmath15 , then both terms in the denominator do not evaluate to @xmath15 , and vice versa .
hence , the cross ratio is well - defined .
it is also clear from definition that @xmath101 for all @xmath58 , so the cross ratio is a projective invariant .
if @xmath102 are points in @xmath3 that lie in @xmath90 , then we also use the notation @xmath103 when convenient . in the case where the four points @xmath104 lie on a single line , we have the following interpretation of the cross ratio .
[ crossratioline ] let @xmath104 be collinear points in @xmath3 so that no three of the four of them agree , and let @xmath89 be a point that is not collinear with @xmath104 .
let @xmath105 be any affine chart of @xmath3 containing @xmath92 and @xmath106 for @xmath94 .
then @xmath107 where @xmath108 is the euclidean distance between @xmath109 for any choice of euclidean metric on @xmath105 that is compatible with the affine structure .
the proof of the above proposition is another easy computation which we will omit .
in particular , under the hypothesis of proposition [ crossratioline ] , the cross ratio @xmath110 is independent of @xmath92 . in this case
, we will use the notation @xmath111 in place of @xmath110 .
for our purposes , we will mainly be evaluating the cross ratio of points in the boundary of a properly convex domain in @xmath3 .
some properties of the cross ratio used in this setting are listed as the next proposition .
these are simple consequences of proposition [ crossratioline ] , but will be very useful in the proof of our main theorem .
[ crossratioinequality ] let @xmath63 be a properly convex domain in @xmath3 , and let @xmath112 be pairwise distinct points that lie on @xmath113 in that order , so that the lines @xmath90 through @xmath92 and @xmath106 for @xmath94 are pairwise distinct .
also , let @xmath114 and @xmath115 be the closed subintervals of @xmath113 with endpoints @xmath116 , @xmath117 and @xmath118 , @xmath119 respectively , so that @xmath92 does not lie in @xmath114 and @xmath115 .
let @xmath120 and @xmath121 .
( see figure [ convexposition ] . )
then the following hold : 1 .
@xmath122 2 .
3 . @xmath124 , and equality holds if and only if the line through @xmath92 and @xmath50 agrees with the line through @xmath92 and @xmath117 .
4 . @xmath125 , and equality holds if and only if the line through @xmath92 and @xmath65 agrees with the line through @xmath92 and @xmath118 .
5 . @xmath126 , and equality holds if and only if the line through @xmath92 and @xmath50 agrees with the line through @xmath92 and @xmath116 .
@xmath127 , and equality holds if and only if the line through @xmath92 and @xmath65 agrees with the line through @xmath92 and @xmath119 .
.,title="fig : " ] ( -352 , 0 ) ( -473 , 68 ) ( -336 , 85 ) ( -288 , 72 ) ( -369 , 90 ) ( -448 , 80 ) ( -492 , 47 ) using the cross ratio , we can define a metric , called the hilbert metric , on the interior of any properly convex subset of @xmath3 .
let @xmath128 be a properly convex domain .
for any two points @xmath129 , @xmath130 in @xmath63 , let @xmath66 be the line in @xmath3 through @xmath129 and @xmath130 , and let @xmath131 and @xmath132 be the two points where @xmath66 intersects @xmath113 , such that @xmath133 lie on @xmath66 in that order .
the _ hilbert distance _ between @xmath129 and @xmath130 is @xmath134 for any rectifiable path @xmath135 in @xmath63 , we will denote the length of @xmath135 by @xmath136 .
since the hilbert metric is defined using the cross ratio , it is invariant under projective transformations that preserve @xmath63 .
the hilbert metric is a finsler metric , i.e. it is given by a norm @xmath137 on the tangent space at every @xmath138 , which varies smoothly with @xmath50 . to obtain an explicit formula for this norm , choose an affine chart @xmath105 of @xmath3 that contains @xmath139 , and equip @xmath105 with an euclidean metric .
this induces a norm @xmath140 on the tangent space of every @xmath138 . for any tangent vector @xmath141 at a point @xmath138 , let @xmath135 be the oriented line through @xmath50 so that @xmath141 is tangential to @xmath135 at @xmath50 , and let @xmath142 and @xmath143 be the two points where @xmath135 intersects @xmath113 .
then define @xmath144 where @xmath145 , @xmath146 are the euclidean distances between @xmath50 and @xmath142 , @xmath50 and @xmath143 respectively .
one can verify that @xmath147 depends neither on the choice of the affine chart @xmath105 nor on the choice of euclidean metric on @xmath105 , and that this norm gives rise to the hilbert metric on @xmath63 .
the next proposition gives several properties of the hilbert metric .
we will omit the proof as they follow from the properties of the cross ratio discussed in section [ cross ratio ] .
[ hilbertmetricprop ] let @xmath148 be open properly convex domains in @xmath3 such that @xmath149 and @xmath63 is strictly convex .
let @xmath150 be a pair of distinct points in @xmath151 , let @xmath66 be the line in @xmath3 through @xmath129 and @xmath130 , and let @xmath135 be the line segment in @xmath151 between @xmath129 and @xmath130 .
the following hold : 1 .
the line segment @xmath135 is rectifiable , and @xmath152 .
2 . @xmath153 , and equality holds if and only if @xmath154 .
3 . if @xmath32 is a rectifiable path in @xmath63 between @xmath129 and @xmath130 , then @xmath155 and equality holds if and only if @xmath156 .
4 . let @xmath157 and @xmath158 be line segments in @xmath63 with endpoints in @xmath113 . if @xmath113 has regularity @xmath77 , then either @xmath159 or there is a unique pair of points @xmath160 and @xmath161
so that @xmath162 . for any convex @xmath3 surface @xmath163 , we can define the hilbert metric @xmath164 on @xmath63 .
the @xmath35 invariance of the cross ratio implies that this descends to a metric on @xmath80 , also called the hilbert metric , which we denote by @xmath165 .
if @xmath135 is a rectifiable path in @xmath80 , we also denote the length of @xmath135 by @xmath166 . in the case
when @xmath27 is a marked convex @xmath3 surface , we also denote the hilbert metric on @xmath27 by @xmath167 and the length of @xmath135 by @xmath168 . when @xmath0 is a closed surface , the hilbert metrics for convex @xmath3 structures in the fuchsian locus @xmath10 agree with the corresponding hyperbolic metrics .
for the rest of this paper , let @xmath169 denote a closed genus @xmath170 surface with @xmath171 open discs removed such that @xmath172 .
let @xmath173 be a smooth embedding that is @xmath174-injective .
for any @xmath175 , let @xmath176 be a diffeomorphism homotopic to @xmath177 so that it maps the boundary components of @xmath169 to closed lines in @xmath178 equipped with the convex @xmath3 structure @xmath179 . then let @xmath180 be the convex @xmath3 structure on @xmath169 obtained by precomposing the charts in @xmath179 by @xmath181 .
in fact , as a consequence of section 5 of goldman @xcite , we know that every convex @xmath3 structure in @xmath182 can be obtained this way .
thus , @xmath177 induces a surjection @xmath183 . if @xmath184 , then @xmath185 , so we know that @xmath186 by ( 2 ) of proposition [ hilbertmetricprop ] . on the other hand , we have the following proposition , which is an easy consequence of proposition [ hilbertmetricprop ] .
[ uniquegeodesic ] let @xmath187 be any @xmath174-injective embedding , let @xmath188\in{\mathfrak{c}}(m_{g',0})$ ] and let @xmath189:=i^\#({\pazocal{m}}_{g',0})\in{\mathfrak{c}}(m_{g , n})$ ] .
also , let @xmath32 be a closed line in @xmath190 . 1 . @xmath191 .
2 . let @xmath135 be any rectifiable closed curve homotopic to @xmath32 .
then @xmath192 , with equality if and only if @xmath156 .
part ( 2 ) of proposition [ uniquegeodesic ] tells us that even though a compact convex @xmath3 surface @xmath80 equipped with @xmath165 is not a unique geodesic space , the closed curves in @xmath80 have unique length minimizing representatives in their free homotopy classes , namely the closed lines .
thus , from now on , we will refer to the closed lines as closed geodesics .
since we have a hilbert metric on any convex @xmath3 surface @xmath80 , we can define a canonical ( up to scaling ) measure on @xmath80 .
let @xmath80 be a convex @xmath3 surface .
the _ busemann area _
@xmath193 is the 2-dimensional hausdorff measure of the hilbert metric @xmath165 , rescaled so that in the case when @xmath80 is a hyperbolic surface , @xmath193 agrees with the hyperbolic area on @xmath80 . for background on the busemann area
, one can refer to chapter 3 , part 1 of bao @xcite .
the busemann area is a borel measure , and so gives us a notion of area for measurable subsets of @xmath80 .
it lifts to a measure @xmath194 on @xmath195 that is @xmath196-invariant .
if we choose an affine chart @xmath105 of @xmath3 containing @xmath195 and choose an euclidean metric on @xmath105 , then we have the usual lebesgue measure @xmath197 on @xmath105 and hence on @xmath195 .
busemann showed that the busemann area on @xmath195 is absolutely continuous with respect to the lebesgue measure .
in fact , the radon - nikodym derivative of the busemann area with respect to the lebesgue measure at some @xmath198 is @xmath199 , where @xmath200 is the unit ball in the hilbert metric centered at a lift @xmath201 of @xmath50 and @xmath202 is some constant . for more details ,
see busemann @xcite . in his paper @xcite
, goldman gave an explicit parameterization of @xmath182 .
roughly , he did this by first parameterizing the deformation space of convex @xmath3 structures on a pair of pants , and then extending this parameterization to all compact surfaces by specifying how to assemble the pairs of pants together . in this subsection , we will explain how to obtain this parameterization for a pair of pants , and briefly describe how to extend this parameterization to compact surfaces . on a smooth pair of pants @xmath29 ,
choose @xmath203 in @xmath204 corresponding to the three boundary components of @xmath29 , such that @xmath205 .
this choice induces a lamination on every @xmath206 in the following way .
let @xmath207 , @xmath208 , @xmath209 be the repelling fixed points of @xmath210 , @xmath211 , @xmath212 in the gromov boundary @xmath213 of @xmath204 respectively . for any @xmath206 , @xmath214 induces a @xmath215-equivariant injection @xmath216 .
this then gives a lamination of @xmath217 by the @xmath204 orbits of the three lines in @xmath217 between @xmath218 and @xmath219 , @xmath219 and @xmath220 , @xmath220 and @xmath218 .
since the orbits of these three lines are disjoint , the lamination descends to a lamination on @xmath221 with three leaves .
[ idealtriangulation ] the lamination on @xmath221 constructed above is called the _ ideal triangulation _ of @xmath221 corresponding to @xmath210 , @xmath211 , @xmath212 .
goldman s parameterization of @xmath222 is given by the following theorem .
[ goldman ] the deformation space @xmath222 is an open @xmath223-dimensional cell .
furthermore , the map @xmath224 obtained by associating to a convex @xmath3 structure the boundary invariants @xmath225 is a fibration over an open @xmath226-cell with fiber a @xmath7-dimensional open cell
. we will give a summary of parts of goldman s proof as some of these will be used later .
there are two main steps in the proof . in the first step ,
one argues that specifying a marked convex @xmath3 structure on a pair of pants is equivalent to specifying the following data ( see figure [ goldman ] ) , up to equivalence under the action of @xmath35 : 1 .
four closed triangles @xmath227 , @xmath228 , @xmath229 , @xmath230 in @xmath3 such that 1 .
@xmath227 and @xmath231 intersect exactly along an edge for @xmath232 , 2 . for any @xmath233 such that @xmath234 , @xmath231 and @xmath235 intersect exactly at a point , 3 .
@xmath236 is a properly convex hexagon .
2 . three elements @xmath237 such that 1 .
@xmath238 has @xmath239 as its repelling fixed point and @xmath240 , 2 .
@xmath241 has @xmath242 as its repelling fixed point and @xmath243 , 3 .
@xmath202 has @xmath244 as its repelling fixed point and @xmath245 , 4 .
cba = i .
more specifically , we have a bijection @xmath246 where @xmath35 acts on @xmath247 coordinate wise , by the usual left action on the @xmath231 s and conjugation on @xmath248 .
in fact , we can explicitly describe the map @xmath249 ; for any @xmath206 , we have @xmath250 $ ] , where * @xmath238 , @xmath241 , @xmath202 , are the images of @xmath210 , @xmath211 , @xmath212 respectively under @xmath215 , * @xmath227 is the unique triangle whose vertices are the repelling fixed points of @xmath238 , @xmath241 , @xmath202 , which we denote by @xmath251 respectively , and whose interior lies in @xmath217 , * @xmath228 is the unique triangle whose vertices are @xmath129 , @xmath130 , @xmath252 and whose interior lies in @xmath217 , * @xmath229 is the unique triangle whose vertices are @xmath130 , @xmath131 , @xmath253 and whose interior lies in @xmath217 , * @xmath230 is the unique triangle whose vertices are @xmath131 , @xmath129 , @xmath254 and whose interior lies in @xmath217 .
it is easy to see that @xmath255 is in fact a fundamental domain of the action of @xmath204 on @xmath217 .
moreover , the @xmath204 orbits of the edges of @xmath227 give a lamination on @xmath217 , which descends to the ideal triangulation of @xmath221 corresponding to @xmath210 , @xmath211 and @xmath212 .
the first step thus reduces the problem to parameterizing @xmath256 , which is the second step of the proof .
one can parameterize @xmath256 by @xmath257 ( @xmath8 was defined in section [ convexprojectivestructuresonsurfaces ] ) so that the map @xmath258 in the statement of theorem [ goldman ] , when described in this parameterization , is just projection to the first six parameters .
the formal proof that @xmath257 actually parameterizes @xmath256 involves solving a system of equations that one obtains from the data of the configuration of the @xmath231 s along with their interaction with @xmath238 , @xmath241 , @xmath202 . rather than do that , we will simply describe a geometric way to interpret the eight parameters , and refer the reader to section 4 of goldman @xcite for the proof .
( see notation [ hexagonnotation ] ) contained in @xmath217 . ]
any point @xmath250 $ ] in @xmath256 is parameterized by the parameters @xmath259 . here , for all @xmath260
, @xmath72 is the smallest eigenvalue of @xmath61 and @xmath73 is the sum of the other two eigenvalues of @xmath61 .
the first six parameters thus determine the eigenvalue data of the holonomy about each boundary component of the marked projective pair of pants @xmath221 .
in particular , the lengths ( in the hilbert metric @xmath261 ) of the boundary components of @xmath221 can be obtained explicitly from these six parameters . indeed ,
if @xmath26 is the boundary component of @xmath221 corresponding to @xmath260 , one can easily compute that the three eigenvalues for @xmath61 are @xmath262 listed in increasing order , so @xmath263 describing the geometrical significance of the last two parameters @xmath264 and @xmath265 is less straightforward . before we do that
, we shall introduce some notation that will be used in the rest of the paper .
[ hexagonnotation ] for any @xmath250\in{\pazocal{q}}$ ] , let @xmath266 be the hexagon @xmath267 ( this is well - defined up to translation by a projective transformation ) . denote its vertices by @xmath268 in that order , where @xmath269 are the repelling fixed points of @xmath238 , @xmath241 , @xmath202 respectively , and @xmath252 , @xmath253 , @xmath254 .
goldman computed at the end of section 4 of @xcite the following cross ratios in terms of the parameters of his parameterization : @xmath270 notice that if we fix the first six parameters , then these cross ratios depend only on @xmath264 ( and not @xmath265 ) , and are strictly increasing with @xmath264 .
moreover , all three of them converge to @xmath24 as @xmath264 converges to @xmath15 and grow arbitrarily large as @xmath264 converges to @xmath14 .
thus , we can think of @xmath264 as the parameter controlling " these three cross ratios .
observe that if we pick any two sets of four pairwise distinct points @xmath271 and @xmath272 in @xmath3 such that no three of @xmath271 lie on the same line and no three of @xmath272 lie on the same line , then there exists a unique projective transformation @xmath58 such that @xmath273 for all @xmath177 .
this implies that any equivalence class @xmath274\in{\pazocal{q}}$ ] has a representative such that @xmath275^t$ ] , @xmath276^t$ ] , @xmath277^t$ ] and @xmath278^t$ ] . for this representative
, we can then compute that @xmath279^t,\\ e&=&[t:-1:\frac{(e , c , b , f)_a}{2}]^t,\end{aligned}\ ] ] in fact , this is how the @xmath265 parameter in the goldman parameterization is defined .
( see section 4 of goldman @xcite for the computation . )
moreover , the ray with source @xmath131 through @xmath280 and the ray with source @xmath129 through @xmath132 are determined entirely by @xmath264 because @xmath281 and @xmath282 depend only on @xmath264 and the points @xmath283 are fixed .
we can then think of @xmath265 as the parameter that determines where @xmath280 lies along the ray with source @xmath131 through @xmath280 , and this determines where @xmath132 is because we know the cross ratio @xmath284 . by goldman s proof of theorem [ goldman ] we now have an identification between the three spaces @xmath222 , @xmath256 and @xmath257 , so we will blur the distinction between them in the rest of this paper .
the next definition gives names to the parameters described above . in the above coordinate system for @xmath222
, the first six parameters @xmath285 are called the _ boundary invariants _ and the last two parameters @xmath286 are called the _
internal parameters_. goldman showed that every convex projective structure on @xmath169 is obtained by gluing @xmath287 pairs of convex projective pairs of pants along their boundaries , and that there are @xmath7 dimensions worth of ways to glue any two such boundaries together .
in fact , he gives an explicit parameterization of the possible ways to do such a gluing by the parameters @xmath288 .
we will call these parameters the _ twist - bulge parameters_. since these parameters do not feature much in our paper , we will not say more about them . choose a pants decomposition @xmath6 for @xmath169 , i.e. a system of @xmath289 pairwise non - intersecting , homotopically non - trivial , simple closed curves in @xmath169 .
this system of curves decomposes @xmath169 into @xmath287 pairs of pants .
hence , to parameterize @xmath182 , we need @xmath289 pairs of boundary invariants @xmath290 ( one pair for each simple closed curve in @xmath6 ) , @xmath291 pairs of twist - bulge parameters @xmath292 ( one pair for each simple closed curve in @xmath6 that is not a boundary component ) and @xmath287 pairs of internal parameters @xmath293 ( one pair for each pair of pants ) . this implies that @xmath182 is a @xmath294-dimensional cell .
next , we will describe an order @xmath295 rotational symmetry of the hexagon @xmath266 for any @xmath221 in @xmath222 ( see notation [ hexagonnotation ] ) .
since the goldman parameters do not behave very well under this rotational symmetry , we will also give a slightly different parameterization of the @xmath7-dimensional open cell fiber in theorem [ goldman ] in order to exploit this symmetry to simplify the proof of our result .
consider any properly convex hexagon @xmath296 in @xmath3 with vertices @xmath268 in that order .
( see figure [ goldman ] . ) for any @xmath297 such that @xmath298 , let @xmath299 , where 1 .
@xmath300 , 2 .
@xmath301 lie along @xmath302 in that order .
there are thirty such cross ratios .
also , for any @xmath206 , define @xmath303 .
we will be using some of these thirty cross ratios to give lower bounds for the lengths of closed curves .
the main reason we chose these thirty cross ratios is that they have relatively simple closed form expressions in terms of the goldman coordinates ( and in the new coordinates as well , which we will see later ) . for any
@xmath304 define the real valued functions @xmath305 @xmath306 @xmath307 with domain @xmath308 .
note that for any @xmath309 , each @xmath310 is strictly increasing with image @xmath311 .
moreover , for any @xmath312 , we have that @xmath313 , @xmath314 , @xmath315 by ( [ importantcrossratio ] ) .
this implies the following easy consequence , which we record as a lemma .
[ symmetrys ] for any @xmath312 , we have @xmath316 this gives a coordinate free and symmetric description of the parameter @xmath264 .
( the sense in which this description is symmetric will be justified later . )
unfortunately , we do not have such a symmetric description for @xmath265 , so we need to replace @xmath265 with a new parameter .
this motivates the next lemma .
[ symmetryr ] for any @xmath206 , we have @xmath317 let @xmath312 .
we can compute ( see section [ crossratiocomputationsandformulas ] ) that @xmath318 this , together with lemma [ symmetrys ] , proves the lemma . taking the three equal expressions in lemma [ symmetryr ] as our new eighth parameter , which we denote as @xmath319
, we get our reparameterization of @xmath222 .
[ newparameterization ] the map @xmath320 given by @xmath321 , where @xmath322 is a diffeomorphism .
the inverse map is given by @xmath323 for the rest of the paper , the parameterization we use for @xmath324 will be the one where the last coordinate is @xmath319 given in proposition [ newparameterization ] .
next , we will carefully describe the symmetric property of @xmath264 and @xmath319 that we mentioned above .
there is a natural @xmath325 action on @xmath256 which cyclically permutes @xmath326 and @xmath248 , i.e. if @xmath327 , then @xmath328 = [ \delta_0,\delta_2,\delta_3,\delta_1 , b , c , a].\ ] ] we can interpret this action in the following way . consider the marked convex @xmath3 structure @xmath221 and the hexagon @xmath329 .
the marking endows the vertices of @xmath266 with a labeling as described in notation [ hexagonnotation ] .
then the action of @xmath170 is simply a cyclic relabeling of the vertices of @xmath330 .
( see figure [ relabel ] . )
the next proposition computes this action in terms of our parameterization .
\mapsto [ \delta_0,\delta_2,\delta_3,\delta_1 , b , c , a].$ ] ] [ symmetrysr ] let @xmath331 be such that @xmath328 = [ \delta_0,\delta_2,\delta_3,\delta_1 , b , c , a].\ ] ] then in the new parameterization of @xmath256 , we have @xmath332 since @xmath170 sends @xmath238 to @xmath241 , @xmath241 to @xmath202 and @xmath202 to @xmath238 , it is clear that @xmath333 for some @xmath334 .
first , we will show that @xmath335 .
let @xmath336=(r , s , r),\ ] ] @xmath337=(r',s',r'),\ ] ] and note that @xmath338 .
this means that @xmath339 , so @xmath335 because @xmath340 as functions , and they are both injective .
next , we show that @xmath341 .
observe : @xmath342 @xmath343 thus , @xmath344 this shows that with our choice of parameterization of the @xmath7-cell fibers of @xmath258 in theorem [ goldman ] , @xmath325 acts as the identity on the the fibers of @xmath258 , and only permutes the boundary invariants of @xmath222 .
it is in this sense that the parameters @xmath264 and @xmath319 are symmetric . in this subsection
, we will give a brief description of some dynamics that naturally occurs in our set up .
suppose first that @xmath80 is a closed convex @xmath3 surface . since the hilbert metric is a finsler metric , @xmath80 induces a geodesic flow @xmath345 on @xmath346 , the unit tangent bundle of @xmath80 .
we will now define the topological entropy of this flow .
let @xmath347 be a flow on a compact manifold @xmath61 .
choose a metric @xmath132 on @xmath61 , and for each @xmath348 , define @xmath349 by @xmath350 one can verify that @xmath351 is in fact a metric on @xmath61 . for
any @xmath18 and any @xmath352 , consider the set of all open covers of @xmath61 satisfying the following property : every open set in the open cover has a @xmath351 diameter of at most @xmath353 . then define @xmath354 to be the size of such an open cover with the fewest number of open sets .
the _ topological entropy _ of the flow @xmath347 is the quantity @xmath355 it is known that the topological entropy is in fact independent of the choice of the metric @xmath132 , and thus depends only on the flow and the topology of @xmath61 . for more details , one may refer to chapter 3.1 of hasselblatt - katok @xcite . in the case when @xmath356 and @xmath347 is the geodesic flow of the hilbert metric on @xmath80 ,
this quantity is interesting because we can think of it as a measure of how different @xmath80 is from a hyperbolic surface . by crampon @xcite , we know that in the case when @xmath0 is a closed surface , @xmath357 $ ] for any @xmath81\in{\mathfrak{c}}(m)$ ] , and @xmath358 if and only if @xmath27 lies in the fuchsian locus of @xmath1 . thus , if we have a sequence in @xmath1 on which the topological entropy converges to @xmath15 , then the dynamics of the geodesic flow is becoming less and less like that of a hyperbolic surface as we move along this sequence . by theorem 1.1 of benoist @xcite , we know that @xmath347 is anosov
. moreover , the topological transitivity of the action of @xmath76 on the set of pairs of distinct points on @xmath87 implies that the periodic points for @xmath347 are dense in @xmath346 .
theorem b of bowen @xcite then allows us to compute the topological entropy of @xmath347 by the formula @xmath359 where @xmath360 is the number of closed orbits of @xmath347 with period at most @xmath265 . in the more general case
when @xmath163 possibly has boundary , we use the following generalization of the geodesic flow of the hilbert metric . define @xmath361 to be the set of points @xmath362 in @xmath346 such that the geodesic through @xmath363 tangential to @xmath141 has endpoints in the limit set of @xmath364 in @xmath68 .
this is also known as the non - wandering set of the geodesic flow .
it is easy to see that @xmath361 is compact . for any compact projective surface @xmath80 ,
define the _ geodesic flow _ of @xmath80 to be the geodesic flow @xmath347 of the hilbert metric on @xmath80 restricted to the subset @xmath361 of @xmath346 .
denote the topological entropy of the geodesic flow of @xmath80 by @xmath365 . for any convex @xmath3 structure
@xmath81 $ ] on @xmath0 , we define @xmath366 . when @xmath80 is closed , @xmath367 , so @xmath368 .
moreover , even in the case when @xmath80 is not closed , @xmath361 is a hyperbolic set for the geodesic flow @xmath347 of @xmath80 . to see this ,
consider two copies of @xmath80 and the obvious pairing of the boundary components of these two copies .
choose twist - bulge parameters for each of these pair of boundary components to glue the two copies of @xmath80 together .
let @xmath369 be the closed convex @xmath3 surface obtained this way and let @xmath49 be the geodesic flow of @xmath369 ( acting on @xmath370 ) . then note that @xmath361 is a @xmath49-invariant subset of @xmath370 , and @xmath371 is exactly @xmath347 .
the hyperbolicity of @xmath361 then follows immediately from the anosovness of @xmath49 .
this allows us to use a result by pollicott ( see theorem 8 of @xcite ) to compute the geodesic flow of @xmath80 by the formula @xmath372 where @xmath373 is the number of closed geodesics in @xmath80 with length at most @xmath265 .
for the rest of the paper , we will use the set up we now describe .
choose once and for all the following : 1 .
a pants decomposition @xmath374 for @xmath375 , 2 . a set of generators @xmath210 , @xmath211 and @xmath212 for @xmath204 such that @xmath205 , 3 .
a diffeomorphism @xmath376 , where @xmath377 are the closures of the connected components of @xmath378 the description of the goldman parameterization for a pair of pants in section [ goldmanparameters ] tells us that this data gives us a parameterization of @xmath1 .
we will now list a couple of definitions to simplify the statement of our theorem .
a closed curve @xmath32 in @xmath0 is _ typical _ if @xmath32 is not homotopic to a multiple of any @xmath379 .
the set of homotopy classes of typical oriented closed curves in @xmath0 is denoted by @xmath380 .
[ goldmandef ] a sequence @xmath381 in @xmath222 is a _ goldman sequence _ if 1 . there are constants @xmath382 and @xmath383 , @xmath384 , such that for all @xmath13 , the lengths of the boundary components of @xmath385 are bounded between @xmath382 and @xmath383 .
2 . for any compact set @xmath386 ,
there is a positive number @xmath296 such that if @xmath387 , then @xmath388 .
a sequence @xmath11 in @xmath1 is a _ goldman sequence _ if for all @xmath389 , the sequence @xmath390 ( see section [ thehilbertmetric ] ) in @xmath222 is a goldman sequence .
let @xmath391 be the function that sends every @xmath392 to the length of the shortest typical closed curve in @xmath27 .
the main theorem of this paper is the following .
[ mainthm1 ] let @xmath393 be a goldman sequence in @xmath1
. then 1 .
@xmath394 2 .
@xmath395 before we begin the proof of this theorem , we will state and prove some of its corollaries .
the first of these is a generalization of ( 1 ) of theorem [ mainthm1 ] .
[ firstcor ] let @xmath11 be a sequence in @xmath1 such that @xmath396 in @xmath222 is a goldman sequence .
let @xmath32 be a closed curve in @xmath0 that can not be homotoped to be disjoint from @xmath397 and let @xmath398 be the geodesic representative of @xmath32 in @xmath12
. then @xmath399 .
since @xmath32 can not be homotoped to be disjoint from @xmath397 , we know that @xmath32 is typical and one of the following must hold : * for all @xmath13 , @xmath398 is contained in @xmath400 , * for all @xmath13 , there is a closed subsegment @xmath401 of @xmath398 such that @xmath402 , the endpoints of @xmath401 lie in @xmath403 , and @xmath401 is not homotopic relative endpoints to a subset of @xmath403 .
if the former holds then ( 1 ) of proposition [ uniquegeodesic ] implies that @xmath404 . by ( 1 ) of theorem [ mainthm1 ] , @xmath405 .
suppose instead that the latter holds .
let @xmath406 and @xmath407 be the two boundary components of @xmath385 that contain the endpoints of @xmath401 ( possibly @xmath408 ) and let the endpoints of @xmath401 in @xmath406 and @xmath407 be @xmath409 and @xmath410 respectively .
parameterize @xmath401 , @xmath406 and @xmath407 by the unit interval @xmath411 $ ] so that @xmath412 , @xmath413 , @xmath414 , @xmath415 , and observe that either @xmath416 or @xmath417 is a typical closed curve in @xmath385 .
here , @xmath418 is concatenation and the inverse is reversing the parameterization .
assume without loss of generality that @xmath419 is typical .
then @xmath420 i.e. @xmath421 thus ( 1 ) of theorem [ mainthm1 ] implies that @xmath422 as another consequence of ( 1 ) of theorem [ mainthm1 ] , we can also say how the maximum injectivity radius and the busemann area of a convex @xmath3 structure on a closed surface degenerate along a goldman sequence .
these results are listed as corollary [ finalcor ] . to prove these results , we need the following lemma . [ growingball ]
let @xmath80 be a closed convex @xmath3 surface and let @xmath363 be a point in @xmath80 where the injectivity radius is maximized .
let @xmath423 be the injectivity radius at @xmath363 .
then there are pairwise distinct line segments @xmath424 , @xmath425 , @xmath426 , @xmath427 in @xmath80 of length @xmath423 such that 1 .
@xmath424 and @xmath425 have the same endpoints 2 .
@xmath426 and @xmath427 have the same endpoints 3 .
@xmath363 is a common endpoint for all four segments 4 . the closed curves @xmath428 and @xmath429 are not homotopic relative to @xmath363 , and neither of them are homotopically trivial . for any point @xmath430 ,
let @xmath431 be the injectivity radius at @xmath432 and let @xmath433 be any lift of @xmath432 in @xmath63 .
observe that there is some @xmath434 such that @xmath435 also , for all @xmath436 , we know that @xmath437
. note that @xmath363 exists because @xmath80 is closed .
choose a lift @xmath438 of @xmath363 in @xmath63 , and suppose for contradiction that up to taking inverses , there is a unique @xmath439 such that @xmath440 .
this is an open condition , so there is a neighborhood @xmath105 of @xmath363 with the following property : for all @xmath441 and up to taking inverses , @xmath439 is the unique element such that @xmath442 .
let @xmath443 , @xmath444 and @xmath445 be the three fixed points of @xmath446 in @xmath3 , where @xmath443 is the attracting fixed point and @xmath444 is the repelling fixed point . by renormalizing , we can assume that @xmath447 , g_1 ^ 0=\left [ \begin{array}{c } 0\\ 1\\ 0 \end{array } \right ] , g_1 ^ -=\left [ \begin{array}{c } 0\\ 0\\ 1 \end{array } \right ] , { \widetilde{p}}=\left [ \begin{array}{c } 1\\ 1\\ 1 \end{array } \right].\ ] ] this implies that @xmath448\ ] ] for some @xmath449 , @xmath197 , @xmath450 such that @xmath451 , @xmath452 . define @xmath453 to be the closed triangle @xmath454:x , y , z\geq 0\bigg\},\ ] ] choose @xmath455 such that @xmath456\ ] ] is a lift of some @xmath441 , and note that @xmath438 and @xmath457 both lie in @xmath453 .
let @xmath66 be the line in @xmath3 that contains @xmath438 and @xmath458 , and let @xmath459 to be the line that contains @xmath457 and @xmath460 .
then define @xmath461 and @xmath462 to be the line segments of @xmath66 and @xmath459 that lie in @xmath63 and have endpoints in @xmath68 .
one can easily compute that @xmath463:a , b\in{\mathbb{r}}\bigg\ } , l'=\bigg\{\left[\begin{array}{c } a+b\lambda\\ ( a+b\mu)\beta\\ a+b\nu \end{array } \right]:a , b\in{\mathbb{r}}\bigg\}.\ ] ] it is then easy to see that @xmath66 and @xmath459 intersect at the point @xmath464.\ ] ] since @xmath465 and @xmath466 , this point does not lie in @xmath453 ( and hence not in @xmath63 ) , so @xmath461 and @xmath462 do not intersect . by applying proposition [ crossratioline ] and proposition [ crossratioinequality ] ,
it is clear that @xmath467 ( see figure [ enlargingrp ] ) .
however , this contradicts the maximality of the injectivity radius at @xmath363 . ]
thus , there are at least two group elements @xmath446 , @xmath468 in @xmath196 such that @xmath469 and @xmath470 let @xmath471 , @xmath472 , @xmath473 , @xmath474 be the midpoints of the line segments in @xmath63 between @xmath438 and @xmath458 , @xmath475 , @xmath476 , @xmath477 respectively
. then for @xmath478 , let @xmath479 , @xmath480 be the line segments in @xmath63 between @xmath438 and @xmath481 , @xmath482 respectively . define @xmath483 , @xmath484 and @xmath485 for @xmath478 , where @xmath486 is the covering map .
it is clear that @xmath424 , @xmath425 , @xmath426 , @xmath427 are pairwise distinct and that ( 1 ) , ( 2 ) and ( 3 ) hold . to get ( 4 ) , simply note that the closed curves @xmath428 and @xmath429 correspond to @xmath446 and @xmath468 in @xmath196 respectively . since @xmath487 and @xmath488 , the two closed curves can not be homotopic relative to @xmath363 , and neither of them are homotopically trivial . in the proof of corollary [ finalcor ] , we will also make use of the following well - known result by benzcri @xcite .
[ benzecri ] ( benzcri ) define @xmath489 and equip @xmath490 with the hausdorff topology .
then the natural action of @xmath491 on @xmath490 is cocompact .
[ finalcor ] let @xmath0 be a closed surface .
let @xmath492 be the function that maps each @xmath392 to the maximal injectivity radius over all points in @xmath27 , and let @xmath493 be the function that maps each @xmath392 to the busemann area of @xmath27 . if @xmath11 is a goldman sequence in @xmath1 , then 1 .
@xmath494 2 .
@xmath495 .
proof of ( 1 ) .
let @xmath27 be any marked closed convex @xmath3 surface and let @xmath363 , @xmath496 and @xmath497 for @xmath478 be as defined in lemma [ growingball ] .
let @xmath498\to l_i\cup l_i'$ ] be a parameterization such that @xmath499 .
since @xmath428 and @xmath429 are not homotopic relative @xmath363 and neither of them are homotopically trivial , either @xmath500 or @xmath501 is typical , so @xmath502 .
since this is true for all @xmath392 , ( 1 ) of theorem [ mainthm1 ] implies that @xmath494 .
proof of ( 2 ) .
let @xmath409 be a point in @xmath12 where the injectivity radius is maximized .
it is sufficient to show that @xmath503 .
choose a lift @xmath504 of @xmath409 in @xmath505 .
by theorem [ benzecri ] , we can assume that the sequence @xmath506 converges to some @xmath507 in @xmath490 . by ( 1 )
, we know that @xmath508 . hence , the sequence @xmath509 also converges to @xmath507 , so @xmath510 the next corollary gives a positive answer to a question posted by crampon and marquis ( question 13 of @xcite ) .
let @xmath0 be a closed surface .
they asked if there is , for any number @xmath25 $ ] , a diverging sequence @xmath11 in @xmath1 such that @xmath511 .
previously , nie proved in @xcite that there exists a diverging sequence @xmath11 in @xmath1 such that @xmath395 .
[ cramponquestion ] suppose that @xmath0 is a closed surface .
for any number @xmath25 $ ] , there is a diverging sequence @xmath11 in @xmath1 such that @xmath511 .
consider any diverging sequence @xmath512 in @xmath10 corresponding to pinching the curves in the pants decomposition of @xmath0 , i.e. the largest and smallest eigenvalues of the holonomy about the closed curves in the pants decomposition converge to @xmath24 as @xmath13 approaches @xmath14 .
let @xmath513 be the internal parameters for @xmath514 . by section 4 of goldman @xcite , we know that @xmath515 for all @xmath389 and for all @xmath13 .
now , for each @xmath13 , consider the one parameter family @xmath516 in @xmath1 , where all the boundary invariants and twist - bulge parameters for @xmath517 are the same as that of @xmath518 , but the internal parameters of each @xmath519 is @xmath520 .
observe that @xmath521 .
also , theorem [ mainthm1 ] implies that @xmath522 . since @xmath523 is continuous ( see section 3 of sambarino @xcite or proposition 8.3 of crampon @xcite ) , we know that there is some @xmath524 such that @xmath525 .
let @xmath526 .
then the sequence @xmath11 is diverging because the curves in the pants decomposition of @xmath0 are getting pinched , and @xmath511 .
the rest of this paper will be devoted to proving theorem [ mainthm1 ] .
we want to find a way to decompose every closed geodesic on a marked convex @xmath3 surface into line segments , so that we can bound the length of each line segment from below by a number that depends on the goldman parameters . we will give a rough description of the idea behind this decomposition before formally describing it .
let @xmath392 be a marked convex @xmath3 surface .
since we have chosen a pants decomposition @xmath6 on @xmath0 , the marking induces a pants decomposition on @xmath27 , which we also denote by @xmath6 .
any closed geodesic @xmath32 on @xmath27 can be thought of as a union of two types of line segments .
the first type are line segments that wind around " collar neighborhoods of the simple closed curves in @xmath6 ( see figure [ winding ] ) and the second type are line segments that go between " these collar neighborhoods ( see figure [ gobetween ] ) .
the length of the first type of line segments can be bounded below by a multiple ( depending on the number of times they
wind around " the collar neighborhood ) of the infimum of the lengths of the simple closed curves in @xmath6 . on the other hand
, we will later show that the length of the second type of line segments can be bounded below by some combination of the logarithm of the thirty cross ratios described in section [ areparameterization ] .
to formally describe how to decompose a closed geodesic into these types of line segments , it is convenient to use an ideal triangulation @xmath527 , which we will now define . for each @xmath528 , let @xmath529 ( see section [ thehilbertmetric ] ) .
we can then construct the ideal triangulation @xmath530 on @xmath531 corresponding to @xmath210 , @xmath211 , @xmath212 for each @xmath389 .
( see definition [ idealtriangulation ] . ) , which lifts to a lamination @xmath532 on @xmath533 .
( this is partially drawn in figure [ goldman ] . )
define the lamination @xmath534 on @xmath27 , which lifts to a lamination @xmath535 on @xmath78 .
we call every element in @xmath535 or @xmath527 an _ edge _ of @xmath535 or @xmath527 respectively .
also , we say two edges in @xmath527 _ share a common vertex _ if they have lifts to @xmath535 that share a common vertex . before we proceed to describe the decomposition of closed geodesics into line segments , we will first take a closer look at the ideal triangulation and some related structure .
observe that the three line segments in @xmath536 with endpoints @xmath131 and @xmath129 , @xmath129 and @xmath130 , @xmath130 and @xmath131 ( see figure [ goldman ] ) are lifts of the three leaves of @xmath531 .
moreover , for each boundary component @xmath135 of @xmath531 , there are exactly two leaves in @xmath530 that accumulate to @xmath135 . using this
, we can construct the following : * ( @xmath537 and @xmath538 ) let @xmath432 be any endpoint of any leaf in @xmath532 , and define @xmath537 to be the stabilizer of @xmath432 in @xmath76
. let @xmath539 and @xmath540 be two open triangles in @xmath541 that share an edge and have @xmath432 as a common vertex , and let @xmath235 be the closure of @xmath542 in @xmath533 for @xmath543 . then define @xmath538 to be the interior of @xmath544 .
( see figure [ constructions ] . )
+ , @xmath545 , @xmath546 , @xmath547 , @xmath548 . ]
* ( @xmath549 , @xmath550 and @xmath551 ) choose @xmath552 that is not a boundary component of @xmath27 , and let @xmath553 be the two pairs of pants given by @xmath6 that contain @xmath554
. in @xmath531 , exactly one of the three leaves of @xmath530 , call it @xmath555 , does not accumulate to @xmath554 .
similarly , let @xmath556 be the unique leaf of @xmath557 that does not accumulate to @xmath554 .
then , among the open line segments in @xmath558 that intersect @xmath554 and have one endpoint in @xmath555 and one endpoint in @xmath556 , choose a length minimizing one and call it @xmath549 ( see figure [ dk ] ) .
let @xmath546 be a lift of @xmath549 to @xmath78 , and it is clear that the endpoints of @xmath546 lie on lifts @xmath559 and @xmath560 of @xmath555 and @xmath556 respectively .
let @xmath561 and @xmath562 be the line segments in @xmath563 joining an endpoint of @xmath559 to an endpoint of @xmath560 such that @xmath564 .
( see figure [ constructions ] . )
then let @xmath565 and @xmath566 , where @xmath567 is the covering map .
( see figure [ dkdk ] . ) * ( @xmath568 and @xmath545 ) choose @xmath552 that is not a boundary component of @xmath27 and let @xmath569 be a lift of @xmath554 to @xmath78 .
let @xmath568 be the stabilizer in @xmath76 of @xmath569 , and let @xmath555 , @xmath556 be the lines in @xmath527 that contain the endpoints of @xmath549 .
choose a lift @xmath546 of @xmath549 that intersects @xmath569 and let @xmath559 , @xmath560 be the lifts of @xmath555 , @xmath556 that contain the endpoints of @xmath546 .
then define @xmath545 ( see figure [ constructions ] ) to be the open convex subset of @xmath78 bounded by the @xmath568 translates of @xmath559 and @xmath560 . .
] and @xmath551 . ]
[ thetwogammas ] let @xmath531 and @xmath570 be the two pairs of pants given by @xmath6 that contain @xmath554 , and choose a lift @xmath569 of @xmath554 in @xmath78 . by conjugating
, we can assume that @xmath569 is the axis of both @xmath571 and @xmath572 for some @xmath573 .
denote by @xmath574 and @xmath575 the repelling fixed points of @xmath571 and @xmath572 respectively . then @xmath576 and @xmath577 .
[ dkdkdk ] observe that @xmath545 contains @xmath569 and all the @xmath568 translates of @xmath549 , @xmath550 and @xmath551 .
moreover , if @xmath61 is a generator of @xmath568 , then for any @xmath578 , the subset of @xmath545 that is bounded between @xmath579 and @xmath580 ( resp . @xmath581 and @xmath582 ; @xmath583 and @xmath584 ) contains exactly one @xmath568 translate of @xmath585 and one @xmath568 translate of @xmath586 ( resp . @xmath549 and @xmath586 ; @xmath549 and @xmath585 ) . [ gammaonomega ] observe that @xmath255 is a fundamental domain of the action of @xmath587 on @xmath533 . in particular , for any @xmath588 , we have @xmath589
. also we can easily describe the action of @xmath537 on @xmath538 .
let @xmath590 and @xmath591 be the two triangles adjacent to @xmath592 that have @xmath432 as a vertex and let @xmath593 and @xmath594 be the other triangles that have @xmath432 as a vertex and are adjacent to @xmath590 and @xmath591 respectively . if @xmath595 , @xmath596 are the two possible generators of @xmath537 , then @xmath597 .
next , we will describe how to decompose any typical oriented closed geodesic @xmath32 in @xmath27 using the ideal triangulation @xmath527 .
choose a parameterization for @xmath32 and define the following .
( here , we abuse notation by denoting the image of a curve by the curve itself . )
* let @xmath598 .
these are the points in @xmath599 that are mapped via @xmath32 to points of intersection of the curve @xmath32 with @xmath527 .
* for any @xmath600 , define @xmath601 to be the unique line in @xmath527 that contains @xmath602 . also , define @xmath603 to be the line @xmath601 equipped with the orientation so that @xmath32 passes from the left to the right of @xmath603 at @xmath432 .
* let @xmath604 .
these are the points in @xmath599 that are mapped via @xmath32 to points where @xmath32 intersects the curves in @xmath6 .
the orientation of @xmath32 induces a cyclic order on @xmath605 .
note that the only elements in @xmath605 that do not have a successor or a predecessor in this cyclic order are the points in @xmath606 .
thus , we can also define the following .
* let @xmath607 be the bijection which takes any point in @xmath608 to its successor .
* define @xmath609 of @xmath599 , where @xmath610 the compactness of @xmath32 implies that @xmath611 is a finite set .
also , the cyclic order on @xmath612 induces a cyclic order on @xmath611 . for any pair of distinct points @xmath613 in @xmath599 , define @xmath614 ( resp .
[ p , q ] ) to be the open ( resp
. closed ) subinterval of @xmath599 in the clockwise direction from @xmath363 to @xmath432 .
is a crossing point and @xmath26 is a crossing segment , title="fig : " ] ( -377 , 102 ) ( -360 , 110 ) ( -590 , 78 ) ( -603 , 67 ) ( -467 , 52 ) a point @xmath432 in @xmath611 is called a _ crossing point _ if the three edges @xmath615 , @xmath601 and @xmath616 do not share any common vertices .
the closed subsegment @xmath617}$ ] is called a _ crossing segment _ corresponding to @xmath601 ( see figure [ crossing segment ] ) , and the triple @xmath618 is called a _ crossing triple _ for @xmath27 .
one should think of the crossing segments as line segments that go between " collar neighborhoods of the simple closed curves in @xmath6 .
since @xmath6 decomposes @xmath27 into @xmath287 pairs of pants , and each @xmath619 has twelve crossing triples , we have the following lemma .
[ crossingtriple ] let @xmath0 be a closed surface of genus @xmath170 with @xmath171 disjoint open discs removed .
then any @xmath392 has @xmath620 crossing triples .
since the crossing points for a fixed parameterized closed curve @xmath32 lie in @xmath599 , they have a natural cyclic order .
so , by choosing a crossing point @xmath621 of @xmath32 , we can enumerate the other crossing points of @xmath32 according to the cyclic order .
let @xmath622 be the set of crossing points of @xmath32 enumerated as described .
the pair @xmath623 is called a _ pants changing pair _ if for all @xmath624 , @xmath625 .
the _ pants changing segment _ corresponding to @xmath623 is the closed subsegment @xmath626}$ ] .
( see figure [ pants changing segment ] . )
the pair @xmath627 is called a _ looping pair _ if there is some @xmath624 such that @xmath628 .
the _ looping segment _ corresponding to @xmath627 is the closed subsegment @xmath626}$ ] .
( see figure [ looping segment ] . )
one should think of the pants changing segments and looping segments as line segments that wind around " the collar neighborhoods of the simple closed curves in @xmath6 .
the pants changing segments wind around while moving between pairs of pants , while the looping segments wind around while staying in the same pair of pants .
is a pants changing pair , @xmath629 is a pants changing segment , title="fig : " ] ( -371 , 45 ) ( -322 , 117 ) ( -351 , 100 ) ( -617 , 89 ) ( -517 , 90 ) ( -605 , 77 ) ( -460 , 77 ) is a looping pair , @xmath629 is a looping segment , title="fig : " ] ( -420 , 87 ) ( -337 , 97 ) ( -340 , 60 ) ( -648 , 58 ) ( -520 , 58 ) ( -550 , 56 ) ( -468 , 49 ) on @xmath599 , one can also visualize the pre - images ( under @xmath32 ) of the crossing segments as neighborhoods of the crossing points , while the pre - images of the pants changing segments and looping segments contain the intervals between subsequent pairs of crossing points .
note that even though we chose a parameterization of @xmath32 to make the above definitions , the pants changing , looping , crossing segments and the crossing triples are in fact independent of the choice of parameterization .
if @xmath629 is a pants changing segment , then @xmath629 intersects exactly one @xmath554 in @xmath6 .
define @xmath630 for any @xmath631 , let @xmath632 be the line in @xmath633 that contains @xmath602 and let @xmath634 be the line @xmath632 equipped with the orientation so that @xmath32 ( equipped with its orientation ) passes from the left to the right of @xmath634 at @xmath432 .
the natural order on @xmath635 induced by the orientation of @xmath32 allows us to define the finite sequence @xmath636 .
if @xmath629 is a looping segment , define @xmath637 note that the edges in @xmath638 share a common vertex , or equivalently , accumulate to a common boundary component @xmath135 of @xmath639 .
moreover , the @xmath640 are oriented so that either all of them point towards @xmath135 or all of them point away from @xmath135 .
since @xmath635 has a natural order induced by the orientation of @xmath32 , we can define the finite sequence @xmath641 . in either case
, @xmath642 is designed to keep track of the amount of twisting " @xmath32 does while moving between crossing points .
let @xmath643 be the finite sequence that is obtained from @xmath642 by forgetting the orientation of the terms of @xmath642 .
again , observe @xmath642 and @xmath643 are independent of the choice of parameterization for @xmath32 .
the decomposition described in section [ setup ] begs the following question .
if we are given the cyclic sequence of crossing triples for some @xmath644 ( recall that @xmath645 is the set of typical closed geodesics in @xmath27 ) , together with the looping or pants changing segments between every pair of crossing points for @xmath32 , can we recover @xmath32 ?
the answer to this question , as we will see later , is yes .
however , the way this question is currently posed is a little awkward because while the cyclic sequence of crossing triples for @xmath32 is a combinatorial object ( there are finitely many possibilities for each entry of this cyclic sequence by lemma [ crossingtriple ] ) , the cyclic sequence of looping or pants changing segments between every pair of crossing points for @xmath32 is not .
hence , we want to replace the latter with something more combinatorial in nature .
it turns out that there are two rather natural ways to do so .
these give two different combinatorial descriptions of @xmath32 , which we call @xmath646 and @xmath647 respectively .
the goal of this subsection is thus to formally define @xmath646 and @xmath647 , and answer the above question about @xmath646 and @xmath647 , i.e. can we recover @xmath32 from @xmath646 or @xmath647 , and if not , how much information do we lose by describing @xmath32 using @xmath646 or @xmath647 ?
we shall start with @xmath646 .
let @xmath644 and let @xmath648 be the cyclic sequence of crossing points along @xmath32 .
also , let @xmath649 be the looping or pants changing segment associated to the pair @xmath623 and let @xmath650 . then define _
@xmath646 _ be the cyclic sequence @xmath651 , where each @xmath652 is the tuple @xmath653 , the next proposition gives a positive answer to the question asked above , i.e. we can recover @xmath32 from @xmath646 .
[ phietaetaequivalence ] if @xmath654 are such that @xmath655 , then @xmath656 .
before we begin the formal proof which is rather technical , let us first see why this proposition is morally true . if we look in @xmath78 , the condition that @xmath655 should imply that the lifts of @xmath32 and @xmath33 are two lines that pass through the same triangles of the ideal triangulation @xmath535 .
this means that their endpoints in @xmath87 are the same , so they have to be equal .
the general strategy is to show that if @xmath655 , then @xmath32 is homotopic to @xmath33 .
this allows us to conclude that @xmath656 by ( 2 ) of proposition [ properties ] .
enumerate the crossing points of @xmath32 and @xmath33 by @xmath648 and @xmath657 respectively , so that if @xmath658 and @xmath659 are the induced enumeration of the terms in @xmath646 and @xmath660 respectively , then @xmath661 for all @xmath662 .
let @xmath663 and @xmath664 be the crossing segments corresponding to @xmath665 and @xmath666 respectively .
also , let @xmath649 and @xmath667 be the pants changing segment or looping segment corresponding to the pair @xmath623 and @xmath668 respectively .
since @xmath669 , @xmath670 and @xmath671 for all @xmath662 , we can define the following . for any @xmath662 ,
let @xmath672\to { \pazocal{m}}\ ] ] be the subsegments of @xmath673 , @xmath674 , @xmath675 respectively such that @xmath676 for all @xmath662 , let @xmath677}$ ] , @xmath678}$ ] , @xmath679}$ ] and @xmath680}$ ] .
note that the line segments @xmath681,@xmath682 , @xmath683 and @xmath684 all lie in a triangle that is the closure of a connected component of @xmath685 .
the same is also true for the line segments @xmath686,@xmath687 , @xmath684 and @xmath688 .
thus , @xmath681 is homotopic relative endpoints to @xmath689 and @xmath686 is homotopic relative endpoints to @xmath690 . since @xmath691 and @xmath692
, we have that @xmath663 is homotopic relative endpoints to @xmath693 . also , since @xmath694 , we know that @xmath649 is a looping segment ( resp . pants changing segment ) if and only if @xmath667 is a looping segment ( resp .
pants changing segment ) .
define @xmath695 as in ( [ fi1 ] ) if @xmath649 is a pants changing segment or as in ( [ fi2 ] ) if @xmath649 is a looping segment , and let @xmath696 be the corresponding object for @xmath667 .
enumerate @xmath697 and @xmath698 according to the order induced on @xmath695 and @xmath696 by the orientations of @xmath32 and @xmath33 respectively .
define @xmath699 , @xmath700 , @xmath701 , @xmath702 , and if @xmath649 is a pants changing segment , define @xmath703 , @xmath704 , @xmath705 , @xmath706 . note that for all @xmath707 , @xmath708 if @xmath649 is a looping segment and @xmath709 if @xmath649 is a pants changing segment .
thus , for @xmath707 , we can define @xmath710\to n$ ] to be the subsegment of @xmath711 or @xmath712 such that @xmath713 and @xmath714 . also , for @xmath715 , define @xmath716}$ ] and @xmath717}$ ] .
if @xmath649 is a looping segment , then for the same reasons as above , @xmath718 is homotopic relative endpoints to @xmath719 . in the case that @xmath649 is a pants changing segment , let @xmath554 be the unique closed geodesic in @xmath6 that intersects @xmath649 .
then @xmath554 is also the unique closed geodesic that intersects @xmath667 .
let @xmath720 be the two pairs of pants given by the pants decomposition @xmath6 that contain @xmath554 , and let @xmath721 , @xmath722 be the leaves in @xmath723 , @xmath724 respectively that do not accumulate to @xmath554 . for simplicity , we will assume that @xmath725 ; the proof in the case when @xmath726 is similar . for all @xmath715 , note that @xmath718,@xmath727 , @xmath728 and @xmath729 all lie in the closure of one of the two quadrilaterals in @xmath27 bounded by the lines @xmath550 , @xmath551 , @xmath721 , @xmath722 .
this means that @xmath718 is homotopic relative endpoints to @xmath719 . in either case , since @xmath649 is the concatenation of the @xmath718 and @xmath667 is the concatenation of the @xmath727 in the obvious order , we see that @xmath649 is homotopic relative endpoints to @xmath730 .
finally , note that @xmath32 is the cyclic concatenation @xmath731 and @xmath33 is the cyclic concatenation @xmath732 . by what we proved above
, the cyclic concatenation @xmath732 is homotopic to the cyclic concatenation @xmath731 , so @xmath32 is homotopic to @xmath33 .
since @xmath32 and @xmath33 are closed geodesics , we deduce from ( 2 ) of proposition [ properties ] that @xmath656 . for any @xmath644 ,
let @xmath733 , @xmath734 and @xmath735 be the set of all crossing segments , pants changing segments and looping segments of @xmath32 respectively .
it is easy to see that the typicalness of @xmath32 implies @xmath733 and @xmath736 are nonempty and finite . from @xmath32
, we can also obtain a second piece of combinatorial data @xmath647 , which we will now define .
let @xmath737 be the function defined as follows .
if @xmath629 is a pants changing segment of @xmath32 intersecting @xmath554 , @xmath738 and if @xmath629 is a looping segment of @xmath32 , @xmath739 let @xmath644 and let @xmath648 be the cyclic sequence of crossing points along @xmath32 .
also , let @xmath649 be the looping or pants changing segment associated to the pair @xmath623.then define _ @xmath647 _ be the cyclic sequence @xmath740 , where each @xmath741 is the tuple @xmath742 and the cyclic order here is induced by the cyclic order on the set of crossing points @xmath743 .
unlike @xmath646 , @xmath647 does not give a complete combinatorial description of @xmath32 , i.e. there are distinct typical closed geodesics @xmath32 and @xmath33 in @xmath27 such that @xmath744 .
for example , if @xmath644 has a pants changing segment @xmath629 intersecting @xmath554 such that @xmath745 , then we can do a dehn twist about @xmath554 to obtain a new curve @xmath746 such that the corresponding pants changing segment @xmath747 of @xmath33 does not intersect @xmath549 . note that in this case , @xmath748 , so @xmath744 , but @xmath749 .
however , @xmath647 is useful because we can obtain a lower bound on the length of @xmath32 in terms of the goldman parameters for @xmath27 and other data which depends only on @xmath647 .
we will demonstrate how to do this in section [ lowerboundlengths ] . in the remainder of this subsection , we will show that even though we can not recover @xmath32 from @xmath647 , describing @xmath32 using @xmath647 only loses us a bounded amount of information " in the following sense .
let @xmath750 and @xmath751 .
then we have the following two maps : @xmath752 @xmath753 as mentioned above , @xmath347 is a bijection but @xmath49 is not .
however , we can obtain an upper bound on the size @xmath754 which depends only on the length of the cyclic sequence @xmath647 .
we will now construct this upper bound .
[ pantschanging ] let @xmath629 be a pants changing segment of @xmath644 , and let @xmath554 be the unique closed geodesic in @xmath6 that intersects @xmath629 .
then @xmath755 and @xmath756 suppose first that @xmath629 intersects @xmath549 transversely .
then @xmath757 is finite , and by remark [ dkdkdk ] , the three quantities @xmath758 are all at most @xmath24 .
the inequalities in the lemma then follow from the definition of @xmath759 . in the case
when @xmath629 does not intersect @xmath549 transversely , we have that @xmath760 , so @xmath761 .
it is then clear that the required inequalties hold in this case as well .
[ looping ] let @xmath629 be a looping segment of @xmath32 . then @xmath762 where @xmath635 is defined by ( [ fi2 ] ) .
consider any lift @xmath763 of @xmath629 in @xmath78 . since @xmath629 is a looping segment
, @xmath763 intersects the lines in @xmath535 finitely many times , and all the lines in @xmath535 that intersect @xmath764 share a common endpoint , @xmath432 .
this implies that @xmath629 lies in a single pair of pants @xmath570 given by @xmath6 , and that @xmath764 lies in @xmath765 .
moreover , by remark [ gammaonomega ] , any other lift @xmath766 of @xmath629 to @xmath767 which intersects @xmath763 must also lie in @xmath765 .
in fact , @xmath768 for some @xmath769 , where @xmath61 is a generator of @xmath537 .
thus , the number of self intersections of @xmath629 is the number of positive integers @xmath770 such that @xmath771 is nonempty .
it follows from the definition of a looping segment that @xmath772 . also , by the description of the action of @xmath537 on @xmath765 given in remark [ gammaonomega ]
, it is clear that if @xmath773 for @xmath624 , then @xmath774 intersects @xmath763 for @xmath775 .
this implies that @xmath776 , so the inequalities in the lemma hold . on the other hand , if @xmath777 for @xmath624 , then there are two possible cases .
let @xmath778 and @xmath779 be the two endpoints of @xmath763 so that @xmath780 and @xmath779 lie on the same line in @xmath781 .
let @xmath135 be the axis of @xmath61 and let @xmath782 .
it is clear that @xmath778 , @xmath779 and @xmath780 lie on @xmath783 .
if @xmath779 lies between @xmath778 and @xmath780 on @xmath783 , then @xmath774 intersects @xmath763 for @xmath784 .
( see figure [ keven ] . )
however , if @xmath780 lies between @xmath778 and @xmath779 on @xmath783 , then @xmath774 intersects @xmath763 for @xmath775 . in either case
, @xmath759 is either @xmath770 or @xmath785 , so the required inequalities still hold . and @xmath786 .
] the above two lemmas describe the discrepancy between the information contained in @xmath646 and @xmath647 . with these
, we can prove the main proposition of this subsection .
[ phitopsi ] if @xmath644 has @xmath787 crossing points , then any @xmath746 such that @xmath744 also has @xmath787 crossing points . moreover , there are at most @xmath788 closed geodesics in @xmath645 that have the same image as @xmath32 under the map @xmath49 .
the first statement follows immediately from the definition of @xmath647 . to prove the second statement
, we will pick any sequence @xmath789 in @xmath790 and reconstruct @xmath646 for all typical closed curves @xmath32 in @xmath791 such that @xmath792 .
then we show that for each @xmath789 , the number of possible @xmath646 that we can construct is at most @xmath788 .
the fact that @xmath347 is a bijection will then imply the proposition .
choose any @xmath793 in @xmath790 , where each @xmath741 is the tuple @xmath794 .
here , @xmath795 , @xmath796 , @xmath797 are lines in @xmath798 equipped with an orientation , and @xmath799 .
suppose @xmath644 such that @xmath792 .
we can assume without loss of generality that @xmath800 , with @xmath801 , i.e. for all @xmath662 , @xmath802 , @xmath803 and @xmath804 .
first , note that @xmath789 contains sufficient information to determine if a pair @xmath623 is a looping pair or a pants changing pair : @xmath623 is a looping pair if and only if there is some pair of pants @xmath805 for @xmath27 such that @xmath806 and @xmath807 are lines in @xmath723 , and there are lifts of the lines @xmath806 , @xmath808 , @xmath809 , @xmath807 to @xmath527 that share a common vertex .
furthermore , if @xmath623 is a pants changing pair and @xmath649 is the corresponding pants changing segment , then @xmath789 determines the unique @xmath554 in @xmath6 that intersects @xmath649 .
indeed , if @xmath720 are the pairs of pants given by @xmath6 containing @xmath810 and @xmath811 respectively ( it might be that @xmath726 ) , then @xmath554 is the common boundary component of @xmath805 and @xmath812 that is bounded away from both @xmath813 in @xmath805 and @xmath814 in @xmath812 .
suppose now that @xmath623 is a pants changing pair and @xmath554 is the unique curve in @xmath6 obtained in the previous paragraph . if @xmath649 is the pants changing segment of @xmath32 corresponding to the pants changing pair @xmath623 , then @xmath815 .
hence , lemma [ pantschanging ] implies that @xmath816 has to be a finite sequence that alternates between @xmath550 and @xmath551 , so that if @xmath817 and @xmath818 are the number of times @xmath550 and @xmath551 respectively occur in @xmath816 , then @xmath819 , @xmath820 and @xmath821 .
one can easily verify that there are at most eighteen possibilities for @xmath816 .
moreover , the orientation of @xmath822 and the sequence @xmath816 determines @xmath823 uniquely . explicitly , let @xmath805 be the pair of pants for @xmath27 that contains @xmath673 and let @xmath824 and @xmath825 be the two boundary components of @xmath805 that @xmath826 accumulates to , such that @xmath822 points from @xmath824 to @xmath825 .
also , let @xmath554 be the third boundary component of @xmath805 that is not @xmath824 or @xmath825 .
assume without loss of generality that @xmath550 accumulates to @xmath824 and @xmath551 accumulates to @xmath825 .
if the first term of @xmath816 is @xmath551 , then we orientate @xmath550 towards @xmath824 and @xmath551 towards @xmath825 to obtain @xmath823 . if the first term of @xmath816 is @xmath550 , then we orientate @xmath550 away from @xmath824 and @xmath551 away from @xmath825 to obtain @xmath823 .
these orientations are forced onto us by the requirement that there is a geodesic that passes from the left to right across each oriented line in @xmath823 .
in particular , if we know @xmath647 , then there are only at most eighteen different possibilities for @xmath823 for each @xmath177 . on the other hand ,
if @xmath623 is a looping pair , then @xmath815 , so lemma [ looping ] , tells us that @xmath816 has to be a sequence that alternates between @xmath810 and @xmath827 , with either @xmath828 , @xmath829 or @xmath830 terms . as in the previous paragraph , given the orientation on @xmath831 and the sequence @xmath816 , there is a unique way to orient the terms of @xmath816 to recover @xmath823 .
explicitly , let @xmath805 be the pair of pants given by @xmath6 containing @xmath649 , and let @xmath554 be the boundary component of @xmath805 that every line in @xmath816 accumulates to .
if @xmath831 is oriented towards @xmath554 , then we orient every term in @xmath816 towards @xmath554 , and if @xmath831 is oriented away from @xmath554 , then we orient every term in @xmath816 away from @xmath554 .
in particular , this tells us that there are at most three different possibilities for @xmath823 . putting all of these together
, we get that if @xmath789 has @xmath832 pants changing pairs and @xmath833 looping pairs , then @xmath834 since @xmath347 is a bijection , we are done . in this subsection
, we will show that we can obtain a lower bound for the length of any @xmath644 in terms of the goldman coordinates and combinatorial data encoded by @xmath647 . as a consequence , we prove ( 1 ) of theorem [ mainthm1 ] .
note that to prove ( 1 ) of theorem [ mainthm1 ] , we actually only need to obtain a lower bound for crossing segments and show that this lower bound goes to @xmath14 as we deform along goldman sequences . however , we need estimates of the lengths of the pants changing segment and looping segments to prove ( 2 )
. first , we will give the length lower bound for pants changing segments and looping segments .
given a pair of boundary invariants @xmath835 for a marked convex @xmath3 surface @xmath836 , let @xmath135 be the simple closed curve in @xmath27 corresponding to @xmath835 .
we know that the length of @xmath135 in the hilbert metric is given by ( see equation [ lengthofc ] ) @xmath837 thus , we can define the function @xmath838 given by @xmath839 .
we want to bound the lengths of the pants changing segments and looping segments of @xmath32 from below by a constant multiple of @xmath840 , where the constant depends only on @xmath647 . but first
, we need the following lemma .
[ matchingloops ] let @xmath629 be a looping segment of @xmath644 .
let @xmath763 be a lift of @xmath629 to @xmath78 , and let @xmath841 be the points in @xmath763 that also lie on other lifts of @xmath629 , enumerated according to the orientation on @xmath629 . then @xmath461 is even and the covering map @xmath842 satisfies @xmath843 for any @xmath844 .
let @xmath531 be the pair of pants given by @xmath6 that contains @xmath629 , let @xmath763 be a lift of @xmath629 to @xmath78 and let @xmath533 be the lift of @xmath531 to @xmath78 that contains @xmath763 . since @xmath629 is a looping segment , all the lines in @xmath535 that intersect @xmath764 share a common endpoint in @xmath87 .
let this common endpoint be @xmath432 , and it is clear that @xmath764 lies in @xmath845 .
since @xmath432 is the image of either @xmath846 , @xmath847 or @xmath848 under a deck transformation , we can assume without loss of generality that @xmath849 , @xmath847 or @xmath848 .
we will do the proof for @xmath849 ; the other two cases are similar .
let @xmath850 ( @xmath210 and @xmath851 are defined at the start of section [ maincontent ] ) .
the properness of @xmath78 and the fact that it is invariant under @xmath852 implies that @xmath533 is contained in a triangle @xmath853 whose vertices are the three fixed points of @xmath852 . via a projective transformation
, we can assume that the attracting , repelling and third fixed point of @xmath852 are @xmath854 $ ] , @xmath855 $ ] and @xmath856 $ ] respectively , and @xmath857\in{\mathbb{r}}{\mathbb{p}}^2:x , y , z>0\}$ ] .
goldman computed in section 1.10 of @xcite that the @xmath852 orbit of any @xmath858\in t$ ] lies on the curve @xmath859}:=\bigg\{[x , y , z]\in t:\bigg(\frac{x}{x_0}\bigg)^{\log(\frac{\nu}{\mu})}\bigg(\frac{z}{z_0}\bigg)^{\log(\frac{\mu}{\lambda})}=\bigg(\frac{y}{y_0}\bigg)^{\log(\frac{\nu}{\lambda})}\bigg\},\ ] ] where @xmath449 , @xmath197 , @xmath450 are the eigenvalues of @xmath852 for @xmath854 $ ] , @xmath856 $ ] and @xmath855 $ ] respectively .
one can easily verify that the curves @xmath860}$ ] for @xmath858\in t$ ] foliate @xmath853 , and that any line intersects each of these curves at most twice . if @xmath861 , then there is some deck transformation @xmath61 such that @xmath862 .
however , since @xmath863 and @xmath864 lie in @xmath865 , we know @xmath61 is of the form @xmath866 for some integer @xmath770 .
the previous paragraph then proves that for any self - intersection @xmath363 of @xmath629 , there are exactly two points in @xmath841 that are mapped to @xmath363 by @xmath867 , so @xmath461 is even . now , suppose that there is some @xmath868 such that @xmath869 . by the pigeon hole principle , one of the following must hold : 1 .
there is some @xmath870 and some @xmath871 such that @xmath872 .
2 . there is some @xmath873 and some @xmath874 such that @xmath872 . either way
, this implies that the curves @xmath875 and @xmath876 must intersect , contradiction . with this , we can obtain the required lower bound stated in the next proposition . in the case when @xmath27 lies in the fuchsian locus
, this proposition follows from the fact that the projection of @xmath877 onto a geodesic is @xmath24-lipschitz .
[ betalowerbound ] if @xmath629 is a pants changing segment or looping segment for a typical oriented closed geodesic @xmath32 in a marked convex @xmath3 surface @xmath836 , then @xmath878 first , consider the case when @xmath629 is a pants changing segment .
let @xmath805 and @xmath812 be the two pairs of pants given by @xmath6 such that @xmath879 and let @xmath554 be the unique element in @xmath6 that intersects @xmath629 .
also , let @xmath880 and @xmath881 be the lines in the ideal triangulation of @xmath805 and @xmath812 respectively that do not accumulate to @xmath554 .
if @xmath629 intersects @xmath549 non - transversely or @xmath629 intersects @xmath549 at most once , then @xmath882 and the required lower bound clearly holds . if @xmath629 intersects @xmath549 transversely and at least twice , then @xmath883 .
enumerate @xmath884 according to the total order induced by the orientation on @xmath629 .
let @xmath885}$ ] and let @xmath886\to{\pazocal{m}}$ ] be the subsegment of @xmath549 with @xmath887 and @xmath888
. since @xmath401 and @xmath889 are distinct line segments sharing the same endpoints , they can not be homotopic relative endpoints by ( 3 ) of proposition [ properties ] .
hence , @xmath890 is a non - contractible simple closed curve in the topological annulus @xmath891 , so it is homotopic to @xmath554 .
this implies that @xmath892 on the other hand , since @xmath549 is a distance minimizing line segment between @xmath880 and @xmath881 , we know that @xmath893 .
thus , @xmath894 for all @xmath895 , so @xmath896 next , consider the case when @xmath629 is a looping segment .
enumerate the set @xmath897 according to the orientation on @xmath32 .
the proof of lemma [ matchingloops ] gives us that @xmath898 .
define @xmath899}$ ] for @xmath900 by lemma [ matchingloops ] , we know that @xmath901 for all @xmath844 , so for @xmath902 , @xmath718 and @xmath903 are two line segments that share their endpoints , but do not intersect anywhere else . in particular , @xmath904 is a non - contractible simple closed curve .
let @xmath531 be the pair of pants given by @xmath6 that contains @xmath629 , and note that ( 2 ) of proposition [ properties ] implies that @xmath904 has to be homotopic to a boundary component of @xmath531 . by using this same argument on the line segment @xmath905 and the trivial line segment that is just the endpoint of @xmath905
, we see that @xmath905 is also homotopic to a boundary component of @xmath531 .
thus , @xmath906 for @xmath907 and @xmath908 .
this implies that @xmath909 now , we want to find the corresponding length lower bound for the crossing segments of @xmath32 .
to do so , we first need to classify the crossing segments of @xmath32 into two types .
let @xmath665 be a crossing point of @xmath32 and let @xmath663 be the corresponding crossing segment .
then let @xmath910 be the triangle in @xmath27 bounded by @xmath673 , @xmath674 and @xmath663 , and let @xmath911 be the triangle in @xmath27 bounded by @xmath675 , @xmath674 and @xmath663 .
consider any crossing segment @xmath663 of @xmath644 , oriented according to the cyclic orientation on @xmath32 .
then @xmath663 is said to be of _ z - type _
if @xmath910 is to the left of @xmath663 and @xmath911 is to the right of @xmath663 .
it is said to be of _ s - type _ if @xmath910 is to the right of @xmath663 and @xmath911 is to the left of @xmath663 .
( see figure [ sz ] . )
it is clear that every crossing segment is either of s - type or z - type , and reversing orientation preserves the type .
observe also that if @xmath623 is a looping pair , then the crossing segment @xmath663 for @xmath665 is of s - type if and only if the crossing segment @xmath912 for @xmath913 is of z - type .
for the rest of the paper , we will simplify notation by writing @xmath914 as @xmath915 .
( see section [ areparameterization ] . )
define the function @xmath916 given by @xmath917 , where @xmath918 is the minimum of the following six numbers : @xmath919 and @xmath920 is the minimum of the following six numbers : @xmath921 the next proposition provide length lower bounds for the crossing segments of @xmath644 .
[ alphalowerbound ] let @xmath644 and let @xmath922 be the crossing points of @xmath32 , ordered cyclically according to the orientation of @xmath32 .
also , let @xmath406 be the crossing segment corresponding to @xmath409 and let @xmath407 be the looping or pants changing segment corresponding to @xmath923 . 1 .
if @xmath923 is a looping pair , then @xmath924 .
2 . if @xmath32 is contained in some pair of pants @xmath531 given by @xmath6 , then @xmath925 .
3 . suppose that @xmath926 and @xmath923 are both pants changing pairs .
let @xmath927 and @xmath410 be the unique points in @xmath606 ( see section [ setup ] ) such that @xmath928 and @xmath929 .
define @xmath930 to be the subsegment @xmath931 , where @xmath932 is the closed subsegment containing @xmath933 , with endpoints @xmath927 and @xmath410 .
if @xmath934 for all @xmath379 , then @xmath935 in this proof , we will use the following notation .
let @xmath63 be a properly convex subset of @xmath3 and let @xmath936 be any two distinct points in the closure @xmath67 of @xmath63 in @xmath3 .
let @xmath937_\omega$ ] denote the closed line segment in @xmath67 with endpoints @xmath131 , @xmath129 .
if @xmath936 are two distinct points in @xmath68 , choose an oriented affine chart @xmath105 containing the closure of @xmath63 , and define @xmath938 to be the oriented open subsegment of @xmath68 that goes from @xmath131 to @xmath129 in the clockwise direction in @xmath105 along @xmath68 .
proof of ( 1 ) . assume without loss of generality that @xmath406 is of s - type and @xmath939 is of z - type .
let @xmath531 be the pair of pants given by @xmath6 containing @xmath406 and @xmath939 , and consider the lamination @xmath532 of @xmath533 ( drawn partially in figure [ goldman ] ) .
there is a deck transformation that sends @xmath940 to either @xmath941_{\omega_{\pazocal{m}}}$ ] , @xmath937_{\omega_{\pazocal{m}}}$ ] or @xmath942_{\omega_{\pazocal{m}}}$ ] , so one of the following must hold because @xmath406 is of s - type : a. there is a lift @xmath943 of @xmath406 with endpoints in @xmath944_{\omega_{\pazocal{m}}}$ ] and @xmath942_{\omega_{\pazocal{m}}}$ ] .
b. there is a lift @xmath943 of @xmath406 with endpoints in @xmath945_{\omega_{\pazocal{m}}}$ ] and @xmath941_{\omega_{\pazocal{m}}}$ ] . c. there is a lift @xmath943 of @xmath406 with endpoints in @xmath946_{\omega_{\pazocal{m}}}$ ] and @xmath937_{\omega_{\pazocal{m}}}$ ] . .
] suppose first that ( i ) holds .
then by applying proposition [ crossratioinequality ] , we see that @xmath947 .
( see figure [ estimate1 ] . )
also , it is clear that the line segment in @xmath78 containing the segment @xmath943 and with endpoints in @xmath87 has one endpoint @xmath50 in @xmath948 and one endpoint @xmath65 in @xmath949 . using proposition [ crossratioinequality ] again
, we see that if @xmath65 lies in @xmath950 , then @xmath951 , and if @xmath65 lies in @xmath952 , then @xmath953 .
( see figure [ estimate2 ] and figure [ estimate3 ] . ) thus , @xmath954 the same argument shows that if ( ii ) holds , then @xmath955 and if ( iii ) holds , then @xmath956 hence , @xmath957 , where @xmath918 is the minimum of the six numbers listed in ( [ list1 ] ) . . ] . ] similarly , since there is a deck transformation that sends @xmath958 to either @xmath941_{\omega_{\pazocal{m}}}$ ] , @xmath937_{\omega_{\pazocal{m}}}$ ] or @xmath942_{\omega_{\pazocal{m}}}$ ] and @xmath939 is of z - type , one of the following must hold : a. there is a lift @xmath959 of @xmath939 with endpoints in @xmath960_{\omega_{\pazocal{m}}}$ ] and @xmath961_{\omega_{\pazocal{m}}}$ ] .
b. there is a lift @xmath959 of @xmath939 with endpoints in @xmath962_{\omega_{\pazocal{m}}}$ ] and @xmath963_{\omega_{\pazocal{m}}}$ ] . c. there is a lift @xmath959 of @xmath939 with endpoints in @xmath964_{\omega_{\pazocal{m}}}$ ] and @xmath965_{\omega_{\pazocal{m}}}$ ] . using the same arguments as in the previous paragraph
, we can show that if ( i ) , ( ii ) or ( iii ) holds , then @xmath966 @xmath967 or @xmath968 respectively .
thus , @xmath969 , where @xmath920 is the minimum of the six numbers listed in ( [ list2 ] ) .
putting our lower bounds for the lengths of @xmath406 and @xmath939 together , we get @xmath970 proof of ( 2 ) .
since @xmath32 is contained in @xmath531 , we know that @xmath32 has at least two crossing points . thus , @xmath971 proof of ( 3 ) .
let @xmath972 and @xmath973 be the two curves in @xmath6 that contain @xmath927 and @xmath410 respectively , parameterized by @xmath974\to n$ ] , @xmath975 for @xmath976 .
note that either @xmath977 or @xmath978 is typical and their geodesic representatives both lie in some pair of pants given by @xmath6 .
either way , we can apply ( 2 ) to get @xmath979 which implies that @xmath980
. we can finally give the lower bound on the length of any @xmath644 , as promised at the start of this subsection .
[ mainestimate ] let @xmath644 . if @xmath27 is such that @xmath934 for any @xmath379 ,
then @xmath981 where @xmath787 is the number of crossing points for @xmath32 .
let @xmath648 be the crossing points for @xmath32 , ordered according to the orientation on @xmath32 .
define @xmath982 and note that @xmath983 and @xmath984 are disjoint and @xmath985 .
let @xmath986 be the set of interiors of all domains for the pants changing segments of @xmath32 , the looping segments of @xmath32 , the subsegments @xmath930 of @xmath32 for @xmath987 , and the subsegments of @xmath406 for @xmath32 if @xmath988 .
one can easily verify that @xmath986 is an open cover for @xmath599 and that every point in @xmath599 is contained in at most five different elements of @xmath986 .
this implies that @xmath989 by proposition [ betalowerbound ] , we know @xmath990 and ( 3 ) of proposition [ alphalowerbound ] implies @xmath991 also , if we define @xmath992 , observe that @xmath993 .
hence , ( 1 ) of proposition [ alphalowerbound ] implies @xmath994 combining equations ( [ lowerbound3 ] ) and ( [ lowerbound4 ] ) gives @xmath995 the theorem then follows from the inequalities ( [ lowerbound1 ] ) , ( [ lowerbound2 ] ) and ( [ lowerbound5 ] ) .
the next lemma is the main computation in this subsection .
this lemma , together with theorem [ mainestimate ] , implies ( 1 ) of theorem [ mainthm1 ] .
[ kgoestoinfty ] for any goldman sequence @xmath11 in @xmath1 , we have @xmath996 to prove the lemma , it is sufficient to show that for any goldman sequence @xmath997 , we have that @xmath998 where @xmath999 is the minimum of the following six numbers @xmath1000 and @xmath1001 is the minimum of the following six numbers @xmath1002 we can compute ( see appendix [ crossratiocomputationsandformulas ] ) each @xmath1003 explicitly in terms of @xmath1004 and the internal parameter @xmath319 for @xmath385 ( see section [ areparameterization ] ) .
this gives us @xmath1005 where arithmetic in the subscripts is done in @xmath325 .
thus , @xmath1006 where @xmath1007 let @xmath1008 it is now sufficient to show that for all @xmath1009 , the limits as @xmath13 approaches infinity of @xmath1010 , @xmath1011 and @xmath1012 are infinity .
( note that @xmath319 and the @xmath1013 depend on @xmath13 . )
let @xmath1014 be the goldman parameters for @xmath385 .
since @xmath1015 is a goldman sequence , every subsequence of @xmath1015 has a further subsequence , also denoted @xmath1015 , such that either @xmath1016 , @xmath1017 , @xmath1018 or @xmath1019 .
we need to show that along each of these subsequences , @xmath1010 , @xmath1011 and @xmath1012 grow to infinity .
suppose first that @xmath1016 or @xmath1017 .
the condition that the lengths of the boundaries curves of @xmath385 are bounded away from @xmath15 and @xmath14 imply that @xmath1020 or @xmath1021 respectively for @xmath232 . since @xmath1022 it is clear that if @xmath1016 or @xmath1017 , then @xmath1023 . in the case
when @xmath1024 is neither infinite nor zero , the condition that the lengths of the boundaries curves of @xmath385 are bounded away from @xmath15 and @xmath14 implies that @xmath1025 , @xmath1026 and @xmath1027 are bounded away from @xmath24 and @xmath14 .
thus , if @xmath1018 , the inequality @xmath1028 implies that @xmath1029 . on the other hand ,
if @xmath1019 , the inequality @xmath1030 implies that @xmath1029 .
we have thus shown that for any goldman sequence @xmath1015 , @xmath1029 . using similar arguments
, we can likewise show that @xmath1031 and @xmath1032 .
the work in section [ lowerboundlengths ] gives us a lower bound on the length of all typical closed geodesics @xmath644 , which depends only on the data @xmath647 and grows to infinity along goldman sequences .
moreover , by section [ combinatorialdescriptionsclosedgeodesics ] , we also have an upper bound on the number of typical closed geodesics with the same @xmath647 .
we will now show that these bounds are strong enough to show that the topological entropy converges to zero along goldman sequences , thereby proving ( 2 ) of theorem [ mainthm1 ] .
suppose that @xmath392 is such that @xmath934 for all @xmath379 . by theorem [ mainestimate ]
, we know that if @xmath1033 , then the number of crossing points of @xmath32 is at most @xmath1034 . applying proposition [ phitopsi ] , we have that for @xmath1035 , @xmath1036 . ) if @xmath644 has @xmath787 crossing points , then lemma [ crossingtriple ] implies that there are at most @xmath1037 possibilities for what the cyclic sequence of crossing triples of @xmath32 can be .
also , if @xmath1038 , and @xmath32 has @xmath787 crossing points , then theorem [ mainestimate ] implies that @xmath1039 thus , we have the inequality @xmath1040 now , let @xmath1041 be the set of oriented closed geodesics in @xmath27 .
if we assume that @xmath1035 , then ( [ inequal1 ] ) and ( [ inequal2 ] ) imply that @xmath1042 where @xmath1043 is an integer such that @xmath1044 for all @xmath1045 .
thus , @xmath1046 so by taking the limit supremum as @xmath853 approaches infinity on both sides , we get that @xmath1047 for any @xmath392 such that @xmath934 for all @xmath379 .
consider a goldman sequence @xmath1048 in @xmath1 . by lemma [ kgoestoinfty ] ,
which tells us @xmath1049 , we have that @xmath1050 recall that there are constants @xmath384 such that @xmath1051 .
hence , to prove ( 2 ) of theorem [ mainthm1 ] , it is thus sufficient to show the following technical proposition , which we prove in appendix [ proof of proposition technical ] .
[ technical ] for any constant @xmath66 and any increasing sequences @xmath1052 and @xmath1053 such that @xmath1054 , we have that @xmath1055 where @xmath1056 is a number in @xmath1057 such that @xmath1058 for all @xmath1059 .
here , we will list the formulas for some of the thirty cross ratios @xmath1060 as defined at the start of section [ areparameterization ] , namely those we used in the proof of our result .
the formulas will be listed in both the goldman parameters @xmath1061 defined in section [ goldmanparameters ] and the new parameters @xmath1062 defined in section [ areparameterization ] .
also , we will demonstrate the computation to obtain one of these , namely @xmath1063 .
the rest of the cross ratios are computed using the same algorithm . to simplify the formulas ,
we will write @xmath1064 as @xmath1013 for @xmath232 . we compute @xmath1063 using definition [ cross ratio def ] .
since the cross ratio is invariant under the action of @xmath35 we can choose a normalization so that @xmath1065 , b=\left [ \begin{array}{c } 0\\ 1\\ 0 \end{array } \right ] , c=\left [ \begin{array}{c } 0\\ 0\\ 1 \end{array } \right ] , f=\left [ \begin{array}{c } 2\\ 2\\ -1\end{array } \right].\ ] ] then in the goldman parameters , we have @xmath1066 , e=\left [ \begin{array}{c }
t\\ -1\\ \frac{\rho_1}{2 } \end{array } \right].\ ] ] ( this was computed in section 4 of goldman @xcite ) . by the definition of the cross ratio , we
can then compute @xmath1067\wedge\left [ \begin{array}{c } 0\\ 1\\ 0 \end{array } \right]\wedge\left[\begin{array}{c } 1\\ 0\\ 0 \end{array } \right]}{\left [ \begin{array}{c } -1\\ \frac{\rho_3}{t}\\ \frac{\rho_2}{2 } \end{array } \right]\wedge\left [ \begin{array}{c } 0\\ 1\\ 0 \end{array } \right]\wedge\left [ \begin{array}{c } 2\\ 2\\ -1\end{array } \right]}\cdot\frac{\left [ \begin{array}{c } -1\\ \frac{\rho_3}{t}\\ \frac{\rho_2}{2 } \end{array } \right]\wedge\left [ \begin{array}{c } 0\\ 0\\ 1 \end{array } \right]\wedge\left [ \begin{array}{c } 2\\ 2\\ -1\end{array } \right]}{\left [ \begin{array}{c } -1\\ \frac{\rho_3}{t}\\ \frac{\rho_2}{2 } \end{array } \right]\wedge\left [ \begin{array}{c } 0\\ 0\\ 1 \end{array } \right]\wedge\left[\begin{array}{c }
1\\ 0\\ 0 \end{array } \right]}\\ & = & \frac{\det\left [ \begin{array}{ccc } -1&0&1\\ \frac{\rho_3}{t}&1&0\\ \frac{\rho_2}{2}&0&0 \end{array } \right]}{\det\left [ \begin{array}{ccc } -1&0&2\\ \frac{\rho_3}{t}&1&2\\ \frac{\rho_2}{2}&0&-1\end{array } \right]}\cdot\frac{\det\left [ \begin{array}{ccc } -1&0&2\\ \frac{\rho_3}{t}&0&2\\ \frac{\rho_2}{2}&1&-1 \end{array } \right]}{\det\left [ \begin{array}{ccc } -1&0&1\\ \frac{\rho_3}{t}&0&0\\ \frac{\rho_2}{2}&1&0 \end{array } \right]}\\ & = & \frac{-\frac{\rho_2}{2}}{1-\rho_2}\cdot\frac{2+\frac{2\rho_3}{t}}{\frac{\rho_3}{t}}\\ & = & \frac{\rho_2(t+\rho_3)}{\rho_3(\rho_2 - 1)}.\end{aligned}\ ] ] now , we will give the list of formulas for the cross ratios we used in the paper . @xmath1068
in this appendix , we will prove proposition [ technical ] , which we restate here for the reader s convenience .
[ technical2 ] for any constant @xmath66 and any increasing sequences @xmath1052 and @xmath1053 such that @xmath1054 , we have that @xmath1055 where @xmath1056 is a number in @xmath1057 such that @xmath1058 for all @xmath1059 .
first , we will fix @xmath1069 to be @xmath1070 and compute @xmath1071 where @xmath1072 is a number in @xmath1073 such that @xmath1074 for all @xmath1075 . the main tool to compute ( [ limitbound1 ] )
is the asymptotic equality commonly known as stirling s formula , which we state here . for fixed @xmath1078 and @xmath66
, we will at times denote @xmath1079 by @xmath1080 to simplify notation . to use stirling s formula
, we need to know how @xmath1080 and @xmath1072 vary with @xmath13 .
hence the following lemma .
proof of ( 2 ) .
suppose for contradiction that @xmath1095 . by choosing a subsequence
, we can assume that @xmath1096 , then @xmath1097 , so we have @xmath1098 however , if @xmath1096 and @xmath1097 , then the right hand side of the inequality ( [ qbound1 ] ) converges to @xmath24 as @xmath1099 .
this then implies that @xmath1100 which is a contradiction .
proof of ( 4 ) . by ( [ qbound2 ] ) , we know that @xmath1101 , so @xmath1102 .
since @xmath1103 , this proves @xmath1104 .
it is clear that @xmath1105 , so suppose for contradiction that @xmath1106 . by taking a subsequence
, we can assume that @xmath1107 , which implies that @xmath1108 . by ( 3 ) and ( 4 ) of lemma [ stirlingconditions ]
, we know that @xmath1114 and @xmath1115 . also , ( 1 ) and ( 2 ) of lemma [ stirlingconditions ] allow us to apply stirling s formula to obtain @xmath1116 i.e. @xmath1117 for each @xmath177 such that @xmath1135 , choose a subsequence @xmath1136 of @xmath1053 such that @xmath1137 exists and @xmath1138 ( we can do this by ( 4 ) of lemma [ stirlingconditions ] ) .
proposition [ computation0 ] then tells us that @xmath1139 since @xmath1140 , we know that @xmath1141 .
thus , by choosing a subsequence of @xmath1052 , we can assume that @xmath1142 since @xmath1143 , this implies that @xmath1144 .
this means @xmath1145 also , @xmath1146 for all @xmath177 , so @xmath1147 next , we show that @xmath1148 . since @xmath1144 , this is clear in the case when @xmath1149 . in the case when @xmath1150,@xmath1151 for large enough @xmath177 . by taking limit supremum , @xmath1152 since @xmath1153 for all @xmath177
, we have that @xmath1154 s. wolpert , _ geometry of the weil - petersson completion of teichmller space _ , in : surveys in differential geometry , viii : papers in honor of calabi , lawson , siu and uhlenbeck , international press , ( 2003 ) .
|
let @xmath0 be a compact surface of negative euler characteristic and let @xmath1 be the deformation space of convex real projective structures on @xmath0 . for every choice of pants decomposition for @xmath0 , there is a well known parameterization of @xmath1 known as the goldman parameterization . in this paper , we study how some geometric properties of the real projective structure on @xmath0 degenerate as we deform it so that the internal parameters of the goldman parameterization leave every compact set while the boundary invariants remain bounded away from zero and infinity .
|
first discovered by @xcite , strong x - ray bursts are believed to occur due to thermonuclear explosions in the bottom helium - reach layers of the atmosphere accumulated by a neutron star during the accretion process in close binary system . since then dozens of burster - type x - ray sources were found .
one of the distinctive feature of type i x - ray bursts is the sudden and abrupt ( @xmath1 s ) luminosity increase ( expansion stage ) followed by exponential decay ( contraction stage ) .
energy released in x - ray radiation during the first seconds greatly exceeds the eddington limit for layers above the helium burning zone which are no longer dynamically stable .
super - critically irradiated shells of atmosphere start to move outward , producing an expanding wind - like envelope .
the average lifetime of an x - ray bursts is sufficient for steady - state regime of mass loss to be established when the local luminosity throughout the most of the atmosphere is equal or slightly greater than the eddington limit . during the last two decades the problem of determining properties of radiatively driven winds during x - ray bursts
was subjected to extensive theoretical and numerical studies .
various theories were put forward with gradually increasing level of accuracy of the problem description , but only a few approaches addressed the case of considerably expanded photosphere under influence of near - eddington luminosities @xcite .
see @xcite for a detailed review of x - ray burst study during 80 s and the beginning of 90 s . similarly to the problem of accretion flows , notion of the existence of sonic point in continuous flow became a natural starting point in the analysis of wind flows from stellar objects .
@xcite , hereafter ehs , investigated the structure of the envelopes with steady - state mass outflow and pointed out the higher eddington luminosity in the inner shells due to the prevalent higher temperatures and correspondingly lower compton scattering opacities .
they showed that the product of opacity and luminosity remains almost constant throughout the atmosphere which is the key assumption of the model .
the existence of wind - like solutions for critically irradiated atmospheres was proved .
@xcite , hereafter t94 , studied analytically spectral shapes of the expansion and contraction stages of bursts .
he showed how ehs s approach to hydrodynamic problem can be greatly simplified with the sonic point condition properly calculated and tied with conditions at the bottom of the envelope .
@xcite applied the t94 model to extract the neutron star mass - radius relations from the observed burst spectra in 4u 1820 - 30 and 4u 1705 - 44 .
@xcite , hereafter ntl , adopted a high accuracy numerical approach to the problem of x - ray burster atmosphere structure based on the moment formalism @xcite .
they integrated a self - consistent system of frequency - independent , relativistic , hydrodynamical and radiative transfer equations over the whole atmosphere including the inner dense helium - burning shells .
three important characteristic of x - ray burst outflow were obtained in this work : the helium - burning zone temperature was maintained approximately at the level of @xmath2 k , the temperature of the photosphere was shown to depart appreciably from the electron temperature and to stay constant at the outer shells , and the existence of the maximum and the minimum values of the mass loss rate was found .
one of the goals of these studies was to provide the algorithm of determination of the compact object characteristics by analyzing observational data .
with the advent of high spectral and time resolution observational instruments ( such as chandra , rxte , usa , xmm - newton missions ) the task of obtaining a suitable tool for fitting the energy spectra became extremely important . despite numerous earlier studies of x - ray burst observations ,
recent developments have shown a growing interest of the astrophysical community in this area @xcite . obviously , the problem of radiative transfer in relativistically moving media is very complicated one and under rigorous consideration it must be solved numerically . in this paper
we develop an alternative approach which allows both numerical and analytical solutions and successfully accounts for all crucial physical processes involved .
we show how this problem under some appropriate approximations yields the spectrum of radiation from spherically symmetric outflows in an analytical form .
we concentrate on the case of extended atmosphere with inverse cubic power dependence of the number density on radius , which is more appropriate for the expansion stage but can also be employed for description of the contraction as a sequence of models with decreasing mass - loss rate .
we represent a numerical approach to the problem which then provides the validation of our analytical description .
we adopt the general approach formulated in ehs and developed in t94 .
the problem of determining profiles of thermodynamic variables of steady - state radiatively driven outflow was solved in t94 .
the problem is reduced to the form of a first order differential equation , which allows easy and precise numerical solution . for a completeness
we present this method in section 2 . using atmospheric profiles obtained for different neutron star configurations we solve the problem of radiative transfer by relaxation method on an energy - opacity logarithmic grid .
we perform temperature profile correction by applying temperature equation to the obtained spectral profiles .
the basic formulae are given in section 3.1 .
then the analytical description of the problem is represented in detail .
the analytic solution of radiative transfer equation on the atmospheric profile @xmath0 is presented in t94 . here
we review the solution by carrying out the integration without introducing any approximations . in section 4
we compare and match our analytical and numerical results to describe the behavior of free parameter .
we finalize our work by examining the properties of our analytic solution , combine it with the results of section 4 and construct the final formula for fitting the spectra in section 5 .
the discussion of our method along with some other important issues concerning the problem being solved are presented in section 6 .
conclusions follow in the last section .
as we already mentioned , that the calculation of x - ray burst spectra can be treated as a steady state problem . to justify this assumption one has to compare the characteristic times of phenomena considered .
time scale for the photosphere to collapse can be estimated as follows @xmath3 here @xmath4 denotes the sonic point radius , which is adopted as the outer boundary of the photosphere throughout this paper .
dimensionless luminosity is @xmath5 , where eddington luminosity is given by @xmath6 opacity @xmath7 is expressed by the compton scattering opacity with klein - nishina correction by @xcite @xmath8 @xmath9 with @xmath10 being the helium abundance , and @xmath11 .
it is exactly this temperature dependence of the opacity that is responsible for the excessive radiation flux , which appears to be super - eddington to the outer less hot layers of the atmosphere . in the framework of strong x - ray bursts
the following condition are usually satisfied : @xmath12 km @xmath13 , @xmath14 . putting @xmath15 results in a time of collapse of the order of several seconds , the observed time which a type i x - ray burst usually lasts . for evaluation of the time for photons to diffuse through the photosphere , we note that a number of scattering events is @xmath16 [ see , for example , @xcite ] , where @xmath17 is the total opacity of the photosphere , which is @xmath18 .
the time for a photon to escape is @xmath19 this indicates that the hydrodynamic structure develops at least ten times slower than the photons diffusing time through the photosphere .
although these time scales can become comparable in cases of greatly extended atmospheres , generally steady - state approximation is acceptable .
the problem of mass loss as a result of radiatively driven wind was formulated by ehs . for convenience of the reader
we summarize all equations important for the derivations in the following sections and refer the reader to ehs paper for details .
the system of equations describing steady - state outflow in spherical symmetry consists of a well known euler ( radial momentum conservation ) equation @xmath20 the mass conservation law @xmath21 the averaged radiation transport equation in the diffusion approximation @xmath22 and the entropy equation @xmath23 where @xmath24 and @xmath25 are , respectively , the pressure , the density , the temperature , the specific entropy , and the diffusive energy flux flowing through a shell at @xmath26 . the outflowing gas is taken to be ideal .
dimensionless coordinate @xmath27 , which is the ratio of the radiation pressure @xmath28 , to the gas pressure @xmath29 , is introduced by @xmath30 where @xmath31 and @xmath32 are the mean molecular weight and the mass of proton , respectively .
then the other thermodynamic quantities are expressed in terms of @xmath27 and @xmath33 as @xmath34 @xmath35 @xmath36,\ ] ] @xmath37 where @xmath38 is the specific enthalpy .
integrals of equation ( [ hydro5 ] ) and ( [ hydro4 ] ) give energy flow and mass flux correspondingly @xmath39 @xmath40 to make two more integrations , which can not be performed analytically , the constancy of @xmath41 , which stands for the integral of equation ( [ hydro6 ] ) , over the relevant layers is assumed . in ehs
this assumption is confirmed numerically .
we can also justified by the following consideration . at the near - eddington regime
the radiation pressure @xmath42 is much greater than the pressure of gas almost everywhere except for the innermost layers adjacent to the helium - burning zone . neglecting the gas pressure in equation ( [ hydro4 ] ) and multiplying it by @xmath43 we get @xmath44 here we moved first two terms of ( [ hydro4 ] ) to the right hand side and used equation ( [ hydro6 ] ) to express third term by @xmath45 . for inner and intermediate layers of the atmosphere
the last term in ( [ hydro14_1 ] ) is negligible and equation reduces to @xmath46 .
this term can become considerably large for the outermost layers where @xmath47 must exceed @xmath48 .
this also is in agreement with observations of x - ray bursts from which super - eddington luminosities are inferred . for the sake of analytical consideration we consider @xmath41 to be constant throughout the whole atmosphere , and the third integral
is @xmath49 replacing @xmath25 of equation ( [ hydro7 ] ) with equation ( [ hydro15 ] ) , the fourth integral is obtained as @xmath50 boundary conditions need to be imposed at the bottom and the outer boundary to determine the four integration constants @xmath51 , @xmath52 and to obtain a specific solution . at the bottom of the atmosphere close to the helium - burning zone
there should be a point where gas and radiation pressure are equal . as another important numerical result ehs showed that near the neutron star surface the temperature and radius profiles level off with respect to @xmath27 , so there is always a point where @xmath53 and @xmath54 is well approximated by the radius of the neutron star @xmath55 .
however , @xmath56 can not be considered as a real temperature of helium - burning shell at the bottom of the star surface because thermonuclear processes are not included in the model .
rigorous account of helium - burning ntl shows that temperature of burning shells vary in small range of values . to obtain the outer boundary condition the concept of sonic point
is used . for the solution to be steady - state and to have finite terminal velocity it should pass sonic point where @xmath57 @xmath58 where @xmath59 is a quantity related to the ratio of the energy flux to the mass flux ( see formula 22 , below ) . in ehs
this formula contains a typo .
we give a proper derivation of this form for @xmath60 in appendix c. t94 has shown how the treatment of the hydrodynamic problem can be reduced to a cauchy problem with the boundary condition determined at the sonic point .
this treatment provides a high - accuracy method of obtaining the hydrodynamic solution .
the crucial point is to relate the position of the sonic point with the values of the velocity and the thermodynamic quantities before solving the set of appropriate hydrodynamic equations .
the profile of the expanded envelope is then obtained as a result of the integration of a single first - order ordinary differential equation ( ode ) from the sonic point inward up to the neutron star surface . for completeness we present the details of this approach .
at the bottom of the atmosphere the potential energy per unit mass of the gas , @xmath61 , is significantly greater than the kinetic energy , @xmath62 , and enthalpy .
therefore , by ignoring these terms in equation ( [ hydro14 ] ) , we obtain the value of the mass flux @xmath63 the inner boundary condition ( [ inbcond ] ) , the integral ( [ hydro16 ] ) , and equation ( [ hydro11 ] ) for entropy can be used to find the temperature distribution with respect to @xmath27 @xmath64,\ ] ] and @xmath65 the condition at the sonic point ( [ hydro21 ] ) allows us to find the constant @xmath66 @xmath67 combining the mass and energy conservation laws ( [ hydro13 ] ) and ( [ hydro14 ] ) , the specific enthalpy and density equations ( [ hydro10 ] ) and ( [ hydro12 ] ) and eliminating the radial coordinate @xmath26 between equations ( [ hydro13 ] ) and ( [ hydro14 ] ) yield the following dependence of the velocity derivative @xmath68 with respect to @xmath27 @xmath69.\ ] ] derivation of equation ( [ hydro26 ] ) is given in appendix d. by imposing boundary conditions at the bottom of the extended envelope ( at the neutron star surface ) and at the sonic point , we can determine the four integration constants necessary to obtain a specific solution .
one can note the obvious fact that the bottom of the envelope can not serve as a starting point of integration of equation ( [ hydro26 ] ) as long as @xmath70 , which introduces uncertainty .
fortunately , we can calculate parameters at the sonic point in the framework of our problem description by solving a nonlinear algebraic equation , which involves only @xmath71 , the ratio of the radiation pressure @xmath28 to the gas pressure @xmath29 at that point .
substitution of the radial coordinate @xmath72 and velocity @xmath73 from the definition of the sonic point position ( [ hydro21 ] ) , and the sonic point density , @xmath74 from equation ( [ hydro10 ] ) , we find @xmath75 and the expression for the temperature given in equation ( [ hydro24 ] ) into the mass conservation law ( [ hydro13 ] ) , after some algebra give an equation for the value of @xmath76 @xmath77^{2/3}\frac{t_{b}}{0.149\,(8 - 5y_{he})^{5/3}y_s y_s^{1/\lambda+2/3}}\right\}+1.\ ] ] here @xmath78 and @xmath79 are the neutron star surface radius and temperature in units of @xmath80 cm and @xmath81 kelvin respectively .
since @xmath60 is expressed in terms of @xmath71 ( eq . [ hydro22 ] ) , equation ( [ hydro27 ] ) can be solved to determine the value of @xmath71 .
knowledge of @xmath71 can then , by substitution in equation ( [ hydro24 ] ) , yield the value of the temperature at the sonic point @xmath82 and then @xmath73 from equation ( [ hydro21 ] ) .
it is now possible to relate @xmath73 to @xmath82 , @xmath56 , @xmath4 , @xmath83 , thus obtaining the analytical expression for the various dynamical quantities at the sonic point in terms of the values of the parameters associated with the boundary conditions . to obtain the solution of the hydrodynamical problem for a particular set of input parameters we use a standard matlab / octave package function minimizators and ode solvers
the radiation field of x - ray burst atmosphere may be described by the diffusion equation , written in spherical geometry , with the kompaneets s energy operator ( see t94 ) : @xmath84 @xmath85 where @xmath86 is a dimensionless frequency , @xmath87 being the effective temperature ; @xmath88 and @xmath89 are the coefficients of free - free absorption and thompson scattering , respectively , whose ratio is given by @xcite @xmath90 with @xmath91 here @xmath92 is the gaunt factor @xcite @xmath93 @xmath94 is the macdonald function , and @xmath95 denotes the free - fall acceleration onto the neutron star surface , in units of @xmath96 cm s@xmath97 . we combine equation ( [ rad1 ] ) with the outer boundary condition of zero energy inflow @xmath98 and the condition of equilibrium blackbody spectrum at the bottom of the photosphere , which is represented in a dimensionless form as @xmath99 we will make use of the temperature equation , which is obtained by integration of ( [ rad1 ] ) over frequency .
the opacity operator vanishes as a result of the total flux conservation with respect to optical depth , leaving us with @xmath100 @xmath101.\ ] ] in the condition of the extended photosphere of x - ray bursts , the density usually is very low and the left - hand side of the last equation can be neglected reducing the last equation to the formula for temperature @xmath102 we will use the last relation to produce a corrected temperature profile for the photosphere where it departs significantly from that given by the hydrodynamic solution .
hydrodynamic profiles calculated in section 2 show that during the expansion stage in the vicinity of the sonic point @xmath103 . considering this relation to be true throughout the entire envelope , according to the mass conservation law , we can write @xmath104 where @xmath105 is the mass loss rate and @xmath73 is the velocity of gas at the sonic point . in this case
opacity can be expressed as @xmath106 noting that in this case @xmath107 we can rewrite the radiation transfer equation in the form @xmath108 the boundary conditions are given by @xmath109 at the inner boundary of photosphere , and @xmath110 at the sonic surface .
the ratio @xmath111 can be written in the form @xmath112 @xmath113 where @xmath114 , and @xmath115 .
stated in this way the problem of radiative transfer allows an analytical approach .
the solution of the radiative transfer equation ( [ rad8 ] ) is @xmath116,\ ] ] where @xmath117 is the modified bessel function of the first kind , and @xmath118 details of derivation of this formula are given in appendix a. the next step is to find the color temperature and to determine the thermalization depth @xmath119 where the boundary condition ( [ rad9 ] ) is valid . for saturated comptonization
, the occupation number behaves in accordance with bose - einstein photon distribution @xmath121 which might be described as a diluted blackbody spectrum or as a diluted wien distribution . at
first we evaluate the color temperature assuming a blackbody spectral shape .
we look for the solution of the form @xmath122 the solution , which is described in detail in appendix b , gives for @xmath123 @xmath124^{7/4}\right).\ ] ] as long as @xmath125 for @xmath126 , there is radiation equilibrium for optical depths deeper than the photospheric envelope .
the temperature equation in the zone @xmath127 reads @xmath128 where @xmath129 and @xmath130 ( @xmath131 ) is the optical depth coordinate at the outer boundary of the expanded atmosphere , @xmath132 ( see eq .
35 ) . this equation can be rewritten as follows @xmath133 neglecting @xmath130 with respect to @xmath134 and making use of taylor expansion ( [ b6 ] ) of @xmath123 we get a constant value of the temperature @xmath135 using the notation @xmath136 formula ( [ rad11 ] ) becomes @xmath137 where @xmath138 while we can write for @xmath119 @xmath139^{-4/7}=\left[\frac{6}{49}p^2d\right]^{-2/7}=\frac{t}{t_0}\left[\frac{6}{49}p^2\right]^{-2/7}d_0^{-2/7}.\ ] ] substituting it into ( [ rad15 ] ) we get for @xmath140 @xmath141^{-2/7 } \frac{d_0^{-2/7}}{\tau_{ns}^{1/2}}.\ ] ] assuming the same electron number density as in ( [ rad6 ] ) we than express opacity at the neutron star surface @xmath142 in the form @xmath143 if we use the dependence of input parameters @xmath95 and @xmath105 on @xmath144 and @xmath10 , the next useful equations for the color ratio @xmath140 , color temperature @xmath145 , and thermalization depth @xmath119 are found : @xmath146 @xmath147 @xmath148 here @xmath149 is the sonic point velocity in units of @xmath150 cm / s .
these relations present the final results of our analytical approach .
there is still a lack of completeness due to the presence of @xmath151 and @xmath73 in the left - hand sides of this system of equations .
parameter @xmath151 and sonic point velocity are not independent parameters of the problem , but at this point they can not be inferred from further analytical consideration .
fortunately , these quantities can be quite well approximated by a power dependence from @xmath152 and @xmath10 , which is done in the next chapter .
to confirm the validity of our analytical approach and to examine the behavior of @xmath151 and @xmath73 in dependence of different input parameters of the problem , we perform numerical modeling of the steady - state radiative transfer process .
the whole procedure consisted of three steps .
first , for particular model of neutron star , i. e. for a given mass and radius , we obtain a set of model atmospheres for a chosen set of bottom temperatures .
these solutions provide us with runs of thermodynamical and hydrodynamical profiles , sonic point characteristics , masses of the extended envelopes and their loss rates .
second , we solve radiative transfer equation ( [ rad1 ] ) on each model atmosphere obtained with the relaxation method ( e.g. press et al .
1992 ) on an energy - opacity grid using logarithmic scale on both dimensions .
the energy range included 500 grid points . the number of grid points in opacity varied between 100 and 300 .
the opacity domain included the range @xmath153 , where @xmath154 is opacity at the sonic point and @xmath155 was taken large enough to meet safely inequality @xmath156 .
we used the mixed outer boundary condition ( [ rad3 ] ) .
spectrum at the inner bottom of the photosphere was taken a pure black body @xmath157 .
numerical calculations of frequency - dependent radiation field consisted of two runs of our relaxation code .
the first run was performed on temperature continuum , obtained from the hydrodynamical solution ( see section 2.5 ) .
then we calculated a spectral temperature profile using formulae ( [ rad5 ] ) which exhibits a quite expected behavior . at some region
this corrected profile departs from initial temperature profile and levels off at some constant value in absolute agreement with analytic result of section 3.2 .
it is also in a qualitative agreement with the ntl self - consisted calculation of radiation driven wind structure of x - ray burster .
a second run is performed on the corrected profile to get more reliable spectrum shape . at the final third step we compared analytical and numerical solutions .
the sonic point provided a natural position to match numerical and analytical solutions . combining the sonic point parameters , calculated through the solution of equation ( [ hydro27 ] ) and using relation ( [ rad7 ] ) we get for the opacity at the sonic point @xmath158 we calculated and and plotted fluxes given by both methods at the sonic point .
a particular value of parameter @xmath151 for analytic model was obtained by matching value of @xmath145 and corrected level of numerically achieved value of photospheric temperature .
we obtained results for approximately 150 different sets of values @xmath159 and @xmath10 .
examples of numerical calculations of spectra for different neutron star models and fitting them with analytical shapes are presented in figures 1 and 2 .
analytical and numerical shapes match quite well in the wide range of neutron star surface temperatures and both show two distinctive features of outgoing spectrum of expansion stage : diluted black - body like high - frequency component and power law soft excess at the lower part .
dependence of the sonic point opacity presented by ( [ num1 ] ) describes correctly the dilution process indicating the assumption of atmosphere structure adopted at the analytical model is correct .
tables 1 and 2 , which summarize results for two different neutron star models , are given in order to compare our results with more rigorous calculations ( ntl ) .
taking mass - loss rate as an input parameter , ntl obtained profiles of different quantities throughout the whole atmosphere .
they argue that the temperature of burning shell is maintained around @xmath2 k for all models .
the temperature of photons departs appreciably from the temperature of ambient matter above photospheric radius and stays practically constant indicating that radiation becomes essentially decoupled from expanded media .
we changed the bottom temperature in a wide range of values and inferred the mass - loss rate , the mass of envelope , etc .
our results are in qualitative agreement with ntl . the crucial physical parameters which define the main spectral signatures are the photospheric radius @xmath160 and its temperature @xmath145 .
runs of the atmospheric profiles obtained by both approaches are quite similar , although @xmath161 in ntl results is usually @xmath162 greater than in our models .
this difference is explainable .
we match isothermal levels given by numerical and analytical calculations and define the obtained value as a photospheric temperature .
this is the lowest estimation , because the temperature profile starts to grow before the bottom of the photosphere is reached .
ntl define @xmath161 as a matter temperature at @xmath160 .
a temperature level calculated at the thermalization depth @xmath119 should compensate the considered difference .
the difference in density profiles , which can achieve a factor of two , will affect the spectrum only in the soft part ( @xmath163 kev ) where the normalization of the power law component can be changed .
this fact does not diminish the validity of our results .
the soft component of the spectrum can be represented as an independent fitting shape with a normalization included as an additional fitting parameter .
this matter is not crucial at the moment due to the restricted spectral capabilities of current x - ray observing facilities .
one can also notice a quick decrease of the envelope mass , and point out a wide variation of @xmath56 .
this discrepancy can be explained by differences in model formulations .
specifically , ntl included helium - burning shells into the model and put the inner boundary condition on the `` real '' neutron star surface while our model stops where radiation and gas pressures are equal ( @xmath164 ) , which is close but still outside of the helium - burning shell . in our approach , part of the bottom of the atmosphere
is left out .
in fact , the lower the mass - loss rate , the greater the portion of mass missing beyond the point where @xmath164 .
this is clearly seen from the tables .
the temperature at the bottom can be considered as an `` effective '' instead of the real temperature of helium - burning zone . as we have already point out that one needs to know the dependencies of @xmath73 and @xmath151 on input parameters to complete the analytical description and thus to employ these results to the fitting of observational x - ray spectra .
analysis of @xmath73 and @xmath151 runs show that @xmath165 and @xmath166 are linear functions of @xmath167 , @xmath168 , @xmath169 and @xmath170 .
we combine all experiments and fit @xmath149 and @xmath151 with a model @xmath171 by the least squares method to get @xmath172 @xmath173 with maximum errors of parameters less than 1% .
the ranges of parameters included in fitting are 0.3 - 7.0 for @xmath79 , 0.6 - 2.0 for @xmath78 , 0.8 - 2.7 for @xmath174 and 0.3 - 1.0 for @xmath10 .
these results can be used to substitute @xmath151 and @xmath149 in equations ( [ rad20])-([rad22 ] ) .
now we have consistent system of equations , which should yield x - ray spectrum of burster in the form of function of only input physical parameters , i. e. neutron star mass , radius , surface temperature and elemental abundance .
the fact that spectra obtained are a black - body like almost everywhere except for small values of energies allows us to proceed with simplification of the formula ( [ rad12 ] ) .
first we note that due to ( [ rad17 ] ) and smallness of @xmath175 @xmath176 for the soft part of spectrum . here
@xmath177 , @xmath178 and @xmath179 are defined in formulae and correspondingly . because @xmath180 is large for small values @xmath175 we can use approximation of the modified bessel function of the second kind for large arguments @xmath181 and rewrite equation ( [ rad12 ] ) as follows @xmath182=\ ] ] @xmath183,\ ] ] where @xmath184 here we rewrite the dilution factor in terms of opacity using relation ( [ rad13 ] ) .
clearly , the second term in the parenthesis of formula ( [ fin11 ] ) is significant only where @xmath185 becomes small ( @xmath175 becomes large ) and the spectrum shape `` adjusts '' to the black - body component . in turn , the first term of equation ( [ fin11 ] ) represents the power law component of the lower part of spectrum with the slope 6/7 , which can be shown by simple similarity ( see also t94 ) @xmath186 another important advantage of this term is that it vanishes for large values of @xmath175 .
this fact gives us opportunity to construct convenient and accurate formula for observational spectra fitting .
we drop the second term in equation ( [ fin11 ] ) and adjust to the diluted black - body shape by means of quadratic power combination as follows @xmath187^{1/2}.\ ] ] comparison of shapes given by formula ( [ fin12 ] ) with exact solution ( [ rad12 ] ) shows that they deviate from each other by less than 2% which is more than acceptable in contemporary astrophysical observational data analysis . using the explicit form of @xmath185 and the form of outgoing flux equation ( [ fin12 ] )
can be rewritten in the form @xmath188^{8/7}\right\}^{1/2}.\ ] ] equation ( [ num1 ] ) yields useful relationship for the dilution coefficient in the manner similar to ( [ rad20])-([rad22 ] ) @xmath189 substituting results of parameter fitting ( [ num2])-([num3 ] ) , we get @xmath190 @xmath191 here , again , @xmath78 , @xmath79 and @xmath174 are the neutron star radius , surface temperature and mass in units of @xmath80 cm , @xmath81 kelvin and solar mass respectively .
now we have provided our spectrum profile ( [ fin13 ] ) with expressions for parameter @xmath151 and the dilution factor .
these three formulae constitute the final analytical results of this paper .
the model and derivations presented above assume that plasma consists of fully ionized hydrogen and helium . in reality
, this assumption can be too simplistic .
for instance , in the case of the recently discovered super - burst @xcite , a sufficient fraction of material should be represented by heavier elements .
these long and powerful bursts are also considered to be due to the nuclear runaway burning in the carbon `` ocean '' under the neutron star surface . in this section
we discuss how our model can be adjusted for study of this phenomenon .
the approach as a whole does not change , but some formulae have to be modified in order to account for the different plasma composition .
first , we note that for plasma which consists of a single ionized element , we have for the mean molecular weight @xmath192 and for the electron number density @xmath193 where @xmath194 and @xmath195 are the atomic weight and the atomic number of the corresponding element . in the general case of heterogenous elements ,
each represented by weight abundance @xmath196 , we write @xmath197 @xmath198 in the hydrodynamic part of this study , these modifications will affect only the form of the terms and factors containing @xmath10 . in the radiation transfer section
, the form of @xmath111 will require more careful treatment .
according to @xcite , the free - free absorption coefficient is @xmath199 where @xmath200 in the case of a hydrogen - helium plasma this factor is conveniently represented just by gas density , i. e. @xmath201 which yields the equation . in general , one should use expressions and to find the correct form of @xmath111 relevant to the specific chemical composition . to be more instructive we conduct such a modification for the case when the plasma has a substantial carbon fraction . using , and , we write for hydrogen - helium - carbon gas @xmath202 @xmath203 and @xmath204
correspondingly , in all formulae the factor @xmath205 will be replaced by @xmath206 and @xmath207 by @xmath208 .
additionally , the right - hand side of equation has to be multiplied by the factor of @xmath209 .
clearly , this modification will add the fifth free parameter @xmath210 to the model .
using the general methodology outlined in this paper one should be able to produce solutions for the parameter @xmath151 and the dilution factor . the problem which can arise from the inclusion of heavy elements is the possibility for heavy ions to be only partly ionized .
the ionization degree can also vary throughout the atmosphere . because full ionization and constancy of the gas s chemical composition are the basic assumptions of the adopted approach , we can not explicitly include the effect of ionization in our model .
instead , it can be accounted for in a manner similar to our temperature profile correction .
first , the approximate atmospheric profiles can be obtained by assuming full ionization .
then , the ionization degree can be calculated by solving the saha equation , and using this solution as a zero - order approximation of the atmospheric temperature and the electron number density profiles .
finally , one should proceed by solving the hydrodynamic problem , in which the partial ionization of heavy elements is taken into account . for reasons mentioned above ,
it is also a problem to include the proper physics for the transport of heavy nuclei to the outer layers .
two major processes can contribute to this element flow .
bulk motion mixing should dominate in the convection zone close to the bottom of the atmosphere . in the outer layers , a strong radiative push should govern the process , because of the large resonance cross - sections of the heavy elements .
the general problem of heavy ions mixing is quite difficult and requires a rigorous approach , which is out of scope of this paper . as far as the boundary conditions are concerned , modeling of carbon nuclear flashes will require higher bottom temperatures .
temperature of the carbon burning zone is argued to be about @xmath211 k [ see @xcite ] , which is close to the upper boundary for the bottom temperature @xmath56 used in our calculations .
no peculiarities of the approach were detected in the case of very high bottom temperatures .
extremely high temperatures will require the correct form of the opacity coefficient @xmath7 @xcite , instead of equation , which represents a simplified formula for @xmath7 in the case of modest temperatures .
another important issue is the correct accounting for the line emission of heavy elements which is detected in the spectral analysis of super - bursts .
@xcite argued that this phenomena is due to reflection from the accretion disk during the burst .
one can estimate the disk heating time by using the standard shakura - sunyaev accretion disk model @xcite and the fact that approximately 10% of the burst luminosity is absorbed by the inner part of the disk @xcite .
_ simple estimates give a timescale of less than a second assuming a mass - accretion rate of the order of eddington or less for the disk accretion regime , and a burst luminosity greater than 5% of eddington , which is detected during several thousand seconds of observation of the super - burst in 4u 1820 - 30 _ consequently , the observed spectral feature of the k@xmath212 line should rather be generated in the burst atmosphere than in the disk .
the disk gains the temperature of the x - ray radiation very quickly .
unfortunately , the origin and behavior of the spectral line features still remain unexplained .
the authors plan to inlcude the spectral line effect in the relaxation method , in order to calculate the line emission during the x - ray burst and to compare this with the observed spectra .
relativistic effects are usually negligible during the strong x - ray burst due to the significant radial expansion and the fact that the outgoing spectrum formation occurs at the outer layers of the atmosphere .
general relativistic effects become important at the contraction stage when the extended envelope recedes close to the ns surface [ see @xcite ] .
haberl & titarchuk ( 1995 ) applied the full general relativity approach for a derivation of ns mass - radius relation in 4u 1820 - 30 using exosat observations and the t94 model .
this paper follows a common idea of the last decade to fit observational and numerical spectra with some model , mostly black - body shapes , to obtain spectral softening / hardening factors @xcite .
we improve this technique in several ways .
we use for fitting more realistic non - blackbody spectral profile , which accounts for the observed power law soft excess of x - ray burster spectra .
the temperature profile is corrected by solving the temperature equation .
the existence of the isothermal photosphere during x - ray bursts is confirmed numerically and analytically .
finally , we analytically obtain the multiplicative ( dilution ) factor which is not a parameter of fitting anymore but self - consistently incorporated in the model .
we show how the theoretical study of radiatively driven wind phenomenon can produce useful techniques for analyzing observational data .
it can fulfill the needs of new emerging branches of observational x - ray astronomy such as a very promising discovery of super - bursts @xcite , which exhibit photospheric expansion and spectral modifications relevant to extended atmospheres .
we present the analytical theory of strong x - ray bursts , which include effects of comptonization and free - free absorption .
partly presented in some earlier publications , this area of the study of the x - ray burst spectral formation was lacking a detailed and self - consistent account .
we use numerical simulation to validate our analytical theory and to link our solution to energy axes .
we show how this information can be extracted from spectral data .
we provide the analytical expression for the x - ray burst spectral shape , which depends only upon input physical parameters of the problem : neutron star mass , radius , surface temperature and elemental abundance .
expressions for color ratios and dilution coefficient are also given .
authors thank peter becker for valuable comments and suggestions which improved the paper .
we are greatful to menas kafatos for encouragement and to center for earth science and space research ( gmu ) for the support of this research .
we appreciate the thorough analysis of the presented work by the referee .
ebisuzaki , t. , 1987 , , 39 , 287 ebisuzaki , t. , hanawa , t. , & sugimoto , d. 1983 , , 35 , 17 ( ehs ) greene , j. , 1959 , , 130 , 693 grindlay , j. e. , & heise , j. 1975 , , no.2879 haberl , f. , & titarchuk , l.g . 1995 , a&a , 299 , 414 kuulkers , e. , homan , j. , van der klis m. , levin , w.h.g .
, mendez , m. 2001 , accepted to lapidus , i. , i. 1991 , , 377 , l93 lapidus , i. , i. , sunyaev , r. , a. , 1985 , , 217 , 291 lewin , w. h. g.,van paradijs , j. , & taam , r. e. 1993 , space sci .
, 62 , 223 london , r. , a. , taam , r. , e. , howard , h. , w. 1986 , , 306 , 170 nobili , l. , turolla , r. , & zamperi , l. 1991 , , 383 , 250 nobili , l. , turolla , r. , & lapidus , i. i. 1994 , , 433 , 276 ( ntl ) paczynski , b. , 1983 , , 267 , 315 press , w. h. , teulkolsky , s.a . , vettering , w.t . , &
flannery b.p .
1992 numerical recipes + ( cambridge : university press ) rybicki , g. b. , & lightman a.p .
1979 radiative processes in astrophysics + ( new york : wiley ) thorne , k. s. 1981 , , 194 , 439 titarchukk , l. g. 1994 , , 429 , 340 ( t94 ) strohmayer e. t. , brown f. e. , ( 2001 ) , in press ( astro - ph/0108420 ) .
shakura , n. , i. , sunyaev , r. , a. , 1973 , , 24 , 337
we look for the solution of equation ( [ rad8 ] ) in the form @xmath213 the basic idea is to separate the high - frequency ( diluted black body ) and the low - frequency @xmath214 parts of spectrum , where different physical processes dominate .
kompaneets operator @xmath215 acting upon black body shape vanishes and we neglect @xmath216 , which allows us to get the solution of radiative transfer problem analytically . at this point
@xmath119 is a parameter of the problem .
the algorithm of determination of @xmath119 will be described separately . substituting ( [ a1 ] ) into ( [ rad8 ] ) we found for @xmath214 @xmath217,\ ] ] with a boundary condition @xmath218 the solution satisfying this condition
is presented by @xmath219 where @xmath220 and @xmath221 is the wronskian @xmath222 thus the product @xmath223 functions @xmath224 and @xmath225 are @xmath226 @xmath227 where @xmath228 and @xmath229 are modified bessel function of the first and the second types respectively .
the function @xmath230 in ( [ a2 ] ) is the right hand side of equation ( a2 ) , namely @xmath231.\ ] ] we introduce a new variable @xmath232 @xmath233 and rewrite solution ( [ a2 ] ) as @xmath234dt.\ ] ] using the properties of modified bessel functions @xmath235 we evaluate the integrals in ( [ a3 ] ) @xmath236 @xmath237 finally @xmath238 takes the form @xmath239=\ ] ] @xmath240.\ ] ] the last formula is a solution of equation ( [ rad8 ] ) . we can simplify this form by noting that we are interested in the solution in the outer layers of atmosphere ( emergent spectrum ) where @xmath241 and @xmath242 , and we can use the asymptotic form for small arguments @xmath243 making this substitution and putting the result into expression for @xmath244 we find that second term in @xmath214 cancels with diluted blackbody term in @xmath244 , which takes the form @xmath245.\ ] ]
substituting relation ( [ rad14 ] ) into the equation of radiative diffusion , multiplying it by @xmath246 and integrating over the energy range from @xmath247 to @xmath248 we get @xmath249\int_{x_*}^\infty\frac{x^2}{e^x-1 } \frac{\alpha_{ff}}{\alpha_t}\,dx,\ ] ] where integrals can be approximated as @xmath250 and , noting that @xmath251 we obtain @xmath252 here we used the fact that @xmath253 . the equation for @xmath123
gets the form @xmath254=\tilde{d}\tau^{3/2}[r(\tau)-1].\ ] ] boundary conditions for this equation are @xmath255 a general solution of equation ( 29 ) is @xmath256 where @xmath257 . in derivation of this formula
we take into account a well known theorem from ode theory that general solution of inhomogeneous ode is the sum of a general solution of the corresponding homogeneous ode and some particular solution of the inhomogeneous ode , which is chosen equals to unity in our case .
the second boundary condition and the fact that @xmath258 leave only @xmath259 nonzero , and the first boundary condition gives us the value for @xmath259 , namely @xmath260 where we put @xmath261 .
then @xmath123 reduces to @xmath262.\ ] ] taylor series expansion of @xmath263 over @xmath264 yields for @xmath123 useful relation @xmath265
we can rewrite the partial derivative in ( [ hydro21 ] ) using the obvious relation @xmath266 differentiating the equation of state ( [ hydro9 ] ) we obtain derivatives of pressure @xmath267 and differentiating ( [ hydro24 ] ) with respect to @xmath268 we get @xmath269 from the other hand differentiation of ( [ hydro10 ] ) gives us @xmath270 combination with the previous equation it yields : @xmath271 @xmath272 now , combining all found derivatives we have @xmath273t\ ] ] or in a more compact form @xmath274t,\ ] ] which is , in fact , a sonic point condition ( [ hydro21 ] ) .
we derive an expression for the derivative of velocity @xmath275 through @xmath276 and @xmath27 .
we substitute the temperature profile found in ( [ hydro24 ] ) to ( [ hydro6 ] ) to obtain @xmath268 as a function of @xmath27 @xmath277.\ ] ] using this expression for @xmath278 and equation ( [ hydro13 ] ) we get @xmath279v^{-\frac{1}{2}}.\ ] ] then we get derivatives @xmath280,\ ] ] @xmath281 differentiation of ( [ hydro24 ] ) also gives us @xmath282\left(-\frac{1}{\lambda y}-\frac{4}{\lambda}\right)=-\frac{t}{\lambda}\left(4+\frac{1}{y}\right).\ ] ] by a combination of all these derivatives we obtain @xmath283^{-1}.\ ] ] substitution of it into ( [ hydro6 ] ) yields @xmath284}.\ ] ] but from ( [ hydro14 ] ) we also have @xmath285 equating the last two expressions for @xmath47 we finally find @xmath275 as @xmath286.\ ] ] here @xmath78 and @xmath79 represent neutron star radius and the bottom temperature of atmosphere in terms of @xmath80 cm and @xmath81 k correspondingly . [email protected]@.lcccc 7.0 & 0.197 & 0.099 & 45.1 & 1&56 & 93.9 & 173&3 & 0.21 & 1.31 & 57.9 & 4.16 + 6.5 & 0.208 & 0.105 & 44.7 & 1&64 & 87.2 & 128&9 & 0.22 & 1.39 & 51.6 & 3.71 + 6.0 & 0.221 & 0.111 & 44.1 & 1&75 & 80.5 & 93&6 & 0.24 & 1.48 & 45.6 & 3.28 + 5.5 & 0.236 & 0.118 & 43.3 & 1&87 & 73.8 & 66&1 & 0.25 & 1.58 & 39.9 & 2.88 + 5.0 & 0.252 & 0.127 & 42.4 & 2&02 & 67.1 & 45&1 & 0.27 & 1.70 & 34.5 & 2.50 + 4.5 & 0.272 & 0.137 & 41.5 & 2&20 & 60.4 & 29&6 & 0.29 & 1.84 & 29.4 & 2.14 + 4.0 & 0.295 & 0.148 & 40.3 & 2&42 & 53.7 & 18&5 & 0.32 & 2.01 & 24.6 & 1.80 + 3.5 & 0.324 & 0.163 & 39.0 & 2&71 & 46.9 & 10&9 & 0.35 & 2.22 & 20.1 & 1.48 + 3.0 & 0.359 & 0.180 & 37.4 & 3&09 & 40.2 & 5&86 & 0.39 & 2.50 & 16.0 & 1.19 + 2.5 & 0.406 & 0.204 & 35.6 & 3&60 & 33.5 & 2&83 & 0.44 & 2.85 & 12.2 & 0.92 + 2.0 & 0.470 & 0.236 & 33.4 & 4&34 & 26.8 & 1&16 & 0.51 & 3.36 & 8.83 & 0.67 + 1.75 & 0.510 & 0.257 & 32.2 & 4&85 & 23.5 & 0&68 & 0.55 & 3.69 & 7.29 & 0.56 + 1.5 & 0.565 & 0.284 & 30.8 & 5&53 & 20.1 & 0&37 & 0.61 & 4.12 & 5.86 & 0.45 + 1.25 & 0.634 & 0.318 & 29.2 & 6&44 & 16.8 & 0&179 & 0.68 & 4.69 & 4.53 & 0.36 + 1.1 & 0.686 & 0.344 & 28.0 & 7&18 & 14.8 & 0&108 & 0.74 & 5.12 & 3.80 & 0.30 + 1.0 & 0.727 & 0.365 & 27.2 & 7&78 & 13.4 & 0&074 & 0.78 & 5.47 & 3.33 & 0.27 + 0.9 & 0.775 & 0.389 & 26.3 & 8&51 & 12.1 & 0&049 & 0.84 & 5.87 & 2.89 & 0.23 + 0.8 & 0.831 & 0.417 & 25.3 & 9&40 & 10.7 & 0&031 & 0.90 & 6.36 & 2.46 & 0.20 + 0.7 & 0.898 & 0.451 & 24.3 & 10&5 & 9.4 & 0&018 & 0.97 & 6.95 & 2.06 & 0.17 + 0.6 & 0.981 & 0.492 & 23.0 & 12&0 & 8.0 & 0&010 & 1.06 & 7.69 & 1.68 & 0.14 + 0.5 & 1.086 & 0.545 & 21.6 & 14&0 & 6.7 & 0&005 & 1.17 & 8.65 & 1.33 & 0.11 + 0.4 & 1.224 & 0.614 & 19.9 & 17&0 & 5.4 & 0&002 & 1.32 & 9.95 & 1.01 & 0.09 + 0.3 & 1.417 & 0.711 & 17.8 & 21&9 & 4.0 &
0&001 & 1.53 & 11.8 & 0.71 & 0.07 + [email protected]@[email protected]@[email protected] 7.0 & 0.248 & 0.110 & 45.1 & 2&12 & 56&2 & 115&8 & 0.242 & 1&80 & 53&4 & 3.53 + 6.5 & 0.261 & 0.116 & 44.5 & 2&24 & 52&2 & 86&1 & 0.256 & 1&90 & 47&7 & 3.15 + 6.0 & 0.277 & 0.123 & 43.6 & 2&40 & 48&2 & 62&6 & 0.271 & 2&02 & 42&2 & 2.80 + 5.5 & 0.294 & 0.131 & 42.6 & 2&58 & 44&1 & 44&2 & 0.288 & 2&16 & 36&9 & 2.46 + 5.0 & 0.313 & 0.139 & 41.5 & 2&80 & 40&1 & 30&2 & 0.308 & 2&32 & 31&9 & 2.15 + 4.5 & 0.337 & 0.150 & 40.3 & 3&07 & 36&1 & 19&8 & 0.331 & 2&52 & 27&2 & 1.84 + 4.0 & 0.364 & 0.162 & 39.0 & 3&40 & 32&1 & 12&4 & 0.359 & 2&75 & 22&8 & 1.56 + 3.5 & 0.398 & 0.177 & 37.4 & 3&82 & 28&1 & 7&27 & 0.394 & 3&04 & 18&7 & 1.29 + 3.0 & 0.440 & 0.196 & 35.7 & 4&36 & 24&1 & 3&93 & 0.437 & 3&41 & 14&9 & 1.04 + 2.5 & 0.495 & 0.220 & 33.8 & 5&12 & 20&1 & 1&90 & 0.494 & 3&89 & 11&4 & 0.81 + 2.0 & 0.570 & 0.254 & 31.5 & 6&21 & 16&1 & 0&78 & 0.572 & 4&58 & 8&24 & 0.59 + 1.75 & 0.620 & 0.276 & 30.2 & 6&97 & 14&0 & 0&46 & 0.623 & 5&03 & 6&80 & 0.50 + 1.5 & 0.682 & 0.304 & 28.7 & 7&97 & 12&0 & 0&25 & 0.687 & 5&61 & 5&47 & 0.40 + 1.25 & 0.763 & 0.339 & 27.0 & 9&33 & 10&0 & 0&121 & 0.770 & 6&38 & 4&24 & 0.32 + 1.1 & 0.824 & 0.366 & 25.9 & 10&4 & 8&83 & 0&073 & 0.833 & 6&96 & 3&56 & 0.27 + 1.0 & 0.872 & 0.388 & 25.1 & 11&3 & 8&03 & 0&050 & 0.883 & 7&43 & 3&12 & 0.24 + 0.9 & 0.928 & 0.413 & 24.2 & 12&4 & 7&22 & 0&033 & 0.940 & 7&98 & 2&71 & 0.21 + 0.8 & 0.993 & 0.442 & 23.3 & 13&7 & 6&42 & 0&021 & 1.008 & 8&63 & 2&31 & 0.18 + 0.7 & 1.072 & 0.477 & 22.2 & 15&4 & 5&62 & 0&012 & 1.089 & 9&43 & 1&94 & 0.15 + 0.6 & 1.169 & 0.520 & 21.0 & 17&6 & 4&82 & 0&007 & 1.188 & 10&4 & 1&59 & 0.13 + 0.5 & 1.292 & 0.575 & 19.7 & 20&6 & 4&01 & 0&003 & 1.313 & 11&7 & 1&26 & 0.10 + 0.4 & 1.455 & 0.647 & 18.1 & 24&9 & 3&21 & 0&001 & 1.479 & 13&5 & 0&95 & 0.08 + 0.3 & 1.682 & 0.749 & 16.2 & 32&0 & 2&41 & 0&0005 & 1.710 & 16&0 & 0&67 & 0.06 +
|
we present an analytical theory of thermonuclear x - ray burst atmosphere structure .
newtonian gravity and diffusion approximation are assumed .
hydrodynamic and thermodynamic profiles are obtained as a numerical solution of the cauchy problem for the first - order ordinary differential equation .
we further elaborate a combined approach to the radiative transfer problem which yields the spectrum of the expansion stage of x - ray bursts in analytical form where comptonization and free - free absorption - emission processes are accounted for and @xmath0 opacity dependence is assumed .
relaxation method on an energy opacity grid is used to simulate radiative diffusion process in order to match analytical form of spectrum , which contains free parameter , to energy axis .
numerical and analytical results show high similarity .
all spectra consist of a power - law soft component and diluted black - body hard tail .
we derive simple approximation formulae usable for mass - radius determination by observational spectra fitting .
|
it is not easy to describe the hadroproduction of @xmath3 mesons within a perturbative qcd framework .
the problem is that , due to the @xmath10 quantum numbers , it is not possible to directly form the colourless @xmath3 meson by gluon - gluon fusion . the simplest possibility is to produce a colour - octet quark - antiquark pair ( @xmath11 ) and then to emit an additional gluon , which carries away the colour , as shown in fig .
[ fig : pqcd]a .
this is often referred to as the colour - singlet mechanism ( csm ) .
however the corresponding cross section is suppressed by the small qcd coupling @xmath12 , and by the additional phase space factor associated with the extra gluon emission . as a result
the lo qcd prediction @xcite is found to be about an order of magnitude lower than the experimental yield of @xmath3 mesons .
an alternative and more phenomenological approach is provided by the colour evaporation model ( cem ) @xcite . here
the quarkonium production cross section is an ( a priori unknown ) fraction of the @xmath13 heavy quark cross section integrated over the @xmath14 invariant mass up to the threshold for producing a pair of the lightest heavy flavour mesons .
there are no constraints on the colour or spin of the @xmath13 pair , the transition from colour - octet @xmath15 to colour - singlet quarkonium is assumed to take place by the ` evaporation ' of soft gluons .
the fraction of @xmath13 pairs that materialise as a particular quarkonium state , @xmath16 for example , is assumed to be universal and is adjusted to give the best fit to existing data . cross section . ] despite its phenomenological success , the cem has no firm theoretical foundation . in practice
one would not expect the evaporation of soft gluons to take place independently of the particular collision environment , and there is no reason why such soft interactions would not modify the quarkonium production properties , and in particular its collision energy dependence .
this is precisely what happens in the theoretically more rigorous formalism for quarkonium production that we introduce below .
= 0.6 the currently most popular and widely used description of quarkonium production is based on a nonrelativistic qcd ( nrqcd ) effective field theory approach @xcite , which retains features of both the csm and cem . here
@xmath13 pairs are produced via a hard partonic ( short - distance ) subprocess , in both colour - singlet and colour - octet states , and non - perturbative universal matrix elements describe the ( large - distance ) transition of the @xmath13 pair into particular quarkonium states .. the cross section of quarkonium @xmath17 production is written in the schematic form @xmath18)~\langle o^h[n ] \rangle , \label{oh}\ ] ] where @xmath19 denotes the set of colour and angular momentum quantum numbers of the @xmath13 pair , and @xmath20 is the cross section of the @xmath13 pair production in a hard subprocess .
the non - perturbative transition from the @xmath13 state @xmath19 into the quarkonium state @xmath17 is described by a long - distance matrix element @xmath21 \rangle$ ] .
these matrix elements are taken as parameters to be determined by fits to experimental data . in this way , it is possible to compensate the low value of the hard subprocess cross section , @xmath22)$ ] , by the large fitted value of the matrix element @xmath21 \rangle$ ] . despite some phenomenological success , a detailed proof of the factorisation formula ( [ oh ] ) is lacking , and in particular it is expected to break down at small values of the quarkonium transverse momentum . predictions for the _ total _ quarkonium cross section in the nrqcd approach must therefore be treated with caution .
since the mass of the @xmath3 meson is not particularly small , it would be desirable to be able to describe the @xmath23 transition within a perturbative qcd framework .
in other words , we would like to consider explicitly an extra gluon exchange ( which was _ hidden _ in the value of the non - perturbative matrix element @xmath21 \rangle$ ] in the case of the colour - octet mechanism , i.e. where the @xmath3 meson is formed from the colour - octet @xmath13 pair after some non - perturbative interaction described by the @xmath21 \rangle$ ] matrix element . the corresponding lowest order in @xmath12 diagrams
are shown in fig .
[ fig : h ] . in comparison with the @xmath24 amplitude , the contributions of fig .
[ fig : h ] contain an extra loop factor with a small coupling @xmath12 .
here we investigate the possibility that this suppression is compensated by the large number of graphs where the additional ( third ) gluon , needed to form the @xmath3 , is absorbed by different parton - spectators . as viewed from the collinear approximation , the amplitude shown in fig .
[ fig : pqcd]b , and fig .
[ fig : h ] , corresponds to the next - to - next - to - leading order ( nnlo ) contribution to the cross section ( as compared with lo @xmath25 production ) .
however the second @xmath26-channel gluon may be absorbed by _ any _ parton spectator ; that is the amplitude is enhanced by the parton multiplicity , @xmath27 .
therefore it may be considered as the lo amplitude in the bfkl approach .
the aim of this paper is to evaluate the numerical size of this enhanced ` nnlo ' contribution , and to see if it can remove the large discrepancy between the perturbative qcd prediction and the inelastic @xmath3 hadroproduction data .
= 0.8 in section 2 we calculate the amplitude of the process shown in fig .
[ fig : h ] using the non - relativistic @xmath28 vertex that was proposed in ref . @xcite . due to the non - relativistic wave function of the @xmath3 ,
there is practically no integration over the quark loop in figs .
[ fig : h]a , b .
indeed , the @xmath3 vertex ( i.e. the @xmath3 wave function integrated over the relative momenta of the charm quarks ) , together with the two nearest @xmath29-quark propagators is g ( + m ) _ , where the index @xmath30 corresponds to the @xmath3 polarization vector ; and @xmath31 and @xmath32 are the 4-momentum and the mass of the @xmath29-quark ( @xmath33 is the momentum of the @xmath3 ) . the constant @xmath34 may be expressed in terms of the electronic width @xmath35 of the @xmath36 decay @xmath37 where @xmath38 is the mass of the @xmath3 meson , and the electromagnetic coupling @xmath39 . in section 3
we compute the total cross section of _ prompt _ inelastic @xmath3 hadroproduction at collider energies .
that is , in that section , we neglect the additional @xmath3 yield coming from @xmath40-quark or @xmath41 decays .
we find that the lo result agrees with the tevatron data rather well .
the transverse momentum distributions of prompt @xmath3 , @xmath42 and the @xmath43 production are presented in sections 4 and 5 , together with a brief discussion of the @xmath3 polarization .
alternative possibilities to produce the @xmath3 are considered in section 6 .
one such mechanism is to create a colour - octet @xmath44 pair , which then transforms to a colour singlet by rescattering via gluon exchange .
another is the associative production of a @xmath3 and @xmath44 pair .
the yield from this latter possibility is small .
however the first mechanism , where two gluons in a symmetric colour - octet @xmath26-channel state , @xmath2 , belong to two different pomerons , may dominate at asymptotically large energies .
the energy and rapidity dependence of @xmath3 ( and @xmath4 ) production is given in section 7 .
section 8 contains our conclusions .
we compute the matrix element of the hard subprocess from the diagrams of fig . [
fig : h ] in the lo collinear approximation .
thus incoming particles , with momenta @xmath45 and @xmath46 , are taken to be on - mass - shell , transversely - polarized gluons .
the calculation of the amplitude shown in fig .
[ fig : h]a is similar to the computation of the amplitude for diffractive @xmath3 photoproduction @xcite . due to the non - relativistic nature of the @xmath3 wave function ,
the difference between the quark momenta @xmath47 and @xmath48 is very small , that is @xmath49 .
therefore , following ref .
@xcite , we may take @xmath50 , and include the integration over the quark loop momentum in the @xmath51 coupling @xmath34 , which is normalized to the width of @xmath36 decay , see ( [ gee ] ) .
we expect that the main contribution will come from the region where the rapidity difference between the final gluon @xmath52 and @xmath3 meson is rather large ; that is @xmath53 . in this limit , the amplitude @xmath54 , corresponding to the diagram fig .
1a , is a^a= 8d^abcdl^2_t g(4_s)^5/2 , [ aa ] where @xmath55 and @xmath56 are respectively the polarization vectors of the @xmath3 and the gluon with momentum @xmath45 , and @xmath57 is the momentum transferred through the pair of @xmath26-channel gluons .
@xmath58 is the colour factor , where the indices @xmath59 are the colours of the two incoming gluons and the final gluon .
bearing in mind possible confinement effects ( and to avoid the logarithmic infrared singularity as @xmath60 ) we introduce a cutoff ( or effective gluon mass ) @xmath61 in the denominator of ( [ aa ] ) .
since the two @xmath26-channel gluons are in a symmetric colour - octet state , the amplitude @xmath62 has positive signature .
therefore it has a small real part , @xmath63 . to obtain the whole amplitude , @xmath62
, we have to add the contribution of the diagram shown in fig .
[ fig : h]b , which is of the form a^b= -8d^abcdl^2_t g(4_s)^5/2 , [ ab ] where the minus sign reflects the ` negative ' colour charge of the antiquark .
we also have to account for the contributions where the gluon @xmath45 couples to the antiquark , and not to the quark .
that is , we must include the graphs with the opposite direction of the ` arrows ' in the quark loop .
thus the differential cross section is @xmath64 where the total amplitude @xmath65 . for transverse ( with respect to the @xmath66 plane ) @xmath3 inelastic production the trace tr@xmath67=s^2m(e\cdot\varepsilon)$ ] ,
while tr@xmath67=s^2m(e\cdot q_t)/m$ ] when the vector @xmath55 corresponds to a longitudinally polarized @xmath3 meson .
thus , after averaging over the incoming gluon transverse polarizations , @xmath68 , the ratio of longitudinal ( @xmath69 ) and transverse ( @xmath70 ) cross sections becomes @xmath71
in the lo collinear approximation , the cross section of inelastic prompt @xmath3 production is of the form @xmath72 where @xmath73 is the centre - of - mass rapidity of the @xmath3 meson , @xmath74 , @xmath75 is the initial hadron - hadron energy squared and @xmath76 are the densities of gluons in the incoming hadrons ( @xmath77 ) . for a fixed rapidity of the @xmath3 meson , the integral over @xmath78 is equivalent to the integration over the mass @xmath79 of the @xmath80 system , where @xmath81 .
the hard subprocess cross section @xmath20 includes the contribution of the diagrams fig .
[ fig : h ] , where the single gluon @xmath45 comes from the beam side , plus the ` inverse ' contribution , in which the single gluon comes from the target .
there is no interference between the original ( fig .
[ fig : h ] ) and the ` inverse ' amplitudes , due to the different colour structure of single and double gluon exchange .
note that single and double gluon exchange correspond , respectively , to antisymmetric and symmetric colour octets . in the small @xmath82 region , and at relatively low scales ,
the gluon distribution behaves like @xmath83 , where the power @xmath84 , while the ` hard ' subprocess cross section @xmath20 does not depend on @xmath85 for @xmath86 .
thus the integral over @xmath78 takes the form @xmath87 .
the main contribution comes from the lowest values of @xmath88 , which correspond to @xmath89 , where @xmath90 .
however the essential interval of integration available at collider energies ( @xmath91 ) is quite large .
this large integration interval partly compensates for the small loop ( @xmath12 ) factor in the @xmath92 amplitude obtained from the diagrams of fig .
[ fig : h ] .
moreover for an inelastic process we have to allow for the emission of additional ( secondary ) @xmath85-channel gluons from the symmetric octet @xmath93 ( see fig .
[ fig : s ] ) .
this leads to a power growth of the ` hard ' subprocess cross section , @xmath94 , as the function of subenergy @xmath79 .
the power behaviour of @xmath95 is driven by the intercept of the amplitude with the four @xmath26-channel gluons ( the so - called quarteton ) . within the leading logarithm approximation ( @xmath96 )
this intercept was evaluated in @xcite .
it was shown that numerically the value of @xmath97 is , to within 10% , equal to that for the usual bfkl pomeron ( two - gluon ) exchange . as a consequence we have to extend the @xmath98 integration over the whole kinematically available rapidity interval @xmath99 .
then the prompt @xmath3 cross section becomes @xmath100 with @xmath101 , and @xmath102 .
we take the same scale , @xmath103 , for both gluons ; different scales are equivalent to a nlo correction to the lo formula .
we introduce the factor @xmath104 to exclude the contribution from the graphs in which the third gluon couples to partons with large @xmath105 in the proton fragmentation regions , that is to allow for the fact that parton densities at large @xmath82 are kinematically suppressed .
induces a nll correction ; that is , it is equivalent to an @xmath12 term ( without a @xmath106 factor ) in the bfkl amplitude . indeed , in the bfkl expansion , the contribution of fig .
[ fig : pqcd]b to the cross section behaves as an @xmath107 term as compared to that of fig .
[ fig : pqcd]a . thus the variation of @xmath108 corresponds to an @xmath12 term without a @xmath109 factor .
this does not mean , of course , that the variation of @xmath108 reproduces the _ whole _ nll contribution to the bfkl intercept . ]
= 0.5 the formulae in eqs .
( [ sigf ] ) and ( [ dsdt ] ) , together with the amplitudes of ( [ aa],[ab ] ) , enable the inelastic @xmath3 cross section to be predicted . in this way , we find that the cross section at the tevatron energy @xmath110 tev is , with the corresponding lo ( one loop ) qcd coupling @xmath12 with @xmath111 mev . ] ( |y|<0.6 ) 2.7 ( pqcd estimate ) [ sig ] in the central rapidity interval @xmath112 ( integrated over the transverse momentum @xmath113 ) . here
we have taken , which occurs in the amplitudes , was first estimated in ref .
@xcite , where it was introduced to describe the photon spectra in @xmath114 decay .
a recent evaluation , together with a collection of previous determinations , is presented in table 15 of ref .
@xcite . based on this table
we choose @xmath115 gev , with a possible uncertainity covered by the interval @xmath116 gev . ]
@xmath115 gev .
this prediction , ( [ sig ] ) , is to be compared with the lastest experimental measurement @xcite ( |y|<0.6 ) = 4.1 ^+0.6_-0.5 ( cdf experiment ) [ expt ] the uncertainties of the prediction , ( [ sig ] ) , are as follows .
\(i ) the choice of @xmath61 . if , instead of 0.8 gev , we were to take @xmath61 equal to 0.5 or 1 gev then @xmath117 becomes 4.0 or 2.0 @xmath118b respectively .
\(ii ) the choice of the factorization and renormalization scales . for different scales
@xmath119 of @xmath120 and @xmath121 , we obtain @xmath122 and @xmath123b , respectively .
\(iii ) an unknown k - factor to account for nlo and higher pqcd corrections .
\(iv ) the uncertainty in the incoming gluon distribution , which is not well constrained at low @xmath82 and rather low scales .
\(v ) the choice of the cut - off @xmath108 .
the variation of the value of @xmath108 plays the role of nll corrections in the bfkl approach .
+ taking all these into account , the expected accuracy of the prediction is about a factor of 2@xmath1243 in either direction or even worse .
unfortunately we can not use the amplitudes of eqs .
( [ aa],[ab ] ) directly to calculate the @xmath113 distribution of the produced @xmath3 mesons . in the case of diffractive @xmath3 photoproduction
it is known that the interference between the two lowest - order diagrams , fig .
[ fig : h]a and fig .
[ fig : h]b , leads to a dip in the @xmath113 distribution .
indeed , the differential cross section goes to zero at @xmath125 @xcite .
however this dip disappears after one includes the leading logarithmic ( @xmath126 ) correction @xcite .
the problem is that we can not resum the analogous corrections in our case , since the corresponding ` quarteton ' eigenfunctions ( coming from 4 gluons in the @xmath26-channel ) are not yet known .
therefore we consider a very simple parametrization @xmath127 motivated by dimensional counting , which accounts for the dimension of @xmath128 .
again we take @xmath129 . the distribution ( [ qt ] ) is normalised by equating its @xmath130 integral to that of ( [ sigf ] ) . in this way
the effective gluon mass , @xmath61 , enters the calculation .
the result , shown in fig . [ fig : qt ] , is in reasonable agreement with the tevatron data @xcite . in comparison with the colour octet model , where the @xmath3 is created in the fragmentation of a gluon jet and the expected distribution
is @xmath131 , in the case of the fig . [
fig : s ] subprocess the @xmath113 distribution ( [ qt ] ) at large @xmath113 is steeper , @xmath132 .
this is similar to the distribution in the colour - singlet model @xcite .
however in contrast with the colour - singlet mechanism ( fig . [ fig : pqcd]a ) here the hadronic transverse energy flow , which compensates the @xmath113 of the @xmath3 meson , is distributed over a larger rapidity interval in the form of a larger number of gluonic minijets ( see fig .
[ fig : s ] ) .
= 0.8 according to ( [ dslt ] ) we expect the @xmath3 mesons to be transversely polarized at small @xmath113 , and longitudinally polarized at large @xmath113 , that is @xmath133 .
may also be affected by the contribution of the longitudinally polarised incoming gluons ( @xmath45 in fig .
[ fig : h]a ) , see for example ref .
@xcite . ] at present the data are only available in the interval of @xmath134 gev , where the observed @xmath3 is approximately unpolarized .
the parameter @xmath135 of ref .
@xcite corresponds to a small transverse polarization at lower @xmath113 values .
however , as @xmath113 increases @xmath136 changes sign and for @xmath137 gev clearly indicates longitudinal polarization of the @xmath3 .
this is qualitatively consistent with our expectations .
in contrast , colour - octet models of prompt @xmath3 production lead to transverse polarization at large @xmath113 @xcite .
the above formalism can be applied to the production of other quarkonium states by simply changing the mass and width of the @xmath10 heavy @xmath13 resonance .
thus , without any free parameters , we can predict the cross section for inclusive prompt quarkonium production .
the results for @xmath42 and for the upsilon states are compared with the tevatron data @xcite in fig . [
fig : qt ] and fig .
[ fig : up ] respectively .
we use the effective gluon mass @xmath61 and the scale that were appropriate for the description of the @xmath3 data . in this way
we remove a large part of the uncertainty in the prediction for @xmath42 .
the good agreement with the @xmath42 data should therefore be regarded as support for our perturbative qcd approach .
= 0.8 the comparison of the prompt @xmath4 predictions with the tevatron data , shown in fig .
[ fig : up ] , is complicated , however , since only half of the total @xmath139 yield arises from prompt production @xcite .
moreover , the @xmath140 states have a large branching fraction of radiative decays to @xmath141 ; about twice as large as those to @xmath139 . bearing in mind these complications , and the uncertainties in the predictions , the agreement with the data is better than may have been expected .
the predicted cross sections for prompt @xmath142 central production at the tevatron energy , @xmath143 = 1.96 tev , are .
d / dy_y=0 = 2.2 , 0.6 ; 40 , 12 , 9 nb , ( tevatron ) respectively ; and correspondingly at the lhc energy , @xmath143 = 14 tev , .
d / dy_y=0 = 8.1 , 2.5 ; 310 , 100 , 80 nb . (
besides the diagrams shown in fig .
[ fig : s ] , where two @xmath26-channel gluons in a symmetric colour octet state are placed rather close to each other in the impact parameter ( @xmath144 ) plane , there may also be a contribution from diagrams like those shown in fig .
[ fig : oct ] . this may be viewed as the production of a colour - octet @xmath44 pair that subsequently changes colour via rescattering . here
the two @xmath26-channel gluons in the amplitude belong to two different pomerons , that is to two different parton showers . a convenient way to calculate such a contribution is to use the agk cutting rules @xcite , that is to calculate diagram fig .
[ fig : oct]b and then use the relation @xmath145 strictly speaking , the two gluons in the lower ( left or right ) parts of fig .
[ fig : oct]b can not form a colour - singlet ( pomeron ) state .
on the other hand , these two gluons are in a colour _ symmetric _ state , and therefore we may use the agk relation ( [ agk ] ) to simplify the calculation .
= 0.8 the amplitude corresponding to fig .
[ fig : oct]b is very similar to that of fig .
[ fig : s ] .
the main difference is that now the density of the second @xmath26-channel gluon is given by an independent gluon distribution @xmath146 .
that is , we need the probability to find two gluons @xmath147 . on the other hand these two gluons are uniformly distributed over the whole transverse area occupied by a proton .
therefore the integration over the transverse momentum @xmath113 is limited by the ` elastic ' slope @xmath148 , which may be taken from the slope observed for diffractive @xmath3 photoproduction ; that is @xmath149 gev@xmath150 @xcite .
thus we can consider the lowest - order amplitudes given by eqs .
( [ aa],[ab ] ) at @xmath151 , and take the integral @xmath152 .
the factor @xmath153 , together with two logarithmic integrations , @xmath154 , in each amplitude is given by the first iteration of the dglap equation , that is @xmath155 , becomes the gluon distribution @xmath156 in the cross section .
next we have to account for an extra 1/2 in the colour factor , which is cancelled by the factor 2 in the agk relation ( [ agk ] ) .
thus , finally , we obtain @xmath157 .
\label{s4a}\ ] ] using mrst2001 lo gluons @xcite , we find that both in the gluon distribution and in the qcd coupling @xmath12 . ]
@xmath158b , at the tevatron energy @xmath159 tev .
this is smaller than the main contribution discussed in section 3 .
however at larger energies the ` two pomeron ' cross section will grow faster than the contribution of section 3 .
indeed @xmath160 in the central region of small @xmath73 , where @xmath161 , whereas the cross section ( [ sigf ] ) is proportional to @xmath162 only .
in particular , at the lhc energy @xmath163 tev we expect in the centre of the rapidity plateau @xmath164b , while the cross section ( [ sigf ] ) is @xmath165b . at first sight , the transverse momentum distribution of the @xmath3 for the contributions of fig . [
fig : oct ] should be peaked at low @xmath113 , namely @xmath166 .
this is true for the contribution of the diagram fig .
[ fig : oct]b , but not for the inelastic process , fig .
[ fig : oct]a .
the emission of intermediate gluons spreads out the @xmath3 distribution from the inelastic process , so that we expect a spectra @xmath167 analogous to ( [ qt ] ) .
we must of course take care of possible double counting .
note that the amplitudes shown in fig .
[ fig : oct]a and fig .
[ fig : s ] are very similar , and the gluon densities given by the global parton analyses do not distinguish the gluons coming from one or more parton showers .
therefore , to be conservative , in what follows we consider only the contribution of fig . [
fig : s ] . a possible contribution of fig .
[ fig : oct]a is well within the uncertainties of the lowest - order in @xmath12 calculations .
another possibility is to consider the production of a @xmath3 meson together with a @xmath44 pair , as shown in fig .
[ fig : cc ] . in this case
the expected hard cross section at the tevatron is about @xmath169 nb , leading to @xmath170 for inelastic hadroproduction , the contribution coming from this mechanism is not large , of the order of 1% .
however in @xmath171 annihilation , associative production is much more important , see for example ref .
indeed , in @xmath171 annihilation there is no suppression for the production of the first @xmath44-pair . on the other hand for time - like annihilation kinematics the third @xmath26-channel gluon ( @xmath172 in fig .
[ fig : h ] ) can couple to the nearest parton only .
thus we loose the enhancement caused by a large parton multiplicity , that is the factor @xmath173 in ( [ sigf ] ) . without this factor
the contributions of single ( fig .
[ fig : h ] ) and associative ( fig .
[ fig : cc ] ) production are expected to be of the same order ( in agreement with the measurements of the belle collaboration for @xmath171 annihilation @xcite ) .
= 0.5 note that if the inelastic @xmath3 hadroproduction were to originate from ` colour - octet ' dynamics , that is the main yield of @xmath3 were to come from gluon @xmath174 fragmentation @xcite , then we would expect the same ratio of ` direct ' to ` associative ' contributions in @xmath171 annihilation , as that for hadroproduction , contrary to our perturbative qcd predictions .
finally we have the possibility of non - prompt @xmath3 production .
experimentally it is observed that a fraction of @xmath3 mesons originate from @xmath176 and @xmath177 decays .
we do not discuss here the details of these production mechanisms . however in order to predict the _ total _ @xmath3 yield we estimate these contributions using the experimental data of refs .
@xcite and collinear factorization .
symbolically , the relation that we use is of the form production , despite the fact that it can not be formed by gluon - gluon fusion .
the dominant mechanism for @xmath178 production is the process with antisymmetric colour - octet @xmath179 exchange .
however , since this exchange corresponds to gluon reggeization , it has the same energy ( and @xmath82 ) dependence as the exchange of a single gluon .
] d / dy = x_1 g(x_1 ) x_2 g(x_2 ) , [ eq : fac ] where @xmath180 and the normalisation of @xmath20 is adjusted to fit the @xmath181 and @xmath175 data .
since the energy dependence of the cross section is driven simply by the @xmath82 dependence of the gluon densities , the experimental data at one fixed energy are sufficient to estimate these non - prompt contributions at other energies production in the forward hemisphere . therefore to normalize the @xmath3 cross section , we use ( [ eq : fac ] ) to calculate the integrated cross section . ] .
in comparison with the usual perturbative qcd colour - singlet ( fig . [ fig : pqcd]a ) and colour - evaporation mechanisms , where the energy dependence is essentially determined by the product of the gluon densities , see ( [ eq : fac ] ) , our perturbative contribution to the cross section ( fig . [
fig : pqcd]b ) is enhanced by an additional log@xmath75 factor , @xmath99 in ( [ sigf ] )
. therefore the energy dependence of these two contributions is different , as can be seen in fig .
[ fig : e1 ] . to demonstrate the uncertainty arising from the choice of the cut - off @xmath108 in the additional logarithm @xmath99 in the computation of the new perturbative contribution
, we show predictions for the two values , @xmath108 = 0.1 ( lower curve ) and 0.5 ( upper curve ) .
= 0.9 note that the amplitude of the new subprocess is calculated at lowest order in @xmath12 .
it is not unusual to have a next - to - leading order @xmath182 factor of the order of 2 for the production of relatively low - mass states , as in drell - yan production for example . on the other hand ,
we see no reason for a strong energy dependence of such a @xmath182 factor .
so , noting the good agreement with the tevatron data , we expect our predictions at the lhc energy to be reliable .
the energy behaviour of the total inelastic @xmath3 cross section in the fixed - target / isr to tevatron energy interval is shown in fig .
[ fig : e2 ] , together with the components from the following subprocesses : @xmath183 and @xmath184 .
we see that the cross section is slightly underestimated at the lower energies , where a contribution initiated by the @xmath185 subprocess , that is by secondary reggeons , may have some influence .
indeed , it was noted in ref .
@xcite that the difference between the @xmath3 production cross sections from @xmath186 and @xmath187 interactions indicates a noticeable @xmath185 contribution at the lower energies .
= 0.9 in fig .
[ fig : y ] we show predictions for the rapidity distributions of @xmath3 and @xmath139 production at both the tevatron ( 1.96 tev ) and lhc ) in the pseudorapidity ranges @xmath188 and @xmath189 respectively .
this gives non - zero acceptance for the @xmath3 down to @xmath190 , see figs .
60 and 63 of ref .
( 14 tev ) energies . in table [ table1 ] we compare our results for the total @xmath3 and @xmath139 cross sections at lhc with the predictions of the cem ( taken from ref .
note that our predicted cross sections are systematically larger because of the steeper @xmath191 dependence caused by the @xmath192 factor in eq .
( [ sigf ] ) .
@xmath193{0ex}{5ex } \\
\hline \sigma_{j/\psi}\mbox{(prompt ) } \ \
( \mu{\rm b } ) & 80 & 33 \\
\sigma_{j/\psi}\mbox{(total ) } \ \
( \mu{\rm b } ) & 150 & 54
\\ \sigma_{\upsilon(1s)}\mbox{(prompt ) } \ \
( \mu{\rm b } ) & 2.5 & 0.4 \\\hline \end{array}\ ] ] = 0.9
we have calculated the prompt hadroproduction of @xmath194 and @xmath43 states within a perturbative qcd framework , without any non - perturbative contributions ( such as occur explicitly in the nrqcd colour - octet model and implicitly in the cem ) .
recall that the original colour - singlet lo perturbative contribution ( based on the subprocess @xmath195 ) falls well short of the data .
however , here , we have studied another perturbative qcd contribution , which turns out to be dominant .
the basic subprocess is @xmath196 , see figs .
[ fig : h],[fig : s ] or fig .
[ fig : pqcd]b .
this contribution is enhanced , particularly at high energies , since the additional @xmath26-channel gluon can couple to a large number of parton spectators .
the uncertainties of such a computation are listed at the end of section 3 .
they are not small . however , with our natural choices of scale and of the effective gluon mass , we successfully describe the available high - energy rhic and tevatron @xmath3 data .
in addition , without any new parameters , we obtain an excellent description of the @xmath42 data , and even a satisfactory description of @xmath4 production . there is additional qualitative support for the @xmath196 mechanism coming from the measurement of the @xmath3 polarization at the tevatron .
this mechanism predicts a longitudinal polarization at large @xmath113 , in agreement with the data , whereas the colour - octet model leads to a transverse polarization @xcite .
we thank anatoli likhoded , lev lipatov and orlando villalobos - baillie for useful discussions . adm
thanks the leverhulme trust for an emeritus fellowship and mgr thanks the ippp at the university of durham for hospitality .
this work was supported by the uk particle physics and astronomy research council , by a royal society special project grant with the fsu , by grant rfbr 04 - 02 - 16073 and by the federal program of the russian ministry of industry , science and technology ss-1124.2003.2 .
h1 collaboration : s. aid et al .
, nucl . phys . *
b472 * ( 1996 ) 3 ; c. adloff et al . , eur .
j. * c10 * ( 1999 ) 373 ; phys . lett .
* b483 * ( 2000 ) 23 ; + zeus collaboration : j. breitweg et al .
, z.phys . *
c75 * ( 1997 ) 215 ; eur .
j. * c6 * ( 1999 ) 603 .
|
we consider the prompt hadroproduction of @xmath0 and the @xmath1 states caused by the fusion of a symmetric colour - octet state , @xmath2 , and an additional gluon .
the cross sections are calculated in leading - order perturbative qcd .
we find a considerable enhancement in comparison with previous perturbative qcd predictions . indeed
, the resulting cross sections are found to be consistent with the values measured at the tevatron and rhic , without the need to invoke non - perturbative ` colour - octet ' type of contributions .
ippp/04/58 + dcpt/04/116 + 17th september 2004 + * inelastic @xmath3 and @xmath4 hadroproduction * v.a .
khoze@xmath5 , a.d .
martin@xmath6 , m.g .
ryskin@xmath5 and w.j .
stirling@xmath7 + @xmath6 department of physics and institute for particle physics phenomenology , + university of durham , dh1 3le , uk + @xmath8 petersburg nuclear physics institute , gatchina , st .
petersburg , 188300 , russia + @xmath9 department of mathematical sciences , university of durham , dh1 3le , uk +
|
the presence of energetic particles in the universe has been know for over a century as cosmic rays ( crs ) and for a comparable time as the agents producing non - thermal electromagnetic radiation from long wave radio to gamma - rays . in spite of accumulation of considerable data on spectral and other characteristics of these particles ,
the exact mechanisms of their production remain controversial . although the possible scenarios of acceleration have been narrowed down , there are many uncertainties about the details of individual mechanisms . nowadays
the agents commonly used for acceleration of particles in astrophysical plasmas can be classified in three categories , namely static electric fields ( parallel to magnetic fields ) , shocks and turbulence . as we will try to show there are several lines of argument indicating that turbulence plays an important role in all these scenarios .
in addition , because of the large values of ordinary and magnetic reynolds numbers , most flows in astrophysical plasmas are expected to give rise to turbulence . the generation and evolution ( cascade and damping ) of plasma turbulence in astrophysical sources is an important aspect of particle acceleration that will be dealt with in other papers of this proceedings .
similarly , electric field and shock accelerations will be discussed by other authors in this proceedings as well . in this paper
we discuss particle acceleration by turbulence or plasma waves which is commonly referred to as * stochastic acceleration ( sa for short)*. fermi ( 1949 ) was the first author to propose sa as a model for production of crs , whereby charged particles of velocity @xmath2 , scattering with a rate @xmath3 in random collisions with moving magnetized clouds of average speed @xmath4 , gain energy at a rate @xmath5 mainly because energy gaining head on collisions are more numerous than energy losing trailing ones . nowadays this class of models are often called _ second order fermi _ process . soon after , several authors proposed plasma waves or magnetohydrodynamics ( mhd ) turbulence as the scattering agents ( see e.g. sturrock 1966 ; kulsrud & ferarrri 1971 , and references therein ) . in a later paper fermi ( 1954 )
proposed acceleration of particles scattering back and forth between two ends of a contracting magnetic bottle , where the particles gain energy at every scattering so that the acceleration rate is equal to @xmath6 , that is linearly with the velocity ratio , hence the name _ first order fermi_. for particle velocities @xmath7 this is a much faster rate .
it is also well known that a particle crossing a convergent flow , e.g. a shock with velocity @xmath8 , gains momentum @xmath9 . for this reason acceleration by shocks
is also referred to as first order fermi acceleration .
however , as described below , this a somewhat of a misnomer because the actual is proportional to the square of the velocity ratio .
ever since late 1970 s when several authors ( krymsky 1977 ; axford 1978 ; bell 1978 ; blandford & ostriker 1978 ) demonstrated that a simple version of this process can reproduce the observed cr spectrum , shock acceleration has been the most commonly invoked process .
however , as we will discuss below , in the last couple of decades there has been a renewed interest in sa by turbulence , specially for electrons in radiation producing astrophysical sources from solar flares to clusters of galaxies . in the next section we review the relative merits of different acceleration model and in 3 we give the general formalism used in the sa and other models . in
4 we will describe application of the sa model to solar flares .
a brief summary and conclusions are presented in 5 .
we are interested in comparing the various ways of production of energetic particles_starting from a relatively cool " background plasma usually having a thermal or maxwellian distribution _ with density @xmath10 and temperature @xmath11 .
clearly this must be the first stage of any acceleration process , where the coulomb collisions , with mean free path @xmath12 , may be an important , if not the dominant energy loss and particle scattering process .
in fact a thermal spectrum requires a high rate for these collisions involving both electrons and protons ( and heavier ions ) .
thus , the first hurdle that any acceleration mechanism must overcome is this loss process .
particles energized at this stage may escape the acceleration region with an energy dependent escape time @xmath13 favoring escape of the higher energy particles , thus resulting in a population of nonthermal particles , which are observed directly or through the radiation they produce .
the escaping particles may be re - accelerated by other mechanisms possibly in collisionless surroundings with size @xmath14 . on the other hand , in a
, i.e. when @xmath15 is larger than the dynamical timescale of the system then , in general , it is difficult to produce a substantial nonthermal tail . as shown in petrosian & east ( 2008 ) irrespective of the rate or energy dependence of the acceleration process a substantial , if not the bulk , of the energy input goes into heating of the plasma rather than producing a nonthermal electron tail .
the most commonly acceleration mechanisms used for analysis of astrophysical sources are the following : static electric fields * e * parallel to magnetic fields can accelerate a particle with charge @xmath16 and velocity @xmath17 with the energy gain rate of @xmath18 .
if we define the dreicer field @xmath19 v / cm for solar flare conditions ) that results in an energy gain of @xmath20 per mean free path , then the energy change over a distance @xmath21 is given by = * e*_dm_ec^2ktnl2.510 ^ 23 ^-2 .
[ efield ] thus , sub - dreicer fields can accelerate electrons to relativistic energies only if they extend over large column depths or @xmath22 . since the rate of energy gain per unit length is independent of particle mass ( because as defined @xmath23 ) , for acceleration of proton to relativistic regime ( @xmath24 ) we need a column depth of @xmath25 .
for example , in solar flare coronal loops , with @xmath26 and column depth @xmath27 , particles can be accelerated up to only 10 s of kev , far below the required 10 s of mev electrons or @xmath28 gev protons .
column depths and temperatures in astrophysical sources ( e.g. @xmath29 and @xmath30 k for intra cluster medium ; @xmath31 and @xmath32 k in typical galactic hii regions , etc ) are also not sufficient for acceleration of electrons or protons to relativistic energies .
super - dreicer fields can accelerate particles to higher energies but since now the acceleration rate is higher than coulomb energy loss rate this can lead to runaway particles and an unstable bump - in - tail distribution which will give rise to turbulence .
in addition , it is difficult to sustain large scale electric fields in a highly conducting ionized plasma unless the resistivity is anomalously high ( tsuneta 1985 ; holman 1985 ) .
after the pioneering work by speiser ( 1970 ) it was assumed that the electric fields induced by reconnection are the agents of acceleration ( see also litvinenko 1996 , 2003 ; la rosa et al .
2006 ) but recent particle - in - cell ( pic ) and mhd simulations of reconnection ( drake et al .
2006 ; see also cassak et al .
2006 ; zenitani & hoshino 2005 ) present a more complicated picture and show that turbulence may be an important ingredient .
_ thus , electric fields can not be the sole agent of acceleration , but they may produce turbulence , which can accelerate particles _ and possibly enhance the reconnection rate , as suggested by lazarian & vishniac ( 1999 ) . as mentioned above there are two types of fermi acceleration , first order in a shock and second order in a turbulent plasma . as also indicated above the former
has been invoked often in astrophysical situations primarily because it is believed to be a faster mechanism of acceleration and the environment surrounding a supernova shock seems well suited for production of cr protons .
however , there are many shortcomings in the original elegant models developed in late 1970 s for this mechanism .
since then a great deal of work has gone into the the development of this model and in addressing its shortcomings .
these include injection of seed particles , losses ( specially for electrons ) , and escape and nonlinear effects ( see drury 1983 ; blandford & eichler 1987 ; jones & ellison 1991 ; malkov & drury 2001 ; diamond & markov 2007 ; beresnyak et al .
more importantly , a _ shock by itself can not accelerate particles _ and requires scattering agents that can cause the repeated passages across the shock front , especially for a parallel shock with magnetic field parallel to flow velocity .
the most likely agent is turbulence and the rate of acceleration is governed again by the scattering rate by turbulence , and the acceleration rate , @xmath33 ( see below ) , is no longer first order in the velocity ratio .
although there are indications that magnetic field and turbulence may be generated by the upstream accelerated particles ( see e.g. bell 1978 ) , many details of the microphysics remain unsolved .
two stream ( or another plasma ) instability is a possible mechanism for generation of turbulence but there are indication that this may be suppressed in a turbulent medium ( yan & lazarian 2002 ) .
exact determination of the scattering rate requires knowledge of the intensity and spectrum of the turbulence which determine @xmath34 but are essentially unknown .
usually bohm diffusion is assumed .
second order fermi or sa process , on the other hand , occurs always at some level because turbulence in addition to scattering can also accelerate particles directly . as is well known , relativistic particles in weakly magnetized plasma ( e.g. that in a supernova shock ) are scattered at a faster rate than the rate of acceleration by turbulence , so that acceleration by a shock is deemed to be faster .
this , however , is not always the case , specially for acceleration of electrons in radiating sources . as we will discuss in more detail below , pryadko & petrosian ( 1997 ; pp97 ) showed that at low energies and/or in strongly magnetized plasmas the acceleration or energy diffusion rate by turbulence exceeds the scattering rate and therefore exceeds the acceleration rate by a shock .
thus , under these circumstances the main objection of slowness of sa does not apply . for example , in the case of solar flares hamilton & petrosian ( 1992 ) show that sa by a modest level of whistler waves can accelerate the background particles to high energies within the desired time ( see also miller and reames 1996 ) .
_ we can conclude then that irrespective of which process of acceleration is at work , turbulence always has a major role .
_ moreover , as we will see in the next section , in practice , mathematically , there is little difference between first and second order fermi acceleration ( see e.g. jones 1994 ) . there are , of course , other acceleration processes similar to those discussed above on the microphysics level but phenomenologically different .
one such process is that proposed by drake et al .
( 2006 ) occurring via the interactions of particles with the `` islands '' produced in their pic simulations .
another is the process proposed by fisk & glockler ( 2010 ) for acceleration in the solar wind .
these will be discussed in other sections of these proceedings .
interactions of particles with turbulence , which is a common ingredient of all acceleration processes , are dominated by many weak rather than few strong scattering events . in this case
the fokker - planck formalism provides the best description of the particle kinetics which can have different forms depending on circumstances .
the basic equations here are well known and have been described in many papers in the past . in this section
we briefly review these equations and point out two important features of plasma wave - particle interactions not broadly known or acknowledged .
these features have important effects on the relative importance of first and second order fermi acceleration , and on the relative acceleration rates of electrons vs protons , @xmath0he vs @xmath1he and other ions .
most astrophysical plasmas are strongly magnetized so that the gyro - radius of particles ( with mass @xmath35 ) , @xmath36 , is much smaller than the scale of the spatial variation of the field . in this case particles
are tied to the magnetic field lines and instead of dealing with temporal evolution of the distribution of energetic particles in six dimensions ( 3 space , 3 momentum ) one deals with the gyro - phase averaged distribution which depends only on three variables ; spatial coordinate @xmath37 along the field lines , the momentum @xmath38 , the pitch angle or its cosine @xmath39 , and of course also time . and @xmath39 , or particle energy ( in @xmath40 units ) @xmath41 and @xmath39 , instead of @xmath42 and @xmath43 . ]
then , the evolution of the particle distribution , @xmath44 , can be described by the fokker - planck equation as particles undergo stochastic scattering and acceleration by interaction with plasma turbulence ( with diffusion coefficients @xmath45 and @xmath46 ) and suffer losses ( with rate @xmath47 ) due to other interactions with the plasma particles and fields .
the particles may also gain energy in presence of shocks or large scale electric fields ( with the rate @xmath48 ) : @xmath49+{\partial \over\partial \mu}\left[d_{\mu\mu}{\partial f\over\partial \mu } + d_{\mu p}{\partial f\over\partial p}\right ] -{1\over p^2}{\partial\over\partial p}(p^2\dot{p}f ) + \dot{s}. \label{fpeq}\end{aligned}\ ] ] here @xmath50 is the net momentum change rate and @xmath51 is a source term , which could be the background thermal plasma or some injected spectrum of particles . the effect of the magnetic field convergence or divergence can be accounted for by adding @xmath52 to the right hand side . and if there are large scale flows with velocity @xmath4 and spatial gradient @xmath53 along the field lines , e.g. around a shock front , their effects can be accounted for by adding the term ( p^3f ) - usf [ flowdiv ] to the right hand side . in general , this complete equation is rarely used to determine the particle distribution @xmath54 . instead the following approximations are used to make it more tractable .
_ pitch - angle isotropy : _ if the pitch angle diffusion rate is high so that the scattering time @xmath55 is shorter than all other time scales ( @xmath56 , etc , where @xmath21 is the size of the interaction region ) , then the pitch angle distribution of the particles will be nearly isotropic .
if we define f(t , s , p ) _
-1 ^ 1 df(t , s , p , ) , q(t , s , p ) _
-1 ^ 1 d(t , s , p , ) , then the kinetic equation simplifies to the _ diffusion - convection equation _ ( see , e.g. kirk et al . 1988 ; dung & petrosian 1994 ) @xmath57 where @xmath58 implies pitch angle averaged values and the three transport coefficients are related to the diffusion coefficients as @xmath59 this approximation is generally valid for high energy particles interacting with alfven waves in a plasma with relatively low magnetization , i.e. alfven velocity @xmath60 , where @xmath61 is the magnetic field strength and @xmath62 is the gas density , .
] so that r_1d_pp/(p^2d_)=/~(v_a / v)^21 .
[ r1 ] however , as mentioned above , pp97 have shown that , in certain situations ( low energies and high magnetic fields ) , the energy diffusion rate @xmath63 may exceed the pitch angel diffusion rate @xmath64 so that @xmath65 , invalidating the above approximation . in this case
, the momentum diffusion is the dominant term and the kinetic equation ( [ fpeq ] ) again simplifies to @xmath66 where now one must include the @xmath39 dependence of all terms and of the distribution function
. in general , there will be significant acceleration only if the energy diffusion and acceleration times are shorter than the crossing time @xmath67 .
in addition , if the scattering time @xmath68 , though now longer than momentum diffusion time , is also shorter than @xmath69 , and/or if the other coefficients ( most importantly those related to acceleration ) vary slowly with @xmath39 , then the particle distribution will again be nearly isotropic can also contribute to the scattering rate , specially at low energies , and help to isotropize the pitch angle distribution . ] and we can use pitch angle averaged distribution @xmath70 and coefficients d_pp= 12_-1 ^ 1 dd_pp ( ) , [ avdpp ] and ( similarly)@xmath71 , @xmath72 and @xmath73 . as described in the caption of figure [ figr1 ] ( right ) use of these averaged quantities is justified .
in addition , now the term @xmath74 should be replaced by the spatial diffusion term @xmath75 .
thus , we can ignore @xmath39 dependences , in which case the two equations are almost identical : they have different spatial dependence terms ( e.g. terms involving @xmath76 ) , but more importantly , have two different momentum diffusion forms ; one for the case @xmath77 in equation ( [ kpp ] ) and the second for @xmath78 given by equation ( [ avdpp ] ) . _
spatial homogeneity : _ a second simplification , which applies to both @xmath77 and @xmath79 cases , can be used if the acceleration region is homogeneous ( i.e. @xmath80 ) , or if one deals with a spatially unresolved source where one is interested in spatially integrated equations . in this case
it is convenient to define @xmath81 $ ] and replace the spatial diffusion term ( or the advection term in equation [ [ kelowe ] ] ) plus other terms involving @xmath82 by an escape term with an escape time @xmath83 defined by dv 4p^2s(_ssfs - f1p^2p(p^3_sp))= n(t , e)(e ) .
[ tesc ] then we obtain the well - known equation [ homog1 ] n t = e
( d_eee n ) - e ( [ a(e ) - e_l ] n ) - n t_esc + , with t_esc=+^2 , [ keall ] where @xmath84 is the energy loss rate .
this clearly is an approximation with the primary assumption being that the transport coefficients vary slowly spatially , e.g. @xmath85 ) , @xmath86 , etc .
note that here @xmath87 and @xmath88 represent the rates of injection and escape of particles in and out of the acceleration site , and that @xmath89 as defined can account for the spatial diffusion term in equation ( [ kehighe ] ) when @xmath90 and the advection term in equation ( [ kelowe ] ) when of @xmath91 .
this equation is fairly general and can handle different acceleration scenarios .
for example for the sa by turbulence the energy diffusion and the direct acceleration coefficients are related as ) is written as @xmath92 in which case we need to add @xmath93 to the right hand side of the direct acceleration term @xmath94 . ] d_ee = v^2|d_pp and a(e)=(d_ee / e)[(2 ^ 2 -1)/(^2 + ) ] , where @xmath95 is equal to @xmath96 for the isotropic case ( equation [ kehighe ] ) and is equal to @xmath97 for equation ( [ kelowe ] ) . as stressed above turbulence
is present in all acceleration scenarios so these energy diffusion and the direct acceleration rates are the minimum rates .
however if there are other diffusion or acceleration mechanisms we should add their contribution .
for example , if the acceleration volume contains a converging flow ( as in a shock ) with velocity @xmath4 , then there will be additional direct acceleration rate @xmath98 . should be added inside the large parenthesis in equation ( [ tesc ] ) . ] or , at low energies and for a high density plasma , as mentioned above , one should add the effects of pitch angle and energy diffusion due to coulomb collisions .
finally , for completeness we mention that for most astrophysical situations the primary contribution to the loss rate for energetic electrons comes from coulomb collisions at low energies ( which as stated at the outset are essential in establishing a thermal distribution ) and synchrotron and inverse compton at high energies , and for protons from elastic coulomb collisions and inelastic strong interactions with background protons and other ions . in certain cases
there may be catastrophic losses through which particles are taken out of the system .
_ in summary then , it turns out that this most commonly used transport equation in astrophysical problems is a good approximation at all energies and all degrees of magnetization for spatially unresolved sources .
_ there are two noteworthy aspects to the formalism described above .
the first feature is related to the fact that the relative rates of the momentum and pitch angle diffusion , and hence the ratio @xmath99 , vary with plasma conditions and particle energy .
this has two important consequences .
the first is that both rates increase with decreasing energy and/or increasing level of turbulence ( see equation [ [ taup ] ] below ) . as a result sa of low energy thermal particles in magnetized plasma ,
the type usually encountered in astrophysical radiating sources , is not slow as the second order name would imply . in deed , in recent years this has been recognized and sa has found application in many sources involving acceleration of low energy background electrons . the second has to do with the change of the ratio @xmath99 . in figure [ figr1 ] ( left , from pp97 )
we show contour maps of this ratio in the electron energy and degree of magnetization space represented by the parameter @xmath100 , the ratio of plasma to gyro frequency . on the middle panel
we show variation of @xmath99 ( from petrosian & liu , 2004 , * pl04 * ) with energy for several values of @xmath39 for both electrons and protons interacting with parallel propagating plasma waves .
these figures show the regions of the phase space where @xmath78 and where the sa by turbulence is the dominant process .
let us consider a simple non relativistic hydrodynamic shock or a parallel shock with magnetic field parallel to the shock flow velocity . of the diffusion parallel and perpendicular to the field lines
( see e.g. giacalone 2005a & 2005b and references therein ) . ]
the sa rate is @xmath101 while the shock acceleration rate is proportional to fractional energy gain per crossing @xmath102 divided by average crossing time @xmath103 ( krymsky et al .
1979 ; lagage & cesarsky 1983 ; drury 1983 ) so that the shock acceleration rate @xmath104 is no longer a first order mechanism , and the ratio @xmath105 .
this means that when @xmath78 , which is the case at low energies , the sa is the more dominant of the two mechanisms .
thus , the following picture seems to emerge .
when a particle crosses the shock to the downstream turbulent region its interactions with plasma waves increase its energy substantially before it has a chance to cross the shock .
only when its energy has increased to a sufficiently large value to make the ratio @xmath77 , then it can undergo repeated passage across the shock , only when its energy has increased to a sufficiently large value to make a @xmath77 , then it can undergo repeated passage across the shock , hence beginning a second stage of acceleration by the shock .
note also that even at high energies and interactions with low frequency fast modes with phase velocity equal to alfven velocity when @xmath106 is less than one , the ratio of the sa to shock acceleration rates a_sa / a_sh~(v_a / u_sh)^2~1/(_p m^2 ) , [ saoversh ] which could be greater than one for low beta plasmas @xmath107 and for shocks with low mach number @xmath108 .
thus , we can think of the acceleration as a with turbulence providing the initial rise in energy of background plasma till they become energetic enough to be accelerated also by the shock during which sa continues and may not be negligible .
this sort of behavior can be seen in the recent pic simulation by sironi & spitkovsky ( 2009 ) where test particles appear to gain energy gradually in the downstream region till they cross the shock and get a jump in energy ( see their figure 8) .
this hybrid scenario , in a way , also solves the long standing injection problem " in shock acceleration requiring injection of high energy ( i.e pre accelerated ) particles , especially for perpendicular shocks . in the electron energy and parameter @xmath109 for @xmath110 .
the red line shows the @xmath111 contour for @xmath112 .
regions below @xmath111 curves are when acceleration by sa becomes dominant ( from pp97 ) . *
middle : * variation with energy of the ratio @xmath99 for different pitch angles of electrons and protons note that the acceleration rate can exceed scattering up to @xmath113 kev for most pitch angles ( from pl04 ) . *
right : * variations of timescales associated with rates of pitch angle diffusion ( green ) , the simple momentum diffusion ( black ) , and the momentum diffusion for the isotropic case with coefficient @xmath114 ( red ) . the latter is considerably longer than the others at energy ranges when resonance with one mode is dominant .
this effect is much more pronounced for protons than electrons .
note also that the shapes of @xmath115 and @xmath116 associated curves are similar for different pitch angles making use of pitch angled averaged quantities a reasonable approximation . , title="fig:",height=207 ] in the electron energy and parameter @xmath109 for @xmath110 .
the red line shows the @xmath111 contour for @xmath112 .
regions below @xmath111 curves are when acceleration by sa becomes dominant ( from pp97 ) . *
middle : * variation with energy of the ratio @xmath99 for different pitch angles of electrons and protons note that the acceleration rate can exceed scattering up to @xmath113 kev for most pitch angles ( from pl04 ) . *
right : * variations of timescales associated with rates of pitch angle diffusion ( green ) , the simple momentum diffusion ( black ) , and the momentum diffusion for the isotropic case with coefficient @xmath114 ( red ) . the latter is considerably longer than the others at energy ranges when resonance with one mode is dominant .
this effect is much more pronounced for protons than electrons .
note also that the shapes of @xmath115 and @xmath116 associated curves are similar for different pitch angles making use of pitch angled averaged quantities a reasonable approximation . , title="fig:",height=207 ] in the electron energy and parameter @xmath109 for @xmath110 .
the red line shows the @xmath111 contour for @xmath112 .
regions below @xmath111 curves are when acceleration by sa becomes dominant ( from pp97 ) . *
middle : * variation with energy of the ratio @xmath99 for different pitch angles of electrons and protons note that the acceleration rate can exceed scattering up to @xmath113 kev for most pitch angles ( from pl04 ) . *
right : * variations of timescales associated with rates of pitch angle diffusion ( green ) , the simple momentum diffusion ( black ) , and the momentum diffusion for the isotropic case with coefficient @xmath114 ( red ) . the latter is considerably longer than the others at energy ranges when resonance with one mode is dominant .
this effect is much more pronounced for protons than electrons .
note also that the shapes of @xmath115 and @xmath116 associated curves are similar for different pitch angles making use of pitch angled averaged quantities a reasonable approximation . ,
title="fig:",height=207 ] \2 .
the second interesting feature has to do with the difference between the energy diffusion and acceleration rates in the two limiting cases ( equations [ kehighe ] and [ kelowe ] ) derived above . for a resonant interaction with plasma waves in general
the three fokker - planck diffusion coefficients obey the following simple relations ( see e.g. pl04 for parallel and pryadko & petrosian 1999 , for perpendicular propagating waves ) : ( d_pp / p^2):(d_p / p):d_=[x_j^2]:[x_j(1-x_j)]:[(1-x_j)^2 ] with x_j=(v_ph , j / v)^2 [ relds ] where @xmath117 is the phase velocity of the plasma mode ( with frequency @xmath118 and wave vector @xmath119 ) in resonance with the particle of momentum @xmath38 and pitch angle cosine @xmath39 .
this implies that for interaction with a single wave the acceleration rate as given in equation ( [ kpp ] ) , would be zero . )
the expressions in equation ( [ relds ] ) .
] however , in general interactions with many waves contribute to each coefficient so that this rate is never zero , but if the interaction with one mode is dominant then the rate becomes much smaller than normal diffusion rate @xmath120 .
figure [ figr1 ] ( right ) shows the energy dependence at various values of @xmath39 of the acceleration ( and scattering ) times based on the two forms of the acceleration rate ( black vs red curves ) . as evident
there is considerable difference between the two rates specially at low energies , and the differences are more pronounced for protons compared to electrons . as discussed below , this is important for the relative sa rates of different species in particular for electrons vs protons .
we will also show that a similar process affects the relative acceleration of @xmath0he and @xmath1he in solar flares .
over the past several decades there has been considerable discussion of first vs second order acceleration ( see e.g. drury 1983 ) and the role of turbulence in the latter ( see e.g. hall & sturrock 1967 ; kulsrud & ferarrri 1971 ) .
however the success of shock acceleration in producing the observed cr spectrum has relegated sa by turbulence to be a less important process even though the importance of the role played by turbulence in scattering of the energetic particles is fully appreciated . as shown above
, there are many similarities between the two processes , and in some cases sa by turbulence may be the dominant process . this fact has been recognized in more recent times and there has been renewed activity in application of the sa model to several astrophysical sources .
most prominent among these is the solar flare which is the most developed and will be discussed in more detail below . but sa has been applied to nonthermal emission from accretion disks around black holes : e.g. sgr a * black hole in the center of the milky way ( liu et al , 2004 , 2006a and 2006b ) , other active galactic nuclei ( stawartz & petrosian 2008 ) , stellar size black holes ( li & miller 1997 ) , gamma - ray bursts ( lazarian et al .
2003 ) , to supernovae shocks ( scott & chevalier 1975 ; cowsik & sarkar 1984 ; fan et al . 2009 ; virtanen & vainio 2005 ) , to giants radio galaxy lobes ( lacombe 1977 ; achterberg 1979 ; eilek 1979 ) , and to intra cluster medium of clusters of galaxies ( petrosian 2001 , brunetti & lazarian 2007 ) .
the complete development of a solar flare involves many phases .
after a complex preflare build up of magnetic fields , the first phase is the reconnection and the process of the energy release .
the final consequences of this released energy are the observed radiations from radio to gamma - rays , solar energetic particles ( seps ) , and coronal mass ejections ( cmes ) .
the basic scenario for these processes , as depicted by the cartoon in figure [ model ] ( left ) , can be summarized as follows : even though it is generally agreed that the flare energy comes from the annihilation of magnetic fields via reconnection , the exact mechanisms of the release and dissipation of this energy remains controversial .
dissipation can occur via _ plasma heating , particle acceleration or plasma turbulence_. as stated above turbulence is a necessary ingredient for acceleration and , as we will outline below , there is considerable observational evidence favoring sa by turbulence .
thus we believe that most of the magnetic energy is converted into turbulence near or above the top of a coronal loop , which we refer to as the acceleration site or the loop top ( lt ) region .
the turbulence undergoes nonlinear wave - wave interactions causing a dissipationless cascade to smaller scales .
the wave - particle interaction results in damping of the turbulence , heating of the plasma and acceleration of particles .
the accelerated particles are somewhat trapped at the lt because of their short mean free path due to scattering by turbulence which enhances their radiation intensity there .
eventually these particles escape the turbulent lt region .
some escape along open field lines and may undergo further scattering and acceleration by a cme shock during their transport to the earth where they are detected as seps .
most of the particles travel down the legs of the loop and produce the observed flare radiation ; microwaves via synchrotron , hard x - rays ( hxr ) via bremsstrahlung produced by electrons , and gamma - rays via nuclear line excitations and decay of pions ( primarily @xmath121 ) produced by protons , along the loop , but primarily at its foot points ( fps ) .
however most of the energy of the accelerated particles is dissipated in the chromosphere and below via inelastic coulomb collisions .
this deposited energy causes heating and evaporation of the plasma up into coronal loops , which then produces the bulk of the flare radiation in the forms of thermal soft x - rays and optical photons .
the evaporation changes the the density and temperature in the corona which can affect the reconnection , energy release and acceleration processes .
observations of solar flares are reviewed by raymond et al . and some theoretical aspects , namely those on global aspects of energy release and acceleration are reviewed by cargill et al . in this proceedings .
here we discuss the confrontation between the accelerations models and observations .
a successful model must account for all observations .
however , some observation are more critical than others . here
we focus on the following three separate observed characteristics , explanation of which we believe constitutes as minimum requirement for models .
radiative signatures of electrons , primarily from observations .
\2 . relative acceleration of electrons and protons .
isotopic abundance enhancements in seps , especially that of @xmath0he , and the variation of abundance ratio and spectra of @xmath0he and @xmath1he .
we have made extensive comparisons of the sa model with these observations and find numerous signatures of electrons and protons that support the model .
below we present a brief description of the above observations and how the sa model can account for them .
radiation by accelerated electrons produce wealth of observation , and we clearly can not address them all here .
instead we focus on a few critical observations , mainly from _ rhessi _ , that are related to the acceleration process .
@xmath122 one observation which we believe provides the most compelling and direct evidence for the presence of turbulence in the lt ( acceleration ) site is the observation by _ yohkoh _ ( masuda et al . 1994 ) showing a distinct impulsive hxr emission from the lt in addition to the usual fp sources .
this apparently is present in almost all _ yohkoh _ flares ( petrosian et al . 2002 ) and analysis of flares has confirmed this picture ( liu et al .
2003 ; krucker & lin 2008 ) for essentially all limb flares . the fact that we see lt and fp emission but little or none from the legs of the loop requires lingering of electrons in the lt region for times longer than the crossing time @xmath123 .
this requires an enhanced scattering near the lt region .
petrosian & donaghy ( 1999 ) show that coulomb scattering can not be the agent for this trapping because then the electrons will also lose most of their energy in the lt region on the same timescale and never reach the fps .
the most likely scattering agent is turbulence .
this turbulence can also accelerate the electrons .
there may be other acceleration at work too , but as described above , at low energies and for solar flare conditions ( low @xmath124 plasma ) sa by turbulence is the most effective process .
@xmath122 in rare cases , when the fp sources are weak or occulted , one can see a double lt source ( figure [ model ] middle , from liu et al .
2008 ; see also sui & holman , 2003 ) as expected in the model depicted on the left .
this simple picture also predicts a gradual rise of the lt source accompanied with a continuous increase in separation of the fps as the reconnection proceeds and larger closed loops are formed , a feature that has been seen at other wavelength but it is usually difficult to see in hxrs in weak flares because of low signal to noise ratio and in most strong flares because they tend to have complicated loop structures . the nov .
3 , 2003 x - class limb flare consisting of a single loop provided a good opportunity to see this feature in hxrs .
the right panel of figure [ model ] ( from liu et al .
2004 ) clearly shows this behavior .
@xmath122 another important observation by is the relative spectra of lt and fp sources .
the lt source is often dominated by a very hot thermal type emission with a relatively soft tail , while the fp sources consist of harder power laws with little or no thermal part as shown by the example in figure [ ff ] ( left ) .
these are exactly the kind of spectra that come out from models of sa by turbulence shown in the right panel of figure [ ff ] .
figure [ ff ] ( left ) also shows a forward fit of the observed spectra to those obtained from sa model with the specified acceleration parameters .
@xmath122 flares during the impulsive phase often show a soft - hard - soft temporal evolution and sometimes a slower than expected ( assuming only losses ) temperature decline in the thermal decay phase ( mctiernan et al .
the results in figure [ ff ] ( right ) agree with these evolutionary aspects as well .
as can be seen the spectra electrons accelerated by turbulence get harder with increasing value of the wave - particle interaction rate parameter ^-1_p=(/2)f_turb(q-1)(ck_/_e)^q-1 , [ taup ] where @xmath125 is the ratio of the turbulence to magnetic field energy density with wave energy spectral index of @xmath126 for wave vectors @xmath127 .
thus , as the level of turbulence or @xmath128 increases and decreases during the impulsive phase , we go from a thermal to soft - hard - soft nonthermal and back to a thermal phase .
@xmath122 in several _ rhessi _ limb flares jiang et al .
( 2006 ) find that during the decay phase the lt source continues to be confined ( not extend to the fps ) , and that the observed energy decay rate is much lower than the spitzer ( 1962 ) conduction rate .
these observations require suppression of the conduction and a continuous input of energy during the decay phase .
a low level of lingering turbulence can be the agent for both . ) with the indicated acceleration parameters , showing the presence of a quasi - thermal ( lt ) and a nonthermal component .
the solid line gives the sum of the two .
the blue dashes indicate the level of the background radiation ( liu et al .
* right : * the dependence on @xmath129 of the accelerated electron spectrum @xmath130 at the lt ( dotted ) and the effective thick target spectrum @xmath131 at the fps ( solid ) .
higher levels of turbulence produce harder spectra and more acceleration than heating ( from pl04 ) .
, title="fig:",height=264 ] ) with the indicated acceleration parameters , showing the presence of a quasi - thermal ( lt ) and a nonthermal component .
the solid line gives the sum of the two .
the blue dashes indicate the level of the background radiation ( liu et al .
* right : * the dependence on @xmath129 of the accelerated electron spectrum @xmath130 at the lt ( dotted ) and the effective thick target spectrum @xmath131 at the fps ( solid ) .
higher levels of turbulence produce harder spectra and more acceleration than heating ( from pl04 ) .
, title="fig:",height=264 ] flares are generally recognized based on radiative and other ( heating and evaporation ) signatures of the accelerated electrons . as one would expect protons
will also be subjected to same acceleration mechanisms .
the radiative signatures of protons are ( i ) narrow gamma - ray lines in the 1 - 7 mev range arising from de - excitation of nuclei of ions excited by accelerated protons ( or viceversa , which produce broader lines ) , and ( ii ) @xmath132 mev continuum emission from decay of pions produced in @xmath133 interaction ( mainly from decay of @xmath121 s ) .
gamma - rays are generally observed in large and gradual flares , but this is partly due to relatively lower sensitivity of past gamma - ray detectors compared to hxr detectors .
however , _ fermi _ with its superior sensitivity is beginning to observe gamma - rays from modest m - class flares ( see ackermann et al .
the sa model has been the working hypothesis for acceleration of protons as well .
( see pioneering works by ramaty ( 1979 ) and collaborators ; e.g. murphy et al .
1987 ) .
there have been considerable discussions of relative energies of populations of accelerated electrons and protons but most recent analysis of hxr and gamma - ray emission of a large sample of flares by shih et al .
( 2009 ) show a good correlation but with relatively broad dispersion ( see figures 3 of raymond et al . in this proceedings . as this figure shows the mean value of the ratio of energy in accelerated electrons ( obtained from hxr fluxes ) to that of protons ( deduced from gamma - ray fluxes )
is larger than 1 while this ratio in galactic crs presumably accelerated in supernova shocks is much smaller than one . in figure [ evspfig ] ( left )
we show the same distribution from the above mentioned figure ( in blue ) along with the distribution of the same ratio we obtained from observations of sep electrons and protons ( for the same energy ranges , taken from data compiled by cliver & ling 2007 , in red ) .
we will return to the differences between the two histograms below .
but for now we focus on the fact that the acceleration mechanism operating in flares seems to put more energy in electrons over protons compared with the mechanism responsible for acceleration of seps and galactic crs . in shock acceleration models ( with alfvenic turbulence as the source of scattering )
one would expect a more efficient acceleration of protons compared to electrons .
this may indicate that a different mechanisms is at work in solar flares .
as shown below this is another evidence that sa ( rather than a shock ) is the dominant acceleration mechanism in flares . ] ) time scales in unit of @xmath134 for protons ( blue ) and electrons ( red ) showing a large barrier or long acceleration time for protons in the 0.1 to 10 mev range ( dashed region ) due to the dominance of a single resonance mode .
this barrier disappears at low energies once @xmath135 becomes longer than momentum diffusion time @xmath136 , and at high energies where there is more than one dominant resonant mode ( from pl04 ) . *
right : * spectra of accelerated electrons ( red ) and protons ( blue ) at the acceleration site ( dotted ) and the effective thick target fp spectra @xmath131 ( dashed ) for given values of @xmath134 and @xmath137 and for temperature @xmath138 kev which produces more nonthermal electrons compared to protons in their respective observed ranges shown by the dashed areas ( from pl04 ) .
, title="fig:",height=204 ] ] ) time scales in unit of @xmath134 for protons ( blue ) and electrons ( red ) showing a large barrier or long acceleration time for protons in the 0.1 to 10 mev range ( dashed region ) due to the dominance of a single resonance mode .
this barrier disappears at low energies once @xmath135 becomes longer than momentum diffusion time @xmath136 , and at high energies where there is more than one dominant resonant mode ( from pl04 ) . *
right : * spectra of accelerated electrons ( red ) and protons ( blue ) at the acceleration site ( dotted ) and the effective thick target fp spectra @xmath131 ( dashed ) for given values of @xmath134 and @xmath137 and for temperature @xmath138 kev which produces more nonthermal electrons compared to protons in their respective observed ranges shown by the dashed areas ( from pl04 ) . ,
title="fig:",height=204 ] ] ) time scales in unit of @xmath134 for protons ( blue ) and electrons ( red ) showing a large barrier or long acceleration time for protons in the 0.1 to 10 mev range ( dashed region ) due to the dominance of a single resonance mode .
this barrier disappears at low energies once @xmath135 becomes longer than momentum diffusion time @xmath136 , and at high energies where there is more than one dominant resonant mode ( from pl04 ) . *
right : * spectra of accelerated electrons ( red ) and protons ( blue ) at the acceleration site ( dotted ) and the effective thick target fp spectra @xmath131 ( dashed ) for given values of @xmath134 and @xmath137 and for temperature @xmath138 kev which produces more nonthermal electrons compared to protons in their respective observed ranges shown by the dashed areas ( from pl04 ) . ,
title="fig:",height=204 ] in pl04 we address this problem by looking at the details of the sa by parallel propagating waves of a thermal population of electrons and protons .
we find in general that for typical flare conditions this mechanism favors acceleration of electrons than protons .
this is because electrons and protons undergo resonant interactions with different plasma modes of different wave vectors or frequencies . in general protons
have fewer resonances than electrons so that it is more likely that they will have one dominant resonant mode . in this case , as described in the second of * two important features * in 3.2 , and shown in figure [ figr1 ] ( right ) , the rate of acceleration given by equation ( [ kpp ] ) becomes very small hindering the acceleration of protons .
figure [ evspfig ] ( middle ) shows presence of a large barrier against acceleration of proton , manifested by a large acceleration timescale , resulting in acceleration of fewer protons in the energy range from few mev to gev range needed for production of gamma - rays ( lines and continuum ) . as mentioned above , the acceleration rate by a shock also depends on the rate of wave - particle interactions .
however , in this case it depends on the spatial diffusion coefficient @xmath139 which in turn depends only on @xmath115 and not the combination of the diffusion coefficient in equation ( [ kpp ] ) which affects the sa rate .
the right panel of figure [ evspfig ] shows the accelerated electron and proton spectra ( multiplied by square of energy ) and the effective thick - target spectra for the specified values of the basic acceleration parameters ( namely density , magnetic field , temperature and level and spectrum of turbulence represented by the parameter @xmath134 ) which produce more nonthermal electrons than protons in their respective observes ranges shown by the hashed areas .
however , as shown in pl04 , the ratio of electron to proton acceleration varies considerably with the basic acceleration parameters such as @xmath134 , @xmath137 and temperature of the injected particles .
in fact , as can be seen from comparison of spectra shown in the left and middle pane of figure [ evsp ] with that in figure [ evspfig ] left a small change in @xmath140 can produce a large change in the ratio of the energies of the accelerated electrons to protons explaining the broad distributions seen in the left panel of figure [ evspfig ] .
similarly , as evident fro spectra shown in the right panel of figure [ evsp ] higher temperature of background particles also favor the acceleration of protons .
one consequence of these is that the proton acceleration will be more efficient in larger ( and most likely lower @xmath61 value ) loops and at late phases , when evaporation increases the temperature and density @xmath10 ( hence the value of parameter @xmath140 ) .
this can explain the difference in centroids of hxr and gamma - ray emission seen by as described by raymond et al .
in these proceedings
. it should be emphasized , however , that here we have concentrated on the ratios of accelerated electrons and protons in the energy ranges that produce the observed hxrs and gamma - rays ( electrons @xmath141 s of kev , protons @xmath141 mev ) .
but as can be seen from above figures , in general , there are considerable number of accelerated protons at lower energies which do not produce detectable radiation .
we will discuss the role of these particles below .
_ in summary , the sa model can account for various differences seen in acceleration of electrons and protons .
_ but for different values of @xmath142 and @xmath143 showing large variations in the relative accelerations of electrons vs protons . * left : * @xmath144 s@xmath145 , @xmath146 , @xmath138 kev . * middle : * @xmath147 s@xmath145 , @xmath148 , @xmath138 kev . *
right : * @xmath144 s@xmath145 , @xmath146 , @xmath149 kev . as evident higher densities ,
higher temperatures and lower magnetic fields favor acceleration of protons vs electrons and viceversa .
note also that the quasi - thermal proton component of less than one mev , which do not produce gamma - rays , can escape and possibly re - accelerated by a cme shock , and be observed as seps .
this can explain the shift to the right of the sep ( red ) distribution shown in the left panel of figure [ evspfig ] .
( from pl04 . ) , title="fig:",height=207 ]
but for different values of @xmath142 and @xmath143 showing large variations in the relative accelerations of electrons vs protons . *
left : * @xmath144 s@xmath145 , @xmath146 , @xmath138 kev .
* middle : * @xmath147 s@xmath145 , @xmath148 , @xmath138 kev . * right : * @xmath144 s@xmath145 , @xmath146 , @xmath149 kev . as evident higher densities ,
higher temperatures and lower magnetic fields favor acceleration of protons vs electrons and viceversa .
note also that the quasi - thermal proton component of less than one mev , which do not produce gamma - rays , can escape and possibly re - accelerated by a cme shock , and be observed as seps .
this can explain the shift to the right of the sep ( red ) distribution shown in the left panel of figure [ evspfig ] .
( from pl04 . ) ,
title="fig:",height=207 ] but for different values of @xmath142 and @xmath143 showing large variations in the relative accelerations of electrons vs protons . * left : * @xmath144 s@xmath145 , @xmath146 , @xmath138 kev . *
middle : * @xmath147 s@xmath145 , @xmath148 , @xmath138 kev . *
right : * @xmath144 s@xmath145 , @xmath146 , @xmath149 kev . as evident higher densities , higher temperatures and lower magnetic fields favor acceleration of protons vs electrons and viceversa .
note also that the quasi - thermal proton component of less than one mev , which do not produce gamma - rays , can escape and possibly re - accelerated by a cme shock , and be observed as seps .
this can explain the shift to the right of the sep ( red ) distribution shown in the left panel of figure [ evspfig ] .
( from pl04
. ) , title="fig:",height=207 ] it is commonly believed that the observed relative abundances of ions in seps favor the sa model ( _ e.g. _ mason et al .
1986 ; mazur et al .
more recent observations and modelings have confirmed this picture ( see mason et al . 2000 , 2002 ; reames et al .
1994 , 1997 ; ng & reames 1994 ; miller 2002 ) . one of the most vexing problem of seps has been the enhancement of @xmath0he .
observations show a wide range of @xmath0he to @xmath1he ratios ; ranging from photospheric values in gradual - strong flares to values several thousand times larger in impulsive - weak flares .
it should be emphasized that there are not two distinct classes ( impulsive and gradual ) with a well defined bimodal distribution .
rather , as indicated by observations ( ho et al . 2005 ) , there is a broad continuum of events as shown in the left panel of figure [ he ] , going from weak , short duration ( impulsive ) events with strong enrichments at one end to long ( gradual ) , strong and normal abundances events at the other extreme end .
it was recognized early that the unusual charge to mass ratio of @xmath0he could be the cause here .
however , these early works did not provide a satisfactory quantitative explanation . with a more complete treatment of @xmath0he and @xmath1he acceleration , liu , petrosian & mason 2004 and 2006 ( lpm04 , lpm06 )
have demonstrated that sa can indeed explain the extreme enhancement of @xmath0he and can also reproduce the observed @xmath0he and @xmath1he spectra for high enrichment cases .
the reason for success of our approach is in a way similar to our treatment of election versus proton acceleration , where inclusion of resonance interactions with multiple wave modes gives rise to the different rates . in case of @xmath0he and @xmath1he
we also find that , once the effects of ionized he ( @xmath140 particles ) are included in the description of the dispersion relation , the low energy @xmath0he ions have more resonances than @xmath1he ions , as a result of which they are more readily accelerated than @xmath1he . in lpm04 and lpm06
it was shown that with this model we can obtain an excellent fit to the observed spectra of several weak shorter duration events for both @xmath1he and @xmath0he ( which show the characteristic convex spectral shapes ) for reasonable plasma and acceleration parameters .
an example of an excellent fit ( with @xmath150 ) is shown in the right panel of figure [ he ] with the solid lines . on this plot
we also show two other sets of spectra for two different values of the sa rate parameter @xmath151 , which is proportional to the level of turbulence ( see equation [ taup ] ) and perhaps to the overall strength of the flare . as can be seen in all cases most of the @xmath0he
are accelerated into the observed range , owing to the high efficiency of their acceleration , while this is true for @xmath1he only for large values of @xmath151 , i.e. for high levels of turbulence . for small levels
there is a barrier for @xmath1he , similar to that for protons in figure [ evspfig ] , so that most of the @xmath1he appear as a low energy bump ( a quasi - thermal component ) , with only a small fraction reaching the observed range .
however , the number of @xmath1he ions in this range increases rapidly with increasing value of @xmath151 .
this also explains the wide range and the trend of the fluence ratios . as shown in the right panel of this figure
the model predicted ratio decreases with increasing levels of turbulence and hence increasing @xmath1he fluence .
this trend is independent of the other sa model parameters ( like the parameter @xmath140 as can be seen in this figure ) .
it also turns out that this model can reproduce the observed distributions of fluences of both ions ( see petrosian et al .
2009 ) .
he and @xmath1he vs the fluence of @xmath1he showing a wide range of @xmath0he enhancement decreasing with increasing event fluence .
this also shows a broader distribution of @xmath1he than @xmath0he fluences ( from ho et al .
* middle : * model fit to @xmath0he and @xmath1he spectra of sep .
30 , 1999 event observed by _ ace _ showing an excellent fit ( solid lines ) for @xmath152 and the specified values of the rate parameter @xmath151 or the level of turbulence ( @xmath153 ) .
also shown are two sets of model spectra with different levels of turbulence ( or @xmath151 ) .
note that in all three cases almost all of the @xmath0he are accelerated into a nonthermal component while at lower levels of turbulence most of @xmath1he form a low energy bump in the unobserved range with a smaller high energy tail .
but with increasing level of turbulence more @xmath1he ions are accelerated into the observed range ( from petrosian et al .
* right : * model calculation of the variation of the accelerated @xmath0he to @xmath1he fluence ratio ( at @xmath154 mev / nucleon ) with level of turbulence or @xmath151 for three values of @xmath137 ( note that @xmath152 for the black line ) .
note also that the trend and the range of the ratio mimics the observation shown in the left panel ( from lpm06 ) .
, title="fig:",height=204 ] he and @xmath1he vs the fluence of @xmath1he showing a wide range of @xmath0he enhancement decreasing with increasing event fluence .
this also shows a broader distribution of @xmath1he than @xmath0he fluences ( from ho et al .
* middle : * model fit to @xmath0he and @xmath1he spectra of sep .
30 , 1999 event observed by _ ace _ showing an excellent fit ( solid lines ) for @xmath152 and the specified values of the rate parameter @xmath151 or the level of turbulence ( @xmath153 ) .
also shown are two sets of model spectra with different levels of turbulence ( or @xmath151 ) .
note that in all three cases almost all of the @xmath0he are accelerated into a nonthermal component while at lower levels of turbulence most of @xmath1he form a low energy bump in the unobserved range with a smaller high energy tail .
but with increasing level of turbulence more @xmath1he ions are accelerated into the observed range ( from petrosian et al .
* right : * model calculation of the variation of the accelerated @xmath0he to @xmath1he fluence ratio ( at @xmath154 mev / nucleon ) with level of turbulence or @xmath151 for three values of @xmath137 ( note that @xmath152 for the black line ) .
note also that the trend and the range of the ratio mimics the observation shown in the left panel ( from lpm06 ) .
, title="fig:",height=204 ] he and @xmath1he vs the fluence of @xmath1he showing a wide range of @xmath0he enhancement decreasing with increasing event fluence .
this also shows a broader distribution of @xmath1he than @xmath0he fluences ( from ho et al . 2005 ) . *
middle : * model fit to @xmath0he and @xmath1he spectra of sep .
30 , 1999 event observed by _ ace _ showing an excellent fit ( solid lines ) for @xmath152 and the specified values of the rate parameter @xmath151 or the level of turbulence ( @xmath153 ) .
also shown are two sets of model spectra with different levels of turbulence ( or @xmath151 ) .
note that in all three cases almost all of the @xmath0he are accelerated into a nonthermal component while at lower levels of turbulence most of @xmath1he form a low energy bump in the unobserved range with a smaller high energy tail .
but with increasing level of turbulence more @xmath1he ions are accelerated into the observed range ( from petrosian et al .
* right : * model calculation of the variation of the accelerated @xmath0he to @xmath1he fluence ratio ( at @xmath154 mev / nucleon ) with level of turbulence or @xmath151 for three values of @xmath137 ( note that @xmath152 for the black line ) .
note also that the trend and the range of the ratio mimics the observation shown in the left panel ( from lpm06 ) .
, title="fig:",height=204 ] it should , however , be emphasized that even though we obtain the observed low ratio of fluences for stronger events , the spectra of @xmath1he obtained for these events ( e.g. the long dashed line in figure [ he ] , right ) do not agree with the observations ( we discuss a possible remedy for this in the next section ) .
nevertheless , this is a significant breakthrough in understanding of seps .
there is also an increasing enhancement with increasing mass of the ion .
possible explanations of these and other aspects of the enrichments can be found in a recent review by petrosian ( 2008 ; e.g. figure10 ) . clearly more work is required for a complete description of all observed characteristics of seps , some of which may require another mechanism of acceleration as discussed next we have shown that the sa model can reproduce many of the observed features but there are several aspect that need refinements or introduction other processes .
for example , we have seen that even though this model can explain the predominance of the accelerated electrons at the flare site and the broad range of the accelerated electron to proton flux ratios it can not account for the difference between the relative rates of accelerations of electrons and protons as deduced by the radiations they produced at the flare site and that observed in seps which favor proton acceleration .
we have also indicated we can account for the varied observed spectra and the broad range of the isotopic enrichments in particular that of the @xmath0he in weaker , more impulsive flares , while the @xmath1he model spectra for high fluence - long duration events seems quite different than that observed in such events .
the model spectra are softer and unlike the observed ( broken ) power laws .
thus , the spectra of the accelerated particles coming out of the flare site must be modified by a secondary process to agree with observations .
as is well known , gradual strong flares are associated with cmes which has led to the idea that the shock produced by the cme can be responsible for acceleration of seps . however , as it usually the case with shock acceleration , the question of seed particles is uncertain here as well .
it is unlikely that the cold background particles in high corona are the seeds .
tylka & lee ( 2006 ) in their phenomenological study of acceleration by a shock were able to produce the observed spectra of the sep assuming a suprathermal " seed population , extending to higher energies for perpendicular shocks .
it then seems natural to assume that the seed particles are flare accelerated ions which are then re - accelerated by the shock .
the @xmath1he spectra of high fluence events shown in figure [ he ] ( right ) have this kind of characteristics making them good candidates for re - acceleration .
this leads us to the following scenario .
the flare site acceleration is the first and primary stage of acceleration and is common to all events .
a second phase acceleration can occur in the cme shocks which could modify the spectrum of seps escaping the flare site .
the possibility that there may be two acceleration mechanisms at work is not a new idea .
the new aspect of this scenario is that we have a hybrid acceleration model , where the seeds of the second stage ( re)acceleration by the cme shock are the flare site particles accelerated by turbulence .
mev / nucleon ) ions are present upstream of all interplanetary shocks " ( desai et al . 2003 ) . ] as shown above , the spectrum of particles escaping the flare site consists of two components : a low energy quasi - thermal component ( which is below the observable energy range of 0.1 to 10 mev / nucleon ) and a higher energy nonthermal component with high @xmath0he enrichment .
this scenario has the attractive feature that even though the nonthermal tails may be highly enriched ( as in impulsive flares ) , the total ( quasi - thermal plus nonthermal ) number of seed particles injected into the cme shock can have essentially normal abundances and harder ( power - law ) spectra . thus , for events near the gradual end
both components are re - accelerated leading to a near normal abundances . in figure [ reacc ]
we compare the sep observations of the gradual 18 jan .
2000 event with small isotopic enhancement ( points ) with the re - accelerated spectra ( solid lines ) obtained using the flare accelerated spectra ( dashed lines ) as the source term @xmath155 in equation ( [ keall ] ) that are re - acceleration , with an addition of direct acceleration term @xmath156 , presumably due to a cme shock , with of @xmath157 at 0.1 mev / nucleon .
the resultant spectra are harder and closer to normal abundance ratio .
we hasten to add that these are preliminary explorations and the agreement , though not perfect , is very good considering that we have used a simple power law acceleration rate .
a better agreement can be achieved with a more realistic form for the acceleration rate .
a similar scenario can also explain the differences between the distributions of electron to proton energy flux ratio deduced from observations at the flare site and from seps shown in figure [ evspfig ] ( left ) . as discussed in 4.2.2 , in general sa at the flare site
is more efficient in acceleration of electrons than protons that have energies capable of producing the observed gamma - rays .
however , as also emphasized in 4.2.2 , just as is the case for @xmath1he , flare accelerated protons have a substantial low energy component which can be re - accelerated at a cme shock to yield a smaller electron to proton ratio in seps as compared to flares . in a new and possibly far reaching work ( petrosian & chen 2010 )
, we have initiated determination of some of the important parameters of the kinetic equation ( [ keall ] ) ( left ) directly from the observed data instead of the usual forward fitting method , like that shown in figure [ ff ] , where one assumes values and energy dependences for these parameters , calculates the particle distribution and its radiative spectrum , and then fits to the data ( see e.g. park et al .
1997 ) . in our new work ,
using the recently developed _
regularized inversion _
technique by piana et al .
( 2003 and 2007 ) , we are able to determine the energy dependence of the escape and scattering times due to turbulence as shown in figure [ reacc ] ( right ) .
an important aspect of this result is that the escape time increases with energy ( and as a result the scattering time decreases ) relatively rapidly in disagreement with the expected behavior for sa by parallel propagating plasma waves with a kolmogorov type spectrum ( see pp97 or pl04 ) .
the observed behavior requires a steeper than kolmogorov spectrum so that we are dealing with wave vector values in the steep damping range .
this discrepancy could also be an indication that the simple sa model used in above mentioned papers requires modification .
for example , inclusion of the effects of the convergence of the magnetic fields in the lt region , as can be seen in figure [ model ] ( left ) , can produce an escape time similar to the timescale for coulomb collisions which increases with energy . or our simplified description of the escape time as defined in equation ( [ keall ] ) may require modification .
another possibility is that the outflows from the reconnection region may produce standing shocks that can modify the acceleration rate relative to the scattering rate .
these are interesting possibilities that eie intend be explored in future works .
he and @xmath1he spectra of event on 18 jan 2000 with our proposed model , where the flare site accelerated spectra ( dashed lines ) for an intermediate value of @xmath151 ( see figure [ he ] ) are used for re acceleration by a combined shock and sa processes .
* middle : * variation with energy of the escape and scattering times due to turbulence obtained directly from _ rhessi _ data using inversion technique for the nov . 3 , 2003 flare .
in contrast to the model calculated @xmath89 in pl04 , the observed time scale increases with energy ( from petrosian & chen 2010 ) . , title="fig:",height=264 ]
he and @xmath1he spectra of event on 18 jan 2000 with our proposed model , where the flare site accelerated spectra ( dashed lines ) for an intermediate value of @xmath151 ( see figure [ he ] ) are used for re acceleration by a combined shock and sa processes . *
middle : * variation with energy of the escape and scattering times due to turbulence obtained directly from _ rhessi _ data using inversion technique for the nov . 3 , 2003 flare .
in contrast to the model calculated @xmath89 in pl04 , the observed time scale increases with energy ( from petrosian & chen 2010 ) .
, title="fig:",height=264 ]
\1 . we have reviewed merits of different acceleration mechanisms and pointed out that in general turbulence plays a significant role in all of them so that sa of particles by the turbulence , the original mechanism proposed by fermi , is omni - present .
we have shown that at low energies and for highly magnetized plasmas sa by turbulence is the most efficient mechanism and that it can accelerate particles as rapidly as required in most astrophysical radiation sources .
thus , sa is an attractive scenario for acceleration of background thermal particles .
we have also pointed out that in the presence of shocks a hybrid process of initial sa acceleration by turbulence and subsequent re - acceleration by the shock is what may be operating in many situations providing an answer to the long standing question of origin of the seed particles for shock acceleration .
we have shown that a closer analysis of wave - particle interactions in a turbulent plasma also leads to interesting differences between the acceleration rates of electrons and protons and among heavier ions , in particular @xmath0he and @xmath1he .
low density , high magnetic fields preferentially accelerate electrons while the opposite is true for protons . in case of @xmath0he and @xmath1he we find that @xmath0he , by virtue of its resonance with more modes than @xmath1he is accelerated readily in most circumstances , while efficient acceleration of @xmath1he to high energies requires high levels of turbulence .
we have argued that recent high spatial observations at hxrs , specially those by provide strong evidence for the presence of turbulence in the acceleration sites of flares in the corona near the top of the flaring loops .
we have shown that the sa by turbulence can account for high relative efficiency of acceleration of electrons vs protons in solar flares that agrees with their relative numbers as deduced by the hxr emission by electrons and gamma - ray emission by protons .
similarly , we have shown that this model can account quantitatively for the long standing puzzle of extreme enhancement of @xmath0he abundances in the low fluence impulsive sep events relative to that of @xmath1he and reproduce their observed spectra accurately . however , there are some aspects of seps that the sa model can not account for .
one is the fact that the ratio of electron to proton energy ratios in seps is smaller than in flares .
another is that @xmath1he spectra resulting in from sa by turbulence in the flare region for high fluence - gradual events are too soft compared with observations .
we have proposed that the hybrid model described above can resolve these discrepancies , whereby the flare accelerated particles are re - accelerated by cme shocks which are more likely to be present in bigger flares .
acknowledgment : i thank colleagues siming liu and wei liu for production of some of the results presented here and graduate student qingrong chen for help in preparation of this manuscript .
the research providing the result presented here were supported by previous nsf and nasa grants and current nasa grant nnx10ac06 g .
|
the subject of this paper is stochastic acceleration by plasma turbulence , a process akin to the original model proposed by fermi .
we review the relative merits of different acceleration models , in particular the so called first order fermi acceleration by shocks and second order fermi by stochastic processes , and point out that plasma waves or turbulence play an important role in all mechanisms of acceleration .
thus , stochastic acceleration by turbulence is active in most situations .
we also show that it is the most efficient mechanism of acceleration of relatively cool non relativistic thermal background plasma particles .
in addition , it can preferentially accelerate electrons relative to protons as is needed in many astrophysical radiating sources , where usually there are no indications of presence of shocks .
we also point out that a hybrid acceleration mechanism consisting of initial acceleration by turbulence of background particles followed by a second stage acceleration by a shock has many attractive features .
it is demonstrated that the above scenarios can account for many signatures of the accelerated electrons , protons and other ions , in particular @xmath0he and @xmath1he , seen directly as solar energetic particles and through the radiation they produce in solar flares .
|
the spin - dependent ( gamow - teller ) matrix - elements , for transitions between members of the baryon octet @xcite acquired renewed interest after measurements were made of the deep inelastic scattering ( dis ) of polarized leptons by polarized protons and neutrons @xcite , which provided valuable information about the spin structure of the nucleon .
one of the most important quantities measured in polarized dis is the longitudinal spin structure function @xmath2 . in the quark parton model , the spin structure function @xmath2
is directly related to the quark spin densities : @xmath3 , @xmath4 , @xmath5 etc .
where @xmath6 . to deduce the various quark spin densities from the @xmath2 data
, one usually assumes that baryons may be assigned to a su(3 ) flavor octet and uses the relation between the quark spin densities and weak matrix elements @xmath7 and @xmath8 from hyperon semileptonic decays . by using the earlier @xmath0 value ,
the emc data led to the unexpected conclusion @xcite that the quarks carry at most a small part of the spin of either nucleon and furthermore , that there is a significant contribution from `` strange '' quarks , which necessarily come from the `` sea '' of quark - antiquark pairs .
this has led to many different suggestions for resolution of what has come to be called the `` spin crisis '' @xcite . among these
is a suggestion @xcite that the conclusions may be distorted because the @xmath0 value obtained from the hyperon semileptonic decays are based on exact su(3 ) flavor symmetry .
su(3 ) symmetry breaking effects may significantly change the value .
there are many works attempting to evaluate the su(3)-breaking effects in the bag model or in the quark model , by applying center - of - mass corrections @xcite , or by including one - gluon exchange interactions @xcite or both @xcite .
the size of the corrections depends on the model and assumptions used to describe the symmetry breaking effects and on the ` existing ' data to be fitted .
some authors used their own data and concluded @xcite that there is no signal for the breakdown of cabbibo s su(3 ) symmetric description . according to ref.@xcite , however , an overall fit to the existing data using broken su(3 ) scheme is better than that from the assumption of perfect su(3 ) symmetry .
another approach , using the chiral effective lagrangian for baryons , @xcite calculated su(3 ) symmetry breaking corrections to axial currents of the baryon octet arising from meson loops .
the size of corrections was found to be surprisingly large ( the loop correction is almost as large as the lowest order result ) which should already have raised suspicion . in a subsequent paper @xcite ,
including the spin-3/2 baryon decuplet in the intermediate state , the meson loop correction to the axial currents is significantly reduced but still substantial ( @xmath9 ) .
however , corrections due to higher baryon resonances , which in principle should be included in the intermediate states , have been ignored in the calculation and may change the result still further .
thus it appears that the validity of cabbibo s su(3 ) symmetric description is far from settled .
most recently , instead of model - dependent calculations , an approach @xcite based on phenomenological analysis of hyperon beta decay data has been suggested to estimate the su(3 ) symmetry breaking effects .
the authors present evidence for a strong variation of the @xmath0 parameter between various transitions . in section ii
, we consider another approach based on a general discussion @xcite of su(3 ) flavor symmetry and its possible breaking and show that the hyperon beta decay data are adequately represented by at most a small deviation from of cabibbo s su(3 ) symmetric description , which can be well accommodated within the framework of the usual assumption of a small su(3 ) breaking proportional to the mass - difference between strange and nonstrange quarks . in section iii , the consequences for the quark spin distribution in the nucleon are discussed .
a brief summary is given in section iv .
in the quark model , which provides an explicit realization of cabibbo s theory connecting strangeness - conserving and strangeness - changing weak interactions , the primary weak current responsible for transitions between hadrons is : @xmath10 with @xmath11{\rm cos}{\theta_c } + [ \lambda_4 + i\lambda_5]{\rm sin}{\theta_c}\ ] ] where @xmath12 ( i=1,2, ... 8 ) denote the gell - mann matrices and @xmath13 represents the triplet ( @xmath14 , @xmath15 , @xmath16 ) of basic quark fields .
eq.(1 ) requires that weak transition - elements necessarily transform as a component of an su(3 ) octet . if baryons are assigned to a su(3 ) octet , represented in matrix form by : @xmath17 the su(3)-octet matrix elements between baryons can be written , in the symmetric limit ( we are concerned only with the values for @xmath18 , i.e. zero four - momentum transfer ) , as @xmath19_- ) \eqno ( 3)\ ] ] which can also be written as @xmath20 , with @xmath21 and @xmath22 .
however , the su(3 ) flavor symmetry is only approximate for strangeness - changing processes .
if su(3 ) symmetry - breaking effects can not be ignored , the expressions for the matrix - elements must be generalized .
we assume that the breaking of su(3)-flavor symmetry is due to a term which transforms like the eighth generator of su(3 )
. this would be the case , for example , if su(3 ) breaking arose entirely from a mass - difference between strange and ( degenerate @xcite ) non - strange quarks .
to first order in the symmetry - breaking interaction , transforming like @xmath23 , the most general su(3 ) structure of the weak matrix - elements between baryons can be written as @xmath24 @xmath25/2 \eqno ( 4)\ ] ] where the first two terms are the ones given in eq.(3 ) and the others are su(3 ) symmetry - breaking corrections .
the corresponding symmetry - breaking parameters @xmath26 , @xmath27 , @xmath15 , and @xmath28 should be small relative to @xmath29 and @xmath30 for such a perturbative expansion to be valid .
vector coupling constants are not affected to first order @xcite .
for the ratio of axial - vector to vector amplitudes , eq.(4 ) yields @xcite @xmath31 @xmath32 @xmath33 @xmath34 where we have listed only those transitions for which these ratios are relatively well measured @xcite : @xmath35 @xmath36 @xmath37 @xmath38 let us first discuss the su(3 ) symmetry scheme .
fig.1 exhibits the results reported in eqs.(9)@xmath39(12 ) under the assumption that su(3 ) symmetry - breaking effects are negligible , viz .
all breaking parameters are zero : @xmath40 in eqs .
( 5)@xmath39(8 ) .
we see that the @xmath41 ratios for the best - measured transitions ( 9)-(11 ) yield , within the errors , a unique solution for @xmath7 and @xmath8 . while the line corresponding to the central value of @xmath41 for the less accurately measured @xmath42 transition does not pass exactly through the same @xmath43 point , a downward shift of @xmath44 by an amount equal to the quoted error , is sufficient to bring it into agreement with the others .
hence it seems that no significant su(3 ) symmetry - breaking effect is needed to describe the existing @xmath41 data .
it is also interesting to note that the favored solution for @xmath7 and @xmath8 obtained from data ( 9)@xmath39(11 ) is not too different from that predicted by the static su(6 ) symmetric model with suitable relativistic recoil corrections ( @xmath45 25@xmath46 reduction@xcite ) . while there does not seem to be any compelling evidence demanding the inclusion of su(3 ) breaking effects
, it may be worthwhile to see what is obtained if one takes the data , eqs.(9)-(12 ) , at face value and seeks a solution allowing any one of the symmetry breaking parameters in eq.(4 ) to be non - zero .
we search in the three - dimensional space @xmath7 , @xmath8 , @xmath47 ( where @xmath47 denotes one of four possible small symmetry breaking parameters : @xmath26 , @xmath27 , @xmath15 or @xmath28 ) to find the minimum of the quantity @xmath48 .
the results are listed in table i. table i : one - parameter fit @xmath49&\cr \noalign{\hrule } % % % % % % % % % % % % % % % & f & & 0.4581 & & 0.4576 & & 0.4610 & & 0.4721 & & & \cr % 0.459$\pm$0.008 & \cr & d & & 0.7992 & & 0.7943 & & 0.7963 & & 0.7852 & & & \cr % 0.798$\pm$0.008 & \cr & f / d & & 0.573 & & 0.576 & & 0.579 & & 0.601 & & & \cr % 0.575$\pm$0.016 & \cr & $ ( g_a / g_v)_{n\rightarrow p}$ & & 1.2573 & & 1.2573 & & 1.2573 & & 1.2573 & & 1.2573$\pm$0.0028&\cr & $ ( g_a / g_v)_{\lambda\rightarrow p}$ & & 0.723 & & 0.721 & & 0.714 & & 0.724 & & 0.718$\pm$0.015 & \cr & $ ( g_a / g_v)_{\sigma^-\rightarrow n}$ & & $ -$0.343 & & $ -$0.337 & & $ -$0.335 & & $ -$0.343 & & $ -$0.340$\pm$0.017&\cr & $ ( g_a / g_v)_{\xi^-\rightarrow\lambda}$ & & 0.190 & & 0.192 & & 0.208 & & 0.250 & & 0.25$\pm$0.05 & \cr \noalign{\hrule } & $ { \chi}^2 $ & & 1.61 & & 1.42 & & 0.86 & & 0.20 & & & \cr \noalign{\hrule } } } \ ] ] as expected , it takes only a small non - zero value of any of these , to obtain a statistically satisfactory solution .
the fifth column , with a @xmath15-type correction , shows the best agreement between the calculated and the measured @xmath41 ratios , and provides the only indication that inclusion of su(3 ) breaking effects may be required .
the best fits under the assumption that su(3 ) symmetry breaking arises from terms of the type @xmath26 or @xmath27 , yield values which , in view of the quoted errors , are indistinguishable from zero , i.e. do not call for any correction at all .
similarly , the evidence for non - zero @xmath28 is marginal . from the results listed in tables i , taking average of all the results , we obtain @xmath50 these values are consistent with those previously used in the analysis of deep inelastic scattering on polarized nucleons @xcite : @xmath51 for illustration , fig .
2 shows the best fit for @xmath28-type solution . comparing fig . 2 and fig . 1
, one sees that after inclusion of su(3 ) breaking in cabbibo s scheme , the lines corresponding to @xmath52 and @xmath53 are both slightly shifted up and the only significant change is for the line corresponding to @xmath42 . all lines now intersect at one point which gives a unique solution of @xmath7 and @xmath8 for a given parameter set .
similar discussion can be carried out for @xmath26- , @xmath27- and @xmath15-type solutions .
it may be noted that all su(3 ) symmetry - breaking parameters considered in this paper are significantly smaller than the su(3 ) symmetric parameters @xmath7 and @xmath8 . compared to the result given in @xcite , our @xmath0 value for a given symmetry breaking
parameter set is unique for the known baryon decay modes .
it suggests that the entire pattern of existing hyperon semileptonic decay data can be very well described in a framework which is basically su(3 ) flavor symmetry with small su(3 ) symmetry - breaking effects .
therefore no evidence of @xmath54 violation for su(3 ) symmetry in hyperon beta decay data can be found .
as we mentioned in the introduction , the quark spin distributions deduced from the @xmath2 data depend on the @xmath0 ratio . in the qcd corrected quark parton model , we have @xmath55 + { { c_{s}}\over { 9}}\delta\sigma \eqno ( 15)\ ] ] where @xmath56 and @xmath57 represents the fraction of the proton spin carried by all the quarks and antiquarks , i.e. the net total quark helicity . where @xmath58 and @xmath59 , with @xmath60 , are qcd correction coefficients for nonsinglet and singlet terms @xcite . to simplify the notation , we have omitted the variable @xmath61 in the quantities listed above .
it should be noted that the anomalous gluon contributions @xcite and higher twist effects @xcite are not included in ( 15 ) .
the former is still a subject of debate and the latter is expected to be only a small correction at the low @xmath61 value ( for example , the e142 @xmath62 data ) . combining ( 15 ) and the following two relations @xmath63 @xmath64 one obtains @xmath65 + { { c_s}\over 9}\delta \sigma \eqno ( 18)\ ] ] hence the data @xmath66 and @xmath67 deduced from @xmath68 depend on @xmath0 value used as input in ( 18 ) . using @xmath69 , @xmath70 , and @xmath71 the emc data @xcite @xmath72 led to @xmath73 however , if instead , using @xmath74 and the same @xmath75 and @xmath76 as used in the emc analysis @xcite , one obtains @xmath77 one can see that by using a smaller @xmath78 value , @xmath66 increases and the magnitude of @xmath67 decreases . however , in contrast to the change of @xmath67 , the total quark helicity @xmath66 is not sensitive to @xmath78 .
this is consistent with the result given by lipkin @xcite . on the other hand ,
if we use @xmath79 up to @xmath80 and @xmath81 up to @xmath82 as given in @xcite , then ( 20 ) becomes @xmath83 comparing ( 21 ) with ( 20 ) , one sees that @xmath66 significantly increases after inclusion of higher order qcd radiative corrections , which are very important in spin analysis , especially at moderate @xmath61 range where the experiments performed .
most recently , e143 group obtained more accurate data of @xmath84 which gives @xmath85 @xcite with @xmath86 . from this , one obtains @xmath87 the difference between the central values of @xmath66 ( and @xmath67 ) in ( 22 ) and in ( 21 ) is due to that the data are taken at different @xmath61 and they have different qcd correction coefficients @xmath88 and @xmath89 . in obtaining ( 22 )
, @xmath86 has been used , but for ( 21 ) @xmath71 was used . however , considering that the errors in ( 21 ) are quite large , the results given in ( 22 ) and ( 21 ) are consistent within the errors . to avoid possible ambiguity caused by @xmath90 symmetry breaking effects , we may choose to only use the su(2 ) symmetry result ( 16 ) and do not use ( 17 ) .
from ( 15 ) and ( 16 ) , one can obtain a relation between @xmath66 and @xmath67 @xmath91 for the proton and similarly @xmath92 for the neutron , where @xmath93 actually , ( 23 ) and ( 24 ) are not independent , because the bjorken sum rule @xmath94 therefore one can not deduce the @xmath66 and @xmath67 separately , even we have both @xmath84 and @xmath95 data . it should be noted that the data @xmath68 and @xmath62 from the experimental measurements may not satisfy ( 26 ) . hence the r.h.s . of ( 23 )
can be different from that of ( 24 ) . to obtain @xmath66 and @xmath67 separately , we need another relation between these two quantities
. this can be obtained from ( 16 ) and ( 17 ) @xmath96 using most recent e143 data @xmath85 and @xmath97 @xcite , one obtains from ( 23 ) and ( 24 ) @xmath98 and @xmath99 they are shown in fig.3 ( line * 1 * for e143 proton data and line * 3 * for e143 neutron data , where y@xmath100 and x@xmath101 ) .
if we assume that there is no strange quark polarization , @xmath67=0 as predicted by the naive quark model , then @xmath102 from the proton data and @xmath103 from the neutron data .
they are consistent within the errors ( see line @xmath104 and line @xmath105 in fig .
however , using su(3 ) symmetry result eq.(27 ) and combining data ( 9 ) and ( 11 ) , one obtains @xmath106 which is also shown in fig.3 ( line * 2 * ) .
one can see that the strange quark polarization would be negative . from fig .
3 , one obtains that the range of @xmath67 would be @xmath107 if the su(3 ) symmetry is imposed .
it should be noted that if one can trust the earlier @xmath108 and @xmath109 elastic scattering data , @xmath110 @xcite ( which gives @xmath111 for e143 proton data and @xmath112 for e143 neutron data ) , then the su(3 ) symmetry relation ( 30 ) is not necessary .
from a general discussion of su(3 ) symmetry and its breaking , we show that the hyperon beta decay data can be well accommodated within the framework of the usual cabbibo s su(3 ) symmetric description with a small su(3 ) symmetry breaking proportional to the mass difference between strange and nonstrange quarks .
the f / d ratio is not far from the value previously used in the deep inelastic scattering analysis .
hence the result given by using su(3 ) symmetry on hyperon beta decays will not be significantly disturbed by su(3 ) symmetry breaking effects .
it implies that the total quark helicity is still far below naive quark model expectation and the strange quark polarization seems to be negative provided the anomalous gluon contributions and higher twist effects are neglected .
the authors thank p. q. hung for helpful discussions in early stage of this work .
we also thank h. j. weber for comments and suggestions .
this work was supported by the u.s .
department of energy and the institute of nuclear and particle physics , university of virginia , usa . g. altarelli and g. ross , _ phys .
* b212 * , 391 ( 1988 ) + a. v. efremov and o. v. teryaev , _ dubna report _ , jinr - e2 - 88 - 287 , ( 1988 ) + r. d. carlitz , j. d. collins and a. h. mueller , sl phys
b214 * , 229 ( 1988 )
|
it is shown that hyperon beta decay data can be well accommodated within the framework of cabbibo s su(3 ) symmetric description if one allows for a small su(3 ) symmetry breaking proportional to the mass difference between strange and nonstrange quarks .
the @xmath0 ratio does not depend sensitively on the exact form of the symmetry - breaking , and the best fits are close to the value previously used in the analysis of deep inelastic scattering of electrons or muons on polarized nucleons .
the total quark helicity and strange quark polarization in the nucleon are discussed . # 1@xmath1#1
|
over the past decades , together with the development of the theory of quantum information @xcite there has been an increasing effort to find those physical systems where quantum information processing could be implemented . among the many different proposals , devices based on superconducting josephson junctions
are promising candidates in the solid state realm ( see the reviews in ref . ) .
josephson qubits can be categorized into three main classes : charge , phase and flux qubits , depending on which dynamical variable is most well defined and consequently which basis states are used as computational states @xmath0 and @xmath1 .
phase qubits @xcite , subject of the present investigation , in their simplest configuration can be realized with a single current biased josephson junction . for bias
lower that the critical current , the two lowest eigenstates of the system form the computational space .
the application of a current pulse , with frequency which is in resonance with the transition frequency of the two logical states , typically in the microwave range , allows to perform all desired single bit operations .
recent experiments @xcite have realized both single - bit and two - bit gates in capacitive coupled phase qubits . in the experiments conducted so far , motivated by similar approach in nmr , the amplitude of the microwave current used to perform the qubit manipulation has a gaussian shape @xcite .
the importance of achieving fast quantum gates with high fidelity rises the question whether there are modulations , other than gaussian , which lead to higher fidelities .
indeed , gaussian pulses act better when their duration time is longer @xcite , therefore the search for modulations which result in high fidelity gates , even when the duration time of the pulse is short , seems to be necessary .
some theoretical work has already been done in this direction to examine the efficiency of different modulations @xcite . in the present paper
we follow a different approach as compared to ref . and show that by employing the quantum optimal control theory @xcite , we can further improve the ( theoretical ) bounds on the error of gate operations .
quantum optimal control has been already applied to optimize quantum manipulation of josephson nanocircuits in the charge limit @xcite . here
we want to test this method in the opposite regime of phase qubit @xcite and see whether it is possible to find optimal modulations of microwave pulses , with different duration times , which give very good fidelity for single bit operations .
the paper is organized as follows : in sec .
[ model ] we will describe the model for the phase qubit and the hamiltonian used in the rest of the paper . in sec .
[ gate ] we introduce the not quantum gate which we have chosen to optimize .
then a brief introduction to the quantum optimal control algorithm which is used for this work will be given in sec .
the numerical results for a phase qubit will be presented in sec .
the achieved accuracy for desired operation , discussed in sec .
[ optimal ] , is further tested against possible imperfections in the pulses shape ( sec . [ shapes ] ) , presence of off - resonance elements in the hamiltonian ( sec . [ off ] ) and possible presence of the inter - qubit capacitive interaction in multi - qubit systems ( sec .
[ int ] ) .
the specific question of the leakage out of the hilbert space is addressed in sec . [ five ] , where we provide numerical results obtained for a junction with five levels inside its potential .
a summary of the results obtained and possible perspectives of this work will be presented in the concluding remarks in sec .
a phase qubit can be realized by a flux - biased rf squid @xcite , a low inductance dc squid @xcite or a large inductance dc squid @xcite . in its simplest design
a phase qubit consists of a single josephson junction ( fig .
[ fig : junction](a ) ) with critical current @xmath2 and a biasing dc current @xmath3 .
the hamiltonian has the form @xmath4 where @xmath5 and @xmath6 are , respectively , the charging energy and the josephson energy of the junction with capacitance @xmath7 , @xmath8 being the quantum of flux , and @xmath9 represents the josephson phase across the junction . the regime in which the superconducting phase @xmath9 is the appropriate quantum variable
is reached when @xmath10 .
the potential energy of the system , as a function of @xmath9 , has the form of tilted washboard with quantized energy levels inside each well ( fig .
[ fig : junction](b ) ) . when @xmath11 there are few levels inside each well and the two lowest states @xmath0 and @xmath1 , with energies @xmath12 and @xmath13 and transition frequency @xmath14ghz , can be used as computational states .
the transition between the two lowest states is made by use of a microwave current @xmath15 of frequency @xmath16 which is in resonance with transition frequency @xmath17 .
transition to higher states ( @xmath18 ) , which are out of qubit manifold , are off - resonance due to the anharmonicity of the potential well . , resistance @xmath19 and critical current @xmath2 which is biased by a @xmath20 current @xmath3 .
a microwave pulse @xmath21 , with frequency @xmath22 , is applied to make transitions between the two lowest energy levels of the system @xmath0 and @xmath1 .
( b ) by neglecting the resistive branch , the potential energy of the system @xmath23 , as a function of the josephson phase across the junction @xmath9 , has the form of a tilted washboard potential .
this potential is defined by the height of the well @xmath24 and the frequency of the classical oscillations in the bottom of the well .
[ fig : junction ] ] to be effectively used as a two - level quantum system , the junction is biased with a dc current slightly smaller than the critical current @xmath11 . in this regime
the potential energy of the system can be approximated by a cubic potential and the hamiltonian ( [ eq : washboard ] ) becomes : @xmath25 the application of a microwave current @xmath26 is taken into account by adding the linear term @xmath27 to the hamiltonian ( [ eq : washboard ] ) . since the eigenstates of the junction biased with a dc current are used as computational states , it is appropriate to write the full hamiltonian in the basis of the eigenstates @xmath28 of the system with dc bias current . to examine the effect of microwave current , one needs to know the elements of the superconducting phase @xmath9 in this basis . moving to the rotating frame ,
in which the fast oscillations due to @xmath29 do not appear , the hamiltonian @xmath30 in the rotating frame is related to the hamiltonian in laboratory frame @xmath31 via @xmath32 whereas the state of the system in the rotating frame is @xmath33 . by introducing @xmath34 and @xmath35 , and considering only the first three levels in the well , the hamiltonian of the phase qubit in the rotating frame takes the following form @xmath36 here we have set @xmath37 , @xmath38 and we have assumed that off - resonance terms have negligible effect . as we shall see , by a proper choice of @xmath39 and microwave current modulation @xmath40 it is possible to perform single - bit operations on the computational states @xmath0 and @xmath1 .
as one can see from the @xmath41 top - left block of the hamiltonian ( [ eq : rotating2 ] ) , the initial phase of the microwave pulse @xmath39 defines the axis of rotation , in @xmath42-plane of the bloch sphere , for a given state , while the pulse amplitude and duration time define the angle of rotation .
for example , by setting @xmath43 ( @xmath44 ) such block is proportional to the pauli matrix @xmath45 ( @xmath46 ) , _
i. e. _ a rotation around the @xmath47(@xmath48)-axis . in a recent experiment @xcite
a @xmath49 rotation around @xmath47 has been implemented as a part of a sequence of operations to create entanglement between two phase qubits .
this motivates us to set @xmath43 and focus this work on the single - qubit not - gate operation consisting of a @xmath49 rotation around the @xmath47-axis . in the typical experiment a shaped pulse with the following gaussian modulation @xcite @xmath50
is used to induce flips between states @xmath0 and @xmath1 and vice versa .
here @xmath51 , @xmath52 and @xmath53 are , respectively , the amplitude , characteristic width and total width of the pulse , @xmath54 being the cut - off of the pulse in time .
the actual result of the operation can be quantified by the fidelity @xmath55 , where @xmath56 is the desired final state and @xmath57 is the state achieved at the end of time evolution starting from initial state @xmath58 . for a @xmath49 rotation and with a typical cut - off value ( @xmath59 )
the amplitude @xmath60 yields a pretty high fidelity of rotation .
more precisely , fig .
[ fig : error - example ] shows the error @xmath61 for a not - gate operation on an arbitrary superposition @xmath62 , which would result in the state @xmath63 , using a gaussian pulse with cut - off @xmath64 , amplitude @xmath65 and duration time @xmath66 .
the leakage outside the qubit manifold , defined as @xmath67 , is also shown in fig . [
fig : leakage - example ] .
it is worthwhile noting that although the leakage , for long enough pulses , can be of the order of @xmath68 , the error in the not - gate operation is always higher than @xmath69 ( see ref . ) . for a not - gate operation on an arbitrary superposition of states @xmath0 and @xmath1 after applying a gaussian pulse with amplitude @xmath65 , cut - off @xmath64 as a function of the duration time @xmath66 .
[ fig : error - example ] ] and @xmath1 after applying a gaussian pulse with amplitude @xmath65 , cut - off @xmath64 as a function of the duration time @xmath66 .
[ fig : leakage - example ] ]
as we mentioned in the introduction , in this work we use quantum optimal control theory in order to obtain microwave current modulations which give rise to a high - fidelity not - gate operation for a phase qubit . in this section
we briefly review the optimal control algorithm which we have employed to obtain optimized modulations . in general , quantum optimal control algorithms @xcite
are designed to lead a quantum system with state @xmath70 from an initial state @xmath71 to a target final state @xmath56 at time @xmath66 by minimizing a cost functional which is a measure of inaccuracy of reaching the desired final state . if @xmath57 denotes the state achieved at time @xmath66 , one can consider two different cost functionals : * @xmath72 + by minimizing this cost functional , although the population of the desired state @xmath56 will be maximized , the overall phase of this state is not forced to be preserved . *
@xmath73 + minimization of this second cost functional , in addition to maximizing the population of the desired state , preserves its overall phase . in optimal control theory the minimization of the cost functional
is done by updating the hamiltonian of the system , via some control parameters , in an iterative procedure until the desired value of the cost functional is reached .
any specific algorithm which is guaranteed to give improvement at each iteration @xcite is called _ immediate feedback control _ and can be briefly described as follows : assume that the hamiltonian of the system depends on a set of parameters @xmath74 which are controllable . by using a proper initial guess @xmath75 for control parameters , first the state of the system @xmath70 is evolved in time with the initial condition @xmath71 giving rise to @xmath57 after time @xmath66 . at this point
the iterative algorithm starts , aiming at decreasing the cost functional by adding a correction to control parameters in each step . in the @xmath76th step of this iterative algorithm *
an auxiliary state @xmath77 is evolved backward in time starting from @xmath78 reaching @xmath79 .
+ in the case of minimizing @xmath80 , @xmath81 and for minimizing @xmath82 , @xmath83 . *
the states @xmath79 and @xmath84 are evolved forward in time , respectively , with control parameters @xmath85 and @xmath86 . here , @xmath87\end{aligned}\ ] ] are updated control parameters .
@xmath88 is a weight function used to fix initial and final conditions on the control parameters in order to avoid major changes at the beginning and end of time evolution and is an important parameter for the convergence of the algorithm .
these two steps are repeated until the desired value of @xmath80 or @xmath82 is obtained . in order to implement the optimization procedure to a not - gate for any arbitrary superposition of computational states
, one must be able to flip @xmath0 and @xmath1 at the same time ( _ i .
e. _ with same pulse ) making sure that the phase relation between them is preserved .
this is guaranteed by using the following definition of fidelity @xmath89 where @xmath90 and @xmath91 are final states achieved at time @xmath66 after applying the same pulse on initial states @xmath0 and @xmath1 .
the minimization of the cost functional @xmath80 , for flipping at the same time the states @xmath0 and @xmath1 , does not necessarily lead to maximization of the fidelity ( [ eq : not - fidelity ] ) due to possible changes in the phase relation between them .
however if @xmath82 is minimized , the maximal fidelity is also guaranteed . therefore in order to obtain a high - fidelity not - gate it seems more natural to minimize @xmath82 instead of @xmath80 . however , in the following we will show that , although in the ideal case optimized pulses obtained from minimizing @xmath82 result in much higher fidelity , when more realistic cases are considered optimized pulses from minimizing @xmath80 lead to higher fidelities , specially for very short pulses . in this work we often use the error @xmath92 instead of fidelity .
in this section we present the numerical results to show that the quantum optimal control theory allows to optimize the modulation of microwave pulses in order to implement a high - fidelity not - gate .
the optimization is done in the rotating frame and the hamiltonian ( [ eq : rotating2 ] ) is used for time evolution while @xmath93 are calculated by means of perturbation theory . and @xmath1 made by pulses with gaussian modulation ( circles ) and optimized modulation obtained from minimizing @xmath80 ( unfilled triangles ) and @xmath82 ( squares ) in a three - level system .
gaussian pulses have amplitude @xmath65 and cut - off in time @xmath64 .
optimized pulses are obtained after at most 5000 iterations and using gaussian pulses as initial guess .
filled triangles are obtained by applying a @xmath94 phase shift after optimized pulses obtained from minimizing @xmath80 .
[ fig : optimized ] ] [ cols="^,^ " , ] fig .
[ fig : fourier ] shows the fourier transform of the two optimized pulses shown in fig .
[ fig : pulses ] . to filter out the high - frequency components of the optimized pulses we set a cutoff frequency @xmath95 for fourier components and apply the truncated pulses again and obtain the error . fig .
[ fig : cutoff ] shows the error for a not - gate for pulses with different duration times as functions of @xmath95 .
@xmath96 is approximately @xmath97 ghz and @xmath98 is typically @xmath99 of @xmath17 . in our calculation
@xmath100 which means that @xmath101 mhz . in the case of @xmath102 ,
top panel of figure [ fig : cutoff ] makes clear that all important harmonics have frequencies smaller than @xmath103 .
note that the number of harmonics included within the cutoff is equal to @xmath104 so that , for @xmath105 , such number is equal to n times the ratio @xmath106 . as a result it seems that , for all values of @xmath66 considered , about 20 harmonics should be sufficient to reach the smallest value of @xmath102 .
@xmath107 , though , seems to be more sensitive to high - frequency components but still about four orders of magnitude smaller than @xmath102 under the cutoff @xmath108 . for a not - gate with optimized pulses obtained by minimizing @xmath80 ( top panel ) and @xmath82 ( bottom panel ) as a function of the cutoff frequency @xmath95 .
integer values of @xmath104 correspond to the number of fourier components included .
@xmath109 is approximately @xmath110 mhz .
[ fig : cutoff ] ] for a not - gate made by applying microwave pulses with gaussian modulation ( circles ) and optimized modulation ( triangles and squares ) when off - resonance terms are kept .
note that optimized pulses are obtained by excluding off - resonance elements .
bottom panel : the absolute value of error difference @xmath111 obtained by subtracting the curves in the top panel from those in fig .
[ fig : optimized ] .
@xmath66 is the total time width of the pulses .
[ fig : off - res ] ] as we mentioned before , we have assumed that off - resonance elements of the hamiltonian ( [ eq : lab - to - rot ] ) in the rotating frame are negligible and we have used hamiltonian ( [ eq : rotating2 ] ) for calculating the evolution . in this section
we check this assumption by addressing the effect of off - resonance elements by evolving the complete hamiltonian ( [ eq : lab - to - rot ] ) using the optimized pulses obtained using hamiltonian ( [ eq : rotating2 ] ) .
top panel of fig .
[ fig : off - res ] shows the error for a not - gate operation implemented by gaussian ( circles ) and optimized pulses from minimizing @xmath80 ( triangles ) and @xmath82 ( squares ) . for @xmath112
the optimized pulses yield a much higher error , with respect to the case when off - resonance terms are neglected , still showing an improvement of two orders of magnitude if compared to gaussian pulses .
bottom panel of fig .
[ fig : off - res ] shows the absolute value of the error difference @xmath113 obtained by subtracting the error without off - resonance term from the error with off - resonance terms .
these figures make clear that while for gaussian pulses off - resonance terms can be neglected , for optimized pulses , specially those obtained from minimizing @xmath82 , they are very important .
note that , contrary to the ideal case where @xmath107 was about eight orders of magnitude smaller than @xmath102 , under the effect of off - resonance terms , @xmath107 seems to be larger than @xmath102 specially for very short pulses with @xmath114 .
this means that the assumption of ignoring these terms is more accurate when @xmath80 is minimized .
the simpler shape of the optimized pulses obtained from minimization of @xmath80 could be a reason for that .
so far we have considered a single qubit with three energy levels and obtained the modulation of the microwave pulses in order to optimize the not - gate operation for the two lowest energy states @xmath0 and @xmath1 .
it is now interesting to consider the setup @xcite containing two qubits interacting via a capacitor .
the question that we want to address is what happens if these optimized pulses are applied on the first qubit while the interaction with the second qubit is present .
the interaction hamiltonian of a circuit with two identical phase qubits has the following form : @xmath115\end{aligned}\ ] ] where @xmath116 and @xmath117 are josephson phases across the junction @xmath118 and @xmath119 and @xmath120 is the capacitance of the interaction capacitor .
note that the term with second derivative in eq .
( [ eq:2bithint ] ) can be included in the hamiltonians of the uncoupled qubits ( [ eq : washboard ] ) by replacing the charging energy @xmath121 with an effective one @xmath122 , where @xmath123 and @xmath124 .
the hamiltonian can again be written in the basis of the eigenstates of the uncoupled qubits , and the strength of the interaction hamiltonian reduces to @xmath125 .
we move to the rotating frame described by the unitary operator @xmath126 and neglect the off - resonance elements of the resulting hamiltonian . by applying microwave pulse on the first qubit ,
our aim is to perform a not - gate operation on such qubit ( namely , @xmath127 ) .
since @xmath120 is typically of few ff and @xmath7 is of the order of pf @xcite we find @xmath128 which leads to an interaction strength @xmath129mhz .
figure [ fig : interaction ] shows the error as function of time width of the pulse @xmath66 for both gaussian ( circles ) and optimized pulses ( triangles and squares ) .
although the optimized pulses are obtained for a single qubit system , they still result in smaller error at least for short pulses .
these results show the importance of the presence of the capacitive interaction even though the strength of the interaction is small . as it is clear from fig .
[ fig : interaction ] for pulses longer than , approximately , @xmath130 ns ( @xmath131 ) the interaction becomes more effective and the error for gaussian and optimized pulses are very close .
moreover , longer pulses lead to higher value of error contrary to what happens in the case of a single qubit . in this case also for very short pulses ( @xmath132 ) @xmath102 is smaller than @xmath107 and for @xmath133 they are of the same order , although in the ideal case @xmath102 was eight orders of magnitude larger than @xmath107 . for a not - gate on the first qubit implemented by applying gaussian ( circles ) and optimized ( triangles and squares ) pulses on the first qubit of a two - qubit system in presence of capacitive interaction .
gaussian pulses have amplitude @xmath65 and cutoff in time @xmath64 and are used as initial guess in the optimization procedure .
optimized pulses are obtained from the single - qubit setup .
the strength of the interaction is @xmath134 .
bottom panel : the absolute value of error difference @xmath111 obtained by subtracting the curves in the top panel from those in fig .
[ fig : optimized ] .
[ fig : interaction ] ] bottom panel of fig .
[ fig : interaction ] shows the absolute value of error difference @xmath111 which is obtained by subtracting the error of ideal case from the error in presence of interaction .
@xmath111 is almost the same in all three cases .
as explained in section [ model ] , the two lowest energy levels of a current - biased josephson junction can be used as @xmath0 and @xmath1 states of a phase qubit . although it would be desirable to have only two levels inside the potential well of the fig .
[ fig : junction ] , this is not the case in experimental setups .
so far we have included the leakage by considering only an additional third level and showed that it is possible to optimize the pulses in order to gain high fidelity for a not - gate for a single qubit . in typical experiments the number of energy levels inside the well varies between three and five . in order to have a more complete understanding of the leakage , in this section we show some results obtained for a five - level system . since adding more levels to the system decreases the inhomogeneity of the level - spacing we choose @xmath135 . of not - gate made by applying gaussian ( circles ) and optimized ( triangles and squares ) pulses as function of duration time of the pulse @xmath66 .
gaussian pulses have amplitude @xmath65 and cutoff in time @xmath64 and are used as initial guess in optimization .
optimized pulses are obtained after at most @xmath136 iterations .
physical system contains five energy states and @xmath98 is assumed to be @xmath137 .
[ fig:5-optimized ] ] figure [ fig:5-optimized ] shows the error @xmath138 for a not - gate implemented by optimized and gaussian pulses , which are used as initial guess , for different duration times @xmath66 .
similar to the case of three - level system , with same number of iterations , minimization of @xmath80 leads to better results for short pulses while for longer duration times of pulses minimizing @xmath82 results in error of not - gate @xmath139 . in the case of minimizing @xmath80 ,
at least one order of magnitude improvement is achieved for long pulses , although the improvement obtained for pulses with shorter time width are the best . by looking at the average value of @xmath80 for transitions between the states @xmath0 and @xmath1 ( top panel of fig . [ fig:5-phasediff ] ) and the final phase difference between them ( bottom panel of fig .
[ fig:5-phasediff ] ) , one realizes that , as it was observed in three - level system , considerable amount of @xmath102 is due to the final phase difference @xmath140 .
for instance the pulse with @xmath141 results in a phase difference approximately equal to zero and therefore @xmath102 for this pulse is of the order of @xmath142 .
a proper phase shift applied after the not - gate operation will compensate the phase difference between the final @xmath0 and @xmath1 states and consequently attaining a very high fidelity .
two examples of pulses with @xmath143 are shown in figure [ fig:5-pulses ] .
@xmath144 is obtained from minimizing @xmath80 and gives rise to @xmath145 while @xmath146 is supposed to minimize @xmath82 with @xmath147 .
it seems that , compared to three - level system , higher frequencies and amplitudes are needed to reach high fidelity of not - gate . in three - level system
the iterative optimization algorithm is applied at most @xmath148 times to reach such fidelities while with five levels @xmath136 iterations were needed to obtain the results shown in figures [ fig:5-optimized ] and [ fig:5-phasediff ] .
the leakage out of the qubit manifold would be the reason for this . for @xmath149 transitions
after applying the gaussian pulses ( circles ) and optimized pulses ( triangles ) obtained from minimizing @xmath80 .
vertical axis is in logarithmic scale .
bottom panel : the phase difference between final @xmath0 and @xmath1 states in units of @xmath49 .
[ fig:5-phasediff ] ] ( top panel ) and @xmath82 ( bottom panel ) , and the corresponding pulse with gaussian modulation ( solid lines ) used as initial guess in optimization process with duration time @xmath141 in a system with five energy states .
[ fig:5-pulses ] ]
in this paper we have shown that it is possible to optimize single - qubit gates for josephson phase qubits by employing quantum optimal control theory .
we have considered the realistic situation in which , in addition to the two computational basis states @xmath0 and @xmath1 , higher energy states are present , which may lead to leakage . typically microwave pulses with gaussian modulation
are used to induce transition between states @xmath0 and @xmath1 , yielding a quite high fidelity for long pulse durations . for the sake of definiteness , here we have focused on the not - gate single - qubit operation and searched for modulations of microwave pulses which optimize such operation , especially for short - duration pulses .
the numerical results obtained for a three - level system , and neglecting off - resonance terms , demonstrate up to ten orders of magnitude improvement in fidelity of a not - gate operation with respect to those obtained through gaussian modulations . to test the effect of possible imperfections in the pulses shape
, we have studied the behavior of the fidelity as a function of the bandwidth of the pulse generator and showed that frequencies not bigger @xmath119 ghz are needed to gain up to four orders of magnitude improvement .
moreover , we have shown that the off - resonance elements of the hamiltonian , which are usually neglected , can be important for optimized pulses , especially for short pulse duration times , due to the very high fidelity reached .
we have also addressed the effect of the presence of a capacitively - coupled second qubit and showed that , even though the optimized pulses are obtained for a single qubit , they still lead to a high fidelity for a not - gate ( up to two orders of magnitude improvement ) especially for very short pulses . finally , we were able to obtain optimized pulses for a system with 5 energy levels in which the leakage outside qubit manifold is more severe . in conclusion
, the two - interacting - qubit system deserves for sure further attention . on the one hand , in order to improve the fidelity of a single - qubit operation , in the presence of capacitive coupling , it seems that a way to switch the interaction on and off should be found even for optimized pulses of an isolate qubit . on the other hand , obtaining optimized pulses while including the interaction , would be a potential theoretical work to be done .
we would like to acknowledge fruitful discussions with f. w. j. hekking .
we acknowledge support by ec - fet / qipc ( eurosqip ) and by centro di ricerca matematica `` ennio de giorgi '' of scuola normale superiore .
s. m. acknowledges support by eu - project scala .
99 m. a. nielsen and i. l. chuang _ quantum computation and quantum information _
( cambridge , 2000 ) . y. makhlin , g. schn and a. shnirman , , 357 ( 2001 ) .
g. wendin and v. s. shumeiko , fiz .
nizk . temp . * 33 * , 957 ( 2007 ) [ low temp . phys . * 33 * , 724 ( 2007 ) ] .
you and f. nori , phys . today * 58 * , 42 ( 2005 ) j. clarke and f. k. wilhelm , nature * 453 * , 1031 ( 2008 ) j. m. martinis , s. nam , j. aumentado and c. urbina , , 117901 ( 2002 ) . j. m. martinis , s. nam , j. aumentado , k. m. lang and c. urbina , , 094510 ( 2003 ) .
j. claudon , f. balestro , f. w. j. hekking , and o. buisson , , 187003 ( 2004 ) .
m. steffen , m. ansmann , r. mcdermott , n. katz , r. c. bialczak , e. lucero , m. neeley , e. m. weig , a. n. cleland , and j. m. martinis , , 050502 ( 2006 ) .
m. steffen , m. ansmann , r. c. bialczak , n. katz , e. lucero , r. mcdermott , m. neeley , e. m. weig , a. n. cleland , j. m. martinis , science * 313 * , 1423 ( 2006 ) .
m. steffen , j. m. martinis , and i. l. chuang , , 224518 ( 2003 ) .
a. p. peirce , m.a .
dahleh , and h. rabitz , , 4950 ( 1988 ) a. borzi , g. stadler , and u. hohenester , , 053811 ( 2002 ) s.e .
sklarz and d.j tannor , , 053619 ( 2002 ) t. calarco , u. dorner , p. julienne , c. williams , and p. zoller , , 012306 ( 2004 ) .
a. sprl , t. schulte - herbrggen , s. j. glaser , v. bergholm , m. j. storcz , j. ferber , and f. k. wilhelm , , 012302 ( 2007 ) .
s. montangero , t. calarco , and r. fazio , , 170501 ( 2007 ) .
safaei , s. montangero , f. taddei , and r. fazio , , 144522 ( 2008 ) .
see also p. rebentrost and f. k. wilhelm , arxiv:0808.2680v1
r. w. simmonds , k. m. lang , d. a. hite , s. nam , d. p. pappas , and j. m. martinis , , 077003 ( 2004 ) .
r. fazio , g.m .
palma and j. siewert , phys .
lett . * 83 * 5385 ( 1999 ) .
i. r. sola , j. santamaria , and d.j .
tannor , j. phys . chem . * 102 * , 4301 ( 1998 )
|
in a josephson phase qubit the coherent manipulations of the computational states are achieved by modulating an applied ac current , typically in the microwave range . in this work
we show that it is possible to find optimal modulations of the bias current to achieve high - fidelity gates .
we apply quantum optimal control theory to determine the form of the pulses and study in details the case of a not - gate . to test the efficiency of the optimized pulses in an experimental setup
, we also address the effect of possible imperfections in the pulses shapes , the role of off - resonance elements in the hamiltonian , and the effect of capacitive interaction with a second qubit .
|
the virtue of direct photons is that they emerge directly from the interaction of point - like partons in processes like quark - gluon compton scattering ( @xmath2 ) and quark - anti - quark annihilation ( @xmath3 ) .
measurements of direct photons in p+p collisions therefore allow to test perturbative qcd ( pqcd ) without the complication of phenomenological parton - to - hadron fragmentation functions which are needed in the description of high-@xmath1 hadron production .
moreover , direct - photon measurements in p+p collisions provide information about the gluon distribution of the proton due to the large contribution from quark - gluon compton scattering .
this is especially interesting for fractional gluon momenta @xmath4 because this range is not well constrained by other processes like deep - inelastic scattering and drell - yan .
however , direct - photon data from @xmath5 and @xmath6 collisions are not generally used in global qcd fits for the determination of parton distribution functions .
one reason for this are discrepancies between data and pqcd , especially at low energies ( see e.g. figure [ fig : cross_sec_pp]c ) .
the relativistic heavy ion collider ( rhic ) delivers spin polarized proton beams .
a further motivation for direct - photon measurements in phenix is to gain insight into the contribution of gluons to the spin of the proton by comparing direct - photon yields in p+p collisions with equal and opposite proton helicities . in this paper , however , the direct - photon cross - section for unpolarized p+p collisions is presented . the experimental challenge in direct - photon measurements is the subtraction of the significant photon background from hadron decays like @xmath7 and @xmath8 .
the comparison of world data on direct - photon production in @xmath5 and @xmath6 collisions with pqcd calculations raises the question whether there is a systematic pattern of deviation .
it has been argued that measured direct - photon transverse momentum ( @xmath1 ) spectra generally tend to be steeper than pqcd predictions @xcite . in this case
the agreement between data and pqcd can be improved by introducing a transverse momentum ( @xmath9 ) of the partons prior to the hard scattering , in addition to the effects expected in the next - to - leading - order description .
recently published results on direct - photon production in p+p and p+be collisions from the fixed - target experiment e706 provide further support for the idea that @xmath9 enhancement needs to be taken into account to describe the data @xcite .
direct - photon spectra in @xmath0gev p+p collisions were measured at rhic with the highly segmented electromagnetic calorimeter ( emcal ) of the phenix experiment @xcite .
this detector consists of two subsystems : a lead scintillator sampling calorimeter ( pbsc ) and a lead glass cherenkov calorimeter ( pbgl ) .
the emcal covers the pseudorapidity range @xmath10 .
the direct - photon cross - section for unpolarized p+p collisions determined with data from rhic run-3 is shown in figure [ fig : cross_sec_pp ] .
the integrated luminosity for this run was 266nb@xmath11 .
the independent pbsc and pbgl measurements are consistent and agree with nlo pqcd predictions
. there is , in particular , no indication that for the range @xmath12gev/@xmath13 covered by the current measurement an additional transverse momentum @xmath9 is needed in the theoretical description .
experimental and theoretical uncertainties are potentially reduced in the ratio of the direct - photon and neutral - pion cross - sections .
figure [ fig : gam_pi0_ratio ] shows that the measured @xmath14 ratio agrees with nlo pqcd predictions .
the akk fragmentation functions @xcite provide a better description of the phenix @xmath15 spectrum @xcite than the kkp fragmentation functions and this in turn leads to a better agreement between the measured and calculated @xmath14 ratio .
direct - photon measurements in nucleus - nucleus ( a+a ) collisions essentially serve two purposes . at sufficiently high transverse momenta direct photons emerge from initial hard parton - parton scattering , analogous to the production mechanism in p+p collisions .
these hard parton - parton scatterings happen prior to the possible formation of a thermalized quark - gluon plasma ( qgp ) .
however , due to their electromagnetic nature photons essentially do nt interact with the hot and dense plasma .
they can thus be used as a measure of the number of initial hard parton - parton scatterings . the second incentive to measure direct photons in a+a collisions
is the expected production of thermal direct photons in the qgp .
thermal direct photon are predominantly produced in the early hot phase of the evolution of the plasma and thus provide a means to determine the initial temperature of the qgp . in a window
@xmath16gev/@xmath13 thermal direct photons from the qgp are expected to be the dominant direct - photon source in au+au collisions at @xmath17gev .
one of the most important observations at rhic is the suppression high-@xmath1 hadrons relative to a scaled p+p reference .
the suppression is quantified with the nuclear modification factor @xmath18 the factor @xmath19 reflects the increase of the parton luminosity per a+a collision relative to p+p collisions in the absence of nuclear shadowing .
thus , in the absence of initial - state or final - state nuclear effects @xmath20 will be unity for sufficiently large transverse momenta for which particle production is dominated by hard scattering .
the average number @xmath21 of inelastic nucleon - nucleon collisions for a given centrality class is determined with a glauber model calculation . as can seen in figure [ fig : raa ] neutral pions , @xmath22-mesons , and charged hadrons are suppressed by a factor @xmath23 in central au+au collisions at @xmath0gev .
this can be explained by energy loss of fast quarks and gluons in the qgp . in such models the initial rate of hard parton - parton scatterings scale as @xmath24 and
the suppression is only due to the final - state interaction with the medium .
the measurement of direct photons at high @xmath1 provides a unique possibility to test these jet quenching models @xcite .
figure [ fig : raa ] shows that the @xmath20 for direct photons in central au+au collisions is consistent with unity : unlike hadrons , direct photons are not suppressed .
one can conclude that initial hard scattering processes in au+au collisions indeed occur at the rate expected from @xmath25 scaling .
the suppression of hadrons at high @xmath1 is therefore a final - state effect caused by the hot and dense medium produced in these collisions .
this supports models which attribute the hadron suppression to energy loss of partons in the medium and is consistent with the formation of a qgp in au+au collisions at rhic .
s. s. adler _ et al .
_ [ phenix collaboration ] , phys .
lett . * 94 * , 232301 ( 2005 ) s. s. adler _ et al . _
[ phenix collaboration ] , phys .
* 91 * ( 2003 ) 241803 k. okada [ phenix collaboration ] , arxiv : hep - ex/0501066 . w. vogelsang and m. r. whalley , j. phys .
g * 23 * ( 1997 ) a1 .
j. huston , e. kovacs , s. kuhlmann , h. l. lai , j. f. owens and w. k. tung , phys .
d * 51 * ( 1995 ) 6139 l. apanasevich _ et al . _ [ e706 collaboration ] , phys .
d * 70 * ( 2004 ) 092009 b. a. kniehl , g. kramer and b. potter , nucl .
b * 582 * ( 2000 ) 514 s. albino , b. a. kniehl and g. kramer , nucl .
b * 725 * ( 2005 ) 181
|
direct - photon measurements in p+p and au+au collisions at @xmath0gev from the phenix experiment are presented .
the p+p results are found to be in good agreement with next - to - leading - order ( nlo ) perturbative qcd calculations .
direct - photon yields in au+au collisions scale with the number of inelastic nucleon - nucleon collisions and do nt exhibit the strong suppression observed for charged hadrons and neutral pions .
this observation is consistent with models which attribute the suppression of high-@xmath1 hadrons to energy loss of quarks and gluons in the hot and dense medium produced in au+au collisions at rhic .
|
the study of the boundary behavior of holomorphic functions on domains in @xmath2 is a deep and classical branch of complex analysis .
one of the most natural problems in this context is to relate the properties of a function at a boundary point to those of its derivatives close to that point . in this paper , we address such questions for certain rational functions in two complex variables having additional algebraic structure .
as we shall see , these investigations involve a rich and fascinating interplay betwen function theory , harmonic analysis of several complex variables , and classical algebraic geometry in the guise of algebraic curves . in several variables ,
a complicating phenomenon without counterpart in the one - variable setting manifests itself ; namely , a rational function can possess a `` non - essential singularity of the second kind , '' meaning its numerator and denominator vanish at the same point without sharing a common factor . from an analytic standpoint ,
such a boundary singularity affects the boundedness , smoothness , and integrability properties of both the rational function and its derivatives in very nontrivial ways . to understand this problem further
, we are led to develop a precise geometric theory of singularities of such functions using a combination of complex analysis and complex geometry .
we specialize to a specific domain in @xmath3 namely the unit bidisk @xmath4 in this setting , greg knese recently initiated an impressive study of integrability and regularity properties of rational functions , see @xcite .
specifically , his theory gives a method for determining if a given rational function @xmath5 is square integrable on the distinguished boundary of the bidisk @xmath6 , that is if @xmath7 .
the knese integrability theory motivates the question of whether or not the derivatives of a rational function are square - integrable , or in other words , whether they belong to the hardy space @xmath8 . for general rational functions ,
such a question is very challenging .
however , the arguments and methods in @xcite rest on properties of a special subset of rational functions .
specifically , we say a holomorphic function @xmath9 is _ inner _ if @xmath10 for almost every @xmath11 then the subclass of _ rational inner functions _ ( rifs ) consists of inner functions @xmath12 , where @xmath13 and @xmath14 are polynomials in two complex variables .
rational inner functions on the bidisk are necessarily of the form @xmath15 where @xmath13 is a polynomial of bidegree @xmath16 with no zeros in the bidisk , @xmath17 is the reflection of @xmath13 , and @xmath18 is a unimodular constant , see ( * ? ? ?
* chapter 5.2 ) . in
what follows , we specialize to rational inner functions of the form @xmath19 , but the results for rifs with monomial factors follow easily .
inner functions on the unit disk play an essential role in single - variable operator and function theory . indeed ,
many hilbert space contractions are unitarily equivalent to the compression of the shift to a model space , namely a hilbert space associated to an inner function @xcite .
similarly , much of analytic function theory on the disk rests on the classic beurling - lax theorem and inner - outer factorizations of one - variable hardy space functions @xcite .
inner functions on the bidisk are also quite important . as in the one - variable setting , they induce interesting hilbert spaces associated to useful operators , see for example @xcite and the references therein .
although inner - outer factorization fails on the bidisk , every bounded holomorphic function on @xmath20 can still be approximated locally - uniformly by constant multiples of inner functions @xcite .
rational inner functions play a particularly critical role in two - variable function theory . for example , rifs provide key examples of two - variable matrix monotone functions @xcite and provide solutions to every solvable pick interpolation problem on the bidisk @xcite
. moreover rifs and their denominators , called _ stable polynomials _ , have close connections to applications in systems and control engineering , electrical engineering , and statistical mechanics ; see @xcite and the references cited within those papers . because of their critical role in function theory on the bidisk and their particularly nice formulas , we restrict attention to rifs .
specifically , in this paper , we will address two integrability questions related to rational inner functions and their derivatives .
first recall that a holomorphic function @xmath21 on the bidisk belongs to the _ hardy space _
@xmath22 for @xmath23 if @xmath24 here , @xmath25 , denotes lebesgue measure on @xmath26 .
the space @xmath27 is the banach algebra of bounded holomorphic functions on the bidisk .
see @xcite for the basic theory of @xmath0-spaces on polydisks .
since rifs are bounded , they immediately belong to every @xmath22 for @xmath28 we can now pose the following question .
[ q1 ] if @xmath29 is a rational inner function on @xmath20 , under what conditions is @xmath30 for @xmath31 and @xmath32 ? in the unit disk @xmath33
analogous and more general forms of question [ q1 ] have been considered in numerous publications going back to work of ahern and clark @xcite ; see for instance @xcite and the references therein for subsequent developments . in higher dimensions ,
see @xcite , it is known that an inner function on the unit ball @xmath34 whose gradient is in @xmath35 must be constant . furthermore , a number of authors have examined the related question of boundary regularity of bounded holomorphic functions on @xmath20 , see @xcite and the references provided there .
however as far as the authors are aware , question [ q1 ] has been open in the setting of the bidisk . to answer this question ,
we study the geometry of boundary singularities of rational inner functions and characterize derivative @xmath0-inclusion in terms of how the rif s zero set makes contact with @xmath36 we are also interested in the integrability of higher derivatives , in particular the mixed partials @xmath37 .
one reason for this is that weighted area integrability of mixed partials is an equivalent condition for membership in _ dirichlet - type spaces _ @xmath38 .
these are spaces of functions @xmath39 that are holomorphic in the bidisk and satisfy , for @xmath40 fixed , the norm boundedness condition @xmath41 note that @xmath42 coincides with the hardy space @xmath43 , while @xmath44 is the _ bergman space _
@xmath45 , and @xmath46 is the _ dirichlet space _ of the bidisk
. we will refer to the spaces @xmath38 as _ weighted dirichlet spaces _ when we wish to emphasize that we are considering parameters @xmath47 .
dirichlet - type spaces are important for several reasons .
the unweighted dirichlet norm leads to a hilbert function space invariant under the action of automorphisms of the underlying domain @xcite , dirichlet - type spaces are connected with potential theory through the exceptional sets of their elements @xcite , and these spaces provide a natural setting for multivariate operator theory in the form of shift operators @xcite .
a characterization of cyclic polynomials for shifts acting on @xmath38 spaces was obtained in a recent series of papers @xcite , but the contents and structures of dirichlet spaces on the bidisk remain somewhat obscure .
this leads to the following natural question : [ q2 ] if @xmath29 is a rational inner function on @xmath20 , under what conditions is @xmath48 for @xmath49 ?
one can easily identify some rational inner functions that belong to all @xmath38 and whose derivatives are in @xmath22 for every @xmath50 .
the prime examples are * monomials @xmath51 , @xmath52 , * finite blaschke products in one variable and products of such functions , * rational inner functions that are continuous on the closed bidisk , such as @xmath53 these examples illustrate the fact that rifs can be thought of as generalizations of finite blaschke products . since finite blaschke products
are smooth on the closed unit disk they are trivially integrable along with their derivatives .
questions [ q1 ] and [ q2 ] become interesting when we consider rational inner functions with singularities such as @xmath54 this function can not be continuously extended to a neighborhood of the point @xmath55 , yet remains bounded due to the simultaneous vanishing of its numerator and denominator .
the results we obtain in this paper , when applied to this @xmath29 , show that the partial derivatives @xmath56 precisely when @xmath57 and @xmath48 precisely when @xmath58 different rifs will have different @xmath59 and @xmath60 intervals ; the exact cut - off points will depend in a subtle way on the geometric nature of the function s boundary singularities .
we now present the structure of our paper and summarize our main results .
first in section [ sec : prelim ] , we collect information about rifs on the bidisk and their counterparts , called _ rational inner pick functions _ , on the bi - upper half plane .
in particular , we review key functional decompositions on the bidisk called _ agler decompositions _ and useful formulas on the bi - upper half plane called _ nevanlinna representations_. both types of formulas stem from agler model theory , which originated in @xcite and has been further developed in many recent papers including @xcite .
we also review some standard ways to measure non - tangential regularity at points on the distinguished boundary @xmath26 .
one theme of our paper is the usefulness of agler model theory and its related formulas in understanding the singularities and associated boundary behavior of rifs .
agler model theory actually fails for general rational inner functions on the polydisk @xmath61 for @xmath62 , see @xcite , which is why we restrict our analysis to @xmath20 . in section [ sec :
contact ] , we review some important results from the theory of algebraic curves , including a geometric version of puiseux s theorem .
we use these to define the contact order of a rif ; this is the essential new object in our paper and provides a measurement of how singular a rif is at points on @xmath63 . intuitively speaking
, the contact order of a rif measures the rate at which its zero set tends to @xmath26 along one of the coordinate directions .
specifically , denote the zero set of our rational inner function @xmath29 by @xmath64 and define the facial varieties @xmath65 then the @xmath66-_contact order of @xmath29 _ is the maximum number @xmath67 such that there exists a sequence @xmath68 converging to a singular point @xmath69 of @xmath29 and a positive constant @xmath70 such that @xmath71 if @xmath29 is continuous on @xmath72 , its @xmath73- and @xmath74-contact orders are defined to be zero .
some work , detailed in theorem [ lem : contact ] , is required to show that the contact orders of a rif are well defined .
in sections [ sec : dcontact]-[sec : reg ] , we study the @xmath0-membership of rif derivatives , in essence giving a complete answer to question [ q1 ] .
our first characterization rests on the geometry of the zero set of @xmath29 via contact order : let @xmath75 be a rational inner function on @xmath20 .
then for @xmath76 , @xmath77 if and only if the @xmath66-contact order of @xmath29 , denoted @xmath67 , satisfies @xmath78 we then prove in theorem [ thm : hp ] that no rational inner function with singularities on @xmath26 belongs to @xmath22 for @xmath79 , a result which is sharp by the example in .
then in a short section , we show that agler decompositions can also be used to establish the non - integrability of rif derivatives in the hilbert space setting . finally , section [ sec : reg ] details the surprising and counterintuitive observation that imposing higher non - tangential regularity on a rif at a singularity ( called being a @xmath80 point ) decreases the range of exponents for which @xmath81 : let @xmath75 be a rational inner function on @xmath20 with a singularity at @xmath82 .
if @xmath83 is a @xmath80 point of @xmath29 , then at least one of @xmath84 and @xmath85 fails to belong to @xmath22 for @xmath86 this is a corollary of theorem [ thm : bj ] , which characterizes non - tangential regularity via contact order .
importantly , we also derive several results from this @xmath0-theory that give statements purely about the geometry of a rif s singularities ; for example , corollary [ cor : geojulia ] , which is a geometric julia inequality of the same flavor as results in @xcite , says that the @xmath73- and @xmath74-contact orders of a singular rif are both at least @xmath87 .
sections [ sec : dirichlet]-[sec : dincl ] are concerned with rif membership in dirichlet - type spaces . in section [ sec : dirichlet ] , we review preliminary information about the structure of dirichlet - type spaces .
then in theorem [ thm : dp ] , we provide a partial answer to question [ q2 ] by showing how membership in @xmath38 for a certain range of exponents @xmath60 can be deduced from @xmath0-integrability , again via contact order .
specifically , [ [ cor : da ] ] let @xmath75 be a rational inner function on @xmath20 . then for @xmath88 ,
if @xmath89 and @xmath90 , then @xmath91 .
this result follows from the more general theorem [ thm : dp ] , which is stated for anisotropic dirichlet spaces . then by combining agler decompositions with a two - variable version of the local dirichlet integrals of richter and sundberg , we are able to prove in theorem [ thm:1d ] that members of a restricted class of rifs fail to be in @xmath92 , the dirichlet space of the bidisk .
finally , section [ sec : dincl ] gives some simple sufficient conditions for placing a rif in @xmath38 .
section [ sect : examples ] provides a detailed treatment of several explicit rifs to illustrate our results , and the paper closes with a discussion of possible avenues for further research .
let @xmath29 be a rational inner function on @xmath94 . then @xmath95 where @xmath13 is a polynomial of degree @xmath16 with no zeros in the bidisk , @xmath17 is the reflection of @xmath13 , and @xmath18 is a unimodular constant .
we can further assume that @xmath13 is atoral and hence , has finitely many zeros on @xmath63 .
see @xcite for more details .
one can further deduce that @xmath13 has no zeros on @xmath96 since @xmath29 can only have singularities at the zeros of @xmath13 , it will only have finitely many singularities on @xmath97 in what follows , we study the @xmath98 and the @xmath99 membership of rational inner functions and the geometry of their singularities .
these properties are not affected by the existence ( or nonexistence ) of a monomial factor @xmath100 .
thus , in the remainder of the paper , we prove statements about rational inner functions of the form @xmath75 , where @xmath101 .
since @xmath13 and @xmath102 share no common factors ( as @xmath13 is atoral ) , we will also say @xmath103 . the corresponding results for general rational inner functions will follow immediately from our results about @xmath19 rational inner functions .
let @xmath104 denote the upper half plane .
then a two - variable _ pick function _ is a holomorphic function that maps @xmath93 into @xmath104 . a pick function
is called _ inner _ if @xmath105 for almost every @xmath106 .
we will move between @xmath104 and @xmath107 via the following useful conformal maps : @xmath108 ; \quad & & \widetilde{\alpha}\colon \mathbb{d } \rightarrow \pi , \
\widetilde{\alpha}(z ) : = i \left [ \frac{1-z}{1+z } \right ] ; \\ & \beta\colon \pi \rightarrow \mathbb{d } , \ \beta(w):=\frac{w - i}{w+i};\quad & & \tilde{\beta}\colon \pi \rightarrow \mathbb{d } , \
\tilde{\beta}(w ) : = \frac{1+iw}{1-iw}. \end{aligned}\ ] ] then @xmath109 and @xmath110 .
if @xmath29 is a rational inner function on @xmath20 , then we can use these maps to transform @xmath29 into a rational inner pick function @xmath21 on @xmath93 as follows : @xmath111,\ ] ] where @xmath112 if @xmath113 are the singular points of @xmath29 on @xmath26 , then on @xmath114 , @xmath21 only has singularities at the points @xmath115 and on the set @xmath116 moreover , it is easy to see that on @xmath117 the function @xmath21 is continuous as a map into @xmath118 sometimes it is convenient to work directly with @xmath29 , but other times , it will be convenient to work with @xmath21 via .
heuristically , @xmath21 has the advantage that the boundary is flat and the range space is a cone , and @xmath29 has the advantage that the boundary is compact and the range space is closed under multiplication .
every rational inner function @xmath119 with @xmath101 possesses a pair of _ agler kernels _ , namely a pair of positive semidefinite kernels @xmath120 satisfying @xmath121 then is called an _ agler decomposition of @xmath29_. these can be obtained from sums of squares of polynomials . by (
* ? ? ? * sections 5 - 6 ) , there are vectors of polynomials @xmath122 , @xmath123 , @xmath124 ( unique up to left multiplication by a unitary matrix ) that satisfy both @xmath125 and the following condition : if we write @xmath126 and @xmath127 , where each @xmath128 and each @xmath129 is a square matrix - valued polynomial , then @xmath130 and @xmath131 are non - vanishing on @xmath107 . then it is easy to see that the functions @xmath132 are agler kernels of @xmath29 .
moreover , as @xmath133 and @xmath134 satisfy additional term and degree bounds , see the beginning of section @xmath135 in @xcite , we can expand these kernels and write @xmath136 for polynomials @xmath137 with @xmath138 and polynomials @xmath139 with @xmath140 .
here , one can take @xmath141 and @xmath142 to be orthonormal bases of @xmath143 and @xmath144 respectively , where @xmath145 is the hilbert function space with reproducing kernel @xmath146 see @xcite and the references therein for additional material concerning agler decompositions .
all rational inner functions @xmath29 possess some non - tangential regularity at points on @xmath63 . specifically , we define the following behavior : let @xmath147 .
then a point @xmath148 is a called _ b - point for @xmath29 _ if there is a sequence @xmath149 converging to @xmath83 such that the corresponding sequence @xmath150 where @xmath151 for any @xmath152 agler , m.5excarthy , and young introduced b - points in @xcite and used them , along with the related notion of a c - point , to generalize the julia - carathodory theorem to @xmath20 .
we also require the following ( likely well - known ) lemma : [ lem : bpoint ] if @xmath29 is a rational inner function on @xmath20 , then every @xmath82 is a @xmath153-point for @xmath154 let @xmath155
. then @xmath156 is a finite blaschke product on @xmath107 . by carathodory s theorem in (
* vi-4 ) , @xmath157 for every sequence @xmath158 approaching one non - tangentially .
let @xmath159 with @xmath160 be a sequence approaching one , and set @xmath161
. then @xmath162 converges to @xmath83 and @xmath163 so @xmath83 is a b - point of @xmath154 the pick functions @xmath21 given by have very useful representations .
by lemma [ lem : bpoint ] , @xmath164 is a b - point of @xmath29 and so , by proposition @xmath165 in @xcite , there is a @xmath166 such that if @xmath167 converges non - tangentially to @xmath164 , then @xmath168 converges to @xmath169 without loss of generality , assume @xmath170 . using the language of @xcite ,
since @xmath164 is a b - point of @xmath29 , then @xmath171 is a carapoint of @xmath21 and since @xmath172 , @xmath173 where ( as in @xcite ) we interpret equality in terms of non - tangential limits .
then an application of theorem @xmath174 in @xcite gives let @xmath21 be defined as above .
then @xmath21 has a _
type @xmath175 nevanlinna representation _ , namely there exists a hilbert space @xmath176 , a densely defined self - adjoint operator @xmath177 on @xmath176 and a contraction @xmath178 satisfying @xmath179 on @xmath176 , and a vector @xmath180 such that @xmath181 where @xmath182 moreover , since @xmath21 is rational inner , @xmath176 can be chosen to be finite dimensional , so @xmath177 and @xmath178 are matrices .
technically the statement of theorem @xmath174 in @xcite does not include the fact that @xmath176 is finite dimensional .
however , this can be deduced in a straightforward manner . specifically , as given in , every rational inner function has an agler decomposition with kernels coming from finite sums of squares or equivalently , kernels that are reproducing kernels of finite - dimensional hilbert spaces . using this
, one can show that the related herglotz representation from theorem @xmath183 in @xcite is defined on a finite - dimensional hilbert space
. then one can follow the computations in @xcite to show that for rational inner pick functions as given above , these herglotz representations give type @xmath175 nevanlinna representations corresponding to finite - dimensional hilbert spaces .
then for these rational inner pick functions , extends to all @xmath184 .
we first use type @xmath175 nevanlinna representations to characterize the sets where these pick functions can be `` large . '' [ lem : spoke ] let @xmath185 be a rational inner pick function with a type @xmath175 nevanlinna representation as in . then for @xmath186 @xmath187,\ ] ] where @xmath188 are from and @xmath189 are the eigenvalues of @xmath178 .
fix @xmath190 and @xmath184 , and assume @xmath21 satisfies the bounds in . then since @xmath191 we can conclude @xmath192 for @xmath193 then @xmath194 is invertible and its neumann series converges , so @xmath195 from this , we have @xmath196 and the other bound in immediately gives @xmath197 the spoke lemma motivates the following additional definitions .
let @xmath21 be a rational inner pick function as in the spoke lemma and let @xmath198 .
then , @xmath199 as long as @xmath200.\ ] ] the set of points @xmath201 satisfying is called a _ spoke region of @xmath21 _ and is denoted by @xmath202 .
it is the shaded area in figure [ fig : spoke ] .
similarly , an _
@xmath203-spoke _ is the set of points @xmath201 satisfying @xmath204 \text{for any eigenvalue $ t_i$ of $ y$}.\ ] ] if @xmath205 or @xmath206 , this equation simplifies to @xmath207 or @xmath208 respectively , for some positive constant @xmath153 .
if @xmath209 , it simplifies to @xmath210 , for some negative @xmath211 and positive @xmath153 .
nontrivial @xmath212-spoke _ is a spoke that does not correspond to eigenvalues @xmath206 or @xmath213 .
this is equivalent to the spoke not being parallel to one of the coordinate axes .
lastly , we say a point @xmath214 is _ on a nontrivial @xmath203-spoke _ if @xmath215 lies on a nontrivial @xmath212-spoke and if we slice the spoke horizontally and vertically at @xmath215 , one side of the resulting spoke does not intersect any other spoke ( this includes spoke boundaries touching ) .
the non - intersecting side of the spoke is called _ the cut - off spoke associated to _
@xmath215 and is denoted @xmath216 . in figure
[ fig : spoke ] , @xmath215 is on a nontrivial @xmath203-spoke , but @xmath217 is not . ] some rational inner functions @xmath29 possess non - tangential regularity at points on @xmath63 that is stronger than the notion of a b - point . to account for this ,
we recall the definition of the intermediate lwner class , denoted @xmath218 from @xcite .
technically , @xcite defines @xmath218 using behavior at @xmath171 . here
, we send @xmath219 in the conditions and so , define @xmath218 using behavior at @xmath220 as follows : let @xmath221 be nonnegative integer , let @xmath222 , and set @xmath223 .
a two - variable pick function @xmath224 is in the _ intermediate lwner class at @xmath220 _ , denoted @xmath218 , if @xmath225 and if there exists a multi - index system of real numbers @xmath226 such that @xmath227 we follow @xcite and say _ non - tangentially _ means that for each @xmath228 the formula holds for all @xmath229 sufficiently small that also satisfy @xmath230 now we can define the notion of a @xmath80 point : let @xmath29 be a rational inner function on @xmath20 and fix @xmath231 . define @xmath232,\ ] ] so @xmath233 then we say that @xmath83 is a _
@xmath80 point of @xmath29 _ if @xmath234 is in the intermediate lwner class @xmath218 at @xmath220 .
informally speaking , @xmath69 being a @xmath80 point of @xmath29 can be thought of as saying that @xmath29 has a non - tangential expansion to order @xmath235 at @xmath83 , with remainder of order @xmath236 . in particular , the notion of @xmath237 point coincides with that of a b - point as defined in @xcite . to simplify notation
, we often assume that the point @xmath83 under consideration is @xmath164 so @xmath238 , where @xmath21 is defined in and has a type @xmath175 nevanlinna representation as in with associated @xmath176 , @xmath177 , @xmath178 , and @xmath60 .
working through the proof of proposition 3.2 in @xcite , one can prove the following : [ lem : bj ] let @xmath29 be a rational inner function on @xmath20 and assume @xmath239 is a @xmath80 point for @xmath29 .
then @xmath240 where @xmath14 is a polynomial with real coefficients satisfying @xmath241 and @xmath242 is a homogeneous vector - valued polynomial of total degree @xmath243 .
let @xmath75 be a rational inner function on @xmath20 with @xmath103 .
the study of @xmath56 and @xmath244 for some @xmath59 or @xmath60 leads naturally to ideals of polynomials of this form : @xmath245\colon \tfrac{q}{p } \in h^2(\mathbb{d}^2 ) \right\}\ ] ] and the related ideal @xmath246 \colon \frac{q}{p } \in h^{\infty}(\mathbb{d}^2 ) \right\}$ ] . when the polynomials of interest possess a degree bound ,
the following set is also useful : @xmath247\colon \tfrac{q}{p } \in h^2(\mathbb{d}^2 ) \text { and } \deg q \le ( j , k ) \}.\ ] ] some results about these are known .
for example in @xcite , knese showed that the entries of @xmath248 , @xmath249 , and @xmath134 generate @xmath250 and computed the dimensions of many of the @xmath251 sets .
the questions at hand rest on the behavior of @xmath29 at singularities on @xmath252 which are related to two additional topics .
first , recall that a singularity of @xmath29 must occur at a zero of @xmath13 .
any zero of @xmath13 on @xmath63 is also a zero of @xmath102 , which leads to the study of the intersection of the zero sets of @xmath13 and @xmath102 , denoted @xmath253 , @xmath254 , as algebraic curves .
the `` amount '' of intersection at a common zero @xmath255 of two polynomials @xmath13 and @xmath14 is called the _ intersection multiplicity _ and is denoted @xmath256 see @xcite for details about intersection multiplicity and its computation . if @xmath257 , one can show @xmath258 is even ( * ? ? ?
* appendix c ) .
moreover bzout s theorem implies @xmath259 and so , the sum of the intersection multiplicities of common zeros on @xmath252 denoted @xmath260 , is at most @xmath261
. thus , @xmath13 can have at most @xmath262 distinct zeros on @xmath63 .
recall that a polynomial @xmath263 $ ] with @xmath264 is said to _ vanish with order @xmath265 at @xmath266 _ if @xmath267 where the @xmath268 are homogeneous polynomials of degree @xmath269 and @xmath270 . in ( * ? ? ?
* theorem 14.10 ) , knese showed that if @xmath13 vanishes to order @xmath265 at a point @xmath82 , then any @xmath271 vanishes to at least order @xmath265 at @xmath83 .
in what follows , we recall some important results from the theory of algebraic curves .
we use these to define the contact order of a rational inner function , which provides an important measure of how singular the rational inner function is at points on @xmath63 and will play a key role in many of our later arguments and results . in
what follows , we require information about the structure of the zero sets of two - variable polynomials . we collect the needed information here . for additional information or proofs of the stated results ,
see ( * ? ? ?
* chapters @xmath272 and @xmath273 ) .
recall the following standard definitions : @xmath274 $ ] denotes the polynomials in @xmath275 with complex coefficients , @xmath276 denotes the power series in @xmath275 with complex coefficients that converge in a neighborhood of the origin , and @xmath277 $ ] denotes the elements of @xmath276 with a largest power in @xmath278 a function @xmath279 is called a _ unit _ if @xmath280 we say that @xmath281 is _ general in @xmath74 of order @xmath282 _ if writing @xmath283 , @xmath284 observe that @xmath285 as long as @xmath286 and @xmath73 does not divide @xmath224 .
furthermore , we say @xmath287 $ ] is a _ weierstrass polynomial of degree @xmath282 _ if @xmath288 satisfies @xmath289 observe that a weierstrass polynomial of degree @xmath290 is a constant function .
in what follows , we reduce from general functions to weierstrass polynomials using the weierstrass preparation theorem @xcite : [ thm : wp ] let @xmath291 be general in @xmath74 of order @xmath282 .
then , there is a neighborhood @xmath292 of @xmath220 , a weierstrass polynomial @xmath14 of degree @xmath282 , and a unit @xmath279 such that @xmath293 in @xmath292 .
we will combine this with puiseux s theorem .
the version presented here appears in @xcite : [ thm : p ] let @xmath294 $ ] be an irreducible weierstrass polynomial of degree @xmath295 .
then there exists a @xmath296 such that if @xmath297 and if we define the restricted zero set
@xmath298 then there is a @xmath299 that converges in @xmath64 and has the following properties : * @xmath300 for all @xmath301 .
* @xmath302 , sending @xmath303 is a bijection . by combining theorems [ thm : wp ] and [ thm : p ]
, one can give a description of the zero sets of two - variable holomorphic functions near @xmath220 . here
are the details .
[ rem : pt ] let @xmath291 be general in @xmath74 of order @xmath295 . by theorem [ thm : wp ] , there is a weierstrass polynomial @xmath14 of degree @xmath295 , a neighborhood of @xmath220 , and a unit @xmath60 with @xmath293 in this neighborhood . by shrinking the neighborhood if necessary
, we can assume it is of the form @xmath304 , for @xmath305 neighborhoods of zero . by proposition 2 and the subsequent lemma in @xcite ,
there is a unit @xmath306 and a set of irreducible weierstrass polynomials @xmath307 such that @xmath308 on some neighborhood where @xmath306 is nonvanishing . by shrinking @xmath305 if necessary
, we can assume that this neighborhood is still @xmath309 we can also assume each @xmath310 has positive degree @xmath311 because otherwise , it could be absorbed into @xmath312 now for @xmath313 , apply theorem [ thm : p ] to @xmath310 to obtain a radius @xmath314 neighborhoods @xmath315 and @xmath316 containing zero defined using @xmath317 , and a power series @xmath318 converging in @xmath315 and satisfying ( i ) and ( ii ) .
set @xmath319 and define the sets @xmath320 further set @xmath321 by theorem [ thm : p ] , each @xmath322 is exactly the collection of points @xmath323 , for @xmath324 .
set @xmath325 and @xmath326 then @xmath327 then for each @xmath328 , we can set @xmath329 and conclude that all points in @xmath70 are of the form @xmath330 as long as we allow all values for each @xmath331 furthermore , if we shrink @xmath332 and @xmath333 , then the only zeros of @xmath224 in @xmath334 will occur on curves satisfying @xmath335 . given that restriction , each set of curves @xmath336 will be in @xmath70 for @xmath73 sufficiently close to zero .
let @xmath119 be a rational inner function on @xmath20 with @xmath103 and fix @xmath337 .
define the one - variable function @xmath338 then @xmath339 is a blaschke product .
if @xmath29 does not have a singularity on @xmath26 with @xmath74-coordinate @xmath340 , then @xmath341 and @xmath339 has @xmath211 zeroes @xmath342 in @xmath343 in this case , define the quantity @xmath344 observe that @xmath345 measures the distance from the zero set of @xmath339 to @xmath346 if @xmath29 has a singularity on @xmath26 with @xmath74-coordinate @xmath340 , define
@xmath347 we require the following result , which will allow us to rigorously define the contact order of a rational inner function . [ lem : contact ] let @xmath348 be a rational inner function on @xmath20 with @xmath103 and a singularity on @xmath26 with @xmath74-coordinate @xmath349 then there exists a rational number @xmath350 such that @xmath351 for all @xmath337 in a neighborhood @xmath292 of @xmath352 .
the number @xmath353 is called the _ @xmath354-contact order of @xmath29 . _ now we prove theorem [ lem : contact ] : * step 1 . *
we first need to reduce the problem to a local setting .
specifically , let @xmath355 be the singularities of @xmath29 on @xmath26 with @xmath74-coordinate @xmath349 fix a positive @xmath356 . for each @xmath269 ,
let @xmath357 denote the part of the disk with radius @xmath358 and center @xmath359 contained in @xmath360 wherever they are well defined , consider the following variants of @xmath345 : @xmath361 and @xmath362 since the zeros of @xmath363 are continuous functions of @xmath340 ( see the proof of ( * ? ? ?
* theorem @xmath364 ) ) and since @xmath363 only vanishes in @xmath365 , there must be a neighborhood @xmath366 of @xmath352 such that for every @xmath367 , each @xmath368 is well defined @xmath369 thus , we can work directly with each @xmath370 separately . without loss of generality , fix @xmath269 and assume @xmath371 then we need only find a neighborhood @xmath372 of @xmath175 , a small @xmath186 , and a rational number @xmath353 satisfying @xmath373 + * step 2 . * to simplify later computations , we reduce to a problem on @xmath104 near @xmath220 .
recall that @xmath374 is a conformal map with @xmath375 and define the polynomial @xmath376 assume there is a small @xmath186 , a rational number @xmath353 , and a neighborhood @xmath377 of @xmath290 satisfying @xmath378 we will show that implies .
first observe that if @xmath379 , then @xmath380 is bounded and @xmath381 this allows one to compute @xmath382 define @xmath383 and fix @xmath384 .
then since each @xmath385 is nonvanishing on @xmath386 , we can use and the above estimates to conclude @xmath387 which establishes .
+ * step 3 .
* now we prove using puiseux s theorem . since @xmath241 , remark [ rem : pt ] gives neighborhoods @xmath332 and
@xmath333 of @xmath290 , natural numbers @xmath388 , and power series @xmath389 converging on a neighborhood of zero and satisfying @xmath335 such that the set of points @xmath390 is contained in the set of points @xmath391 where each @xmath392 is @xmath311-valued and the curves @xmath393 are contained in @xmath70 for sufficiently small @xmath394 .
write each @xmath395 and let @xmath396 denote the @xmath397 roots of unity . define the index set @xmath398 then we can write the @xmath399 coordinates of points in @xmath70 as @xmath400 now restrict to @xmath401 and observe that for any @xmath402 , the function @xmath403 because otherwise @xmath102 would vanish on an entire curve contained in @xmath63 .
let @xmath404 be the smallest integer @xmath405 such that @xmath406 since each @xmath335 , we know every @xmath407 and so for @xmath408 sufficiently close to zero , @xmath409 define @xmath410 .
choose @xmath186 so that @xmath411 then there is a neighborhood @xmath377 of @xmath290 such that for all @xmath412 , @xmath413 which proves as required .
we can now officially define the contact order of a rational inner function : [ def : contact ] let @xmath348 be a rational inner function on @xmath20 with @xmath103 and let @xmath414 denote the distinct @xmath74-coordinates of the singularities of @xmath29 .
then the _ @xmath73-contact order of @xmath29 _ is the number @xmath415 defined by @xmath416 where the @xmath417-contact order of @xmath29 is defined in theorem [ lem : contact ] .
if @xmath29 has no singularities on @xmath72 , then we set @xmath418 then for each @xmath269 , there is a small neighborhood @xmath419 of @xmath352 such that @xmath420 and , there is at least one neighborhood , say @xmath421 , where the equivalence holds : @xmath422 the _
@xmath74-contact order of @xmath29 _ , denoted @xmath423 , is defined analogously .
an elementary argument shows that the above definition agrees with the definition from the introduction .
the goal of this section is to prove the following theorem : [ thm : contact ] let @xmath75 be a rational inner function on @xmath20 .
then for @xmath76 , @xmath77 if and only if the @xmath66-contact order of @xmath29 , denoted @xmath67 , satisfies @xmath78 we first require the following two lemmas : [ lem : blaschke ] consider a finite blaschke product @xmath424 for @xmath425 then the modulus of the derivative of @xmath426 satisfies @xmath427 we first consider a single blaschke factor @xmath428 then we can compute @xmath429 since @xmath430 for @xmath431 , this immediately implies that holds for blaschke factors .
observe that @xmath432 and so for @xmath431 , @xmath433 which implies .
[ lem : hpderiv ] let @xmath75 be a rational inner function on @xmath20 with @xmath103 and fix @xmath337 such that @xmath363 does not vanish on @xmath434
. then @xmath435 satisfies @xmath436 where @xmath437 is the distance from the zero set of @xmath339 to @xmath438 , rigorously defined in .
first as @xmath439 , we can write the finite blaschke product @xmath440 , for @xmath441 then @xmath442 by lemma [ lem : blaschke ] , we have @xmath443 where the implied constant depends on @xmath59 and @xmath211 .
so , it makes sense to first study @xmath444 for @xmath445 .
observe that if @xmath446 then it should be clear that @xmath447 only depends on @xmath448 and indeed , by a change of variables , we can assume @xmath449 . to emphasize that @xmath60 is now real , we will write @xmath450
. we will show @xmath451 now if @xmath452 , then this is basically immediate .
so , assume @xmath453 writing @xmath454 , we have @xmath455 for @xmath456 , with @xmath426 small but independent of the choice of @xmath457 $ ] .
then an application of trigonometric substitution with @xmath458 gives @xmath459 \setminus ( -b , b ) } \frac{1}{|1- t e^{i\theta } |^{2{\mathfrak{p } } } } d \theta + ( 1-t)^{\mathfrak{p}}\int_{-b}^b \frac{1}{\left ( ( 1-t)^2 + \theta^2 \right)^{\mathfrak{p } } } \ d \theta \\ & \approx ( 1-t)^{\mathfrak{p}}+(1-t)^{\mathfrak{p}}\int_{\text{arctan}\left ( \frac{-b}{1-t}\right)}^{\text{arctan}\left ( \frac{b}{1-t}\right ) } \frac{(1-t ) \sec^2 ( x)}{(1-t)^{2{\mathfrak{p } } } \sec^{2{\mathfrak{p}}}(x ) } \ dx \\ & = ( 1-t)^{\mathfrak{p}}+ ( 1-t)^{1-{\mathfrak{p } } } \int_{-\text{arctan}\left ( \frac{b}{1-t}\right)}^{\text{arctan}\left ( \frac{b}{1-t}\right ) } \cos^{2{\mathfrak{p}}-2}(x ) \ dx\\ & \approx ( 1-t)^{\mathfrak{p}}+ ( 1-t)^{1-{\mathfrak{p } } } \\ & \approx ( 1-|\alpha|)^{1-{\mathfrak{p } } } , \end{aligned}\ ] ] since @xmath460 , where none of the implied constants depend on @xmath461 . by and again using @xmath462 we can compute @xmath463 as desired .
the proof of theorem [ thm : contact ] is almost immediate : we prove the result for @xmath464 assume @xmath465 .
then by lemma [ lem : hpderiv ] , @xmath466 let @xmath415 denote the @xmath73-contact order of @xmath29 and let @xmath414 denote the distinct @xmath74-coordinates of the singularities of @xmath29 .
then by the discussion in definition [ def : contact ] , there is a neighborhood @xmath467 around each @xmath468 such that @xmath469 there is also at least one neighborhood , say @xmath421 , where we have the equivalence @xmath470 writing each @xmath471 and shrinking each @xmath472 if necessary , we can assume that the @xmath472 are disjoint and each @xmath473 for some @xmath474 .
then we have @xmath475 as long as @xmath476 or equivalently , if @xmath477 . now , if @xmath478 , we have @xmath479 which completes the proof .
[ ex : favcontact ] consider the simple rational inner function @xmath480 whose only singularity on @xmath63 is at @xmath164 .
its associated pick function defined as in is @xmath481 to compute the @xmath73-contact order of @xmath29 , fix @xmath337 and observe that @xmath482 writing @xmath483 and assuming @xmath340 is close to @xmath175 , so @xmath484 is close to zero , we can compute @xmath485 then we have @xmath486 similarly , we can compute @xmath487 combining the two equations gives @xmath488 for @xmath340 near @xmath175 .
thus , the @xmath73-contact order of @xmath29 satisfies @xmath489 and by symmetry , @xmath490 theorem [ thm : contact ] now immediately implies that @xmath491 if and only if @xmath492 [ ex : agmcy ] the rational inner function @xmath493 from @xcite is an example with higher contact order at its singularity at @xmath164 .
namely , fixing @xmath494 and solving for @xmath73 , we find that @xmath495 setting @xmath496 and computing @xmath497 and @xmath498 , we obtain @xmath499 meaning that @xmath500 for this example .
using computer algebra , one can also show that @xmath501 .
thus @xmath502 if and only if @xmath503 .
[ rem : hp ] theorem [ thm : contact ] immediately implies that the derivatives @xmath504 for every rational inner function @xmath29 .
because the @xmath98 spaces are nested , this immediately implies that @xmath491 for every @xmath505 moreover , it is worth pointing out that a particularly nice formula holds for each @xmath506 .
[ prop : h1 ] let @xmath75 be a rational inner function on @xmath20 with @xmath507 . then @xmath508 we prove the result for @xmath509 .
fix @xmath337 , so that @xmath510 is a finite blaschke product .
moreover , @xmath511 except at the finite number of points @xmath340 corresponding to the @xmath74-coordinates of singularities of @xmath154 by lemma [ lem : blaschke ] , we know @xmath512 using that and the argument principle , we can compute @xmath513 as desired .
the goal of this section is to prove the following theorem : [ thm : hp ] let @xmath75 be a rational inner function on @xmath20 that does not extend continuously to @xmath514 then @xmath515 for @xmath31 and @xmath516 in example [ ex : favcontact ] , we considered @xmath517 by computing @xmath29 s contact orders and applying theorem [ thm : contact ] , we deduced @xmath491 if and only if @xmath492 thus theorem [ thm : hp ] is sharp in the sense that it can not be improved for the entire class of rational inner functions . combining theorems [ thm : contact ] and [ thm : hp ]
, we obtain a lower bound on the contact orders of a rif in the situation where the function does not extend continuously to @xmath72 .
[ cor : geojulia ] let @xmath518 be a rational inner function on @xmath20 that does not extend continuously to @xmath514 then the contact orders of @xmath29 satisfy @xmath519 for @xmath520 .
this complements results of agler , m.5excarthy , and young , see ( * ? ? ?
* theorem 4.9 , corollary 4.12 ) , which can be interpreted as saying that the maximum of the contact orders of a rif with a singularity on @xmath26 is at least @xmath87 . to prove theorem [ thm : hp ] , we require the following lemma .
it clarifies how a rational inner function @xmath29 can fail to extend continuously to @xmath514 [ lem : limitpoint ] let @xmath348 be a rational inner function on @xmath20 that does not extend continuously to @xmath514 then , there is some @xmath82 such that @xmath521 and a sequence @xmath522 converging to @xmath83 such that @xmath523 does not converge to @xmath524 , where @xmath524 is the non - tangential limit that exists because every @xmath69 is a @xmath153-point of @xmath29 ( see lemma [ lem : bpoint ] ) .
proceeding by contradiction , assume that no such @xmath83 exists . then at each @xmath82 , either @xmath525 or for every @xmath522 converging to @xmath83
, the sequence @xmath523 converges to @xmath524 .
these conditions imply that @xmath29 is a continuous function on @xmath63 .
because @xmath29 is rational inner , we know that the fourier coefficients of @xmath29 satisfy @xmath526 for @xmath527 but then by theorem @xmath528 in @xcite , we can conclude that @xmath29 is continuous on @xmath529 a contradiction . for the remainder of this section , we make the following assumptions : * @xmath75 is a rational inner function on @xmath20 that does not extend continuously to @xmath72 and @xmath21 is its associated rational inner pick function as defined in .
* @xmath164 is the badly - behaved point guaranteed by lemma [ lem : limitpoint ] and @xmath530 , where equality means @xmath175 is the non - tangential limit guaranteed because @xmath164 is a b - point of @xmath29 .
we first study the geometry of the sets where a rational pick function @xmath21 can be large , earlier called spokes .
in particular , we will show that certain level sets can become trapped within a spoke and forced to approach @xmath531 see subsection [ ss : spokes ] for a review of the notation . [ prop : spoke ] assume @xmath29 satisfies assumptions @xmath532 and @xmath533 .
then there is an @xmath186 and a point @xmath215 on a nontrivial @xmath534-spoke such that @xmath535 and @xmath21 is continuous on the cut - off spoke @xmath536 as a function mapping into @xmath537 .
moreover , the connected component of the set @xmath538 defined by @xmath539 that contains @xmath215 , and is called @xmath540 , goes to @xmath171 within @xmath536 . by lemma [ lem : limitpoint ]
, there exists a sequence @xmath541 converging to @xmath164 such that @xmath523 converges to some @xmath542 with @xmath543 shifting to the upper - half plane , define the sequence @xmath544 .
then @xmath545 converges to @xmath171 and @xmath546 converges to some nonzero @xmath547 . in what follows
, we assume @xmath548 is finite , but similar arguments handle the @xmath549 case .
let @xmath550 and consider the spoke region @xmath551 . for @xmath552
sufficiently large , @xmath553 is larger than @xmath358 and @xmath554 is not in a trivial @xmath555-spoke .
this implies that there must be a nontrivial @xmath555-spoke containing a subsequence of @xmath545 .
choose @xmath282 so that @xmath556 is in this nontrivial spoke and satisfies @xmath535 , and @xmath21 is continuous ( as a function into @xmath537 ) on @xmath557 , the cut - off spoke associated to @xmath215 .
now , we show @xmath540 goes to @xmath171 in @xmath557 . without loss of generality ,
assume @xmath557 extends in the negative-@xmath558 , positive-@xmath559 direction .
fix any point @xmath560 such that @xmath561 .
then @xmath21 is continuously differentiable at @xmath217 . for @xmath562 , define the one - variable function @xmath563 then @xmath564 is a one - variable rational pick function and @xmath565 is real for all @xmath558 in some real interval containing @xmath290 . by @xcite , we can write @xmath566 for @xmath567 a nonnegative integer , @xmath568 and @xmath569 positive , and @xmath570 real . then @xmath571 . in particular , at @xmath572 , one can compute @xmath573 therefore , @xmath574 .
in particular , @xmath575 . without loss of generality ,
assume @xmath576 then by the implicit function theorem , there exist open intervals @xmath577 and @xmath221 containing @xmath578 and @xmath579 respectively and a continuously differentiable @xmath580 such that @xmath581 in particular , we can extend @xmath540 in the negative @xmath558-direction from the point @xmath215 .
further , observe that @xmath540 can not cross the boundary of the spoke because @xmath582 immediately across the boundary of the spoke .
now , we need to show @xmath540 approaches @xmath583 in @xmath536 . because @xmath540 is trapped within a nontrivial spoke , it suffices to show the @xmath558-coordinate goes to @xmath584 in the negative @xmath558-direction . by contradiction , assume the part of @xmath540 in @xmath536 has bounded @xmath558-coordinate . because @xmath21 is continuous
, this implies that there is some point @xmath585 on both @xmath540 and @xmath536 with smallest x - coordinate ( so the @xmath558-coordinate will be negative with @xmath586 large ) . by our earlier arguments
, we can apply the implicit function theorem at @xmath585 . if @xmath587 we can extend @xmath540 in the negative @xmath558-direction and so , obtain the needed contradiction .
if @xmath588 we can extend @xmath540 in the positive @xmath559-direction .
if the @xmath558-coordinate is affected , we obtain the needed contraction . therefore , assume that the points in this extension of @xmath540 have @xmath558-coordinate equal to that of @xmath585 .
because @xmath540 will eventually cross the spoke boundary otherwise , there must be a point with maximum @xmath559-value obtained by this extension of @xmath589 applying the implicit function theorem at this point gives the needed contradiction .
now , we consider transformations of nontrivial spokes .
let @xmath186 and let @xmath590 be a rational inner pick function with a type @xmath175 nevanlinna representation as in lemma [ lem : spoke ] .
assume there is a nontrivial @xmath591-spoke with boundary curves @xmath592 , for @xmath211 negative and @xmath426 positive .
consider the conformal map @xmath593 defined by @xmath594 . under @xmath595 , that is , sending @xmath596 , the nontrivial @xmath591-spoke is transformed into a _
nontrivial @xmath597-horn_. the horn has boundary curves @xmath598 and at a fixed @xmath558 , the horn has height @xmath599 as long as @xmath558 is sufficiently close to zero , say in the interval @xmath600 . observe that the conformal map @xmath601 , defined earlier by @xmath602 also satisfies @xmath603 in what follows , we often write @xmath604 for @xmath605 and use similar notation for @xmath606 .
now , we can transfer proposition [ prop : spoke ] to the associated horn as follows : [ lem : horn ] assume @xmath29 satisfies @xmath532 and @xmath533 .
then there is a non - trivial horn with boundaries @xmath607 and a point @xmath608 in the horn such that the connected component of the level set @xmath609 containing @xmath608 , which is called @xmath610 , goes to @xmath220 within the horn . moreover
if @xmath611 , then @xmath612 for all @xmath613 and @xmath614 for all @xmath615 on the boundary of the horn . applying proposition [ prop : spoke ] to @xmath29 gives an @xmath186 and a point @xmath616 such that @xmath617 .
moreover , there is a nontrivial @xmath618-spoke containing @xmath215 such that the connected component @xmath540 of the level set @xmath619 containing @xmath215 goes to @xmath171 in the cut - off spoke @xmath620 set @xmath621 and consider @xmath622 .
as @xmath623 , all of the @xmath618-spokes are contained in @xmath622-spokes .
thus , there is a nontrivial @xmath622-spoke containing @xmath215 such that @xmath540 goes to @xmath171 within the spoke .
let @xmath592 denote the boundaries of this nontrivial @xmath622-spoke containing @xmath215 .
clearly on @xmath538 , @xmath624 , and on the boundary of the nontrivial @xmath625-spoke containing @xmath215 , we have @xmath626 .
now consider the transformation of these objects under @xmath595 .
we obtain a nontrivial horn with boundary curves @xmath607 , a point @xmath627 in the horn , and the level set @xmath628 since @xmath595 is its own inverse .
let @xmath610 be the connected component of @xmath538 containing @xmath608 . by moving @xmath215
further along the @xmath622 if necessary ( so that the part of @xmath540 going towards @xmath171 from @xmath215 remains connected under @xmath595 ) , we can assume @xmath610 approaches @xmath220 within the horn .
recalling the definitions of @xmath606 and @xmath21 from and setting @xmath629 , we have @xmath630 moreover , one can easily show that @xmath612 for all @xmath613 .
similarly , for @xmath615 on the horn boundary , we have @xmath631 the previous equations can be combined to obtain : @xmath632 and so @xmath633 as desired . in this section , we obtain the integral estimates necessarily to prove theorem [ thm : hp ] .
it is easier to integrate over horns in @xmath114 than over regions in @xmath63 .
thus we define @xmath634 . here
@xmath635 extends continuously to @xmath636 , except at a finite number of singular points . to specify the integration region ,
let @xmath637 denote the boundaries of the nontrivial horn guaranteed by lemma [ lem : horn ] .
let @xmath638 denote the point from lemma [ lem : horn ] and set @xmath639 .
then for @xmath640 $ ] , the height of the horn at @xmath558 is approximately @xmath641 and there is a point on @xmath610 with @xmath558-coordinate @xmath558 .
define @xmath642 to be the region in the horn with @xmath558 values in the interval @xmath643.$ ] then we have : for @xmath29 satisfying @xmath532 and @xmath533 and @xmath642 defined above , if @xmath644 for @xmath645 .
the proof is basically just changing variables .
first writing @xmath646 and @xmath647 , we have @xmath648 for each @xmath649 , there are unique @xmath650 such that @xmath651 solving for @xmath652 gives @xmath653 and so @xmath654 analogous formulas hold for @xmath655 changing variables to @xmath558 and @xmath559 gives @xmath656 by the chain rule , @xmath657 then , using the boundedness of @xmath642 , we can conclude that @xmath658 which , coupled with , gives the desired result .
now theorem [ thm : hp ] simply requires the following result : for @xmath29 satisfying @xmath532 and @xmath533 and @xmath642 defined above , @xmath659 first , observe that if @xmath660 \rightarrow \mathbb{r}$ ] is a twice continuously differentiable function , then @xmath661 this follows from the calculus of variations and in particular , the euler - lagrange equations .
see ( * ? ? ?
* chapter 8) . specifically ,
( up to an @xmath662 modification ensuring the lagrangian function is twice continuously differentiable ) one can show that the integral @xmath663 with the constraints @xmath664,$ ] @xmath665 @xmath666 is minimized by the straight line @xmath667 using this @xmath668 gives the lower bound in .
now , recall that @xmath642 is the region in @xmath114 satisfying the bounds @xmath669 and @xmath670 observe that for all but a finite number of @xmath558-values in @xmath671 , the function @xmath672 has a non - vanishing denominator on @xmath671 and so is clearly twice continuously differentiable .
fix such an @xmath673 and let @xmath674 denote the @xmath559-coordinate of the point with @xmath558-coordinate on @xmath610 .
then by the lemma [ lem : horn ] , @xmath675 applying to the real and imaginary parts of @xmath672 separately , we obtain : @xmath676 where the implied constants depends only on @xmath59 , @xmath211 , and @xmath426 . in the second - to - last line
, we also used lemma [ lem : horn ] and the fact that the points @xmath677 are on the boundary of the horn .
then @xmath678 as long as @xmath679 it is worth mentioning that the above proofs do not explicitly discuss the case @xmath680 .
however , by contrapositive , if @xmath681 for some @xmath682 , we also have @xmath683 immediately .
in this section , we provide an alternate proof that , for certain rational inner functions @xmath29 , a partial derivative @xmath84 or @xmath85 is not in @xmath684 .
this proof rests on agler decompositions and related results from @xcite involving polynomial ideals , intersection multiplicities , and vanishing orders ; these were defined in section [ sec : prelim ] .
this alternative proof has the advantage of being short , and it also illustrates , along with our work on local dirichlet integrals in subsequent sections , the usefulness of agler decompositions in the context of examining the integrability of rational inner functions and their derivatives .
first , for @xmath31 define @xmath139 so that @xmath685 then the statement @xmath686 is equivalent to @xmath687 where @xmath688 denotes a polynomial ideal defined in subsection [ ssec : ideal ] . in the case
where the intersection multiplicity @xmath689 is maximal , we have a complete and attractive description of @xmath688 in terms of @xmath690 .
[ lem : p2ideal ] assume @xmath691 $ ] is an atoral polynomial with no zeros on the bidisk and @xmath692 .
further , assume @xmath693 then @xmath694 let @xmath695 be the agler kernels of @xmath696 given in .
by corollary @xmath697 in @xcite , @xmath29 has a unique pair of agler kernels , so ( up to left multiplication by a unitary matrix ) , @xmath698 and @xmath699 by theorem @xmath700 in @xcite , the elements of @xmath701 generate @xmath250 .
now observe that @xmath702 a moment s thought should reveal that if @xmath703 are the entries of @xmath248 and @xmath704 are the entries of @xmath134 , then one can find agler kernels of @xmath705 corresponding to polynomial vectors with entries @xmath706 and @xmath707 respectively .
however , @xmath708 the maximum intersection multiplicity of @xmath709 and @xmath710 on @xmath711 then another application of corollary 13.6 in @xcite implies that @xmath705 has a unique pair of agler kernels .
thus , by theorem 7.1 in @xcite , @xmath712 is generated by @xmath713 which proves the claim . we can now prove the following derivative result : [ thm : notinh2 ] let @xmath75 be a rational inner function on @xmath94 with @xmath103 .
further , assume @xmath693 then at least one of @xmath84 , @xmath85 is not in @xmath714 without loss of generality , assume @xmath13 has a zero at @xmath715 and vanishes there to order @xmath265 .
this means we can write @xmath716 where each @xmath717 is a homogeneous polynomial in @xmath718 of total degree @xmath719 and @xmath270
. then @xmath720 contains a nonzero term involving at least one of @xmath721 , @xmath722 .
assume , without loss of generality , that there is a nonzero term involving @xmath721 .
then , since @xmath723 , @xmath724 , we have @xmath725 the derivative of the nonzero term from @xmath720 involving @xmath721 is a nonzero homogeneous polynomial of total degree @xmath726 , and it will not cancel with any other terms in the derivative , since it is the unique monomial with a given power in @xmath718 .
this implies that @xmath727 vanishes to at most order @xmath726 at @xmath728 and so by theorem 14.10 in @xcite , @xmath729 .
proceeding towards a contradiction , assume @xmath730 and so @xmath731 by lemma [ lem : p2ideal ] , this means we can write @xmath732 for some @xmath733 rearranging terms gives @xmath734 as @xmath13 and @xmath102 share no common factors ( this follows from @xmath13 atoral ) , we can conclude that @xmath13 divides @xmath735 and so there is a polynomial @xmath736 such that @xmath737 as @xmath738 , this immediately implies that @xmath739 a contradiction .
the goal of this section is to prove the following theorem : [ thm : bj ] let @xmath75 be a rational inner function on @xmath20 with a singularity at @xmath82 .
if @xmath83 is a @xmath80 point of @xmath29 , then @xmath740 where each @xmath67 is the @xmath66-contact order of @xmath154 let @xmath75 .
then theorem [ thm : bj ] can be compared to corollary 14.7 in @xcite , which gives a relationship between the intersection multiplicity of @xmath102 and @xmath13 at a singularity of @xmath29 and the level of non - tangential regularity of @xmath29 at the given singularity .
moreover , we may view theorem [ thm : bj ] as a partial higher - order analogue of the geometric julia inequality in corollary [ cor : geojulia ] .
theorem [ thm : bj ] coupled with theorem [ thm : contact ] yields the following : [ cor : bjc ] let @xmath75 be a rational inner function on @xmath20 with a singularity at @xmath82 . if @xmath83 is a @xmath80 point of @xmath29 , then at least one of @xmath84 and @xmath85 fails to belong to @xmath22 for @xmath86 before proceeding , we require the following lemma , which is likely well - known : [ lem : ineq ] if @xmath177 and @xmath153 are self - adjoint @xmath741 matrices and @xmath153 is positive - definite , then @xmath742 observe that since @xmath177 is self - adjoint and @xmath153 is positive - definite , we have : @xmath743 where we used the fact that @xmath744 is self - adjoint to conclude that the moduli of the eigenvalues of @xmath745 are all less than @xmath175 .
now we can proceed with the proof of theorem [ thm : bj ] : let @xmath29 be a rational inner function on @xmath20 . moreover , assume there is a point @xmath82 such that @xmath83 is both a singular point and @xmath80 point of @xmath29 . without loss of generality ,
assume @xmath239 . +
* we will first show that if @xmath746 satisfies @xmath747 and is sufficiently near @xmath164 , then @xmath748 it is easier to prove on @xmath93 .
so , define @xmath224 by @xmath749.\ ] ] then lemma [ lem : bj ] implies that there is a finite - dimensional hilbert space @xmath176 , a self - adjoint operator @xmath177 and a positive contraction on @xmath178 on @xmath176 satisfying @xmath179 such that @xmath240 where @xmath14 is a polynomial with real coefficients satisfying @xmath241 and @xmath242 is a homogeneous vector - valued polynomial of degree @xmath243 .
if @xmath750 and @xmath751 is near @xmath220 , then using lemma [ lem : ineq ] , we have @xmath752 from here , we can conclude that @xmath753 where the first inequality is trivial and the second follows from the above string of inequalities .
now we return to @xmath94 .
let @xmath292 be a small neighborhood of @xmath164 such that if @xmath754 , then @xmath755 satisfies and @xmath756 as in the proof of theorem [ lem : contact ] . then since @xmath757 on @xmath94 precisely when
@xmath750 on @xmath93 , we obtain for @xmath758 + * step 2 .
* now define @xmath759 then by theorem 2.4 in @xcite , @xmath14 is atoral and @xmath760 is rational inner . since by definition , @xmath761
, we also have @xmath762 define the polynomial @xmath763 .
then @xmath764 and by the discussion of puiseux s theorem in remark [ rem : pt ] , there are neighborhoods @xmath332 and @xmath333 containing zero , natural numbers @xmath388 , and power series @xmath389 converging on a neighborhood of zero and satisfying @xmath335 such that the set of points @xmath765 is contained in the set of points @xmath766 here each @xmath392 is @xmath311-valued and the curves @xmath393 are contained in @xmath70 for sufficiently small @xmath394 . writing each @xmath767 and interpreting @xmath768 as an @xmath311-valued function , we can write the @xmath399-coordinates of points in @xmath70 as @xmath769 for @xmath394 sufficiently small and @xmath770 and @xmath771 this immediately gives a parameterization of the zero set of @xmath772 near @xmath164 .
indeed , if @xmath773 and @xmath774 are the obvious translations of @xmath332 and @xmath333 , the set of points @xmath775 must have @xmath721-coordinates satisfying @xmath776 for nonzero constants @xmath777 .
furthermore , each @xmath335 implies that each @xmath778 .
fix @xmath328 and for ease of notation , write @xmath779 without loss of generality , assume @xmath780
. we will show that @xmath781 set @xmath782 , where @xmath783 will be chosen later . then @xmath784 for some nonzero @xmath785 and @xmath786 the @xmath787 roots of unity . because @xmath780 , the list @xmath788 contains exactly the n@xmath789 roots of unity .
moreover each quadrant in @xmath636 contains at least one n@xmath789 root of unity , where we say adopt the convention that points on an axis are said to lie in both quadrants .
thus , we can choose @xmath790 such that either @xmath791 given this @xmath790 and @xmath783 chosen small enough , we can guarantee @xmath792 but since @xmath793 , then @xmath14 vanishes at @xmath794 , a contradiction . now shrinking @xmath795 if necessary and using and the conclusion that each @xmath796
, there is some nonzero @xmath797 such that @xmath798 for all @xmath799 satisfying @xmath793 . + * step 3 .
* choose @xmath800 close to @xmath175 and further , assume @xmath337 .
we will show @xmath801 implying that the @xmath73-contact order of @xmath802 is at least @xmath803 . for @xmath804 ,
set @xmath805 and @xmath806 . then using @xmath337 , we have @xmath807 thus , for @xmath808 in a sufficiently small neighborhood of @xmath164
, we can assume @xmath809 is in the neighborhood @xmath810 , defined in step @xmath811 moreover , @xmath812 if and only if @xmath813 .
then for all @xmath808 in a small neighborhood of @xmath164 satisfying @xmath814 , we can apply and to obtain @xmath815 this finishes the proof of . by theorem [ thm : contact ]
, we can conclude that @xmath816 for all @xmath817 a computation using a change of variables and the chain rule now shows that if @xmath818 , then at least one of the statements @xmath819 must hold .
then another application of theorem [ thm : contact ] gives @xmath740 where @xmath67 is the @xmath66-contact order of @xmath154
recall that the _ dirichlet - type spaces _ @xmath38 are spaces of functions @xmath821 that are holomorphic in the bidisk and for @xmath40 fixed , satisfy @xmath822 we are particularly interested in the _ isotropic _ spaces @xmath38 but it is frequently useful to consider the more general scale of _ anisotropic dirichlet - type spaces _
@xmath820 , furnished with the norm @xmath823 where now @xmath824 .
we collect some basic material on dirichlet - type spaces here , and refer the reader to @xcite for more information concerning @xmath38 and @xmath820 .
the one - variable versions of the dirichlet spaces , consisting of holomorphic functions @xmath825 , will be denoted by @xmath826 and are discussed in the textbook @xcite .
the norm of @xmath827 is given by @xmath828 from the norm definition in , one can see that the polynomials are dense in every @xmath820 .
in addition , any holomorphic function that extends to a strictly larger bidisk belongs to each @xmath820 .
moreover , if each @xmath829 , then @xmath830 , see @xcite , and each one variable @xmath831 embeds in @xmath820 in a natural way .
when @xmath832 , the spaces @xmath820 are banach algebras of functions continuous on the closed bidisk .
functions in weighted dirichlet spaces are not necessarily in @xmath833 , but @xmath38 is contained in @xmath684 whenever @xmath834 .
thus each @xmath835 possesses a radial limit at almost every point of @xmath26 , see @xcite .
if @xmath836 satisfies @xmath837 then @xmath820 has an equivalent integral norm given by @xmath838 where @xmath839 , and @xmath840 is normalized area measure in the disk , see @xcite .
later , we shall make use of the fact that the expression @xmath841 is invariant under the transformation @xmath842 , where for @xmath843 , @xmath844 we say that a ( necessarily holomorphic ) function @xmath845 belongs to the _ multiplier space _ @xmath846 if @xmath847 for all @xmath848 .
the multiplier spaces do not seem to admit a simple characterization for all @xmath824 , but it is known that @xmath849 when @xmath850 ( see @xcite ) , and furthermore , every function holomorphic on a neighborhood of the closed bidisk is a multiplier for each @xmath820 .
an important tool in the study of the dirichlet space on the unit disk is the _ local dirichlet integral _ introduced by richter and sundberg in @xcite ; see @xcite for applications and further developments . for @xmath851 and @xmath852 ,
set @xmath853 with the understanding that @xmath854 if the radial limit @xmath855 does not exist . integrating the local dirichlet integral with respect to @xmath856 , normalized arclength measure , we recover the classical _ douglas formula _ ,
@xmath857 which expresses the dirichlet integral @xmath858 in terms of boundary values of the function @xmath21 . in (
* lemma 3.2 ) , richter and sundberg make the useful observation that if @xmath859 and @xmath860 , then @xmath861 it follows , in particular , that if @xmath29 is an inner function on the disk , then @xmath862 we will develop similar objects on the bidisk but first , derive an equivalent norm for the two - variable dirichlet space @xmath92 : [ lem : norm ] let @xmath21 be a function in @xmath92 with @xmath863 then the expression @xmath864 this implies that doug(@xmath21 ) defines an equivalent norm on @xmath92 .
clearly @xmath865 and as @xmath866 , then lemma @xmath867 in @xcite implies that @xmath868 the formula for @xmath869 is similar .
now we implement a two - variable version of the proof of lemma @xmath867 from @xcite .
specifically , perform the change of variables @xmath870 and @xmath871 for @xmath872 then the integral over @xmath873 in the formula for doug@xmath874 becomes @xmath875 ^ 2 } \int_{[0,2\pi]^2 } \frac { \left | f(e^{i(s_1+t_1 ) } , e^{i(s_2+t_2 ) } ) -f(e^{i(s_1+t_1 ) } , e^{it_2})-f(e^{it_1},e^{i(s_2+t_2 ) } ) + f(e^{it_1},e^{it_2 } ) \right|^2}{|e^{is_1}-1|^2|e^{is_2}-1|^2 } dt ds.\ ] ] just as in the proof of lemma @xmath867 , we can apply parseval s formula to the function @xmath876 to reduce the above integral to @xmath877 ^ 2 } \frac{| e^{is_1k } -1|^2 | e^{is_2\ell}-1|^2}{|e^{is_1}-1|^2|e^{is_2}-1|^2 } ds = \sum_{k , \ell = 1}^{\infty } k
\ell | a_{k\ell}|^2,\ ] ] via another application of parseval s formula .
we define the _ local dirichlet integral for the bidisk _ at a point @xmath878 by setting @xmath879 the integral above is to be interpreted as being infinite if the radial limit @xmath880 does not exist . as in the one - variable case
, @xmath881 implies @xmath882 .
we immediately obtain an analog of .
[ lem : riflocal ] let @xmath75 be a rational inner function on @xmath20 and let @xmath883 . then @xmath884 fix @xmath883 .
then for almost every @xmath885 , the function @xmath886 belongs to the hardy space of the unit disk .
hence by , @xmath887 repeating this argument using the functions @xmath888 and @xmath889 and integrating with respect to @xmath890 , we arrive at the formula @xmath891 now , simply note that for every @xmath892 , the radial limit @xmath893 exists and satisfies @xmath894 ; this follows from lemma [ lem : bpoint ] and proposition @xmath165 in @xcite .
the goal of this section is to prove the following theorem , which connects the inclusion of rational inner functions in @xmath38 spaces to the inclusion of their derivatives in related @xmath98 spaces .
[ thm : dp ] let @xmath75 be a rational inner function on @xmath20 with @xmath101 .
then for @xmath88 , if @xmath89 , then @xmath895 and if @xmath90 , then @xmath896 .
[ rem : prange ] in theorem [ thm : dp ] , the important @xmath59 range is @xmath897 indeed , if we have theorem [ thm : dp ] on this range , for every @xmath29 and @xmath898 , theorem [ thm : contact ] implies that @xmath899 , and so , every @xmath900 .
due to the nested property of our spaces , for @xmath901 , we immediately obtain @xmath902 and @xmath903 . similarly , if @xmath904 , theorem [ thm : dp ] is trivial .
namely , theorem [ thm : hp ] implies that @xmath902 if and only if @xmath29 is continuous on @xmath72 .
so , the statement is vacuous for @xmath29 with singularities on @xmath905 .
conversely , by the continuous edge - of - the - wedge theorem in @xcite , if @xmath75 is continuous on @xmath72 , then @xmath13 has no zeros on @xmath72 .
thus , @xmath29 is holomorphic on an open set containing @xmath72 and belongs to every @xmath906 by remark [ rem : prange ] , the proof of theorem [ thm : dp ] will only address the range @xmath897 theorem [ thm : dp ] also has the following important corollary .
[ cor : da ] let @xmath75 be a rational inner function on @xmath20 .
if @xmath89 and @xmath907 , then @xmath908 . more specifically , if @xmath909 then @xmath910 assume @xmath89 and @xmath907 .
by theorem [ thm : dp ] , the assumptions imply @xmath911 write @xmath912
. then an application of the cauchy - schwarz inequality gives @xmath913 as needed .
we first require the following lemma .
the lemma is likely true for @xmath914 , but as this is not required to prove theorem [ thm : dp ] , we restrict to the interval @xmath915 for an easier proof . [
lem : dp ] let @xmath426 be a finite blaschke product @xmath424 for @xmath425 then @xmath426 satisfies @xmath916 where @xmath917 is the distance from the zero set of @xmath426 to @xmath438 and the implied constant depends on @xmath59 and @xmath918 . since @xmath919
, we have the following equivalent norm for the one - variable space @xmath920 @xmath921 where @xmath922 is normalized area measure on @xmath107 .
see for example , ( * ? ? ?
* chapter 1.6 ) . applying this to @xmath426 , computing @xmath923 , and using each @xmath924 on @xmath107
, we have @xmath925 where the implied constants depend on @xmath926 now we simply need to show that for @xmath445 , @xmath927 and apply the following computation , using @xmath465 : @xmath928 where the implied constants depend on @xmath552 .
so we are done once we show @xmath929 , for an implied constant depending on @xmath59 .
we first go on a slight detour .
observe that @xmath930 using well - known properties of the gamma function @xmath931 , we have @xmath932 for large @xmath405 , say @xmath933 , since @xmath934 for large @xmath405 . returning to @xmath935 ,
observe that @xmath936 which implies that @xmath937 fix @xmath783 such that if @xmath938 , then @xmath939 where our choice of @xmath662 depends on @xmath59 . from this and
, we can conclude that if @xmath940 , then @xmath941 where the implied constants depend on @xmath59 ( and @xmath662 , which depends on @xmath59 ) .
similarly if @xmath942 , we have @xmath943 where the implied constant again depends on @xmath59 ( and @xmath662 ) . from this
, the proof of theorem [ thm : dp ] is mostly an application of theorem [ thm : contact ] : as discussed in remark [ rem : prange ] , we restrict attention to @xmath944 and assume @xmath89 .
then by theorem [ thm : contact ] , @xmath477 .
define a sequence of functions @xmath945 by @xmath946 then for almost every @xmath337 , each @xmath947 has a radial boundary value at @xmath340 , denoted @xmath948 and the function @xmath949 is a blaschke product with @xmath950 furthermore by lemma [ lem : dp ] , @xmath951 where the implied constant depends on @xmath59 but not on @xmath952then we can compute @xmath953 where the last statement uses @xmath477 and follows via the arguments used in the proof of theorem [ thm : contact ] .
in this section , we use local dirichlet integrals and agler decompositions to show that certain rational inner functions fail to be in the two - variable dirichlet space @xmath954 in the following lemma , we obtain an alternate formula for @xmath955 , when @xmath29 is rational inner . to obtain the result ,
let @xmath415 and @xmath423 be the agler kernels defined in with expansions given in .
here , one can take @xmath141 and @xmath142 to be orthonormal bases of @xmath143 and @xmath144 respectively . [
lem : local2 ] for rational inner @xmath75 with @xmath103 , we have @xmath956 for all @xmath883 , where the functions @xmath957 , @xmath958 are from .
moreover , if @xmath29 does not have a singularity at @xmath959 , then also holds for @xmath960 .
first fix @xmath961 since @xmath13 has no zeros on @xmath962 , for every @xmath963 , we can extend the agler kernel formulas to the points @xmath964 and @xmath965 as follows : @xmath966 by manipulating these equations , we can conclude that @xmath967 plugging this equation into the numerator of and using the fact that @xmath968 for @xmath31 , we obtain @xmath969 since @xmath13 is nonzero on @xmath970 , the slice functions @xmath971 , @xmath972 are all in @xmath973 and we can use to obtain .
now , we extend to points on @xmath63 .
fix @xmath959 and assume @xmath29 does not have a singularity at @xmath974 .
we will prove the following string of equalities : @xmath975 equation follows from the derived formula for points in @xmath20 .
so , to prove the string of equalities , we just need to justify the switching of the limits in and .
first consider and define the rational function @xmath976 as follows : @xmath977 for a polynomial @xmath978 of four variables .
moreover , @xmath978 vanishes whenever @xmath979 for @xmath980 so @xmath978 is divisible by both @xmath981 and @xmath982 .
this is a common conclusion we make and follows from an application of hilbert s nullstellensatz .
specifically , hilbert s nullstellensatz implies that @xmath983 is in the ideal generated by each @xmath984 for some positive integer @xmath282 .
this means @xmath984 divides @xmath983 and since each @xmath984 is irreducible , it must actually divide @xmath978 .
this means we can write @xmath985 for a polynomial @xmath986 . notice that for @xmath987 , @xmath988 is exactly the function appearing in the integral .
then , for all @xmath987 , this function is bounded ( independent of @xmath358 ) as long as we restrict the @xmath989 to closed subsets of @xmath63 such that @xmath13 does not vanish at @xmath989 , @xmath990 , @xmath991 for any @xmath989 in the set .
since @xmath13 has a finite number of zeros on @xmath992 for each @xmath783 , there is an open set @xmath993 , which contains any troublesome @xmath989 values and satisfies @xmath994 then , we can use the dominated convergence theorem on @xmath995 to conclude that @xmath996 which in turn gives .
analogous arguments ( writing out the functions of interest , factoring out linear terms to get rational functions that are bounded away from a finite set , and using the dominated convergence theorem ) give . the desired formula for @xmath960 follows immediately .
we can use lemma [ lem : local2 ] to show that certain rational inner functions are not in the two - variable dirichlet space .
[ thm:1d ] let @xmath75 be a rational inner function with @xmath997 , with @xmath998 furthermore assume that @xmath999 , @xmath13 is irreducible , and all of the zeros of @xmath13 on @xmath63 occur at @xmath164 .
then @xmath29 is not in the two - variable dirichlet space @xmath954 by lemma [ lem : norm ] , the function @xmath29 will fail to be in @xmath92 if the integral @xmath1000 is infinite .
specifically , we claim that at least one @xmath1001 for @xmath1002 a basis vector of @xmath144 from . this will prove the desired claim .
first , for @xmath1003 , let @xmath1004 be a basis vector of @xmath144 as in .
then @xmath1005 and @xmath1006 for a three - variable polynomial @xmath1007 with @xmath1008 since @xmath1007 vanishes whenever @xmath805 , it must be divisible by @xmath1009 .
this implies that we can write @xmath1010 for a one - variable polynomial @xmath1011 with @xmath1012 then we can write @xmath1013 for some @xmath1014 and polynomial @xmath1015 with @xmath1016 observe that @xmath1017 because that would imply @xmath1004 is a function of only one variable .
this would give a contradiction because @xmath1018 and @xmath13 is irreducible .
we claim that for some @xmath1002 , we actually have @xmath1019 by way of contradiction , assume not .
fix any @xmath1020 @xmath1021 with @xmath1022 from . since @xmath1023
, there are at least two such basis vectors of @xmath144 .
then there are nonzero constants @xmath1024 and @xmath1025 such that @xmath1026 taking limits as @xmath1027 , this says @xmath1028 and so @xmath1029 for some rational function @xmath1030 with @xmath1031 polynomials . since @xmath1021 and @xmath1004 are linearly independent , @xmath1032 then rearranging terms implies that @xmath1033 since @xmath13 is irreducible with @xmath1034 , this implies @xmath13 divides @xmath1035 .
then we obtain @xmath1036 , a contradiction .
thus , there is some @xmath1037 such that @xmath1038 for @xmath1039 and a polynomial @xmath137 with @xmath1040 and @xmath1041 moreover , since @xmath137 does not vanish near @xmath29 s singularity , its presence does not affect whether the integral given below converges or diverges .
this implies that @xmath1042 where we used hlder s inequality .
if this last integral converged , then @xmath1043 would be in the polynomial ideal @xmath250 .
but , @xmath1044 and so using the notation of ( * ? ? ?
* section 5 ) , this would imply that the set of polynomials @xmath1045 , which by ( * ? ? ?
* cor 13.6 ) and its proof , contradicts that @xmath1046 is maximal .
thus , the integral must diverge , implying that @xmath29 is not in the dirichlet space .
we believe that the conclusion of theorem [ thm:1d ] persists under much weaker assumptions .
for instance , if @xmath29 is a rif that satisfies the hypotheses of theorem [ thm:1d ] and @xmath1047 are fixed , then the rational inner function @xmath1048 has @xmath1049 and has @xmath1050 zeros on @xmath26 located at roots of unity , yet also fails to belong to @xmath92 . to see this ,
observe that the taylor coefficients of @xmath1051 satisfy @xmath1052 and @xmath1053 if @xmath1054 or @xmath1055 .
thus @xmath1056 and the series on the right diverges by theorem [ thm:1d ] .
furthermore , in section [ sect : examples ] , we shall exhibit a rif @xmath119 with a single singularity on @xmath26 that does not belong to the dirichlet space despite having @xmath1057 .
we first state some elementary inclusion results for @xmath820 spaces .
we will use some of these results when we explore specific examples in section [ sect : examples ] .
[ lem : inclusion ] let @xmath1058 . then * @xmath1059 if and only if @xmath1060 and @xmath1061 ; * @xmath1059 if and only if @xmath1062 and @xmath1063 furthermore , if @xmath1064 , where @xmath13 is an atoral polynomial with no zeros in @xmath94 and is not a polynomial in one variable only , then * @xmath1059 if and only if @xmath1065 ; * @xmath1059 if and only if @xmath1066 the first two assertions follow upon considering the power series expansions of first derivatives . to obtain the second assertion , note
that @xmath13 can not vanish in @xmath1067 or @xmath1068 .
this implies @xmath1069 and @xmath1070 are both holomorphic on a neighborhood of @xmath1071 , and hence the functions @xmath1072 are immediately in every @xmath826 .
[ lem : cslemma ] if @xmath1073 , then @xmath1074 . for
a given @xmath60 , write @xmath1075\cdot \left[(k+1)^{\alpha/2}(\ell+1)^{\alpha/2 - 1}|a_{k\ell}|\right].\ ] ] writing @xmath1076 and applying the cauchy - schwarz inequality , we obtain @xmath1077 and the expression on the right - hand side is equal to @xmath1078 . we can now state a sufficient condition for a rational inner function to belong to a weighted dirichlet space .
sometimes it is easier to check this condition for a range of exponents than to determine the exact contact orders of a given rif .
[ prop : da ] let @xmath119 be a rational inner function on @xmath20 and suppose @xmath1079 for some @xmath1080
. then @xmath1081 .
we can write @xmath1082 . since @xmath1083 by the quotient rule
, we have @xmath1084 .
hence @xmath1085 is a multiplier of @xmath1086 and so @xmath84 is in @xmath1086 .
the same reasoning applies to @xmath85 .
next , lemma [ lem : inclusion ] implies that @xmath1087 and @xmath1088 .
then by lemma [ lem : cslemma ] , @xmath1089 , and another application of lemma [ lem : inclusion ] gives @xmath1081 .
we use this result in the next section to exhibit some concrete examples of rational inner functions that belong to a range of weighted dirichlet spaces .
we now illustrate our theorems and questions with several examples .
[ faveex]let @xmath1090 we will now show that @xmath244 if and only if @xmath58 first observe that example [ ex : favcontact ] coupled with corollary [ cor : da ] implies that @xmath244 for @xmath58 to obtain the rest of the claim , observe that @xmath1091 where the coefficients @xmath1092 are given by @xmath1093 for @xmath1094 set @xmath1095 .
then for @xmath552 sufficiently large and @xmath1096 , we can use an asymptotic estimate for the binomial coefficients , see @xcite , to obtain : @xmath1097 fix @xmath552 and restrict to @xmath1098 with @xmath1096 . for @xmath552
sufficiently large , there are approximately @xmath1099 positive integers @xmath405 satisfying @xmath1100 then for @xmath552 large , we have @xmath1101 choosing @xmath282 sufficiently large , we can then compute @xmath1102 this final series diverges for @xmath1103 this shows that @xmath1104 for @xmath1105 as desired .
we now provide a second method using agler kernel arguments to show @xmath1106 . by lemma [ lem : norm ]
, we can work with the douglas integral @xmath1107 and by lemma [ lem : local2 ] , this involves agler kernels .
first , in the notation of lemma [ lem : local2 ] , we have @xmath1108 . after a short computation , we find that @xmath1109 by lemma [ lem : local2 ] , this implies that @xmath1110 it remains to show that the last double integral is infinite .
first , note that @xmath1111 , @xmath1112 , is the reproducing kernel for @xmath684 .
since @xmath1113 for @xmath1114 , this observation implies that @xmath1115 setting @xmath1116 gives @xmath1117 the latter integral diverges , which implies that @xmath1118 and @xmath1119 .
any degree @xmath164 inner function of the form @xmath1120 with @xmath1121 having @xmath1122 , is in @xmath38 if and only if @xmath1123 .
this can be seen by transforming @xmath1124 into the rif treated in example [ faveex ] by precomposing with a suitable pair of mbius transformations , and exploiting the fact that such composition operators are bounded above and below in the integral version of the @xmath38 norm .
the second example is based on @xmath1125 .
the associated rational inner function is @xmath1126 the relevant pick function here is @xmath1127 since @xmath1128 , bzout s theorem says that @xmath1129 .
unlike the previous example , however , @xmath1130 .
namely , the polynomials @xmath13 and @xmath102 have a common zero of multiplicity @xmath87 at @xmath1131 , and further common zeros at @xmath1132 and @xmath1133 , each of multiplicity @xmath175 . theorem [ thm : hp ] guarantees that @xmath1134 .
in fact , the contact orders of @xmath1135 can be computed and are both equal to @xmath87 .
then by theorem [ thm : contact ] , the partial derivatives @xmath1136 are in @xmath1137 precisely when @xmath1138 .
we can also address @xmath8 membership using agler decompositions . by (
* cor.13.6 ) , @xmath1135 does not have a unique agler decomposition .
however , using the methods discussed in appendix @xmath153 from @xcite , we can compute its canonical agler pairs @xmath1139 and @xmath1140 as follows : @xmath1141 and @xmath1142 using theorem @xmath700 in @xcite , it is clear that @xmath1143 one can show that @xmath1144 has the following basis : @xmath1145 moreover , one can compute @xmath1146 one can show that @xmath1147 , and so conclude @xmath1148 using the given basis .
the derivative @xmath1149 appears less amenable to direct analysis as we are lacking a description of the corresponding ideal in the case of deficient intersection multiplicity .
we turn to the question of membership of @xmath1135 in weighted dirichlet spaces .
we address this question by determining a range of values of @xmath60 for which @xmath1150 and applying proposition [ prop : da ] .
we have @xmath1151 now , since there are constants @xmath1152 satisfying : @xmath1153 for @xmath1154 , we have @xmath1155 thus @xmath1156 if and only if @xmath1157 , and moreover by proposition [ prop : da ] , we have @xmath1158 for all @xmath1159 .
this means that we obtain membership for the same range of parameters that we would by using contact order and invoking corollary [ cor : da ] . on the other hand , @xmath1160 even though @xmath689 is not maximal .
the argument is similar to that in the previous example .
namely , by the formula for @xmath248 , the kernel @xmath1161 in includes the term @xmath1162 where @xmath1163 and by direct computation , @xmath1164 then using lemma [ lem : local2 ] , we have @xmath1165 this means that @xmath1166 again invoking the reproducing property in @xmath684 , we see that for @xmath1167 , @xmath1168 finally , we obtain the desired estimate @xmath1169
in this last section , we collect some observations and pose several problems concerning inner functions and their derivatives . in section [ sect : dirichletcontact ]
, we related one - variable dirichlet norms to the contact order of a rif on the bidisk and used this to deduce inclusion in @xmath38 from inclusion of derivatives in @xmath22 .
one could go further and ask whether this implication is reversible , that is , whether membership of @xmath29 in @xmath1170 implies that @xmath1171 . if such a statement were true , then we would have the attractive characterization that @xmath91 if and only if the contact orders of a rif @xmath29 satisfied @xmath1172 .
in particular , @xmath38 would contain no rational inner functions with singularities on the two - torus when @xmath1173 . in this paper
, we have based many of our arguments on the notion of contact orders of a rif , and it seems to the authors that this concept deserves to be explored further .
let us make a few observations here .
first , we note that there exist rifs with arbitrarily high contact order : this follows from the existence of functions belonging to arbitrarily high intermediate lwner classes @xmath1174 @xcite .
moreover , contact order is a conformally invariant quantity as can be seen from the fact that @xmath0-membership of the derivative of a rif is conformally invariant .
this implies that all level sets of a rif , not just @xmath1175 , exhibit the same type of geometry .
a futher study of such geometric issues seems to be of interest . as we noted in the geometric julia inequality [ cor : geojulia ] ,
the contact orders are at least @xmath87 when the rif has a singularity .
however , it is not immediately apparent what general values the contact orders of a rif can take . specifically , in principle
, it might be possible for a rif to have non - integer contact orders , but we do not have an example of such a function .
one can pose the same kind of questions that we have addressed in this paper for rifs for more general classes of bounded functions , for instance , for singular inner functions .
such functions can be constructed as follows .
consider a measure @xmath1176 , singular with respect to lebesgue measure on @xmath26 , whose fourier coefficients satisfies @xmath1177 compute its poisson integral @xmath1178(z_1,z_2)=\int_{{\mathbb{t}}^2}p(r_1 , s_1-t_1)p(r_2 , s_2-t_t)d\mu(t_1,t_2),\quad ( z_1,z_2)=(r_1e^{it_1 } , r_2e^{it_2})\in { \mathbb{d}}^2,\ ] ] and then find its harmonic conjugate @xmath1179 $ ] in @xmath94 to obtain a holomorphic function by setting @xmath1180=\exp(-(p[\mu]+iq[\mu]))$ ] . the resulting function is bounded and has @xmath1181(e^{is } , e^{it})|=1 $ ] almost everywhere by fatou s theorem for poisson integrals in polydisks , see ( * ? ? ?
* chapter 2 ) .
a discussion of the basic properties of singular inner functions , as well as other subclasses of inner functions , can be found in @xcite .
further developments can be found in , for instance , @xcite .
let @xmath1182 be the measure induced by the integration current associated with the subvariety @xmath1183 and define , for @xmath1184 , the singular measure @xmath1185 we have @xmath1186 and @xmath1187 otherwise , so this is an admissible measure .
the corresponding poisson integral is @xmath1188(z_1,z_2)=\sigma+2\sigma\sum_{k=1}^{\infty}(r_1r_2)^k\cos[k(t_1+t_2 ) ] , \quad ( z_1,z_2)=(r_1e^{it_1},r_2e^{it_2})\in { \mathbb{d}}^2.\ ] ] the harmonic conjugate is @xmath1189(z_1,z_2)=2\sigma\sum_{k=1}^{\infty}(r_1r_2)^k\sin[k(t_1+t_2)],\ ] ] and we obtain the analytic function @xmath1190(z_1,z_2)=p[\mu_\sigma](z_1,z_2)+iq[\mu_\sigma](z_1,z_2)=\sigma\left(1 + 2\sum_{k=1}^{\infty}(z_1z_2)^k\right)=\sigma\frac{1+z_1z_2}{1-z_1z_2}.\ ] ] after simplifying and exponentiating , we are left with the singular inner function @xmath1191(z_1,z_2)=\exp\left(-\sigma\frac{1+z_1z_2}{1-z_1z_2}\right).\ ] ] we now note that by construction , the singularity set of @xmath1192 $ ] is a set of positive @xmath60-capacity whenever @xmath1193 , see ( * ? ? ? * section 5 ) , and hence immediately deduce @xmath1192 \notin { { \mathfrak d}_{\alpha}}$ ] for @xmath1193 since exceptional sets of functions in @xmath1086 can not have positive capacity @xcite . on the other hand ,
newman and shapiro @xcite have observed that the taylor coefficients of the ( one - variable ) inner function @xmath1194 can be expressed in terms of certain special functions .
namely , we have @xmath1195 where @xmath1196 are associated laguerre polynomials , and it is known that @xmath1197 for @xmath1184 fixed . using this , we find that @xmath1192=\sum_{k\geq 0}a_k(z_1z_2)^k$ ] has @xmath1198\|^2_{\alpha}\approx \sum_{k=0}^{\infty}(k+1)^{-3/2 + 2\alpha},\ ] ] and thus , for all @xmath1184 , @xmath1192 $ ] belongs to @xmath1199 if and only if @xmath1200 . as the example shows , it is not the size of the support of a singular measure alone that determines whether the associated @xmath1180 $ ] belongs to @xmath38 .
we hope to return to membership problems for singular inner functions on the bidisk in future work . finally ,
all of the problems addressed in this paper can be stated for the @xmath552-dimensional polydisk @xmath1201 and are , to the best of the authors knowledge , completely open .
we should stress that many of the techniques used in this paper are strictly two - dimensional in nature : bzout s theorem , puiseux expansions , agler decompositions , and nevanlinna representations simply do not have exact analogs in higher dimensions .
nevertheless , we suspect that at least some of our results should have @xmath552-dimensional counterparts in terms of objects like the general resolution of singularities , see @xcite for a survey .
the authors thank greg knese and constanze liaw for numerous conversations , comments , and suggestions during the initial stages of this work .
the second and third authors thank institute mittag - leffler , where they spent a productive week engaged in this project .
the third author thanks stefan richter for useful remarks concerning local dirichlet integrals , as well as dan petersen for helpful discussions about algebraic geometric concepts .
j. agler , on the representation of certain holomorphic functions defined on a polydisc , _ topics in operator theory : ernst d. hellinger memorial volume _ , 4766 ,
oper . theory adv .
appl . , * 48 * , birkhuser , basel , 1990 .
a. kummert , 2-d stable polynomials with parameter - dependent coefficients : generalizations and new results .
special issue on multidimensional signals and systems .
ieee trans .
circuits systems i fund .
theory appl .
* 49 * ( 2002 ) , no . 6 , 725731 .
w. rudin , _ lectures on the edge - of - the - wedge theorem _
, conference board of the mathematical sciences regional conference series in mathematics , no .
american mathematical society , providence , r.i . , 1971 .
d. sarason , _ sub - hardy hilbert spaces in the unit disk .
_ university of arkansas lecture notes in the mathematical sciences , 10 . a wiley - interscience publication , john wiley & sons , inc .
, new york , 1994 .
|
we analyze the singularities of rational inner functions on the unit bidisk and study both when these functions belong to dirichlet - type spaces and when their partial derivatives belong to hardy spaces .
we characterize derivative @xmath0 membership purely in terms of contact order , a measure of the rate at which the zero set of a rational inner function approaches the distinguished boundary of the bidisk .
we also show that derivatives of rational inner functions with singularities fail to be in @xmath0 for @xmath1 and that higher non - tangential regularity of a rational inner function paradoxically reduces the @xmath0 integrability of its derivative .
we derive inclusion results for dirichlet - type spaces from derivative inclusion for @xmath0 . using agler decompositions and local dirichlet integrals
, we further prove that a restricted class of rational inner functions fails to belong to the unweighted dirichlet space .
|
with the development of internet , we have entered the era of big data .
it is consequently a natural idea to leverage the large scale yet noisy data on the web for various vision tasks @xcite .
methods of exploiting web images for automatically image dataset construction have recently become a hot topic @xcite in the field of multimedia processing .
existing methods @xcite usually use an [ fig : figure1 ] iterative mechanism in the process of image selection , but these datasets tend to be statistically problematic because of the visual feature distribution of images selected in this way , which is known as the dataset bias problem @xcite .
1 shows the airplane " images from four different image datasets .
we observe some significant differences : pascal @xcite shows airplanes " from the flying viewpoint , while sun @xcite tends to show distant views at the airport ; caltech @xcite has a strong preference for side views and imagenet @xcite is rich in diversity , but mainly contains close - range views .
classifiers learned from these datasets usually perform poorly in domain adaptation @xcite . to address this problem ,
a large number of domain adaptation approaches that explicitly cope with the noisy labels of web images have been proposed for various vision tasks @xcite .
the images are partitioned into a set of clusters ; each cluster is treated as a bag " and the images in each bag as instances " . as a result , these tasks can be formulated as a multi - instance learning ( mil ) problem , and different mil methods have been proposed in @xcite . however , the yield for all of these methods is limited by the restriction of diversity which provided by image search engine with a single query . to obtain high accuracy and diverse candidate images , as well as to overcome the download restrictions of the image search engine , @xcite proposed the use of multiple query expansions instead of a single query to collect candidate images from the image search engine .
the issue remains that these methods still use iterative mechanisms in the process of image selection , which leads to the dataset bias problem @xcite .
motivated by the situation described above , we target the construction of an image dataset in a scalable way , while ensuring robustness and accuracy .
the basic idea is to leverage multiple query expansions for the initial candidate image collection and to use mil methods for selecting images from different distributions . to obtain multiple query expansions ,
we expand each query to a set of query expansions , which results in most of the noisy expansions being filtered out . after obtaining the raw image dataset with unfiltered query expansions ,
mil methods are applied to filter individual and group noisy images . to verify the effectiveness of our proposed approach , we build an image dataset with 20 categories .
we compare the image classification ability , cross - dataset generalization ability and dataset diversity of our dataset with three manually labelled image datasets , cifar-10 , stl-10 and imagenet , to demonstrate the domain robustness of our dataset .
we also report the results of object detection on pascal voc 2007 , and then compare the object detection ability of our method with four baseline methods .
our main contributions are summarized as follows : [ 1 .
] to the best of our knowledge , we are the first to propose the automatic construction of a domain - robust image dataset .
our proposed approach , based on multiple query expansions and multi - instance learning , considers the source of candidate images and retains images from different distributions .
the dataset constructed by our approach thus efficiently alleviates the dataset bias problem .
[ 2 . ] to suppress the search error and unfiltered noisy query expansions induced noisy images , we formulate image selection as multi - instance learning problems and propose to solve the associated optimization problems by the cutting - plane and concave - convex procedure ( cccp ) algorithm , respectively .
we have released our image dataset drid-20 on website : https://drive.google.com / drive / folders/0b7ds7afpuzt1bm pwcfrkcdzwuue?usp = sharing .
we hope the diversity of drid-20 will offer unparalleled opportunities to researchers in the multi - instance learning , transfer learning , image dataset construction and other related fields .
this paper is an extended version of @xcite .
the extensions include : taking both bag level and instance level noisy images into account in the process of image selection instead of only instance level noisy images , we use a combination of bag level and instance level selection mechanisms and achieve better results ; comparing the image classification ability and dataset diversity of our dataset drid-20 with stl-10 , cifar-10 and imagenet ; and increasing the number of categories in the dataset from 10 to 20 , so that our dataset drid-20 covers all categories in the pascal voc 2007 dataset .
the rest of the paper is organized as follows : in section 2 , a brief discussion of related works is given .
the proposed algorithm including query expanding , noisy expansion filtering and noisy image filtering is described in section 3 .
we evaluate the performance of the proposed algorithm against several other methods in section 4 .
lastly , the conclusion and future work are offered in section 5 .
given the importance of labelled image datasets in the area of high - level image understanding , many efforts have been directed toward image dataset construction . in general , these efforts can be divided into three principal categories : manual methods , semi - automatic methods and automatic methods . in the early years , manual labelling was the most important way to construct image datasets .
( e.g. , stl-10 @xcite , cifar-10 @xcite , pascal voc 2007 @xcite , imagenet @xcite and caltech-101 @xcite ) .
the process of constructing these datasets mainly consists of submitting keywords to an image search engine to download candidate images , then cleaning these candidate images by manual annotation .
this method has high accuracy but is labor intensive .
to reduce the cost of manual labelling , a large number of works have focused on active learning ( a special case of semi - supervised method)@xcite@xcite@xcite .
@xcite randomly labelled some seed images to learn visual classifiers .
the learned visual classifiers were then implemented to conduct image classification on unlabelled images and find low confidence images for manual labelling . here
low confidence images are those whose probability is classified into positive and negative close to 0.5 .
the process is iterated until sufficient classification accuracy is achieved .
@xcite presented an active learning framework to simultaneously learn contextual models for scene understanding tasks ( multi - class classification ) .
@xcite presented an approach for on - line learning of object detectors , in which the system automatically refines its models by actively requesting crowd - sourced annotations on images crawled from the web .
however , both manual labelling and active learning require pre - existing annotations , which often results in one of the most significant limitations to developing a large scale image dataset .
automatic methods have attracted more and more attention to decrease the cost of manual annotation @xcite .
@xcite adopted text information to re - rank images retrieved from a web search and used these top - ranked images to learn visual models to re - rank images once again .
@xcite leveraged the first few images returned from an image search engine to train the image classifier , classifying images as positive or negative .
when the image is classified as a positive sample , the classifier uses incremental learning to refine its model . with the increase in the number of positive images accepted by the classifier , the trained classifier will reach a robust level for this query .
@xcite proposed the use of a clustering based method to filter `` group '' noisy images and a propagation - based method to filter individual noisy images .
the advantage of these methods is that the need for manual intervention is eliminated .
however , for methods @xcite , the domain adaptation ability is limited by the restriction of the initial candidate images and the iterative mechanism in the process of image selection . to obtain high diversity candidate images , @xcite proposed the use of multiple query expansions instead of a single query in the process of collecting the initial candidate images , then using an iterative mechanism to filter noisy images .
the automatic works discussed here mainly focus on accuracy and scale in the process of image dataset construction , which often results in poor performance on domain adaptation .
there are many works related to the generation of query expansions and noisy image filtering , though they are not aimed at image dataset construction . since most image search engines restrict the number of images returned for each query , wordnet @xcite and conceptnet @xcite are often used to obtain synonyms to overcome the download restriction of these search engines .
the advantage of wordnet and conceptnet is that synonyms are usually relevant to the given query and almost do not need to be purified .
the disadvantage of wordnet and conceptnet is that both of them are usually not comprehensive enough for query expanding .
worse , the images returned from an image search engine using synonyms tend to experience the homogenization problem , which results in poor performance on domain adaptation .
recent works @xcite have proposed the use of google books ngram corpus ( gbnc ) @xcite to expand query to a set of query expansions .
the google books ngrams corpus covers almost all related queries at the text level .
it is much more general and richer than wordnet and conceptnet .
the disadvantage of using gbnc for query expanding is that it may also generate noisy query expansions .
recently , word embedding @xcite provides a learning - based method for computing the word - word similarity distance which can be used to filter noisy query expansions . in this paper
, we use gbnc to expand the query to a set of query expansions , and then take both word - word and visual - visual similarity distance to filter noisy query expansions . to efficiently ease the dataset bias problem , several authors have developed domain adaptation approaches for vision tasks . @xcite clustered relevant images using both textual and visual features . by treating each cluster as a bag " and the images in the bag as instances " , the authors formulated this problem as a multi - instance learning problem ( mil ) which learns a target decision function for image re - ranking . however , the yield is limited by the restriction placed on the initial candidate images obtained from the internet using a single query . in this paper
, we focus on the mil method , as it retains images from different data distributions while filtering out noisy images .
it can be anticipated that there will be more visual patterns ( responding to different query expansions ) in our work to represent the given query .
in addition , mil methods are applied to filter group and individual noisy images to retain images from different distributions . in return , the constructed image dataset could achieve a better domain adaptation ability than traditional datasets constructed by a single query and an iterative mechanism .
we seek to construct a domain robust image dataset which can be well generalized to unseen target domains . as shown in fig .
2 , we propose three major steps for our web - supervised image dataset construction framework : query expanding , noisy expansion filtering and noisy image filtering . a set of semantically rich expansions are obtained by searching in the gbnc @xcite , from which the visually non - salient and less relevant expansions are filtered by exploiting both word - word and visual - visual similarity . after obtaining the candidate images by retrieving unfiltered expansions with the image search engine
, we treat each unfiltered expansion as a bag " and the images in each bag as instances " .
we then formulate this task as an mil problem with constrained positive bags . using this approach ,
images from different data distributions will be kept while noisy images will be filtered out , and a domain robust image dataset will be constructed .
image datasets constructed by existing methods tend to have high accuracy but usually have weak domain adaptation ability @xcite . to construct a domain - robust image dataset ,
we expand the query ( e.g. , `` horse '' ) to a set of query expansions ( e.g. , `` jumping horse , walking horse , roaring horse '' ) and then use these different query expansions ( corresponding images ) to reflect the different visual patterns " of the query .
we use gbnc to discover query expansions for the given query with parts - of - speech ( pos ) , specifically with noun , verb , adjective and adverb .
our motivation is to identify all related query expansions .
gbnc is much more general and richer than wordnet @xcite and conceptnet @xcite .
using gbnc can help us to find all the expansions ever published for any possible query . through query expanding , we obtain a comprehensive semantic description for the given query .
however , query expanding not only brings all the useful query expansions , but also some noisy query expansions .
these noisy query expansions can be roughly divided into two types : ( 1 ) visual non - salient ( e.g. , `` betting horse '' ) and ( 2 ) less relevant ( e.g. , `` sea horse '' ) . using these noisy query expansions to retrieve images will have a negative impact on dataset accuracy and robustness . from the visual perspective , we want to identify visually salient query expansions and eliminate non - salient query expansions in this step .
the intuition is that visually salient expansions should exhibit predictable visual patterns , hence we use an image classifier - based filtering method . for each query expansion ,
we directly download the top @xmath0 images from the google image search engine as positive images ( based on the fact that the top few images returned from image search engine tend to be positive ) , then randomly split these images into a training set and validation set @xmath1 .
we gather a random pool of negative images and split them into a training set and validation set @xmath2 .
we train a linear support vector machine ( svm ) classifier @xmath3 with @xmath4 and @xmath5 using dense histogram of oriented gradients ( hog ) features .
we then use @xmath6 as validation images to calculate the classification results .
we declare a query expansion @xmath7 to be visually salient if the classification results @xmath8 give a relatively high score . from the relevance perspective , we want to identify both semantically and visually relevant expansions for the given query .
the intuition is that relevant expansions should have a relatively small semantic and visual distance , therefore we use a combined word - word and visual - visual similarity distance - based filtering method . words and phrases acquire meaning from the way they are employed in society . for computers , the equivalent of `` society '' is `` database '' , and the equivalent of `` use '' is `` a way to search the database '' @xcite .
normalized google distance ( ngd ) constructs a method to extract semantic similarity distance from the world wide web ( www ) using google page counts @xcite . for a search term @xmath9 and search term @xmath10 ,
ngd is defined by : @xmath11 where @xmath12 denotes the number of pages containing @xmath9 , @xmath13 denotes the number of pages containing both @xmath9 and @xmath10 and @xmath0 is the total number of web pages searched by google .
we denote the semantic distance of all query expansions by a graph @xmath14 in which each node represents a query expansion and its edge represents the @xmath15 between the two nodes .
we set the target query as center ( x ) and other query expansions have a score ( @xmath16 ) which corresponds to the distance to the target query .
similarly , we represent the visual distance of the query and expansions by a graph @xmath17 in which each node represents a query expansion and each edge represents the visual distance between the query and the expansions .
the feature is a 1000 dimensional bag of visual words based on sift features .
the edge weight @xmath18 corresponds to the euclidean distance .
the semantic distance and visual distance will be used to construct a new two - dimensional feature @xmath19}$ ] .
the problem is to calculate the importance weight @xmath20 and bias penalty @xmath21 in decision function @xmath22 to determine whether or not the expansion is relevant .
there are many methods of obtaining these coefficients @xmath20 and @xmath21 . here
we take linear svm to work around this problem .
although linear svm is not the prevailing state - of - the - art method for classification , we find our method to be effective in pruning irrelevant query expansions .
unfiltered expansions are then used to retrieve the top @xmath23 images from the image search engine to construct the raw image dataset .
regardless of the fact that our method is unable to remove noisy expansions thoroughly in most cases , the raw image dataset constructed by our method still achieves much higher accuracy than directly using the flickr or google image data . besides
, the raw image dataset constructed through unfiltered query expansions has much richer visual patterns .
although the google image search engine ranks the returned images , several noisy images are still included . in addition , a few unfiltered noisy expansions will also bring noisy images to the raw image dataset .
in general , these noisy images can be divided into two types : group noisy images ( caused by unfiltered noisy expansions ) and individual noisy images ( as a result of the error index of the image search engine ) . to filter these group and individual noisy images while retaining the images from different distributions , we use mil methods instead of an iterative mechanism in the process of noisy image filtering . by treating each unfiltered expansion as a bag " and the images corresponding to the expansion as instances " , we formulate a multi - instance learning problem by selecting a subset of bags and a subset of images from each bag to construct a domain robust image dataset for the given query .
since the precision of images returned from the google image search engine tends to have relatively high accuracy , we define each positive bag as at least having a portion of @xmath24 positive instances which effectively filter group noisy images caused by unfiltered noisy query expansions .
we denote each instance as @xmath25 with its label @xmath26 , where i=1, ... ,n .
we also denote the label of each bag @xmath27 as @xmath28 .
the transpose of a vector or matrix is represented by superscript @xmath29 and the element - wise product between two matrices is represented by @xmath30 .
we define the identity matrix as @xmath31 and @xmath32 , @xmath33 @xmath34 denote the column vectors of all zeros and ones , respectively .
the inequality @xmath35{}'\geq \mathbf{0}}$ ] means that @xmath36 for i=1, ... ,n .
the decision function for filtering individual noisy images is assumed in the form of @xmath37 and has to be learned from the raw image dataset .
we employ the formulation of lagrangian svm , in which the square bias penalty @xmath38 and the square hinge loss for each instance are used in the objective function .
the decision function can be learned by minimizing the following structural risk function : @xmath39 @xmath40 @xmath41 where @xmath42 is a mapping function that maps @xmath9 from the original space into a high dimensional space @xmath43 , @xmath44 is a regularization parameter and @xmath45 values are slack variables .
the margin separation is defined as @xmath46 .
@xmath47{}'}$ ] means the vector of instance labels , @xmath48 and @xmath49 satisfies constraint ( 4 ) . by introducing a dual variable @xmath50 for inequality constraint ( 3 ) and kernel trick @xmath51
, we arrive at the optimization problem below : @xmath52 where @xmath53 , @xmath54 and @xmath55{}'$ ] . by defining @xmath56}$ ] as a @xmath57 kernel matrix , @xmath58 and @xmath59 as a
@xmath57 transformed kernel matrix for the augmented feature mapping @xmath60{}'}$ ] of kernel @xmath61 .
( 5 ) can be rewritten as follows : @xmath62 ( 6 ) is a mixed integer programming problem with respect to the instance labels @xmath63 .
we take the label - generating initialize @xmath64 for @xmath65 as @xmath66 , and set @xmath67 ; + use mkl to solve @xmath68 and @xmath69 in ( 10 ) with @xmath70 ; + select most violated @xmath71 with @xmath72 and set @xmath73 ; + repeat step 2 and step 3 until convergence .
+ mmc ( lg - mmc ) algorithm proposed in @xcite to solve this mixed integer programming problem .
we first consider interchanging the order of @xmath74 and @xmath75 in ( 6 ) and obtain : @xmath76 according to the minmax theorem @xcite , the optimal objective of ( 6 ) is an upper bound of ( 7 ) .
we rewrite ( 7 ) as : @xmath77 @xmath71 is any feasible solution in @xmath78 . for the inner optimization sub - problem ,
let @xmath79 be the dual variable for inequality constraint .
its lagrangian can be obtained as : @xmath80 setting the derivative of ( 9 ) with respect to @xmath81 to zero , we have @xmath82 .
@xmath83 is denoted as the domain of @xmath69 , where @xmath69 is the vector of @xmath84 .
the inner optimization sub - problem is replaced by its dual and ( 8) can be rewritten as : @xmath85 or @xmath86 here , we can interchange the order of @xmath87 and @xmath88 because the objective function is concave in @xmath89 and convex in @xmath69 .
additionally , ( 10 ) can be regarded as a multiple kernel learning ( mkl ) problem @xcite , and the target kernel matrix is a convex combination of base kernel matrices @xmath90 . although @xmath78 is finite and ( 10 ) is an mkl problem , we can not directly use existing mkl techniques like @xcite to solve this problem .
the reason is that the exponential number of possible labellings @xmath91 and the fact that the base kernels are also exponential in size make direct mkl computations intractable .
fortunately , not all the constraints in ( 8) are active at optimality , thus we can employ a cutting - plane algorithm @xcite to find a subset @xmath92 of the constraints that can well approximate the original optimization problem .
the detailed solutions of the cutting - plane algorithm for ( 10 ) are described in algorithm 1 .
finding the most violated constraint @xmath71 is the most challenging aspect of the cutting - plane algorithm . according to ( 5 ) ,
the most violated @xmath71 is equivalent to the following optimization problem : @xmath93 we solve this integer optimization problem by enumerating all possible candidates of @xmath71 . here
we only enumerate the possible labelling candidates of the instances in positive bags as all instances in the negative bags are assumed to be negative in our paper .
lastly , we can derive the decision function from the raw image dataset for the given query as : @xmath94 where @xmath95 and @xmath96 .
the decision function will be used to filter individual noisy images in each bag which correspond to unfiltered query expansions .
to filter group noisy images ( caused by unfiltered noisy expansions ) , we represent bag @xmath27 with the compound feature @xmath97 of its first @xmath98 positive instances : @xmath99 with @xmath100 we refer to the instances in @xmath101 as the first @xmath98 instances of @xmath27 according to classifier @xmath102 ( see equation 12 ) .
the closer the images in @xmath27 are to the bag centre , the higher is the probability that these images will be relevant to the bag .
the assignment of relatively heavier weights to images which are a short distance from the bag centre will increase the accuracy of classifying bag @xmath27 as positive or negative , then increase the efficiency of filtering noisy group images .
following @xcite , we assume @xmath103^{-1}}$ ] to be a weighting function , @xmath104 represents the euclidean distance of images @xmath25 from the bag centre , @xmath105 and @xmath106 are scaling and offset parameters which can be determined by cross - validation .
the representation of ( 13 ) for bag @xmath27 can be generalized to a weighted compound feature : @xmath107 with @xmath108 where @xmath109\in \mathbb{r}^{d\times i}}$ ] is a matrix whose columns are the instances of bag @xmath27 , @xmath110^{t}\in \mathbb{r}_{++}^{i}}$ ] are the vectors of weights , and @xmath111 is an indicator function for the first k positive instances of bag @xmath27 .
then classifying rule of bag @xmath27 to be positive or negative is : @xmath112 where @xmath113 is the vector of classifying coefficients , @xmath114 is the feature vector of ( 15 ) , @xmath115 is a vector of latent variables and @xmath116 is the hypothesis space @xmath117 .
the learning problem is to determine the parameter vector @xmath118 .
initialize @xmath118 with svm by setting @xmath119 ; + compute a convex upper bound using the current model for the second term of ( 21 ) ; + minimize this upper bound by solving a structural svm problem via the proximal bundle method @xcite ; + repeat step 2 and step 3 until convergence .
+ given a training set @xmath120 , this is a latent svm learning problem : @xmath121 before solving ( 18 ) , we first solve the classifying rule of ( 17 ) .
it is necessary to solve the below following problem : @xmath122 this is an integer linear - fractional programming problem .
since @xmath123 , ( 19 ) is identical to the relaxed problem : @xmath124 where @xmath125^i}$ ] is a unit box in @xmath126 .
( 20 ) is a linear - fractional programming problem and can be reduced to a linear programming problem of @xmath127 variables and @xmath128 constraints @xcite . in this work , we take the concave - convex procedure ( cccp ) @xcite to solve ( 18 ) .
we rewrite the objective of ( 18 ) as two convex functions : @xmath129-\left [ c\sum_{i\in d_{p } } f_\omega \left ( x_{b_i } \right ) \right ] } \\ \end{aligned}\ ] ] where @xmath130 and @xmath131 are positive and negative training sets respectively .
the detailed solutions of the cccp algorithm for ( 21 ) are described in algorithm 2 .
lastly , we obtain the bag classifying rule as ( 17 ) to filter group noisy images which correspond to unfiltered noisy query expansions . in summary ,
the existing automatic methods reduce the cost of manual annotation by leveraging the generalization ability of machine learning models . however , this generalization ability is affected by both the quality of the initial candidate images and the capability of models to retain images from different distributions .
previous works largely focus on accuracy and scale , and most use an iterative mechanism for the image selection process which often results in the dataset bias problem . to the best of our knowledge , we are the first to propose the automatic construction of a domain - robust image dataset .
we achieve the domain adaptation ability of our dataset by maximizing both the initial candidate images and the final selected images from different data distributions .
since the datasets for existing dataset construction methods @xcite have not been released , we are unable to directly compare our dataset with their extracted datasets .
we therefore systematically compare the image classification ability , cross - dataset generalization ability and dataset diversity of our dataset with three publicly available datasets stl-10 , cifar-10 and imagenet .
the motivation is to verify that a domain - robust image dataset has a better image classification ability on third - party datasets , and to confirm that a domain - robust image dataset has better cross - dataset generalization ability and dataset diversity .
we also report the object detection ability of our dataset and compare our method with four baseline methods @xcite . to facilitate comparison with datasets stl-10 , cifar-10 and imagenet , we choose common categories in these datasets : airplane / aeroplane , bird , cat , dog , horse to construct our dataset .
we also select 15 other categories in pascal voc 2007 to construct our dataset , since most of the existing weakly supervised and web - supervised learning methods are tested on the pascal voc 2007 dataset .
overall , we use the proposed method in this paper to build our dataset , drid-20 , which consists of all 20 categories in the pascal voc 2007 dataset . for each given query ( e.g.,horse " ) in our experiments ,
we first expand the given query to a set of query expansions with pos .
to filter visual non - salient expansions , we retrieve the top @xmath132 images from the image search engine as positive images ( in spite of the fact that noisy images might be included ) . set the training set and validation set @xmath133 , @xmath134 . through experiments ,
we declare a query expansion @xmath7 to be visually salient if the classification results ( @xmath135 ) return a relatively high score .
we have released the query expansions for twenty categories and corresponding images ( original image url ) on website : https://drive.google.com/drive/folders/0b7ds7afpuzt1bmpwc frkcdzwuue?usp = sharing .
to filter less relevant expansions , we select @xmath136 positive training samples from these expansions that have a small semantic or visual distance from these expansions .
we calculate the semantic distance and visual distance between different queries ( e.g. , `` horse '' and `` cow '' ) and obtain the @xmath137 negative training samples .
we do not select the @xmath137 negative training samples from expansions which have a large semantic or visual distance because these expansions have a higher probability of being positive than other different query expansions . here , we set @xmath138 and train a classifier based on linear svm to filter less relevant expansions . the first @xmath139 ( for category `` plant '' expansions , @xmath140 ) images are retrieved from the google image search engine for each unfiltered query expansion to construct the raw image dataset .
we treat unfiltered query expansions as positive bags and images in bags as instances .
we define each positive bag as having at least a portion of @xmath141 positive instances .
negative bags can be obtained by randomly sampling a few irrelevant images that are not associated with the given query .
mil methods are applied to learn the decision function ( 12 ) , and the individual noisy images in each bag are filtered .
the decision function of ( 12 ) is also used to select the most @xmath98 positive instances in each bag , representing this bag for group noisy image filtering .
the value of @xmath98 for different categories may be different .
in general , categories which have larger query expansions tend to select a smaller value .
there are multiple methods for learning the weighting function ( e.g. , logistic regression or cross - validation ) , here we follow @xcite and use cross - validation to learn the weighting function .
we label 10 datasets , each containing 100 positive bags and 100 negative bags .
the positive bags and negative bags each have 50 images .
labelling only needs to be carried out once to learn the weighting function and weighted bag classification rule ( 17 ) . the learned weighted bag classification rule ( 17 )
will be used to filter noisy bags ( corresponding to group noisy images ) . for better comparison with other datasets
, we evenly select positive images from positive bags to construct the dataset drid-20 .
each category in drid-20 has 1000 images , and this dataset has been released publicly on website .
p2.2cm<|p1cm<|p0.7cm<|p0.7cm<|p0.7cm<|p0.7 cm < dataset @xmath142 category & airplane & bird & cat & dog & horse + stl-10 & 1300 & 1300 & 1300 & 1300 & 1300 + cifar-10 & 6000 & 6000 & 6000 & 6000 & 6000 + pascal voc & 238 & 330 & 337 & 421 & 287 + imagenet & 1434 & 2126 & 1083 & 1603 & 1402 + drid-20 & 1000 & 1000 & 1000 & 1000 & 1000 + for dataset image classification ability , cross - dataset generalization ability and dataset diversity comparison , we select five common categories in stl-10 , cifar-10 , pascal voc 2007 , imagenet and drid-20 . for object detection ability comparison
, we use all 20 categories in the drid-20 dataset and pascal voc 2007 .
stl-10 has ten categories , and each category of which contains 500 training images and 800 test images .
all of the images in stl-10 are colour 96@xmath14396 pixels .
we use all the training images and test images in stl-10 to represent the dataset .
the cifar-10 dataset consists of 60000 32@xmath14332 colour images in 10 categories , with 6000 images per category , of which 5000 are training images and 1000 are test images .
similarly , we use all 6000 images in cifar-10 to represent the dataset .
imagenet is an image dataset organized according to the wordnet @xcite hierarchy .
it provides on average 1000 images to illustrate each category .
we use all the images in imagenet for each category to represent the imagenet dataset .
pascal voc 2007 is a benchmark dataset in image classification and object detection which provides the vision and machine learning communities with a standard dataset of images and evaluation procedures .
pascal voc 2007 contains 20 categories , each of which contains training / validation data and test data . for image classification ability , cross - dataset generalization ability and dataset diversity comparison
, we utilize the training / validation data to represent the pascal voc 2007 dataset .
our dataset drid-20 is constructed according to the categories in pascal voc 2007 and has 1000 images in each category . to evaluate the image classification ability , cross - dataset generalization ability and dataset diversity , we resize all the images in stl-10 , imagenet , pascal voc 2007 and our drid-20 to 32@xmath14332 . for all [ t ] [ fig : roi_seg_10000_result ] datasets
, we extract the same histogram of oriented gradient ( hog ) feature and train one - versus - all classifiers .
the detailed number of images in each category for above mentioned experiments is shown in table 1 .
the idea of training detection models without bounding boxes has received renewed attention due to the success of the dpm @xcite detector . to compare the object detection ability of our method with four other baseline methods @xcite , we select pascal voc 2007 as the test data , because recent state - of - the - art weakly supervised and web - supervised methods have been evaluated on this dataset . for each query expansion , we train a separate dpm to constrain the visual variance .
we resize images to a maximum of 500 pixels and ignore images with extreme aspect ratios ( aspect ratio @xmath144 2.5 or @xmath145 0.4 ) . to avoid getting stuck to the image boundary during the latent re - clustering step , we initialize our bounding box to a sub - image within the image that ignores the image boundaries .
following @xcite , we also initialize components using the aspect - ratio heuristic .
some of the components across different query expansion detectors ultimately learn the same visual pattern .
for example , the images corresponding to the query expansion walking horse " are similar to the images corresponding to standing horse " . in order to select a representative subset of the components and merge similar components , we represent the space of all query expansions components by a graph @xmath146 , in which each node represents a component and each edge represents the visual similarity between them .
the score @xmath147 for each node corresponds to the average precision .
the weight on each edge @xmath148 is obtained by running the @xmath149 component detector on the @xmath150 component set .
we solve for the same objective function proposed in @xcite to select the representative components @xmath151 : @xmath152 where @xmath153 is a soft coverage function that implicitly pushes for diversity : @xmath154 [ t ] after the representative subset of components has been obtained , we augment them with method as described in @xcite and subsequently merge all the components to produce the final detector .
we choose pascal voc 2007 as the third - party test data for comparing the image classification ability of our dataset drid-20 with stl-10 , cifar-10 and imagenet . for this experiment , we select five categories that are common to all these datasets : airplane / aeroplane , bird , cat , dog and horse .
we randomly select training images from various datasets for our choice of positive training images .
we choose the same 1000 negative training images for all datasets . for details , we sequentially select [ 200,400,600,800,1000 ] training images from cifar-10 , stl-10 , imagenet and drid-20 as the positive training images , and use 1000 fixed negative training images to learn the image classifiers .
we then test the performance of these classifiers on the corresponding categories of the pascal voc 2007 dataset .
we repeat the above experiment 10 times and use the average performance of image classifiers as the final performance for each dataset .
the image classification ability of all datasets for each category and the entire dataset is shown in figure 3 .
we make the following observations from fig . 3 : \(1 ) it is interesting to observe that the category
airplane " has a relatively higher classification accuracy than the categories bird " , cat " , dog " and horse " with a small amount of training data [ 200,400 ] .
a possible explanation is that the scenes and visual patterns of airplane " are relatively simpler than the categories bird " , cat " , dog " and horse " .
even with a small amount of training data , there are still a large number of positive patterns in both auxiliary and target domains .
that is to say , the samples are densely distributed in the feature space , and the distribution of the two domains overlap much more easily . on the other hand ,
the positive samples from both domains for the categories bird " , cat " , dog " and horse " are distributed sparsely in the feature space .
it is likely that there will be less overlap of the data distributions of the two domains .
\(2 ) cifar-10 exhibits much worse performance on image classification than stl-10 , imagenet and drid-20 according to the accuracy over all five common categories , which demonstrates that the svm classifier learned with training data from the auxiliary domain performs poorly on the target domain .
the explanation is perhaps that the data distributions of cifar-10 are quite different from those of the pascal voc 2007 dataset .
the cifar-10 dataset has a more serious dataset bias problem than stl-10 , imagenet and drid-20 .
\(3 ) we also observe that imagenet is slightly worse than drid-20 in each individual category and in the whole dataset , possibly because the distribution of samples from imagenet is relatively rich .
imagenet is constructed with the goal that objects in images should have variable appearance , positions , [ t ] view points , and poses , as well as background clutter and occlusions .
\(4 ) drid-20 outperforms cifar-10 , stl-10 and imagenet in terms of average accuracy in five common categories , which demonstrates the domain robustness of drid-20 .
the explanation is that drid-20 constructed by multiple query expansions and mil selecting mechanisms has much more visual patterns than cifar-10 , stl-10 and imagenet when given the same number of training samples . in other words
, drid-20 has much richer feature distribution and is more easily overlapped with unknown target domains .
we also report the hardware configuration of our experiment .
we use hp desktop pcs ( 3.2ghz cpu with 8 gbyte ram ) for the image collection .
all the data processing and experiments are performed on an acer workstation ( 3.5ghz cpu , 16 gbyte ram and 4 gbyte vram ) with libsvm @xcite .
cross - dataset generalization ability measures the performance of classifiers learned from one dataset and tested on other datasets .
it indicates the domain robustness of dataset @xcite .
here we compare the cross - dataset generalization ability of our dataset drid-20 with three publicly available dataset cifar-10 , stl-10 and imagenet .
we choose the same five categories ( horse , bird , airplane , cat and dog ) included in all four datasets to verify their cross - dataset generalization ability .
we randomly select 200 images per category from each dataset as the test data . for the choice of training data , we sequentially select [ 200,300,400,500,600,700,800 ] images per category from various datasets as the positive training images and use 1000 fixed negative training images to learn the image classifiers .
the training images in each category are selected randomly and the training images and test images have no duplicates .
the average classification accuracy for five categories ( horse , bird , airplane , cat and dog ) represents the cross - dataset generalization ability of one dataset on another dataset .
when training the image classification model , we set the same options for four datasets ; we set the type of svm as c - svc , the type of kernel as a radial basis function and all other options as the default libsvm options .
the cross - dataset performance of the four datasets and their average performance is shown in figure 4 . by observing figure 4 , we draw the following conclusions : \(1 ) in three of four datasets , the best classification performance with the increase in the number of training images is achieved by drid-20 .
when tested on stl-10 , imagenet and drid-20 , it can be seen that the generalization ability of stl-10 , imagenet and our dataset drid-20 is very close , but drid-20 performs slightly better than stl-10 and imagenet .
in addition , drid-20 outperforms cifar-10 , stl-10 and imagenet in terms of average cross - dataset performance on four datasets , which demonstrates the domain robustness of drid-20 .
a possible explanation is that our drid-20 dataset , being constructed by multiple query expansions , has much more visual patterns or feature distributions than stl-10 or cifar-10 , which just use only one query for candidate image collection . at the same time , mil selection mechanisms maximize the retention of useful visual patterns to represent the drid-20 dataset .
\(2 ) cifar-10 shows poor performance on cross - dataset generalization except on its own dataset .
the explanation is that the data distributions of its auxiliary domain and target domain are quite strongly related , making it difficult for other datasets to exceed its performance when tested on cifar-10 .
all images in cifar-10 are cut to 32@xmath14332 and objects in these images are located in the middle of the image . besides , these images contain relatively small other objects or scenes .
images in stl-10 are 96@xmath14396 and are full size in imagenet and drid-20 .
these images not only contain target objects , but also include a large number of other scenarios or objects .
based on these conditions , although cifar-10 has a better performance on its own domain , it still has a serious dataset bias problem which coincides with its average cross - dataset generalization performance . following @xcite , we compute the average image of each category and measure the lossless jpg file size , which reflects the amount of information in an image .
the basic idea is that a diverse image dataset will result in a blurrier average image , whereas an image dataset with little diversity will result in a more structured , sharper average image .
therefore , we expect the average image of a more diverse image dataset to have a smaller jpg file size .
we resize all images in stl-10 , imagenet and drid-20 to 32@xmath14332 grey images , and create average images for each category from 100 randomly sampled images . fig .
5 compares the image diversity of five common categories in drid-20 , imagenet , and stl-10 , and shows example images .object detection results ( a.p . ) on pascal voc 2007 ( test ) . [ cols="^,^,^,^,^,^,^",options="header " , ] and average images in these datasets . by observing fig .
5(a ) , it can be seen that the average image of drid-20 is blurred and it is difficult to recognize the object , while the average image of imagenet and stl-10 is relatively more structured and sharper .
drid-20 has a slightly smaller jpg file size than imagenet and stl-10 .
this phenomenon is universal for all five categories .
drid-20 is constructed with the goal that images in this dataset should exhibit domain robustness and be able to effectively alleviate the dataset bias problem . to achieve domain robustness , we not only consider the source of the candidate images , but also retain the images from different distributions .
we can see from the above experiments that , with a certain number of samples , drid-20 has much more effective visual patterns and feature distributions than the cifar-10 , stl-10 and imagenet datasets , and thus has better domain adaptation ability .
we report the performance of object detection on the pascal voc 2007 test set .
table 2 shows the results of our proposed method and compares it to the state - of - the - art weakly supervised and web - supervised methods @xcite .
methods @xcite and @xcite have state - of - the - art performance for weakly supervised object detection .
@xcite is trained on manually selected videos without bounding boxes and shows results on 10 out of 20 categories .
@xcite uses weak human supervision ( voc data with image - level labels for training ) and initialization from objectness @xcite .
@xcite takes web supervision and then trains a mixture dpm detector for the object .
@xcite is a fully supervised object detection method and it is a possible upper bound for weakly supervised and web - supervised approaches .
compared to @xcite which uses weak supervision , and @xcite which uses full supervision , the training set of our proposed approach and @xcite do not need to be labelled manually .
nonetheless , the results of our proposed approach and @xcite surpass the previous best results of weakly supervised object detection methods @xcite .
a possible explanation is perhaps that both our approach and that of @xcite use multiple query expansions for candidate image collection , and the training data collected by our approach and @xcite are richer and contain more effective visual patterns . in most cases , our method surpasses the results in @xcite , which also uses web supervision and multiple query expansion for candidate image collection .
the explanation for this is that we use different mechanisms for noisy images removal .
compared to @xcite which takes iterative mechanisms in the process of noisy image filtering , our approach applies an mil method for removing noisy images .
this maximizes the ability to retain images from different data distributions while filtering out the noisy images . by using the same feature and training strategies
, our approach achieves the best performance compared to weakly supervised and web - supervised method @xcite .
the main reason for this is that our training data are generated from multiple query expansions and mil filtering mechanisms , and thus contain much richer and more accurate visual descriptions for these categories . in other words
, our approach discovers many more useful linkages to visual patterns for the given category .
in this paper , we presented a new framework for domain - robust image dataset construction with web images .
three successive modules were employed in the framework including query expanding , noisy expansion filtering and noisy image filtering . to verify the effectiveness of our proposed method
, we constructed an image dataset drid-20 .
extensive experiments show that our dataset has better domain adaptation ability than the traditional manual - labelled datasets stl-10 , cifar-10 and imagenet .
in addition , our data was successfully applied to help improve object detection on pascal voc 2007 , and the results demonstrated the superiority of our method to several weakly supervised and web - supervised state - of - the - art methods .
we have publicly released the drid-20 dataset to facilitate the research in this field .
although good results were obtained , there is still room to improve the proposed dataset construction framework .
for example , we can potentially use more sophisticated approaches to purify noisy query expansions , noisy images and that will be the focus of our future work .
this research was supported by the national natural science foundation of china ( no . 61473154 ) .
|
labelled image datasets have played a critical role in high - level image understanding ; however the process of manual labelling is both time - consuming and labor intensive .
to reduce the cost of manual labelling , there has been increased research interest in automatically constructing image datasets by exploiting web images .
datasets constructed by existing methods tend to have a weak domain adaptation ability , which is known as the `` dataset bias problem '' . to address this issue
, we present a novel image dataset construction framework which can be generalized well to unseen target domains . in specific ,
the given queries are first expanded by searching in the google books ngrams corpus to obtain a richer semantic description , from which the visually non - salient and less relevant expansions are filtered out . by treating each unfiltered expansion as a `` bag '' and the retrieved images as `` instances '' ,
image selection can be formulated as a multi - instance learning problem with constrained positive bags .
we propose to solve the employed problems by the cutting - plane and concave - convex procedure ( cccp ) algorithm .
using this approach , images from different distributions will be retained while noisy images will be filtered out . to verify the effectiveness of our proposed approach , we build a domain - robust image dataset with 20 categories , which we refer to as drid-20 .
we compare drid-20 with three publicly available datasets stl-10 , cifar-10 and imagenet .
the experimental results confirm the effectiveness of our dataset in terms of image classification ability , cross - dataset generalization ability and dataset diversity .
we further run object detection on pascal voc 2007 using our data , and the results demonstrate the superiority of our method to the weakly supervised and web - supervised state - of - the - art detection methods .
shell : bare demo of ieeetran.cls for ieee journals domain robust , multiple query expansions , image dataset construction , mil
|
collisions of heavy nuclei at the relativistic heavy ion collider have created matter with energy densities exceeding the predicted threshold for deconfinement of color charge into a hot dense plasma @xcite . in this quark gluon plasma(qgp ) , quarks and gluons are not bound within hadronic states and the matter behaves collectively .
comparisons with hydrodynamic simulations indicate rapid thermalization of the colliding system into a hot dense nuclear medium .
the produced medium affords an opportunity to study the properties of a new phase of quantum chromodynamics ( qcd ) in an extreme environment .
hard scattering with large momentum exchange between partons in the incoming nuclei is well - described by perturbative qcd ( pqcd ) .
the scattered partons emerge back - to - back in azimuth in the plane transverse to the beam direction , and fragment into a pair of correlated cones of high momentum particles , referred to as jets .
the study of jets and their hadronic fragments in heavy - ion collisions provides insight into the properties of hot dense nuclear matter .
measurements of single high transverse momentum ( @xmath3 ) particles @xcite and correlations between high-@xmath3 particles @xcite have demonstrated that the fast partons embedded in the produced medium dissipate a large amount of their initial kinetic energy . in this paper , we present angular correlations of hadron pairs with both hadrons in the midrapidity range @xmath4 . fragments from the same jet form a peak at small relative azimuthal angle ( @xmath5 ) , i.e. the near - side peak .
pairs composed of one fragment from each jet will appear in an away - side peak at @xmath6 .
past measurements @xcite for hadrons @xmath7 5 gev/@xmath1 have shown that the away - side correlations peak is suppressed relative to baseline measurements in @xmath8+@xmath8 collisions .
the suppression of the away - side jet is a signature of parton energy loss inside the medium .
the same measurements show that near - side jet fragments at large momentum are not suppressed .
this feature of the data is understood to result from the requirement of a large momentum particle in the final state , which creates a bias towards small energy loss either by the preferential selection of hard scatterings near the medium surface @xcite or due to fluctuations in energy loss @xcite .
the detailed mechanism by which partons lose energy when passing through a deconfined medium are not yet fully understood . in pqcd descriptions of the parton - medium interaction
the predicted parton energy loss should scale as the path length squared @xcite . in competing anti - de - sitter space / conformal field theory descriptions characterizing a strongly coupled medium , the energy loss scales as the path length cubed @xcite .
the variation in azimuthal angle of the away - side jet suppression with respect to the reaction plane ( @xmath9 ) is sensitive to the total amount of energy lost by the away - side parton along long paths ( out - of - plane ) or short paths ( in - plane ) through the medium .
the degree to which the away - side jet suppression varies will be determined in part by the path - length scale of energy loss .
single particle observables at high @xmath0 , such as the nuclear suppression with respect to the reaction plane ( @xmath10 ) or the azimuthal anisotropy ( i.e. @xmath11 ) @xcite , are also sensitive to this path - length variation of energy loss .
current pqcd calculations predict a lower @xmath11 than is found in the data and may imply a larger than path length squared dependence to energy loss @xcite .
the reaction - plane dependence of the back - to - back jets provides an additional test on the path length dependence in that the two particle observable selects a different distribution of hard - scattering locations and should probe longer paths through the medium than single particle observables .
the path - length dependence of both single and two particle observables have already been studied through selection of the collision centrality @xcite .
however , centrality selection varies not only the path length , but also other important properties of the medium ( e.g. the overall energy density ) .
selection with respect to the reaction plane more directly varies the path lengths , while leaving the other medium properties unchanged .
in addition to the uncertainties associated with the energy loss mechanisms , many of the details within hydrodynamic simulations of heavy - ion collisions have not been fully constrained and tested by experiment . for instance , one such uncertainty is the geometrical description of the energy deposited by the colliding nuclei which could contribute to the degree of away - side suppression variation with respect to the reaction plane .
two competing descriptions , the glauber model @xcite and the color glass condensate @xcite , predict different azimuthal distributions of matter with respect to the reaction plane . thus the two descriptions give different in - plane and out - of - plane path lengths through the medium .
these descriptions are also used as different starting points to the hydrodynamic evolution of the medium .
other model uncertainties include , but are not limited to , the extent of geometry fluctuations , the time required for thermalization into a hydrodynamic medium , the characteristics of the phase transition to confined hadrons , and the conditions under which those hadrons become free - streaming particles into the vacuum .
these ambiguities in the proper modeling of heavy - ion collisions can result in significant uncertainty in the extracted properties of the medium , such as the shear viscosity @xcite .
in midcentral collisions ( the middle 2060% of the total cross section ) the variation of the away - side suppression is expected to be largest as the collision zone is the most anisotropic .
in contrast , central collisions are much more isotropic and so provide a sample of events with small anisotropy expected in the away - side suppression .
for instance , in the glauber model , midcentral events will have a root - mean - square thickness through the medium of 3.2 fm in the in - plane direction versus 4.8 fm in the out - of - plane direction , which is a 50% variation in path length .
however , for central 020% collisions , the path length through the medium varies from 5.0 fm in the in - plane direction to 5.8 fm in the out - of - plane direction , which is a much smaller 16% variation .
it is notable that the thickness through the medium in midcentral collisions changes more with respect to the reaction plane than it does between central and midcentral collisions where the away - side suppression at large momentum is already known to vary @xcite . also worth noting
is that the largest thickness in midcentral collisions is comparable to the shortest thickness in central collisions .
any prediction for the away - side suppression with respect to the reaction plane will be a convolution of the energy loss and a description of the space - time evolution of the medium . in the limit where the medium is never fully opaque to fast partons and
the energy lost by the typical parton is some fraction of its initial energy , the away - side suppression will increase with angle with respect to the reaction plane .
this results because the average path length through the medium of the recoil parton is longer when out of the reaction plane .
it is possible , in the extreme limit of a medium with a large opaque core and thin transparent corona , the away - side suppression could instead weaken as the trigger particle orientation varies from the in - plane to out - of - plane directions .
the weakening in the thin corona scenario results from two effects ; a larger relative number of scattering centers producing a pair of back - to - back final - state particles in the out - of - plane direction , but also the variation of the trigger particle multiplicity by angle with respect to the reaction plane .
however , it is worth noting that a large core and thin corona is an extreme configuration .
variations within more realistic models of the away - side suppression will be intermediate between these extreme scenarios . in this paper , we present azimuthal correlation measurements between pairs of neutral pion trigger particles ( @xmath12 ) within @xmath13 47 gev/@xmath1 and charged hadron associated partner particles ( @xmath14 ) within @xmath15 34 , 45 , and 57 gev/@xmath1 .
these combinations of final - state particle momentum ranges have previously been shown to be dominated by jet fragmentation as they are above medium - induced two - particle correlations which contribute significantly at lower momenta @xcite .
the low momentum structures ( the `` ridge '' and `` shoulder '' ) may be the result of parton - medium interactions ( e.g. @xcite ) or global correlations from fluctuating initial conditions @xcite . these fluctuations have substantially less impact at large pair momentum where the background contribution becomes small . in this study , as illustrated in fig .
[ fig : phisdef ] , using only particles at large pair momentum , the away - side ( @xmath16 ) suppression by trigger particle orientation with respect to the reaction plane ( @xmath17 ) is presented as a probe of both the mechanism of parton energy loss and the space - time evolution of matter created by the collisions of large nuclei .
( color online ) definition of azimuthal angles .
trigger and associated partner particles are measured at @xmath18 and @xmath19 , respectively . the trigger particle orientations
are taken with respect to the reaction plane , @xmath20 .
the relative azimuthal separation of the trigger particle and the associated partner particle is @xmath21 . ]
a previous measurement @xcite by the star collaboration for 2060% centrality between 46 gev/@xmath1 trigger particles and @xmath22 gev/@xmath1 partner particles for two @xmath23 wide in - plane and out - of - plane selections indicated an increased suppression of the away - side jet for the out - of - plane , but with little significance due to large underlying event subtraction uncertainties .
the new results presented in our paper have sufficient statistics to specify a trend in the away - side suppression in midcentral collisions at larger momentum where subtraction uncertainties are negligible .
the results presented here are based on 3.4 billion minimum - bias au+au events recorded by the phenix detector in 2007 .
comparisons to @xmath8+@xmath8 collisions use previously published measurements from data recorded in 2006 @xcite .
collision centrality was determined by division into percentile of the integrated charge collected by beam - beam counters ( bbc ) @xcite located at @xmath24 between 3.0 and 3.9 .
the timing between the arrival of charged particles in the north and south bbc was used to reconstruct the event position along the collision axis ( @xmath25-vertex ) , and to restrict the event sample to @xmath26 cm of the nominal interaction point of the two beams .
the orientation of reaction - plane azimuthal angle ( @xmath27 ) is reconstructed event - by - event using the quadrupole component ( @xmath28 ) of the charge in the reaction plane ( rxpn ) detector @xcite , located at @xmath24 between 1.0 and 2.8 .
the resolution of the rxpn detector is highest for midcentral collisions ( @xmath2920% ) where both the quadrupole component and the detector occupancy are large .
the set of resolution corrections , @xmath30 , for single particle anisotropies , @xmath31 , where : @xmath32 are estimated from correlations between the independent north ( @xmath33 ) and south ( @xmath34 ) rxpn reaction - plane reconstructions @xcite .
a single fit parameter ( @xmath35 ) is mapped into the resolution corrections via : @xmath36 the fit parameter is extracted from the correlations via : @xmath37 where @xmath38 the set of functions , @xmath39 and @xmath40 , are the even - ordered modified bessel functions and the modified struve functions respectively .
the extracted values used to correct the measured second - order azimuthal anisotropy , are @xmath41 = 0.66(4 ) and 0.66(3 ) for 020% and 2060% collisions , respectively . a 10% systematic uncertainty in 05% collisions and 5% elsewhere accounts for non - flow contributions to the resolution corrections @xcite .
the similar values are a result of the peak in reaction - plane resolution appearing near 20% centrality . a direct inspection of these reaction - plane distributions for events containing a photon above 1 gev/@xmath1 did not reveal significant contributions from jets .
neutral pion trigger particles are reconstructed from photon clusters measured by either lead - glass or lead - scintillator electromagnetic calorimeters ( emcal ) in the two central arms of phenix , in total covering @xmath42 and @xmath43 in azimuth @xcite .
clusters are subject to cuts based on the known response of the emcal , including noisy and low - response towers , as well as shower shape cuts .
neutral pions are identified through the 2@xmath44 decay channel by pairing all photons within an event .
incorrect pairings between photons create a broad combinatorial background under the @xmath45 mass peak .
this background is minimized by requiring the reconstructed mass to lie near the @xmath45 mass peak .
this requirement was 0.1250.160 mev/@xmath46 for central events , but was relaxed to 0.1200.165 mev/@xmath46 in midcentral events where the combinatorial background is lower .
since combinatorial pairs are more often made with the abundant photons found at low energy , the energy asymmetry of the decay ( @xmath47 ) was restricted to be less than 0.5 for 05% central events .
this was also relaxed slowly for more peripheral events until all pairs with asymmetries less than 0.7 were accepted .
the tightness of the cuts was used to control the rate of combinatorial pairings such that @xmath45 trigger particles have a signal - to - background ratio averaged over the mass window of 4:1 in central collisions and 10:1 in midcentral collisions .
charged hadron partner particles are reconstructed in the central arms using the drift chambers ( dc ) with hit association requirements in two layers of multi - wire proportional chambers with pad readout ( pc1 and pc3 ) , achieving a momentum resolution , @xmath48 , of @xmath49 ( gev/@xmath1 ) . only tracks with unambiguous and distinguishable dc and pc1 hit information
projections of these tracks are required to match a pc3 hit within a @xmath50 proximity window to reduce background from conversion and decay products .
a track association to a signal in the ring imaging erenkov detector is used to reject electrons for partner selections below 5 gev/@xmath1 where little signal is produced by charged pions .
within an event , all pairs formed from @xmath45 trigger particles ( @xmath51 = 47 gev/@xmath1 ) and three sets of charged hadron associated partner particles ( @xmath52 = 34 , 45 , 57 gev/@xmath1 ) are measured .
two centrality classes are used : a central selection of 020% collisions and a midcentral selection of 2060% collisions .
trigger particles are separated into six @xmath53 bins in azimuthal angle with respect to the reaction plane , @xmath54 .
the angular resolution of the measured reaction plane , at approximately @xmath55 , is larger than this binning ; consequently , significant smearing takes place between neighboring trigger orientation bins .
pairs within phenix are collected at different efficiencies due to the non - uniform central arm acceptance .
the relative pair efficiencies are corrected by mixed pair distributions in which trigger and partner particles are drawn from different events of the same class ( bins of 5% centrality , 5 cm @xmath25-vertex ) .
the resulting acceptance - corrected distributions are reported as correlation functions , @xmath56 , which are defined as : @xmath57 where @xmath58 is the number of measured pairs per event for either the same or mixed events , as indicated .
double - struck notation ( @xmath59 ) is used here to indicate measured quantities .
representative correlation functions for in - plane and out - of - plane trigger particle orientations are shown in fig .
[ fig : cfs ] .
the full set of the measured correlation functions used in this analysis is shown in figs .
[ fig : cfsall_00_20 ] and [ fig : cfsall_20_60 ] . note that these distributions are not corrected for reaction - plane resolution .
( color online ) correlation functions for the most in - plane , @xmath60=015@xmath61 , ( solid squares ) and out - of - plane , @xmath60=75 - 90@xmath61 , ( open squares ) trigger @xmath45 orientations in central 020% and midcentral 2060% collisions , left and right columns respectively , for 34 gev/@xmath1 partner hadrons .
expected underlying event contributions are shown as solid curves ( see text for details ) . ]
these inclusive pairs are assumed to correlate in one of two ways .
( 1 ) two particles within the same event may correlate trivially by participation in the same collision geometry .
these pairs produce an azimuthal angular correlation from the the single particle anisotropy with respect to the reaction plane .
( 2 ) two particles may also correlate with each other via the same hard - scattering process .
these particles will be fragments from the same ( di)jet . to separate the jet particle pairs from the other background pairs , the two - source assumption is expressed as @xcite : @xmath62 where the jet contribution to the correlation function is contained in @xmath63 .
the remaining harmonic terms describe the background contribution which is complicated by the trigger particle binning with respect to the reaction plane .
the background modulation coefficients ( @xmath64 ) are calculated via : @xmath65 this description of the background accounts for the trigger particle binning and reaction - plane resolution effects on the background shape @xcite .
the trigger particle orientation appears explicitly in terms of the bin center , @xmath9 , and width , @xmath1 .
single particle anisotropy values , @xmath28 and @xmath66 , were measured by correlating the trigger and partner particles with respect to the reaction plane , such that : @xmath67 where the observed anisotropies are corrected for the reaction - plane resolution as described previously in equation [ eq : deltadef1 ] . given sufficient detector resolution and narrowness of the trigger particle orientation binning , the sign of the @xmath68 term in eq .
[ eq : defcf ] will flip sign between in - plane and out - of - plane bins as shown in fig .
[ fig : cfs ] .
this effect is expected as selecting out - of - plane trigger particles should decrease the likelihood of finding a second background particle nearby .
the same is not true for particles correlated via hard scattering .
both the second- and fourth - order anisotropy of the background correlations have been considered as the finite fourth - order contributions were determined to be non - negligible for some trigger particle orientations .
likewise , higher - order terms in the reaction - plane resolution correction are also included . the uncertainties on the reaction - plane resolution corrections ( @xmath69 ) and the observed anisotropies ( @xmath70 , @xmath71 ) are propagated separately as they impact the away - side suppression with respect to the reaction plane in characteristically different ways . the uncertainty in the reaction - plane resolution corrections is fully correlated between trigger orientations . for instance , this uncertainty increases ( or decreases ) both the extracted in - plane and out - of - plane jet yields at @xmath72
however , the uncertainty in the observed anisotropies is fully anti - correlated between trigger orientations .
thus , this uncertainty increases the extracted in - plane yield while decreasing the out - of - plane yield ( or vice versa ) . at large momentum , both of these subtraction uncertainties are small and always dominated by other sources . the subtraction procedure was also examined for contamination of the jet correlations by fakes in the charged tracking , which become significant at large @xmath3 .
the fake high @xmath3 tracks are present only in the partner sample and are largely uncorrelated with trigger particle for the partner @xmath3 presented here .
thus the fake tracks , which are already less influential in events with a high @xmath3 @xmath45 , are subtracted with other uncorrelated pairs as part of the background contribution , so long as the anisotropies are measured with the same particle cuts .
the subtraction robustness against tracking fakes at high @xmath0 was checked by repeating the procedure with a 3@xmath73 pc3 matching requirement . by taking the trigger particle orientation as @xmath74 , the bin width as @xmath75 , and by truncating higher than second - order terms , the functional form of the background in eq .
[ eq : defcf ] reduces to the @xmath76 modulation used in previous trigger particle orientation averaged results such as those found in @xcite .
this property is demonstrated in fig .
[ fig : cfs_ave ] where the trigger particles from all orientations are considered .
correlation functions for trigger @xmath45s averaged over all trigger orientations in central 020% and midcentral 2060% collisions , left and right columns respectively , for 34 gev/@xmath1 partner hadrons .
expected average underlying event contributions are shown as solid curves . ]
the background level , @xmath77 , is determined using the zero yield at minimum ( zyam ) method @xcite . at high-@xmath3 ,
the well - separated near- and away - side jets provide a large angular region at mid-@xmath5 angles with negligible jet signal .
this allows the zyam level to be found with negligible bias and sufficient statistics despite the lower efficiency phenix has for collecting pairs near @xmath78 .
the zyam uncertainty was estimated through simulation of the statistical uncertainties as has been described in @xcite .
the jet contribution , @xmath63 , is then reported as a per - trigger azimuthal yield such that : @xmath79 the efficiency - corrected single particle and pair rates are @xmath80 and @xmath81 respectively .
the single particle partner efficiency , @xmath82 , is estimated in simulations of detector acceptance and occupancy as was done in @xcite . by design
, the trigger particle efficiency cancels in the ratio .
central events , 020% collisions , are analyzed as a cross check against experimental artifacts in midcentral collisions since they have a smaller away - side jet yield .
thus , the central events should exhibit a smaller trigger particle angle variation , require a larger reaction - plane resolution correction , a larger event correlation subtraction , and have increased background in @xmath45 identification .
representative per - trigger azimuthal yields in central collisions for each of the partner momentum selections for the most in - plane and most out - of - plane trigger particle selections are shown in fig .
[ fig : jfs_00_20 ] . figure [ fig : jfsall_00_20 ] shows the full set of the measured per - trigger azimuthal yields used in this analysis for central collisions .
the most in - plane and most out - of - plane trigger - particle orientations select the shortest and longest average path lengths through the medium , respectively , and thus may be expected to have the maximimum differences . on the near - side ,
a jet distribution is clearly observed for each selection . a direct comparison between the most in - plane and most out - of - plane trigger shows no significant variation .
the measurement at mid-@xmath5 demonstrates the good agreement resulting from correct description of the underlying event correlations . on the away - side ,
the jet yield is small due to medium suppression and the statistical precision suffers once the pairs are divided among the various trigger particle orientations .
no evidence of experimental artifacts such as over - subtraction or incorrect description of the background is seen , despite the challenging analysis environment present in central collisions .
integrated near- and away - side per - trigger yields ( @xmath83 ) are calculated within angular @xmath5 windows , as indicated in fig .
[ fig : jfs_00_20 ] , approximating the @xmath84 width of the jet distributions measured in the trigger particle orientation averaged results .
the near - side azimuthal integration windows are @xmath85 ( @xmath86 ) for @xmath87 gev/@xmath1 ( @xmath88 gev/@xmath1 ) .
similarly , the away - side azimuthal integrations windows are @xmath89 ( @xmath90 ) for @xmath87 gev/@xmath1 ( @xmath91 gev/@xmath1 ) .
use of these windows corresponds to an assumption that the jet distributions do not widen significantly at high @xmath3 , as a function of the trigger particle orientation with respect to the reaction plane .
this assumption is supported by the absence of significant centrality dependence in jet correlation widths ( @xmath92 ) for particles at high @xmath3 @xcite . within statistical uncertainties
a constant jet width is consistent with the data .
integrated yields as a function of trigger particle orientation for both the near- and away - side are then corrected for the reaction - plane resolution .
the resolution correction is applied such that : @xmath93 where @xmath83 and @xmath94 are the corrected and uncorrected yields , respectively .
the value of @xmath95 is the observed second - order anisotropy of integrated per - trigger yield with respect to the reaction plane and is determined by fitting the trigger particle orientation dependence of each @xmath96 measurement individually .
this procedure is the similar to the correction of reaction - plane resolution on single particles , here applied to integrated per - trigger pair yields .
the corrected per - trigger yields ( @xmath83 ) are reported as the nuclear jet suppression with respect to @xmath8+@xmath8 collisions , @xmath97 .
the result for central collisions is shown in fig .
[ fig : iaa_00_20 ] .
the variation of the fit used in the resolution correction is the dominant source of @xmath9-correlated uncertainty , having larger impact than the insignificant event anisotropy uncertainties . in the case of zero signal variation with reaction plane orientation , the correction becomes completely correlated with statistical scatter in the uncorrected measurement .
thus , the @xmath9-correlated systematic uncertainty from the resolution correction is conservatively treated as correlated with the statistical uncertainty when computing the final significance of the measured trends .
for the same reason , this source of systematic uncertainty has little correlation between the centrality and momentum selections . for central events
the near - side suppression is consistent with a constant as a function of @xmath9 within the statistical and @xmath9-correlated systematic uncertainties .
the values are also consistent with no suppression when considering the global scale uncertainty that appears on the trigger particle orientation averaged @xmath98 .
on the away - side , there is significant suppression in central events , as evidenced by the trigger particle averaged @xmath98 , but the statistical precision with which to determine the @xmath9 variation is limited .
mid - central events , 2060% collisions , have greater eccentricity and could be expected to show correspondingly larger trigger particle orientation dependence due to path - length variation through the collision zone .
the same set of representative per - trigger azimuthal yields is shown in fig .
[ fig : jfs_20_60 ] for the midcentral selection .
the full set at midcentrality is shown in fig .
[ fig : jfsall_20_60 ] .
again , the near - side jets for the most in - plane and most out - of - plane trigger particle orientations are consistent with each other , a direct indication of little variation with respect to the reaction plane .
the mid-@xmath5 are also in agreement with zero , as before , further demonstrating that the underlying event flow correlations are well described by equations [ eq : defcf]-[eq : flowvariables ] .
in contrast to the near - side , the away - side measurements ( see insets in fig . [
fig : jfs_20_60 ] ) change between the in - plane and out - of - plane trigger particle orientations with the latter having consistently smaller yield for all partner momenta .
the integrated near- and away - side per - trigger jet yields for midcentral collisions are shown in fig .
[ fig : iaa_20_60 ] .
the near - side jet is essentially flat , with negligible suppression ( @xmath99 ) .
the away - side jet yield is increasingly suppressed with increasing @xmath9 .
this falling trend results in only small associated particle yield remaining for out - of - plane trigger particle orientations .
( color online ) angle with respect to the reaction - plane dependence of the nuclear suppression factor , @xmath98 , expressed as the ratio between in - plane and out - of - plane trigger particles from fits to the data in central 020% collisions .
the bars represent total uncertainty taking into account the correlations between sources ( see text for details ) . ]
( color online ) angle with respect to the reaction - plane dependence of the nuclear suppression factor , @xmath98 , expressed as the ratio between in - plane and out - of - plane trigger particles from fits to the data in midcentral 2060% collisions .
the bars represent total uncertainty taking into account the correlations between sources ( see text for details ) . ] [ cols="^,^,^,^,^,^,^,^,^,^,^,^ " , ] in order to quantify the variation and significance of the trigger particle orientation dependencies shown in figs .
[ fig : iaa_00_20 ] and [ fig : iaa_20_60 ] , the ratio of the out - of - plane to in - plane suppression ( @xmath100 ) is constructed . in the ratio , the global scale uncertainties on each measurement cancel .
the @xmath98 values at @xmath9 = @xmath101 ( @xmath102 ) and at @xmath78 ( @xmath103 ) are estimated by both linear and flow - like cosine fits to the trigger particle angle measurements and evaluation at these angles .
the reported ratios are therefore independent of the chosen binning with respect to the reaction plane and the values do not rely heavily on the assumed functional form of the dependence .
the best - fit was determined by @xmath104 minimization in which : @xmath105 where @xmath106 is @xmath107 for the @xmath108 variation of the @xmath9-correlated systematic error @xcite .
as discussed above , the systematic uncertainty is conservatively treated as fully correlated with the statistical uncertainty .
the difference between linear and cosine fits provides only a small source of additional uncertainty due to the unknown true functional form .
the resulting values of @xmath100 and the total uncertainty are shown in figs .
[ fig : iaaratios_00_20 ] and [ fig : iaaratios_20_60 ] .
the average value of @xmath100 across partner momentum is constructed by weighting the individual measurements by the @xmath8+@xmath8 per - trigger yields @xcite . in general , the data are well fit by both the linear and cosine functions , giving reasonable @xmath104 .
no evidence is seen for systematic deviations from either fit within the sizable statistical uncertainties and both forms give similar goodness of fit values .
these values appear along with the @xmath100 ratios in table [ tab : iaa ] . for both central and midcentral collisions ,
the near - side jet yield is independent of trigger particle orientation with respect to the reaction plane within one standard deviation of the experimental uncertainties .
these measurements are consistent with surface bias of the hard scattering center created by the requirement of a trigger particle and a resulting short path length through the collision zone traversed by the near - side parton .
central collisions have insufficient statistics to determine the away - side variation .
however in midcentral collisions where the expectation of surface bias would lead to a large variation in the path length traversed by the away - side parton , the measurements show a significant falling trend with increasing trigger particle angle with respect to the reaction plane .
the suppression of away - side jet fragments in the out - of - plane direction is larger than in the in - plane direction , the out - of - plane away - side jet peak having only @xmath109 of the yield of the in - plane direction .
thus the large variation by angle with respect to the reaction plane is significant .
assuming the modulation to be flow - like ( dominated by the second - order variation ) , the suppression pattern implies @xmath110 .
as the midcentral away - side measurements are consistent between @xmath52 selections within the stated uncertainties , the hint of a rising trend in @xmath52 is not significant .
the values quoted here are consistent with those previously measured in @xcite and provide a factor four better constraint in the @xmath100 ratio .
recent single particle measurements of azimuthal anisotropy at high @xmath3 ( @xmath111 gev/@xmath1 ) found that @xmath112 @xcite .
thus , the away - side per - trigger yields at high @xmath3 favor an anisotropy larger than that measured for the single particles .
however , the difference is marginal and additional measurements will be needed to confirm . shown in fig .
[ fig : theory ] are the ( color online ) away - side @xmath100 ratio for midcentral , 2060% collisions , from fig .
[ fig : iaaratios_20_60 ] .
the solid line shows the results from an energy loss calculation @xcite using two hydrodynamic evolution models @xcite .
the shaded band shows the uncertainty that results from the selection of a particular hydrodynamic evolution ; the lower extent covering @xcite and the upper covering @xcite .
dotted lines show the uncertainty from the initial event geometry ( glauber or cgc ) as calculated within @xcite . ]
results of a monte - carlo energy loss calculation from t. renk @xcite using the time - space evolution provided by two different hydrodynamic simulations @xcite and two initial state descriptions , glauber and cgc .
these particular combinations of a jet energy loss model and collision evolution together predict less variation in the away - side suppression with respect to the reaction plane than is witnessed by the data .
variation of the initial geometry description within @xcite between glauber and cgc produces only small changes in the extracted @xmath98 out - of - plane to in - plane ratio , indicating limited sensitivity to this model parameter of the reaction plane dependent dijet observable .
however , other model parameters that vary between the two hydrodynamic models ( such as the thermalization time and freeze - out temperature ) were found to impact the away - side suppression anisotropy to a greater degree , indicating sensitivity to simulation parameters that are not well - constrained by other measurements .
consequently , these data warrant more detailed study with various energy loss models , and also different space - time evolution models .
we have shown that away - side jet fragment suppression increases substantially with increasing angle with respect to the reaction plane in midcentral au+au collisions at @xmath113 gev .
the away - side yield in the out - of - plane orientation is reduced by a factor of @xmath114 relative to the in - plane direction .
in contrast , the measured near - side @xmath98 is reaction plane independent , and consistent with no suppression .
these results directly show that the energy lost by fast partons in the hot nuclear medium increases as their paths through the medium become long .
a theoretical description of these experimental data implementing an energy loss formalism and a time - space evolution of the collision should be sought in union with other experimental quantities ; such as @xmath115 , @xmath98 , and @xmath115(@xmath9 ) @xcite .
energy loss formalisms that have successfully described the large momentum @xmath115 and @xmath98 may be paired with a particular time - space evolution in also describing the @xmath9 dependence of these same quantities . as shown for the combination above
, the data presented here disagree with the present calculations .
these data should play an important role in constraining simulations of the space - time evolution of heavy - ion collisions and the subsequent extraction of medium properties .
we thank the staff of the collider - accelerator and physics departments at brookhaven national laboratory and the staff of the other phenix participating institutions for their vital contributions .
we acknowledge support from the office of nuclear physics in the office of science of the department of energy , the national science foundation , a sponsored research grant from renaissance technologies llc , abilene christian university research council , research foundation of suny , and dean of the college of arts and sciences , vanderbilt university ( u.s.a ) , ministry of education , culture , sports , science , and technology and the japan society for the promotion of science ( japan ) , conselho nacional de desenvolvimento cientfico e tecnolgico and fundao de amparo pesquisa do estado de so paulo ( brazil ) , natural science foundation of china ( people s republic of china ) , ministry of education , youth and sports ( czech republic ) , centre national de la recherche scientifique , commissariat lnergie atomique , and institut national de physique nuclaire et de physique des particules ( france ) , ministry of industry , science and tekhnologies , bundesministerium fr bildung und forschung , deutscher akademischer austausch dienst , and alexander von humboldt stiftung ( germany ) , hungarian national science fund , otka ( hungary ) , department of atomic energy and department of science and technology ( india ) , israel science foundation ( israel ) , national research foundation and wcu program of the ministry education science and technology ( korea ) , ministry of education and science , russia academy of sciences , federal agency of atomic energy ( russia ) , vr and the wallenberg foundation ( sweden ) , the u.s . civilian research and development foundation for the independent states of the former soviet union , the us - hungarian fulbright foundation for educational exchange , and the us - israel binational science foundation .
the complete set of per - trigger yields and correlation functions used as source material for the analysis of the dependence of the away - side suppression on angle with respect to the reaction plane are shown in figs .
[ fig : jfsall_00_20][fig : cfsall_20_60 ] .
|
pair correlations between large transverse momentum neutral pion triggers ( @xmath0 = 47 gev/@xmath1 ) and charged hadron partners ( @xmath0 = 37 gev/@xmath1 ) in central ( 020% ) and midcentral ( 2060% ) au+au collisions are presented as a function of trigger orientation with respect to the reaction plane .
the particles are at larger momentum than where jet shape modifications have been observed , and the correlations are sensitive to the energy loss of partons traveling through hot dense matter .
an out - of - plane trigger particle produces only @xmath2 of the away - side pairs that are observed opposite of an in - plane trigger particle .
in contrast , near - side jet fragments are consistent with no suppression or dependence on trigger orientation with respect to the reaction plane .
these observations are qualitatively consistent with a picture of little near - side parton energy loss either due to surface bias or fluctuations and increased away - side parton energy loss due to a long path through the medium .
the away - side suppression as a function of reaction - plane angle is shown to be sensitive to both the energy loss mechanism in and the space - time evolution of heavy - ion collisions .
|
the reflectivity study is a very powerful scattering technique @xcite performed at grazing angle of incidence to study the structure of surface and interface of layered materials , thin films when the length scale of interest is in nm regime . utilizing the intrinsic magnetic dipole moment of neutron , the neutron reflectivity study provides magnetic structure in addition with the structural information . in reflectivity analysis ,
the electron density ( ed)/scattering length density ( sld ) ( in case of x - ray / neutron ) of different layers along the depth is estimated by model - fitting the experimental data .
text - based language like _ fortran _ is commonly used to write a simulation and data analysis code . however , recently a graphical language labview ( ( _ _ lab__oratory _ _ v__irtual _ _ i__nstrument _ _ e__ngineering _ _ w__orkbench , from national instruments corp . )
has emerged as a powerful programming tool for instrument control , data acquisition and analysis .
it offers an ingenious graphical interface and code flexibility thereby significantly reducing the programming time .
the user - friendly and interactive platform of labview is utilized , here , to simulate a versatile angle - resolved reflectivity at glancing angles .
there is only a recent report @xcite of labview - based reflectivity simulator for energy - resolved reflectivity study where the effects of absorption and interfacial roughnesses are not included .
the http://henke.lbl.gov/optical_constants/multi2.html[center for x - ray optics ( cxro ) ] provides a similar online simulator , however , the labview - based simulator has complete freedom to customize the programme according to user s choice .
let us start with brief review on theoretical formalisms of specular reflectivity with labview simulated curves .
scattering geometry for specular scan , polarized neutron reflectivity and reflectivity from liquid surface are discussed in the next section .
data inversion technique and the phase problem are discussed then considering the auto - correlation function ( acf ) .
finally , the labview - based reflectivity simulator is discussed in detail .
the basic quantity that is measured in a scattering experiment @xcite is the fraction of incident flux ( intensity of x - ray / number of neutrons ) that emerges in various directions is known as the differential cross - section .
normalizing this quantity by the incident intensity and density of scatterer , one obtains the scattering rate , @xmath0 which is termed as reflectivity ( merely , differed by some constant in different convention and dimensionality ) in nano - scale study of materials at grazing angle . for elastic scattering
, @xmath0 depends only on the momentum transfer , @xmath1 which can be varied by varying the energy ( energy - resolved reflectivity ) of the incident beam _ i.e. _ with a white beam at a fixed grazing incident angle or by varying the incidence angle ( angle - resolved reflectivity ) using monochromatic beam .
the angle - resolved reflectivity which is common in practise for its better resolution is discussed here . in following discussion
we consider , mainly , x - ray reflectivity however the formalism holds exactly the same way for neutron reflectivity considering sld instead of ed .
* the born formalism : * in quantum mechanical treatment , the incident beam , represented by a plane wave emerges out as spherical wave when interacted with the scattering potential . incident plane is related to the emerging spherical wave through the integral scattering equation . in
first born approximation of the integral scattering equation , the scattering amplitude depends on the fourier transforation ( ft ) of the scattering potential .
when the scatterer is composed of homogeneous layers parallel to the x - y plane , the scattering amplitude , in the first - order approximation , depends only on the ft of the gradient of ed along the z - axis , implying the expression for reflectivity @xcite as : @xmath2 where @xmath3 is the ed at depth z ( from top of the sample ) averaged over the x - y plane , @xmath4 is classical radius of electron , or the thomson scattering length .
the ed profile , @xmath3 and the born approximated reflectivity curves for bare si - substrate ( black curve ) , a 100 thick single layer of poly - sterene ( ps ) ( red curve ) and au ( blue curve ) ( @xmath5 ) on si - substrate . for small values of @xmath6 ,
the failure of the model is obvious as @xmath0 becomes more than unity ( grey curves ) .
the oscilations ( kiessig fringes ) with @xmath7 @xmath8 exactly corresponds to the layer - thickness , @xmath9100 .
interesting to note that , the overall level of the reflectivity curve is higher or lower depending upon whether the ed of a layer is greater or less than that of the substrate.,width=8 ] in the simplest possible situation where scattering from a bare substrate is considered , the ed is a step function : @xmath10 obviously , @xmath11 where @xmath12 is the dirac s delta function .
using the integral property of delta fn .
@xmath13 eqn .
( [ born ] ) implies , @xmath14 @xmath15 is known as the fresnel reflectivity which does not hold at small angle as eqn .
( [ fresnel ] ) violates the physical constraint that @xmath16 as @xmath17 .
this limitation is a consequence of neglecting the higher order terms of the scattering integral equation in first - order born approximation . using eqn .
( [ fresnel ] ) , one can rewrite eqn .
( [ born ] ) as @xmath18 when there is a uniform layer of thickness , @xmath19 on a substrate ( refer to fig .
[ born_refl ] ) , the ed are given by : @xmath20 whose derivative is a pair of scaled @xmath21-functions : @xmath22 the integral property of @xmath21-function leads to @xmath23 the reflectivity in this case , on simplification is obtained as @xmath24\ ] ] superposed on the fresnel reflectivity ( @xmath25 ) , it is a sinusoidal curve with a repeat distance of @xmath26 . in fig .
[ born_refl ] , the black curve is the generated fresnel curve following born approximation for si ( @xmath27 @xmath28 ) while the red and blue curves corresponding to a uniform 100 layer of poly - sterene ( ps ) and au ( @xmath5 ) on si , respectively .
the failure of born approximation is obvious for small values of @xmath6 as @xmath0 blows up beyond 1 ( grey region ) .
it is interesting to note that , the overall level of the reflectivity curve is higher or lower depending upon whether the ed of a layer is greater or less than that of the substrate
. moreover , the distinct oscillations ( kiessig fringes ) with @xmath7 @xmath8 exactly corresponds to the layer - thickness , @xmath9100 .
_ generation of born reflectivity , discrete ft and conversion of length - scales : _ the generation of born reflectivity curve is not straightforward following eqn . which in its analytic form demands @xmath29 to be spanned from @xmath30 to @xmath31 whereas in reality the accessible range of @xmath29 is finite .
hence , one needs to perform the discrete differentiation and ft for a given range of @xmath29 ( say , from @xmath32 to @xmath33 with uniform interval of @xmath34 ) to find @xmath35 over a desired range of @xmath6 ( say , from @xmath36 to @xmath37 with interval of @xmath38 ) .
it may be mentioned here that what matters in generating the reflectivity curve is the peaks of @xmath39 corresponding to each interfaces ( starting from the air - film interface to film - substrate interface ) and the relative separation of the peaks .
the last peak in @xmath39 corresponding to assumed finite extend ( say , 100 @xmath40 ) of the substrate does not matter significantly .
it is important to note that one can always perform a coordinate shift in the z - space and also may redefining the z - range ( minimal is the film thickness as @xmath41 varies in between only ) with suitable @xmath34 .
moreover , the selection of @xmath6 values is free to be selected as per interest when one performs the ft manually ( following eqn . ) and @xmath36 or @xmath37 may have no connection with @xmath32 and @xmath33 . for a given set of layer - thicknesses ( @xmath19 ) and corresponding eds ( @xmath41 ) , it is easy to interpolate the ed profile , @xmath42 .
practical surfaces and interfaces are always rough and the effect of roughness may be included in @xmath43 . when the height variation at any surface or interface is assumed to be gaussian
, the edp will be an error - function profile given by @xmath44 where @xmath45 .
the discrete differentiation is performed following the _ forward _ method , _
i.e. _ @xmath46 for j = 0 , 1 , . .
, n-1 ( considering @xmath47 ) or the _ backward _ method , _
@xmath48 for j = 0 , 1 , . .
, n-1 ( considering @xmath49 ) . and the discrete ft is done using the following method : @xmath50 where @xmath51 is a given value of @xmath6 and @xmath52 so , for a set of @xmath6 values , one obtains the reflectivity curve . however , when a standard discrete ft transformer tool or library function ( like a _ black - box _ ) which takes n number ( a convenient choice is a power - of - two _ i.e. _ @xmath53 where @xmath54 is a positive integer which reduces the computation time if _ fast ft ( fft ) _ algorithm is used ) of @xmath55 values as input is used , the usual output is spanned over @xmath56 with @xmath57 and one may need to shift the positive half and the negative half . for standard ft tool with the exponential in the form of @xmath58
the following relation is useful for conversion of the length - scales @xmath59 ( if the exponential has the form of @xmath60 , then @xmath61 ) . * the scattering geometry : * to be more precise , eqn .
( [ born ] ) is the expression for specular reflectivity where the measurement is done in a @xmath62 geometry _
i.e. _ w. r. to the beam , sample is rotated by an angle @xmath63 following a rotation of detector by @xmath64 .
one may define the axes by coinciding a particular axis with the beam direction , in particular when the beam - direction is fixed in space ( for synchrotron beam - lines ) , however , as we are interested in structure of the sample it is convenient to define the axes fixed with the sample . the scattering geometry _
w. r. to _ the reference axes fixed with the sample . a beam ( x - ray / neutron ) incidents at an angle @xmath65 and gets scattered ( reflected / diffracted ) towards ( @xmath66 ) .
@xmath64 represents the deviation of the beam from its incident direction.,width=8 ] fig .
[ scatgeo ] shows a general scattering geometry when reference frame is chosen fixed with the sample . in this geometry , the incident wave vector is given by , + @xmath67 and the scattered wave vector for an angle @xmath68 out - of the plane of reflection is given by , @xmath69 where @xmath70 are the incident and scattered angle w. r. to the sample . for elastic scattering , @xmath71
the transfer wave vector , + @xmath72 using the relation @xmath73 one can obtain the following form for magnitude of the momentum transfer , @xmath74}\\ \nonumber & = & \frac{2\pi}{\lambda}\sqrt{2(1-cos2\theta)}\\ & = & \frac{4\pi}{\lambda}sin\theta \label{n}\end{aligned}\ ] ] here , one may note that the last expression is equivalent to the bragg s condition for @xmath75 where @xmath64 is the deviation of the scattered beam from its initial direction .
one important point is that the expression for @xmath6 , derived from simple vector geometry becomes equivalent , apparently , with the bragg s condition which is essentially involved with interference !
actually , when one considers @xmath76 , the essence of interference effect automatically comes in as the @xmath77 is connected to the @xmath78 by fourier transformation and vice - versa .
another point , obvious from the bragg s condition , @xmath79 , is that when the small length - scale ( atomic spacing of few ) is probed , the bragg peaks should appear at larger angles ( @xmath80 few tens of degrees ) whereas for larger length - scales ( few tens or hundreds of nm ) the peaks should appear at smaller angles ( @xmath80 few degrees only ) the former one is the case of diffraction and the later for reflectivity . sometimes , it is convenient to express @xmath1 in terms of the parallel ( to sample surface ) component , @xmath81 and the perpendicular , @xmath82 . the specular condition , @xmath83 with @xmath84 , implies @xmath85 .
so specular reflectivity can be done either by varying @xmath86 for fixed @xmath63 ( energy resolved ) or by varying @xmath63 for fixed @xmath86 ( angle resolved ) .
+ _ footprint correction : _ one important correction needs to be included , particularly for specular reflectivity at small angles , when the footprint of the beam exceeds the sample size , l. if the beam width is w , then the footprint , f for an incidence angle of @xmath65 is given by f = w / sin@xmath65 , hence , for f@xmath87l the experimental data need to be corrected by multiplying a factor of f / l .
+ _ instrumental resolution : _ another crucial point that one has to consider is the instrument resolution function .
the effect of instrumental resolution can be considered as convolution by the resolution function which in most cases is approximated by a gaussian whose standard deviation is usually determined by the full - width at half - maxima ( fwhm ) of the direct beam in a detector scan .
we consider a gaussian resolution function with suitably defined window [ @xmath88 ( 2 @xmath89 n@xmath90 + 1 ) data points ] for post - processing ( weighted sum ) of the simulated data to include the instrumental resolution effect . _ specular reflectivity from liquid surface / interface _ : schematic of the experimental set up and the beam geometry for reflectivity study from liquid surface .
@xmath91 is the direct beam while oax defines the horizontal plane .
the rotation of the deflector crystal ( black line at o ) about the direct beam defines a cone of semi - opening angle @xmath92 at o. the rotation of deflector crystal , @xmath93 is related to the angle of incidence , @xmath94 through the relation @xmath95 . to bring the point of interest back to its original position
a vertically shift of @xmath96 followed by a horizontally rotation by an angle of @xmath97 $ ] is required.,width=8 ] while discussing about the scattering geometry and experimental set up for reflectivity study of solid samples , we may have a glimpse on the reflectivity study from liquid surface or liquid - liquid interface @xcite which is special because the liquid surface necessarily be horizontal .
moreover , the reflected intensity from liquid surface / interface drops drastically with increase in angle hence a synchrotron beam is preferred .
unfortunately , synchrotron beams are usually fixed in direction and need to incline by bragg - reflection using suitable single - crystal .
usually the deflector crystal is mounted at o , the center of the goniometer stage where the solid samples are mounted and the liquids are placed on an additional stage which can rotate about the vertical axis through o. the rotation of the deflector crystal about the direct beam ( @xmath98 , refer to fig .
[ liqrefl ] ) obeying the bragg condition ( for example , the bragg angle of ge(@xmath99 ) at 18 kev @xmath100 ) causes the bragg - reflected beam to describe a cone of opening angle @xmath101 ( @xmath102 ) . as a result
, the change in angle of incidence causes the point of incidence to shift hence for each angle the point of incidence ( the liquid stage as a whole ) has to be brought back to its original position .
now , the locus of the beam on a virtual cylinder ( @xmath103 : defined by the rotation of the liquid stage about the vertical axis through o ) of radius d is given by @xmath104 + 2y^2+z^2=d^2 $ ] .
if a rotation of the deflector crystal by an amount @xmath93 brings the horizontal beam @xmath105 to @xmath106 which makes an angle @xmath94 with horizontal ( x - y ) plane then considering the vertical shift on a virtual vertical plane ( not on the cylindrical surface ) , one can write , @xmath107 which relates the angle of incidence , @xmath94 to the rotation of the crystal through the relation @xmath95 .
so , for a rotation of the deflector crystal by an amount of @xmath93 , the point of interest should be vertically shifted by an amount of @xmath96 ( for a line joining the origin and a point @xmath108 that makes an angle @xmath94 with x - y plane , it is obvious that @xmath109 hence @xmath110 ) followed by a horizontal rotation of @xmath97 $ ] . to avoid the shifting of the incident spot while changing the incident angle two - crystal assembly
may also be used where the fixed beam passes through one focus of an ellipsoid of revolution and the point of incidence on the liquid surface fixed at another focus while the driving crystals move in a coupled fashion over two circular cross - sections of the ellipsoid of revolution . however , using two crystals involves more loss of beam intensity and more complicity in alignment . * the dynamical theory and parratt formalism : * schematic of reflection and refraction of a wave incident on a smooth surface at an angle @xmath65 .
reflected wave emerges at an angle , @xmath111 and the refracted wave transmitted through the medium at an angle @xmath112 .
@xmath113 , @xmath114 , @xmath115 are the incident , reflected and transmitted wave vectors , respectively.,width=8 ] schematic of multiple reflection from a multilayer film having @xmath116-layers on a substrate with @xmath117 interfaces . for incident wave amplitude
is normalized to unity , @xmath118 . no reflection from the substrate sets , @xmath119.,width=8 ] in classical treatment of scattering ,
the continuity of the electric and magnetic field vectors of the propagating electro - magnetic wave at an interface provides the relation between the reflection ( @xmath120 ) and transmission ( @xmath121 ) coefficients and hence imply the reflectivity , @xmath122 . for a smooth air - medium interface ( refer to fig .
[ fresnel1 ] ) with an incidence angle of @xmath123 ( specular ) , the expression for @xmath120 and @xmath121 , known as fresnel formula is given by @xmath124 where @xmath125 and ( from snell - descartes law ) @xmath126 the refracted wave transmits at an angle @xmath112 through the medium having refractive index , @xmath127 where ed , @xmath41 is redefined as @xmath128 and @xmath129 is the absorption co - efficient .
it is more convenient to use @xmath41 ( electron per @xmath130 ) rather than the dimensionless number @xmath21 because @xmath41 is a @xmath86-independent specific number for a material .
the dynamical calculation , for a thin film consisting of a single layer on a substrate , using matrix transfer method @xcite for each layer which essentially connects the fields between two consecutive layers , yields the following relation @xmath131 where @xmath132 is the coefficient of reflection at the interface between @xmath133-th and @xmath134-th layer .
it may be noted here that the additive term to unity in the denominator corresponds to multiple reflection in the film .
extension to a multi - layer film having @xmath116 homogeneous layers on a substrate , one obtains the recursion relation for the ratio of reflection and transmission coefficients of the j - th layer , as follows @xmath135 where @xmath136 and @xmath137 depth including the j - th layer thickness from top . in our convention
( refer to fig .
[ multirefl ] ) , @xmath138 for air / vacuum and @xmath139 for the substrate .
the expression for @xmath132 is given by @xmath140 with @xmath141 no reflection from the substrate ( assumed to be sufficiently thick ) sets @xmath142 as the start of the recursion .
after @xmath117 iterations , the expression for specular reflectivity is obtained as @xmath143 + _ critical angle : _ it may be noted here that , x - ray , propagating through a medium of higher refractive index , suffers total external reflections when incident on a surface of a material having lower refractive index .
the very concept of total external angle defines an angle below which the x - rays are totally reflected back and termed as critical angle . setting @xmath144 in eqn .
, one obtains @xmath145 _ i.e. _ the critical angle , @xmath146 is related to the ed of the material as @xmath147 typical orders of magnitude are : @xmath148 and @xmath149 so that @xmath150 to @xmath151 . for si , @xmath152 and corresponding @xmath153 with @xmath154 @xmath8 for @xmath155 . for a multilayer film , the overall critical angle
is determined by the layer having the highest value of @xmath21 . in fig .
[ born_refl ] , the critical angle for blue curve ( au on si ) is defined by @xmath156 as @xmath157 whereas for red curve ( ps on si ) the critical angle is defined by @xmath158 as @xmath159 . for incident angle
less than the critical angle _
i.e. _ @xmath160 , the penetration depth is only few nano - meters , however , it increases sharply to several micro - meters as @xmath65 exceeds @xmath146 and x - ray immediately sees the whole layer . actually , there is a so - called _ evanescent wave _ within the refracting medium , propagating parallel to the interface and exponentially decaying perpendicular to it . in grazing incidence diffraction ( gid )
study where only the first few atomic layers are of main interest @xmath65 is kept just below the @xmath146 _ i.e. _ ( @xmath161 ) .
+ the transfer matrix formalism is equivalent to the recursive approach of parratt s @xcite formalism and both ( known as dynamical theory ) are fairly exact since it incorporates the issue of multiple scattering .
the additive term to unity in the denominator of eqn .
, corresponding to multiplicative reflection @xcite , contributes significantly for small incident angles and when ignored the reflectivity expression can be simplified to , @xmath162 where @xmath163 is the thickness of @xmath133-th layer . in further assumption that the incident angle does not change significantly from layer to layer i.e.
@xmath6 is same for all layers , with an additional approximation ( so that first two terms of the binomial expansion of the right hand side of the eqn
. would be sufficient to consider ) , @xmath164 one can write eqn . in a simpler form : @xmath165 where @xmath166 depth including the j - th layer thickness from top .
( [ born2 ] ) is exactly the same with eqn .
( [ born ] ) in the continuous limit . this greatly simplified treatment , known as
the kinematical theory or born approximation reduces the parratt recursive formula to a simple relation . in born approximation ,
the effect of multiple reflection from different layers of a multi - layer film is ignored and also the effect of refraction from one layer to another is not considered by assuming the incident angles to be same for all interfaces .
hence , for a bare substrate where there is no multiple reflection and only one angle of incidence , both the parratt curve and the born curve are expected to be identical and same with the fresnel curve .
however , the third approximation ( [ 3rdbornapprox ] ) makes the difference and fails to restrict the born curve from blowing up at small angles .
formulation of reflectivity expression for scattering of electromagnetic wave from rough surfaces is difficult as the solution of the relevant wave equations turns out to be complicated involving matching of the boundary conditions over random rough surfaces and several simplifying assumptions have to be invoked for their solution .
an exponential damping factor may be included when the roughness ( @xmath167 ) has an error - function profile , as introduced by nvot and croce @xcite to modify eqn .
( [ reflectance ] ) , eqn .
( [ parratt1 ] ) and eqn .
( [ born2 ] ) in the following form : @xmath168 @xmath169 @xmath170 it may be noted here that the roughness term as introduce here carries more weightage for larger @xmath6 hence damps the reflectivity curve slightly more in comparison with roughness introduced following eqn . .
reflectivity curves for a 100 thick single layer of ps on si substrate following different formalism .
electron density , convoluted with the roughness is in the inset .
the region highlighted by the blue circle shows distinct difference between the curves .
the back one is for exact dynamical theory .
the effect of multiple reflection is ignored for green curve .
further approximation ( born ) that the incident angle to be same at all interfaces and @xmath171 is assumed for the red curve.,width=7 ] fig .
[ compare ] compares the reflectivity curves for a 100 thick single layer of ps on si substrate following eqn .
( [ transmatrix ] ) using ( [ reflectance2 ] ) , ( [ parratt2 ] ) and ( [ born3 ] ) .
the inset figure shows the electron density profile with the gaussian roughness ( refer to eqn . ) of the interfaces .
the region highlighted by the blue circle , particularly around @xmath172 @xmath8 , shows distinct difference between the curves .
the back one is for exact dynamical theory .
the effect of multiple reflection causes the reflectivity to grow more than 1 as observed for the green curve . for
the born approximated ( blue ) curve , the deviation is even serious as it assumes further that the incident angle to be the same at all boundaries . +
* polarized neutron reflectivity ( pnr ) : * in addition to the structural information , reflectivity study using the spin polarized neutron ( pnr ) provides the detail of in - plane magnetic ordering in a magnetic thin film .
neutrons arriving at the sample surface are spin polarized either parallel @xmath173 or antiparallel @xmath174 to the quantization axis defined by the applied magnetic field , h ( along the y - direction , refer to fig .
[ scatgeo ] ) . in this case
, the refractive index includes the nuclear and magnetic scattering contributions and expressed as @xmath175 where @xmath176 is the complex nuclear sld and @xmath177 and @xmath178 is the magnetization .
eqn . is same , except the additional magnetic term , with eqn . as @xmath179 .
depending upon the polarization of incident and reflected neutron beam , the measured reflectivities are designated as @xmath180 , @xmath181 , @xmath182 and @xmath183 .
the non - spin - flip ( nsf ) data , @xmath180 and @xmath181 depend on structural as well as magnetic contribution and provide the magnetization parallel to the applied field [ ( @xmath184 , however , the spin - flip ( sf ) intensities , @xmath182 and @xmath183 reflectivities ( usually , these two are degenerate ) are purely of magnetic origin and depend on the average square of the transverse in - plane magnetic moment , @xmath185 @xcite
. one important feature of pnr is that the value obtained for @xmath186 can be calibrated in @xmath187 units because the number of scatterer can be estimated from simultaneous analysis of r@xmath188 and r@xmath189 using eq . .
the pnr data , collected in this geometry , are insensitive to the out - of - plane moment @xmath190 . in pnr study of antiferromagnetic thin - films ,
one obtains additional half - order peaks corresponding to double the lattice spacing in real space of similar spins .
+ _ comparison between x - ray and neutron refractivities : _ the nature of interaction for x - ray and neutron with the scatterer is different as x - ray sees the electron cloud within the sample ( long - range electro - magnetic interaction ) but neutron sees the nucleuses as point scatterer ( short - range strong interaction ) in addition with the magnetic ( long - range electro - magnetic ) structure of the scatterer ( in case of polarized neutron reflectivity utilizing the intrinsic dipole moment of neutrons ) .
even with laboratory - sources of x - ray , the reflectivity as low as @xmath191 and @xmath6 value as high as @xmath1921 @xmath8 can be achieved , however for neutron reflectivity , it is difficult to have @xmath0 values below @xmath193 or @xmath6 beyond 0.2 @xmath8 .
unlike monotonic dependence of electron density over atomic number in case of x - ray , the neutron reflectivity provides reacher structure since the scattering length density varies drastically from element to element even for different isotopes of the same atom and can be negative as well .
* auto - correlation function and the phase problem : * + here , we discuss the difficulty of finding unique ed profile from reflectivity data . as the reflectivity expression
is obtained as modulus - squared in the final step , the information of the phase of the scattered wave is lost in the reflectivity experiments as a result the direct inversion of the data to reconstruct the edp is impossible .
however , supplemented with additional information like prior knowledge of number of layers _ etc . _ from sample preparation makes it possible to extract the edp . + now let us discuss the lack of uniqueness of edp mathematically by considering the auto - correlation function or acf of @xmath194 , which provides a model - independent real - space representation of the information contained in the intensity of reflectivity pattern , defined as , @xmath195 where @xmath196 is the ft of @xmath194 . considering @xmath197 and comparing eqn .
( [ acf ] ) with eqn .
( [ born1 ] ) , one can write , the acf of @xmath198 as follows : @xmath199 = \mathcal{f}^{-1 } \left[\mathfrak{r}(q_z ) \right ] \label{patterson}\ ] ] where @xmath200 is the inverse ft and @xmath201 .
the above expression is known as the patterson function and imply the information regarding the depth of interfaces independent of any model .
however , there is an inherent discrepancy of this data inversion technique due to the so - called phase - loss ambiguity . as a result
some peaks may get introduced additionally or get overlooked in the acf of @xmath198 as pair - wise depth difference does matter in its calculation .
a pair of sharp peaks , say , at @xmath202 and @xmath203 with amplitudes @xmath204 and @xmath205 , respectively , contributes to a symmetric pair of very sharp peaks at @xmath206 with amplitude @xmath207 to @xmath208 along with an additive of @xmath209 to the acf at the origin . for @xmath116 interfaces _
i.e. _ @xmath116 number of peak in @xmath198 will pair - wise contribute to @xmath210 number of peaks towards its acf in general .
the peak for air - sample interface at @xmath211 in @xmath198 , when considered pair - wise with other interfaces , will contribute to @xmath212 peaks at their respective positions .
but , the additional peaks with the possibility of overlap and/or exact cancelation may complicate the analysis unless one has prior information of layers from sample growth .
+ as for example , let us consider two x - ray reflectivity curves looking exactly the same except a small difference near the first deep , marked by the red circle in fig .
the black curve corresponds to a bi - layer film ( @xmath213 and @xmath214 ) having same thickness of 50 each on a substrate ( @xmath215 ) . for green curve
, there is only a single layer ( @xmath216 ) of thickness 100 on the same substrate . at @xmath217 , exact cancelation of the contributions from pairs of @xmath21-functions at ( @xmath211 ; @xmath218 ) and ( @xmath218 ; @xmath219 ) in case of the bi - layer film results in the same acf as for single layer , and hence both reflectivity curves look exactly the same ( refer to fig . [ acf1]b ) ! interestingly , such situation may happen when @xmath220 where @xmath221 . to generate the acf ,
one may directly use the discrete inverse ft formula given by , @xmath222 where @xmath223 which needs the @xmath224 to be defined from @xmath225 with uniform @xmath38 and the obtained acf will be automatically scaled having uniform interval of unity ( @xmath226 ) .
however , the truncation effect due to finite span or window of @xmath6 which introduces ripples in the acf with a characteristic wavelength of @xmath227 will be unavoidable
. the peaks in acf will be distinct and better when @xmath37 as large as @xmath228 and larger the number of data points .
when a standard inverse ft tool is used , it is convenient to interpolate the reflectivity data from @xmath225 to @xmath229 with uniform @xmath57 so that acf ( exchanging the positive half and the negative half ) will be automatically scaled with unit interval between @xmath230 .
* data inversion technique : * a direct data inversion is cumbersome due to many practical difficulties like limited extent of measured value of @xmath6 , statistical noise , incoherent background signal and the blurring from instrumental resolution . + having preliminary idea about the system one can simulate the reflectivity curve close enough to the experimental one by proper choice of parameters . for a close choice of model edp @xmath231 with corresponding simulated reflectivity , @xmath232 one may start iteration@xcite using the following ansatz @xmath233\right]\label{fit } \nonumber\ ] ] where @xmath234 is the experimental reflectivity and @xmath235 is used as the @xmath236 in the next iteration .
fig . [ refl_labview2 ] shows the block diagram of the simulator .
the code is build by drag - n - drop the icons and connecting them by wires through which data flow following the desired logic .
one can easily modify and customize any of such programmes according to their need and preferences .
[ refl_labview1 ] shows the screen - shot of the front panel of the simulator which simultaneously plots specular reflectivity curves for a multi - layer film following the ( i ) exact dynamical theory ( eqn .
( [ transmatrix ] ) with eqn .
( [ reflectance2 ] ) ) and the ( ii ) born theory ( eqn . with restriction @xmath237 ) according to the range and steps of the imported experimental data ( normalised ) or as provided from the front panel corresponding to the model parameters ( thickness , roughness , ed , absorption coefficient ) fed from front panel or imported from file providing the path .
the simulator needs the followings as the input : wavelength ( @xmath86 ) in or energy in kev , range and step of @xmath63 ( or according to the experimental data when imported , providing the path ) , layer detail _ i.e. _ @xmath238 ( with option to be imported , providing the path ) , fwhm of the direct beam , number of data point ( n@xmath90 ) to define the instrumental resolution window and the output path .
it is easy to include repeated layers by just putting @xmath239 for @xmath238 to a particular layer where the repeated layers are intended to be inserted .
the repeated layer detail with number of repeat can be incorporated in another window of the front panel . to delete input data , one has to right click on that particular input @xmath87 data operation @xmath87 delete element@xmath240row@xmath240column .
inclusion of background and footprint effect ( providing the sample size and beam width ) are optional . @xmath0 ( or @xmath241 as selected ) _ vs. _
@xmath6 and @xmath0 ( or @xmath241 as selected ) _ vs.
_ @xmath63 along with the electron density profile ( edp ) including the roughness modification are simultaneously plotted during run as shown in fig .
[ refl_labview1 ] .
the total thickness of the film and the number of layers in between air and substrate is displayed in the front panel during run .
the simulator generates five output files , namely , the parameter file ( _ _ filename__par.txt ) , box edp ( _ filename__boxedp.txt ) and the convoluted edp ( _ filename__convedp.txt ) , the simulation data file with ( _ filename__generated.txt ) and without ( _ filename_.txt ) inclusion of the instrumental resolution effect . while simulating the reflectivity curve for an experiential data
, one can reduce the simulation time using the option of reducing the number of data points by a given factor . a typical reflectivity data ( refer to fig .
[ erst ] ) of a multi - layer of erbium stearate langmuir - blodgett film on si substrate obtained from a lab - source ( rotating copper anode , enraf nonius , model fr591 ) using cu@xmath94 characteristic x - ray is considered here to estimate the parameters by simulating the reflectivity curve . in the out - of - plane direction ,
the film consists of 25 layers of er separated by organic spacer ( stearate tail ) .
the thickness of different layers from the top ( _ i.e. _ from air to substrate ) , used to simulate the curve is as follows : stearate tail ( 22.5 ) + 24 @xmath89 [ er@xmath242-head ( 4.4 ) + stearate tail ( 45 ) ] + er@xmath242-head ( 4.4 ] + sio@xmath243 ( 22 ) .
the total film - thickness is 1234.5 and total number of model - layers is 51 .
corresponding electron density profile with and without the roughness effect is shown in fig .
[ edp ] .
once having a close guess of the parameters , experimental data can be fitted utilizing labview platform with constrained non - linear least - square fit option using the levenberg - marquardt algorithm or the trust - region dog - leg algorithm to optimize the set of parameters for the best fit .
labview - based fitting part is in progress .
a model - independent acf or the patterson function generation programme from the specular reflectivity data which provides an idea about the thickness of layers and the depth of interfaces is also developed .
some programmes are also developed those are useful in furnishing the spec - files .
it may be mentioned here that the stand - alone executable version of the simulator is also built and it needs only the labview run - time environment ( free to download from national instruments ) .
i would like to acknowledge my supervisor prof .
milan k. sanyal for teaching me the fundamentals of reflectivity technique , detail of analysis and providing me the opportunity to carry out scattering measurements .
_ * download link : * _ one needs to run the main programme _
specular reflectivity simulator.vi_ only , however , it needs two sub - programmes namely , _
boxedp_subvi.vi _ and _ convedp_subvi.vi _ ( to be placed in the same folder ) during run .
https://docs.google.com/file/d/0b54_hau9fqrsq190s1h1tdrtsgm/edit?usp=sharing[click here to download the zipped folder ] containing these three programme and other related programme files .
one may use the stand - alone version _
reflgen.exe _ as well . without having full labview software , one needs to install only the labview run - time engine ( free to download from national instruments ) .
the limitation of this version is that the source code _
i.e. _ the block diagram is not available hence can not be edited .
after downloading one should run it to open the actual labview simulator . from _ file
_ @xmath87 _ vi properties _ , one can find the location of the programme to copy and paste it at the desired folder .
the programme named _
bornreflgen_acf _ can be used to generate the born reflectivity and corresponding acf .
to merge reflectivity data , furnish spec - files and for conversion of electron density , the other programmes may be helpful .
99 l. g. parratt , phys . rev . * 95 * , 359 ( 1954 ) .
t. p. russell , mat .
* 5 * , 171 ( 1990 ) .
j. daillant and a. gibaud , _ x - ray and neutron reflectivity : principle and applications , lectures notes in physics _
( springer , 1999 ) .
m. tolan , _ x - ray scattering from soft - matter thin films , 5 - 89 _ ( springer , 1999 )
. j. als - nielsen and d. mcmorrow , _ elements of modern x - ray physics , 61 - 95 _ ( john wiley & sons , 2001 ) j. k. basu and m. k. sanyal , phys .
rep . * 363 * , 1 ( 2002 ) . v. a. kheraj , c. j. panchal ,
m. s. desai and v. potbhare , pramana : j. phys . * 72 * , 1011 ( 2009 ) .
d. s. sivia , _ elementary scattering theory : for x - ray and neutron users , 93 - 112 _ ( oxford , 2011 )
. m. l. schlossman _ et .
al _ , rev .
instrum . *
68 * , 4372 ( 1997 ) .
m. k. sanyal , s. hazra , j. k. basu , and a. datta , phys .
b * 58 * , r4258 ( 1998 ) . h. zabel , k. theis - brhl , and b.p .
toperverg , _ the handbook of magnetism and advanced magnetic materials : vol .
3 - novel techniques , h. kronmller , s.p.s .
parkin ( eds . ) , 2327 - 2362 _ , ( wiley , new york , 2007 ) . s. gayen , m. k. sanyal , a. sarma , m. wolff , k. zhernenkov , and h. zabel , phys . rev .
b * 82 * , 174429 ( 2010 ) .
|
a versatile x - ray / neutron reflectivity ( specular ) simulator using labview ( national instruments corp . ) for structural study of a multi - layer thin film having any combination , including the repetitions , of nano - scale layers of different materials is presented here ( available to download from the link provided at the end ) .
inclusion of absorption of individual layers , inter - layer roughnesses , background counts , beam width , instrumental resolution and footprint effect due to finite size of the sample makes the simulated reflectivity close to practical one .
the effect of multiple reflection is compared with simulated curves following the exact dynamical theory and approximated kinematical theory .
the applicability of further approximation ( born theory ) that the incident angle does not change significantly from one layer to another due to refraction is also considered .
brief discussion about reflection from liquid surface / interface and reflectivity study using polarized neutron are also included as a part of the review .
auto - correlation function in connection with the data inversion technique is discussed with possible artifacts for phase - loss problem .
an experimental specular reflectivity data of multi - layer erbium stearate langmuir - blodgett ( lb ) film is considered to estimate the parameters by simulating the reflectivity curve .
|
on - chip solid - state refrigeration has long been sought for various applications in the sub - kelvin temperature regime , such as cooling astronomical detectors @xcite . in a normal metal - insulator - superconductor ( nis )
junction @xcite , the superconductor density of states gap makes that only high - energy electrons are allowed to tunnel out of the normal metal or , depending on the bias , low - energy ones to tunnel in , so that the electronic bath as a whole is cooled . in sinis devices based on aluminum ,
the electronic temperature can drop from 300 mk down to below 100 mk at the optimum bias point .
while this level of performance has been demonstrated in micron - scale devices @xcite with a cooling power in the picowatt range , a difficulty arises in devices with large - area junctions needed for a sizable cooling power approaching the nanowatt range . for instance , a high - power refrigerator has been shown to cool an external object from 290 mk down to about 250 mk @xcite .
one of the main limitation to nis coolers full performance is the presence in the superconducting leads of non - equilibrium quasi - particles arising from the high current running through the device .
the low quasi - particle relaxation rate and thermal conductivity in a superconductor bound these hot particles in the vicinity of the junction and lead to severe overheating in the superconducting electrodes .
there are several methods for reducing the accumulation of quasi - particles in a superconductor .
for example a small magnetic field @xcite can be used to introduce vortices that trap quasi - particles .
this approach is however not applicable to electronic coolers with large - area junctions since a vortex also reduces the cooling performance if it resides within a junction .
the most common method is to use a normal metal coupled to the superconductor as a quasi - particle trap : quasi - particles migrate to the normal metal and relax their energy there through electron - electron and electron - phonon interaction . in the typical case of a fabrication process based on angle evaporation ,
quasi - particle traps are formed by the structures mirroring each superconducting electrode , sitting on a side of the cooling junction and featuring the same oxide barrier layer .
the trapping efficiency is usually moderate , but can be improved in two ways : the normal metal can be put in direct contact with the superconductor , as out - of - equilibrium quasi - particles would diffuse more efficiently to the trap @xcite , or the trap can be closer to the junction @xcite . in both cases , it is important to prevent inverse proximity effect in the superconductor , which smears locally the superconductor density of states and degrades cooling efficiency .
the existence of an optimum transparency for the interface between the trap and the superconducting lead is therefore expected , but remains to be investigated @xcite . in this paper , we present an effective method to evacuate quasi - particles in a sinis cooler , based on what we call a quasi - particle drain . it is a kind of quasi - particle trap made of a layer of normal metal located at a fraction of superconducting coherence length away from the junction and separated by a thin insulating layer to stop the inverse proximity effect
we compare the cooling performance when varying the quasi - particle drain barrier transparency over a wide range .
the efficiency of the quasi - particle drain is demonstrated through the electronic cooling from 300 to 130 mk at a 400 pw cooling power .
a simple thermal model captures the effect of the quasi - particle drain reasonably well .
4 @xmath0m@xmath1 .
cu appears as orange and the al / almn bilayer as yellow .
the cooler junctions are highlighted by blue dashed lines , and the thermometer junctions are highlighted by red dashed lines .
the circuit connected to the top electrodes probes electron temperature of the normal island cooled by the two current - biased large junctions .
( c ) current - voltage characteristics at a 50 mk temperature of samples a , b , d , e and f with different tunnel barrier thicknesses between the quasi - particle drain and the superconducting leads , see table 1 . ]
we use the fabrication process described in @xcite , in which a sinis cooler is obtained by photo - lithography and chemical etch of a nis multilayer . here , we add a normal metal layer at the bottom , which is used as a quasi - particle drain .
figure 1a false - colored scanning electron micrograph shows a side view of a typical sinis cooler , obtained by cutting it with a focused ion beam . from top to bottom , the 100 nm - thick cu normal metal layer to be cooled is suspended between two 200 nm - thick al superconducting electrodes .
the latter rest on two separate quasi - particle drains made of a 200 nm layer of almn deposited on a si wafer .
we choose almn @xcite as a quasi - particle drain normal material because it acts chemically as al in terms of oxidation and etch .
the layers are separated by two aluminum oxide barriers , which we name the drain barrier between the almn and al layers and the cooler barrier between the al and cu layers .
sample parameters are given in table 1 .
@lllllll sample & drain barrier & cooler barrier & @xmath2 & @xmath3 & color + & ( mbar , second ) & ( mbar , second ) & @xmath4 & @xmath5 & & + a & 1.3 , 10 & 1.3 , 300 & 0.71 & 398 & green + b & 0.26 , 10 & 1.3 , 300 & 1.56 & 382 & blue + c & 0.18 , 1 & 0.8 , 180 & 0.55 & 370 & purple + d & 5@xmath610@xmath7 , 10 & 1.3 , 300 & 1.01 & 228 & gray + e & 0 & 1.3 , 180 & 1.31 & 180 & gray + f & n / a & 1 , 300 & 0.83 & 390 & red + figure 1b is an optical micrograph showing a top view of the cooler . the two nis junctions area are outlined by dashed blue lines .
they are separated by a trench in the al and almn layers , created by chemical over - etch , underneath the array of holes in the suspended cu layer .
each junction has an area of 70 @xmath6 4 @xmath0m@xmath1 and is surrounded by two quasi - particle traps : a side trap made of cu next to it , and a quasi - particle drain made of almn .
two additional small nis junctions connected to the normal metal are used as a sinis thermometer .
electron temperature is accessed by comparing the measured voltage @xmath8 drop under a small bias current ( typically 10 na ) to a calibration against the cryostat temperature .
the current flowing through a nis junction with a voltage v writes @xmath9},\ ] ] where @xmath10 $ ] is the normalized superconductor density of states , @xmath11 is the superconducting gap , @xmath12 is the normal state junction resistance , and @xmath13 are the fermi - dirac energy distributions of electrons in s and n , respectively . if leakage is negligible , a low sub - gap current then means a low electronic temperature .
figure 1c shows the current voltage characteristics ( iv curves ) of different samples measured with a standard current - biased 4-probe technique in a dilution cryostat at 50 mk , with a focus on the low - bias regime .
the two innermost curves stand for samples e and d , which have no drain barrier and a very thin drain barrier , respectively .
superconductivity in the al layers is then affected by a strong inverse proximity effect , which results in a depressed critical temperature and a low superconducting gap @xmath11 so that @xmath3 = 180 @xmath0ev and 228 @xmath0ev respectively . as samples a , b , and c are fabricated using a higher oxidation pressure for the drain barrier , they have a typical value for @xmath3 of about 350 @xmath0ev and a ratio of minimum conductance to normal - state conductance of about @xmath14 , indicating that inverse proximity effect is weak .
these samples show a sharp iv characteristic , with sample c ( not shown ) behaving almost identically to sample b. the sole difference between samples a and b is the drain barrier . having a thinner barrier
, sample b exhibits less current at a given sub - gap bias , less overheating from quasi - particles in the superconductor .
the drain barrier lets quasi - particles get efficiently trapped in the drain , while it stops the inverse proximity effect . in a normal metal ,
the main heat flux from the electron system to the environment is through the coupling with the phonon system .
we estimate the cooling power from the electron - phonon coupling power using @xmath15 where @xmath16 wk@xmath17m@xmath18 is the electron - phonon coupling constant in cu @xcite , @xmath19 @xmath0m@xmath20 is the island volume for all samples , @xmath21 is the measured electron temperature and @xmath22 is the phonon temperature , both in the normal island .
we assume here that @xmath23 , which leads to an upper limit on the estimation of the cooling power .
figure 2a displays the calculated electron - phonon coupling power @xmath24 ( solid lines ) and the total joule power ( dashed lines ) @xmath25 measured at the optimum bias , the bias at which electronic cooling is maximum .
the typical power scale for all our present coolers is in the nanowatt range .
figure 2 b presents the most important result of our work , namely the behavior of the normal island electronic temperature at optimum bias as a function of the bath temperature .
all coolers have the largest temperature drop at a bath temperature around 300 mk . at lower temperatures ,
heating above the bath temperature is observed , which arises from the sub - gap current . having only the side trap and no quasi - particle drain
, sample f shows a poor cooling from 300 mk down to only 230 mk .
carrying a quasi - particle drain , sample a cools to 200 mk . with a reduced drain barrier
, sample b cools to 160 mk . in sample c
, we obtain the best cooling by further reducing the drain and the cooler barriers : from 300 mk , sample c cools down to 132 mk with a 400 pw cooling power .
most importantly , from a bath temperature of 250 mk , sample c cools down to below 100 mk , see the arrow in .
the electronic temperature of the superconducting electrodes can be accessed by balancing the normal metal electron - phonon coupling power ( eq .
2 ) with the nis junction cooling power at a voltage bias @xmath26 : @xmath27}.\ ] ] figure 2 b upper part ( dashed lines ) displays the electronic temperature of the superconductor @xmath28 derived when assuming again @xmath23 in the normal metal .
although the latter assumption calls for further discussion , the trend remains that the superconductor gets significantly overheated .
in order to describe further the thermal transport in our devices , we consider a one - dimensional multilayered thermal model .
it is a set of coupled heat equations for different systems .
the model geometry shown in figure 3a includes a normal metal island ( n ) , superconducting leads ( s ) , side traps ( st ) , and quasi - particle drains ( d ) , similar to the geometry of the studied coolers ( figure 1a ) .
although a non - equilibrium description of a biased nis junction is possible @xcite , we assume for simplicity every part of the device can be described by local temperature .
we also neglect inverse proximity effect in the superconductor density of electronic states , as can be justified by the sharpness of measured ivs . in the following equations
, we will make use of the quantities @xmath29 , @xmath30 and @xmath31 that are the local current @xmath32 , cooling power @xmath33 and electron - phonon coupling power @xmath24 per unit area in the @xmath34 plane , see fig . 3a .
in addition , @xmath35 is the temperature , @xmath36 is the thickness and @xmath37 is the thermal conductivity in the element @xmath38 . here ,
temperature gradients are in the @xmath39-direction , charge and heat currents through the junctions ( @xmath29 and @xmath30 ) are in the @xmath40-direction , electron - phonon couplings @xmath24 are in the bulk . in the normal metal , the temperature gradient
is determined by the electronic thermal conductivity , the local electron - phonon coupling plus , in the region n2 in contact with the junction , the local cooling power as @xmath41 in the superconductor , the cooling junction injects a heat current equal to the joule power plus the cooling power .
the effect of the drain in region s1 or of the side trap in region s2 is described using the expression of the heat flow through a nis junction at zero bias @xmath42 .
weak electron - phonon coupling in s1 and s2 is neglected here , and @xmath43 the drain is submitted to the heat coming from the superconductor through the drain barrier . as it is made of a normal metal , electron - phonon coupling is taken into account as @xmath44 and similarly for the side trap @xmath45 we solved these coupled differential equations numerically @xcite using measured parameters from samples a , b , and f at 300 mk . from the measured electrical conductivity , we find that our sputtered cu films have a residual resistance ratio of about 1.5 , which is limited by disorder in the film @xcite .
the deduced value of @xmath46 wk@xmath47m@xmath48 ( @xmath49 ) is in agreement with the tabulated value in @xcite .
we use @xmath50 @xcite and @xmath51 wk@xmath17m@xmath18 @xcite .
we take into account the exponential decay of @xmath52 with temperature @xcite . for the cooler barrier ,
we use @xmath12 = 500 @xmath53m@xmath1 , close to the measured value for sample a ( 400 @xmath53m@xmath1 ) and b ( 700 @xmath53m@xmath1 ) . based on the different properties of samples a and b ( table 1 ) , we use drain barrier resistivity @xmath54 = 500 @xmath53m@xmath1 for sample a and @xmath54 = 10 @xmath53m@xmath1 for sample b. note that @xmath54 is the only differing parameter between samples a and b. this guess ( based on the value of the cooler barrier and the prediction in @xcite ) is necessary as we can not measure @xmath54 directly . in solving for sample
f , equation ( 6 ) and the terms for @xmath55 in equation ( 5 ) are ignored as f does not have a quasi - particle drain . solving these equations
gives a complete temperature profile : @xmath21 , @xmath28 , @xmath55 , and @xmath56 of the device .
figure 3b compares @xmath21 as a function of bias voltage from the modeling results ( dashed lines ) with the experimental data ( solid lines ) at 300 mk .
the good match between the two confirms that our simple model captures the essential physics of the device : a thinner drain barrier is the single parameter that enhances the performance of the cooler .
figure 3c shows the calculated temperature profile in the superconductor ( solid lines ) and in the drain ( dashed lines ) at optimum bias .
it consistently shows that superconducting electrodes get overheated over a typical length scale of about 50 @xmath0 m from the junction .
carrying a weak drain barrier transparency , sample a has a local superconductor temperature @xmath28 well above the drain temperature @xmath55 . with an improved barrier transparency in sample b
, its superconducting electrodes are well thermalized by the drains .
one obtains @xmath57 at distances x of about 20 @xmath0 m away from the junction .
the behavior is consistent with the magnitude of electronic cooling observed , thus demonstrating the effect of an improved drain barrier transparency for a good efficiency of the quasi - particle drain .
we have designed and studied electronic coolers capable to cool an electronic bath from 250 mk , the base temperature of a he@xmath20 cryostat , down to below 100 mk , the working regime of a dilution cryostat . with a fine - tuned barrier ,
the quasiparticle drain is efficient in thermalizing nis junctions .
the related geometry does not impose any limit in making the junction larger , thus opening the possibility to obtain a cooling power well above the present level of 400 pw .
the fabrication is low cost and involves only photolithography , the device is of high quality and robust . on this basis
, we are developing a platform that integrates coolers and sensors on a single chip , which is of great potential for astrophysics and other low temperature applications .
we acknowledge the support of the european community research infrastructures under the fp7 capacities specific programme , microkelvin project number 228464 , the epsrc grant ep / f040784/1 , and the academy of finland through its ltq coe grant ( project no .
samples are fabricated in the micronova nanofabrication center of aalto university .
we thank d. gunnarsson for help with the sputter .
30 s. h. moseley , aip conf .
proc . * 1185 * , 9 ( 2009 ) .
p. l. richards , j. appl .
phys . * 76 * , 1 ( 1994 ) .
n. a. miller , g. c. oneil , j. a. beall , g. c. hilton , k. d. irwin , d. r. schmidt , l. r. vale , and j. n. ullom , appl .
. lett . * 92 * , 163501 ( 2008 ) .
m. nahum , t. m. eiles and j. m. martinis , appl .
. lett . * 65 * , 3123 ( 1994 ) .
j. t. muhonen , m. meschke , and j. p. pekola , rep .
. phys . * 75 * , 046501 ( 2012 ) .
f. giazotto , t. t. heikkila , a. luukanen , a. m. savin and j. p. pekola , rev .
phys . * 78 * , 217 ( 2006 ) .
j. p. pekola , t. t. heikkila , a. m. savin , j. t. flyktman , f. giazotto , and f. w. j. hekking , phys .
* 92 * , 056804 ( 2004 ) .
s. rajauria , p. s. luo , t. fournier , f. w. j. hekking , h. courtois , and b. pannetier , phys .
* 99 * , 047004 ( 2007 ) .
p. j. lowell , g. c. oneil , j. m. underwood , and j. n. ullom , appl .
. lett . * 102 * , 082601 ( 2013 ) j. t. peltonen , j. t. muhonen , m. meschke , n. b. kopnin , and j. p. pekola , phys .
b * 84 * , 220502 ( 2010 )
. i. j. agulo , l. kuzmin , m. fominsky , m. tarasov , nanotechnology * 15 * , s224 ( 2004 ) .
g. c. oneil , p. j. lowell , j. m. underwood , and j. n. ullom , phys .
b * 85 * , 134504 ( 2012 ) .
a. luukanen , a. m. savin , t. i. suppula , j. p. pekola , m. prunnila , and j. ahopelto , ltd-9 aip conf . proc . * 605 * , 375 - 378 ( 2002 ) . v. j. kauppila , h. q. nguyen and t. t. heikkila , arxiv:1304.1288 , under review in phys .
b. h. q. nguyen , l. m. a. pascal , z. h. peng , o. buisson , b. gilles , c. b. winkelmann , and h. courtois , appl .
. lett . * 100 * , 252602 ( 2012 ) .
al doped mn at 2500 ppm .
we have tested down to 50 mk that its resistivity @xmath58 0.036 @xmath59 m (= @xmath60 ) .
a. m. clark , a. williams , s. t. ruggiero , m. l. van den berg , and j. n. ullom , appl .
. lett . * 84 * , 625 ( 2004 ) .
f. c. wellstood , c. urbina , and j. clarke , phys .
b * 49 * , 5942 ( 1994 ) .
m. meschke , j. p. pekola , f. gay , r. e. rapp , and h. godfrin , j. low temp . phys . * 134 * , 1119 ( 2004 ) .
j. voutilainen , t. t. heikkila , and n. b. kopnin , phys .
b * 72 * , 054505 ( 2005 )
. s. rajauria , h. courtois , and b. pannetier , phys .
b * 80 * , 214521 ( 2009 ) .
comsol multiphysics is a computer software developed by comsol , www.comsol.com .
l. h. qian , q. h. lu , w. j. kong , and k. lu , scr . mater . * 50 * , 14071411 ( 2004 ) .
a. v. timofeev , m. helle , m. meschke , m. m " ott " onen , and j. p. pekola , phys .
lett . * 102 * , 200801 ( 2009 ) .
|
the performance of hybrid superconducting electronic coolers is usually limited by the accumulation of hot quasi - particles in their superconducting leads .
this issue is all the more stringent in large - scale and high - power devices , as required by applications . introducing a metallic drain connected to the superconducting electrodes via a fine - tuned tunnel barrier
, we efficiently remove quasi - particles and obtain electronic cooling from 300 mk down to 130 mk with a 400 pw cooling power . a simple thermal model accounts for the experimental observations .
|
active galactic nuclei ( agns ) , such as quasars and seyfert galaxies , are powered by gravitational accretion of matter onto supermassive black holes in the central engines .
the energy conversion in agns is more efficient as implied by high flux variability and short variability timescales @xcite .
accretion of matter onto black holes can have high energy release efficiency @xcite .
broad - line regions ( blrs ) in agns are photoionized by the central radiation of accretion disks .
the broad emission line variations will follow the ionizing continuum variations due to the photoionization process ( e.g. * ? ? ? * ; * ? ? ? * ) .
the reverberation mapping model was tested with the reverberation mapping observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a review about the reverberation mapping research studies is given by ( * ? ? ? * and references therein ) .
a general assumption for the cloud ensemble is often virial equilibrium .
the blrs are mainly dominated by the gravitational potentials of the supermassive black holes .
this would imply keplerian orbits of clouds , and this was supported with evidence for the keplerian motions of the blr clouds within the well - studied seyfert 1 galaxy ngc 5548 ( e.g. * ? ? ?
* ; * ? ? ?
the virialization assumption has been commonly and widely accepted for the reverberation mapping observations to estimate the black hole masses @xmath3 .
this treatment neglects the contribution of a radiation pressure to the dynamics of clouds .
as the central ionizing source is photoionizing the blr clouds , the central radiation will produce a radiation pressure on the clouds .
the radiation pressure will counteract a part of the gravitational force of a black hole .
thus , the clouds undergo a decrease of the effective gravity of the black hole , and then the orbital velocities of the clouds will decrease for the same orbital radii . in general , this will decrease the widths of broad emission lines , and then lead to underestimate the black hole masses based on the reverberation mapping method .
the radiation pressure force may contribute significantly to the cloud motion , and in this case the clouds on bound orbits could be significantly sub - keplerian @xcite
. however , most of the photoionized gas in the blr follows keplerian orbits @xcite .
@xcite studied the effect of radiation pressure on black hole virial mass estimates for narrow - line seyfert 1 galalxies , and the black hole virial masses might be significantly underestimated if the radiation pressure is neglected .
@xcite found that the radiation pressure force is not important in @xmath8 agns with @xmath9 .
the simple virial mass estimates can give a reasonable approximation to @xmath3 even when the radiation pressure force is important @xcite .
in the reverberation mapping researches ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , the radiation pressures of the central engines were not considered in estimating @xmath3 for the broad - line agns .
the continuum radiation pressures due to the free electron thomson scattering and the recombination and re - ionization of hydrogen ions are usually believed negligible or not considered . in this paper
, we will investigate the influences of the continuum radiation pressures on the black hole mass estimates for agns with high accretion rates , including quasars and broad - line seyfert 1 galaxies .
the structure of this paper is as follows .
section 2 presents model .
section 3 presents applications .
section 4 is for discussion and conclusions . in this work ,
we assume the standard @xmath10cdm cosmology with @xmath11 , @xmath12 = 0.30 , and @xmath13= 0.70 .
several assumptions are made .
first , the clouds mainly consist of hydrogen and are partially ionized .
second , the clouds are blobs at the distance @xmath14 from the central black hole .
dynamical equilibrium corresponds to circular orbits ( e.g. * ? ? ?
third , the clouds are in the circular orbits .
fourth , the clouds are optically thin to the thomson scattering .
a cloud has a mass @xmath15 , where @xmath16 is the total number of hydrogen in the cloud and @xmath17 is the mass of hydrogen atom .
the cloud is subject to the gravity of the central black hole with a mass @xmath3 , and the gravity @xmath18 is @xmath19 where @xmath20 is the gravitational constant , and @xmath14 is the orbital radius of this cloud .
the cloud is exerted with a centrifugal force @xmath21 @xmath22 where @xmath23 is the circular velocity .
here , we consider two origins of continuum radiation pressure forces on a cloud : the thomson scattering of the central source radiation by free electrons , and the recombination and re - ionization of hydrogen ions that will continue to absorb ionizing photons to compensate for the recombination . the recombination line photon number per unit time , i.e. , the recombination rate of the ionized hydrogen , is @xmath24 , where @xmath25 is the hydrogen recombination coefficient , @xmath26 is the number density of the free electrons , @xmath1 is within @xmath27 , and @xmath28 is the number of the hydrogen ions . on the hydrogen ionizing timescale @xmath29 , the neutral hydrogen atoms from the recombination are re - ionized into the hydrogen ions with a number @xmath30 , where @xmath31 .
thus , the total continuum radiation pressure force on the cloud due to the above two origins is @xmath32 where @xmath33 is the ratio of the ionizing luminosity @xmath34 to the central source luminosity @xmath35 , @xmath36 is the thomson cross - section , @xmath37 is the speed of light , @xmath38 is the ionized number of hydrogen atoms in the cloud , @xmath39 is the flux - weighted average of the hydrogen bound - free absorption cross - section , and @xmath40 is the hydrogen ion recombination timescale in the cloud .
the two continuum radiation pressure forces in equation ( 3 ) are similar to those in equation ( 9 ) in @xcite that focused on stability of blr clouds .
the resultant of forces is @xmath41 . as @xmath42
, we have @xmath43 where @xmath15 is used in the deduction , @xmath44 , @xmath45 is the black hole mass estimated as @xmath35 is not considered , and @xmath0 is the black hole mass counteracted by the continuum radiation pressure due to the central engine luminosity .
the cross - section @xmath39 is equal to @xmath46 where @xmath47 , @xmath48 is the frequency of 13.6 @xmath49 photon , @xmath50 is the photoionization cross - section of hydrogen , @xmath51 is the ionizing radiation flux , and @xmath52 is the total radiation flux . for @xmath53 @xcite , @xmath54 .
the ionizing timescale is @xmath55 the recombination timescale of ionized hydrogen is @xmath56 where @xmath57 is the free electron number density and @xmath58 is the number density of hydrogen .
the hydrogen recombination coefficient @xmath59 was given with a set of formulae in @xcite .
an analytic expression of @xmath25 to the excited levels of hydrogen is given by @xcite @xmath60 \rm{\/\ cm^3 \/\ s^{-1}},\ ] ] where @xmath61 , and @xmath62 in units of @xmath63 is the electron temperature of clouds . as @xmath64 , equation ( 8) gives @xmath25 to an accuracy of better than 1 percent relative to the result of @xcite .
the photoionized gas in clouds is at @xmath65 and produces optical and ultraviolet emission lines of quasars @xcite .
photoionization governs the thermodynamics of the clouds that have a stable equilibrium temperature of order @xmath66 , independent of their locations @xcite . in the case of @xmath65 , @xmath67 . for the blrs of agns
, @xcite presented the calibrated radius - luminosity relation @xmath68 where @xmath14 and @xmath69 are in units of light days and @xmath70 , respectively . the radiation flux @xmath71 . in general
, there is @xmath72 that means a partially ionized cloud .
@xmath73 means a fully ionized cloud , and @xmath74 means a fully non - ionized cloud .
first , we consider a case that a fraction of hydrogen is ionized , and we will take @xmath75 ( case a ) .
the cloud may be mostly ionized , and we will take @xmath76 ( case b ) .
the ratio of @xmath77 has an average @xmath78 @xcite .
a typical value of @xmath79 was obtained for the blrs of agns @xcite .
the luminosity @xmath35 is estimated with @xmath80 @xcite .
@xmath81 is used for the ultraviolet ( uv ) continuum ( e.g. * ? ? ?
the range of @xmath72 is suggested by the corresponding variations between the broad - line and continuum light curves in the reverberation mapping observations ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
in the following section , we take @xmath82 , @xmath79 , @xmath83 , and @xmath53 to estimate @xmath0 for cases a and b.
we collect 40 high accretion rate agns , including quasars and broad - line seyfert 1 galaxies , with the black hole virial masses @xmath2 estimated from the reverberation mapping method ( see table 1 ) .
table 1 consists of four parts .
the first part contains 17 agns , collected from @xcite , with relatively high accretion rates .
the second part contains 13 agns with super - eddington accreting massive black holes ( seambhs , * ? ? ?
the first results from a new reverberation mapping campaign are reported in @xcite for the seambh agns .
the new reverberation mapping observations of the seambh agns are presented in @xcite and @xcite .
the third part contains 9 quasars in the canada - france high-@xmath4 quasar survey ( cfhqs ) at redshift @xmath84 6 , accreting close to the eddington limit @xcite .
the last one contains 2 agns from other works .
there is one overlap agn ( see table 1 ) .
the cfhqs quasars have @xmath2 and luminosity @xmath85 .
the optical luminosity @xmath86 around 5100 @xmath87 at the rest frame of source is used to estimate @xmath35 with @xmath88 @xcite . as in @xcite
, @xmath81 is used for the cfhqs quasars .
the estimated results of @xmath0 are presented in table 1 .
figure 1 shows the comparisons of @xmath0 to @xmath2 . for case a ,
four agns with seambhs and three cfhqs quasars have @xmath89 ( see figure 1a and table 1 ) . for case
a , 9 agns match @xmath7 ( see figure 1a ) . for case
b , all agns match @xmath89 , and 30 agns follow @xmath90 ( see figure 1b ) .
these comparisons show that the continuum radiation pressures of the central engines have a significant influence on the black hole virial mass estimates on the basis of the reverberation mapping method .
the masses @xmath0 are sensitive to the percent of ionized hydrogen atoms @xmath1 in the clouds .
for some agns , such as iras f12397 , iras 04416 , and j1641 + 3755 , the continuum radiation pressures can counteract about 3040 percent gravitational forces of @xmath3 ( case a ) . in the plots of @xmath0 versus @xmath2 ,
the seambh agns , the cfhqs quasars , and the rest of agns populate in the narrow zones ( see figure 1 ) . from this view point , the seambh agns at @xmath91 and the cfhqs quasars at @xmath92 are not essentially different .
this means that their central black holes are likely accreting at the very high rates , regardless of the environments surrounding these agns , but maybe related to the environments surrounding the central black holes , such as dusty tori and accretion disks .
also , some of the rest of agns are at the similar accretion rates to those of the seambh agns and the cfhqs quasars . whether or not the continuum radiation pressure is considered to estimate @xmath3 depends on the ratio of @xmath93 .
the radiation pressure must be considered as @xmath94 that will lead to underestimate @xmath3 by a factor @xmath95 , i.e. , @xmath96 . when @xmath97 , @xmath3 will be underestimated by a factor @xmath98 , i.e. , @xmath99 .
thus , the importance of the continuum radiation pressure can not be neglected in estimating @xmath3 for the high accretion rate agns .
there is a undetermined parameter @xmath1 , the ionization percent of the blr clouds , in equations ( 3 ) and ( 4 ) .
two values of @xmath100 and @xmath101 are assumed to estimate @xmath0 .
these assumptions may lead to some unreasonable results . following the strmgren radius of a strmgren sphere of ionized hydrogen around a young star @xcite , an ionization depth of a cloud
is defined as the position where the recombination rate equals the ionization rate , and the hydrogen atoms are fully ionized . as
the recombination and ionization are in equilibrium , the ionizing photon number arriving at one cloud per unit time equals the recombination line photon number per unit time .
the ionizing photon number arriving at the cloud per unit time is @xmath102 , where @xmath103 is the cross - section area of the cloud , @xmath104 is the ionizing photon number emitted by the central engine per unit time ( @xmath105 the mean energy of photons ) , @xmath106 is the ionization parameter of hydrogen , and @xmath58 is the hydrogen number density of the cloud . at the same time , the ionized region of the cloud produces the recombination line photon number per unit time @xmath107 , where @xmath26 is the electron number density in this region , and @xmath108 is the fully ionized volume of the cloud .
thus , we have @xmath109 because of the ionization balance .
the ionization depth of the cloud is @xmath110 .
the cloud has @xmath111 , where @xmath112 and @xmath113 are the volume and thickness of the cloud , respectively .
the ionization parameter has @xmath1141 for emission line gas in quasars and seyfert galaxies @xcite .
a gas density of @xmath115@xmath116 is required by a very large lyman continuum optical depth ( e.g. * ? ? ?
the gas density values are the same as those in @xcite for quasars and seyfert galaxies .
@xcite got @xmath117 .
a density of @xmath118 is set by the presence of broad semi - forbidden line c iii@xmath119\lambda 1909 $ ] and the absence of broad forbidden lines such as [ o iii@xmath119\lambda \lambda 4363,4959,5007 $ ] @xcite .
the typical values are @xmath120 and @xmath114 ( e.g. * ? ? ?
* ; * ? ? ?
the cloud size of @xmath121@xmath122 was constrained by the smoothness of the emission - line profiles @xcite .
@xcite showed a universal ionization parameter @xmath114 in the inner photoionized layer of the blr clouds , independent of luminosity and distance .
the radiation pressure confinement of the photoionized layer appears to explain the universality of the blr properties in agns , the similar relative line strength over the vast range of @xmath123@xmath124 @xmath70 in luminosity @xcite .
a widely accepted power - law relation has been established between the luminosity @xmath125 and the blr size @xmath14 , as @xmath126 @xcite .
this power - law relation spans over a range of @xmath127 in @xmath125 @xcite .
so , @xmath129 will be independent of @xmath86 , @xmath34 , and @xmath14 .
this independence may lead to a universal ionization parameter as suggested in @xcite .
the typical values of @xmath120 and @xmath114 are taken to estimate @xmath130 , and then @xmath1 with @xmath131 and @xmath132@xmath133 .
we have @xmath134 @xmath135 , and then @xmath136 0.20.6 .
thus , it is basically reasonable to take @xmath75 and 0.5 in the calculations of @xmath0 .
so , the masses counteracted by the continuum radiation pressures of the central engines are not negligible compared to , or are comparable to the black hole virial masses at least for some agns ( see figure 1 ) .
this counteracting effect of the continuum radiation pressure is significant for quasars at @xmath92 ( see figure 1 ) .
j0100 + 2802 at @xmath137 has a bolometric luminosity of @xmath138 and a black hole mass of @xmath139 , and it is the most luminous quasar known at @xmath140 @xcite .
it has a @xmath141 as @xmath75 and @xmath142 as @xmath143 .
these @xmath0 for this object are comparable to the black hole virial mass @xmath144 .
so , j0100 + 2802 will have @xmath145 .
this larger black hole mass further gives rise to the most significant challenge to the eddington limit growth of black holes in the early universe @xcite .
there are the same cases for the cfhqs quasars as in j0100 + 2802 ( see figure 1 ) .
their masses @xmath3 are larger than the virial masses @xmath2 .
@xcite suggested for @xmath146 quasars that the supermassive black holes in the early universe likely grew much more quickly than their host galaxies .
the larger black hole masses @xmath3 of the quasars at @xmath92 further strengthen this suggestion .
thus , it is important to consider the radiation pressure effect on the black hole mass estimates for the agns with the high accretion rates .
this importance of the radiation pressure is consistent with that suggested by @xcite who argued for narrow - line seyfert 1 galalxies , and seems to be inconsistent with the suggestions in @xcite and @xcite .
an assumption of isotropy in equations ( 3 ) and ( 4 ) is made for the central source luminosity @xmath35 .
this assumption might be significantly influence the estimated values of @xmath0 , and then the results of this paper .
the anisotropic illumination of a blr by the central radiation was discussed by @xcite , and recently was investigated by @xcite .
the central radiation of quasars are anisotropic ( e.g. * ? ? ?
@xcite derived bolometric corrections based on the theoretical accretion disk models of @xcite , and obtained an anisotropic correction factor of @xmath147 , where @xmath148 and @xmath149 is the total source luminosity assuming isotropy .
@xcite got a factor of @xmath150 for quasars .
so , we take @xmath151 , i.e. , @xmath152 to re - estimate @xmath0 , and the re - estimated @xmath0 are compared to @xmath2 in figure 2 .
considering the anisotropy of the central source luminosity @xmath35 , the continuum radiation pressure effects on the black hole masses @xmath3 can not be still neglected for the close- and super - eddington limit accretion rate agns .
thus , the anisotropy of the central source luminosity could not influence significantly the main results assuming isotropy .
@xcite proposed the seambh agns as the most luminous standard candles in the universe .
since a part of the seambh agns and the cfhqs quasars have the same behaviors in the plots of @xmath0 versus @xmath2 , the cfhqs quasars likely have the potential to be the most luminous standard candles in the universe .
this potential will extend the redshifts of the most luminous standard candles from @xmath153 up to @xmath92 , and will be important to study the universe because of extending significantly the cosmic distance beyond the range explored by type ia supernovae @xcite .
an average column density of a cloud is @xmath154 ( @xmath120 and @xmath155@xmath156 ) . the electron thomson scattering optical depth of the cloud is @xmath157 , and the cloud is optically thin , which confirms the fourth assumption of the model
. the ionizing optical depth of the cloud is @xmath158 , and the cloud is optically thick to the ionizing photons .
the cloud has the strmgren radius smaller than its thickness .
so , the cloud is partially ionized , confirming the first assumption of the model , and this is consistent with the constraints set by the reverberation mapping observations . the radiation pressure force ratio of the second to the first term in equation ( 3 ) has averages @xmath1596.3 for agns in table 1 .
the first term of the free electron thomson scattering is comparable to the second term of the recombination and re - ionization of the hydrogen ions .
the second term becomes more important as @xmath1 increases .
the partly ionized clouds will be ionized more as @xmath34 increases , and the second term would be more important as the central engines become brighter .
the radiation pressure forces due to absorption of ionizing photons and thomson scattering were used to modify the expression for the black hole virial mass @xmath2 ( see equation 5 in * ? ? ? * ) . as @xmath160@xmath161 , the modified @xmath2 becomes larger by a factor of @xmath162100 for the agns accreting around the eddington limit . in consequence , j0100 + 2802 at @xmath137 will have @xmath3 @xmath163@xmath164 that is nearly impossible to the eddington limit growth of black holes in the early universe because of the larger initial masses of primary black holes .
if considering the line - driven radiation pressure force @xcite , the radiation pressure force due to the gas opacity will be @xmath165 times that due to the electron scattering opacity @xcite .
this will bring a larger mass @xmath166 @xmath0 @xmath167@xmath168 for j0100 + 2802 .
these extremely large corrected masses of the black hole in j0100 + 2802 indicate some problem of their treatments of the radiation pressure forces on the blr clouds in agns , or the model of formation and evolution of black hole , or the model of the universe . in this paper , we investigate the influences of two continuum radiation pressures , usually believed negligible or not considered , of the central engines in agns on the black hole mass estimates with the reverberation mapping method or the descendent methods .
the continuum radiation pressure forces are from two origins : the free electron thomson scattering of the central radiation , and the recombination and re - ionization of the ionized hydrogen .
the radiation pressures depend on a parameter @xmath1 , the ionized percent of the clouds in a blr .
the counteracted black hole masses by the radiation pressures @xmath0 are compared to the black hole virial masses @xmath2 for 40 agns with the high accretion rates .
the masses @xmath0 are sensitive to @xmath1 . as @xmath143 , @xmath7 for all the agns in figure 2 . as @xmath75
, the radiation pressures can counteract at least about 3040 percent gravitational forces of the black holes for some agns ( see figure 1 ) .
four seambh agns at @xmath169 and five cfhqs quasars at @xmath170 are around the line @xmath171 for @xmath75 ( see figure 1a ) .
thus , the continuum radiation pressures of the central engines have to be considered in estimating the black hole masses for the agns accreting around the eddington limit , regardless of the redshifts or the surrounding environments of agns . the most luminous quasar j0100 + 2802
likely has the same case as the nine agns ( see figures 1 and 2 ) .
a part of the seambh agns and the cfhqs quasars is blended with the non - seambh and non - cfhqs agns ( see figures 1 and 2 ) .
the close- and super - eddington limit accreting agns are not different from the rest of agns .
the anisotropy of the central source luminosity could not influence significantly the main results assuming isotropy .
the force multiplier , the ratio of gas opacity to electron scattering opacity @xcite , will be needed for the agns with @xmath7 due to the radiation pressures in equation ( 3 ) .
though , some extremely large masses are derived from the force multiplier @xmath172 for the black hole in j0100 + 2802 ( @xmath173@xmath168 ) . in future , the force multiplier , due to the photoionization absorption , and the resonance and subordinate line absorption , could be calculated with the spectral simulation code cloudy described by @xcite , and with detailed parameters of the blr structure , the cloud distribution , the gas density of cloud , the chemical abundances of gas , the central continuum , and the observed emission lines , for each high accretion rate agns .
thanks the national natural science foundation of china ( nsfc ; grants 11273052 and u1431228 ) for financial support . j.m.b . acknowledges the support of the nsfc ( grant 11133006 ) .
thanks the financial supports of the youth innovation promotion association , cas and the project of the training programme for the talents of west light foundation , cas .
mrk 335 & 0.026&@xmath174 & @xmath175 & 1 & @xmath176 & @xmath177 + pg 0026 + 129&0.142 & @xmath178 & @xmath179 & 1 & @xmath180 & @xmath181 + pg 0052 + 251&0.155 & @xmath182&@xmath183 & 1 & @xmath184 & @xmath185 + 3c 120 & 0.033&@xmath186&@xmath187 & 1 & @xmath188 & @xmath189 + pg 0844 + 349&0.064&@xmath190&@xmath191 & 1 & @xmath192 & @xmath193 + mrk 110 & 0.035&@xmath194&@xmath195 & 1 & @xmath196 & @xmath197 + pg 0953 + 414&0.234&@xmath198&@xmath199 & 1 & @xmath200 & @xmath201 + pg 1211 + 143&0.081&@xmath202&@xmath203 & 1 & @xmath204 & @xmath205 +
pg 1226 + 023&0.158&@xmath206&@xmath207 & 1 & @xmath208 & @xmath209 + pg 1229 + 204&0.063&@xmath210&@xmath211 & 1 & @xmath212 & @xmath213 + ngc 4593 & 0.009 & @xmath214&@xmath215 & 1&@xmath216 & @xmath217 + pg 1307 + 085&0.155&@xmath218&@xmath219 & 1 & @xmath220 & @xmath221 + ic 4329a & 0.016&@xmath222&@xmath223 & 1 & @xmath224&@xmath225 + mrk 279 & 0.030&@xmath226&@xmath227 & 1 & @xmath228&@xmath229 + pg 1613 + 658&0.129 & @xmath230&@xmath231 & 1 & @xmath232 & @xmath233 + pg 1700 + 518&0.292&@xmath234&@xmath235 & 1 & @xmath236 & @xmath237 + ngc 7469 & 0.016 & @xmath238&@xmath239 & 1 & @xmath240 & @xmath241 + @xmath242mrk 335 & 0.026&@xmath243&@xmath244 & 2&@xmath245 & @xmath246 + mrk 1044 & 0.017 & @xmath247&@xmath248 & 2&@xmath249&@xmath250 + mrk 382 & 0.034&@xmath251&@xmath252 & 2&@xmath253&@xmath254 + mrk 142 & 0.045&@xmath255&@xmath256 & 2&@xmath257&@xmath258 +
* iras f12397&0.043&@xmath259 & @xmath260 & 2&@xmath261&@xmath262 + mrk 486&0.039&@xmath263&@xmath264 & 2 & @xmath265&@xmath266 + mrk 493&0.031&@xmath267&@xmath268 & 2&@xmath269&@xmath270 + * iras 04416&0.089&@xmath271 & @xmath272 & 2&@xmath273&@xmath274 + sdss j075101&0.121&@xmath275&@xmath276 & 2&@xmath277&@xmath278 + * sdss j080101&0.140&@xmath279&@xmath280 & 2&@xmath281&@xmath282 + sdss j081441&0.163&@xmath283&@xmath284 & 2&@xmath285&@xmath286 + sdss j081456&0.120&@xmath287&@xmath288 & 2&@xmath289&@xmath290 + * sdss j093922&0.186&@xmath291&@xmath292 & 2&@xmath293&@xmath294 + * * * * * j0210 - 0456 & 6.438 & @xmath295@xmath296 & @xmath297 & 3 & @xmath298 & @xmath299 + j2329 - 0301 & 6.417 & @xmath300@xmath296 & @xmath301 & 3 & @xmath302 & @xmath303 + j0050 + 3445 & 6.253 & @xmath304@xmath296&@xmath305 & 3 & @xmath306 & @xmath307 + j0221 - 0802 & 6.161 & @xmath308@xmath296&@xmath309 & 3 & @xmath310 & @xmath311 + * j2229 + 1457 & 6.152 & @xmath312@xmath296&@xmath313 & 3 & @xmath314 & @xmath315 + j1509 - 1749 & 6.121 & @xmath316@xmath296 & @xmath317 & 3 & @xmath318 & @xmath319 + j2100 - 1715 & 6.087 & @xmath320@xmath296 & @xmath321 & 3 & @xmath322 & @xmath323 + * j1641 + 3755 & 6.047 & @xmath324@xmath296 & @xmath325 & 3 & @xmath326 & @xmath327 + j0055 + 0146 & 5.983 & @xmath328@xmath296&@xmath329 & 3 & @xmath330 & @xmath331 + * * *
|
in this paper , we investigate the influences of two continuum radiation pressures of the central engines on the black hole mass estimates for 40 active galactic nuclei ( agns ) with high accretion rates .
the continuum radiation pressure forces , usually believed negligible or not considered , are from two sources : the free electron thomson scattering , and the recombination and re - ionization of hydrogen ions that continue to absorb ionizing photons to compensate for the recombination .
the masses counteracted by the two radiation pressures @xmath0 depend sensitively on the percent of ionized hydrogen in the clouds @xmath1 , and are not ignorable compared to the black hole virial masses @xmath2 , estimated from the reverberation mapping method , for these agns . as @xmath1 increases , @xmath0 also does .
the black hole masses @xmath3 could be underestimated at least by a factor of 3040 percent for some agns accreting around the eddington limit , regardless of redshifts of sources @xmath4 .
some agns at @xmath5 and quasars at @xmath6 have the same behaviors in the plots of @xmath0 versus @xmath2
. the complete radiation pressures will be added as agns match @xmath7 due to the two continuum radiation pressures . compared to @xmath2 ,
@xmath3 might be extremely underestimated if considering the complete radiation pressures for the agns accreting around the eddington limit .
|
the discovery of high temperature superconductivity in the iron - based superconductors@xcite has attracted much attention in many aspects .
not only the high @xmath3 itself , but also a number of experiments indicating non - universality of the superconducting gap function , such as sign reversing , anisotropy , or the presence of nodes , suggest an unconventional pairing mechanism .
most probable candidate for such an unconventional mechanism is the pairing mediated by spin fluctuations , where the superconducting gap changes sign between the disconnected fermi surfaces , namely , the so - called @xmath4 pairing@xcite .
back in 2001 , one of the present authors proposed that spin fluctuation mediated pairing in systems with nested disconnected fermi surfaces may give rise to a very high @xmath3 superconductivity@xcite .
the idea is that the repulsive pairing interaction mediated by spin fluctuations can be fully exploited without introducing nodes of the superconducting gap on the fermi surfaces .
although the fermi surface of the iron - based superconductors does resemble the proposed fermi surface configuration , there are some important differences .
one is that the fermi surface in the iron - based superconductors has multiple orbital characters , and second is that there can be frustrations in the sign of the gap function , which can give rise to gap nodes on the fermi surface . in the present paper ,
we focus on this gap - sign frustration problem , and propose that the fermi surface configuration that gives the least frustration provides the optimal condition for high @xmath3 in the iron - based superconductors . in ref.@xcite ,
two of the present authors pointed out that maximizing the fermi surface multiplicity leads to the optimization for @xmath3 , but considering the frustration problem studied in the present paper , the `` sweet spot '' for high @xmath3 is further limited to a narrow window of the fermi surface configuration .
let us first describe the band structure of lafeaso .
lafeaso takes a layered structure , where fe atoms form a square lattice in each layer , sandwiched by as atoms with tetrahedral coordination .
we use the band structure obtained from first principles@xcite to construct the maximally localized wannier functions@xcite .
these wannier functions have five orbital symmetries ( @xmath5 , @xmath6 , @xmath7 , @xmath1 , @xmath8 ) , where @xmath9 refer to those for this unit cell with two fe sites as shown in fig.[fig1](a ) .
the two wannier orbitals in each unit cell are equivalent in that each fe atom has the same local arrangement of other atoms .
we can then take a unit cell that contains only one orbital per symmetry by unfolding the brillouin zone , and an effective five - band model on a square lattice is obtained , where @xmath10 and @xmath11 axes are rotated by 45 degrees from @xmath12-@xmath13 .
( fe ) , @xmath14 ( as below the fe plane ) and @xmath15 ( above fe ) .
( b ) the band structure ( left ) of the five - band model for lafeaso , and the fermi surface ( right ) at @xmath16 for 10 percent electron doping .
the main orbital characters of some portions of the bands and the fermi surface are indicated .
the dashed horizontal line in the band structure indicates the fermi level for 10 percent electron doping .
the short arrow in the band structure indicates the position of the dirac cone closest to the fermi level .
the gray areas in the fermi surface around the zone corners represent the @xmath17 fermi surface , which is barely absent for 10 percent electron doping .
( c ) the portion of the band that has mainly the @xmath1 orbital character .
[ fig1],width=302 ] in fig.[fig1](b ) ( right ) , the fermi surface for 10 percent electron doping is shown in the two - dimensional unfolded brillouin zone .
the fermi surface consists of four pieces : two concentric hole pockets ( denoted here as @xmath18 , @xmath19 ) centered around @xmath20 , two electron pockets around @xmath21 @xmath22 or @xmath23 @xmath24 , respectively . besides these pieces of the fermi surface
, there is a portion of the band near @xmath25 that touches the @xmath26 , so that the portion acts as a `` quasi fermi surface @xmath27 '' around @xmath25 . as for the orbital character , @xmath28 and portions of @xmath29 near brillouin zone edge
have mainly @xmath6 and @xmath7 character , while the portions of @xmath29 away from the brillouin zone edge and @xmath17 have mainly @xmath1 orbital character ( see also fig.[fig1](c ) ) .
the band structure and the fermi surfaces of the iron - based superconductors are sensitive to the lattice structure . in this section ,
we consider hypothetical lattice structures of lafeaso , where we fix the bond length at its original length and vary the bond angle @xmath28(fig.[fig2](a))@xcite .
we first neglect the three dimensionality ( out - of - plane hoppings ) , and consider a two dimensional model .
the fermi surface is obtained for 10 percent electron doping .
when the bond angle is large , two hole fermi surfaces , @xmath18 and @xmath19 are present around the wave vector ( 0,0 ) , while the @xmath17 around @xmath25 is missing . as @xmath28 decreases , the @xmath17 fermi surface appears around @xmath25 , and there are now three hole fermi surfaces .
this appearance of the additional fermi surface has been noticed as an effect of increasing the pnictogen height @xcite .
when @xmath28 is further reduced , the @xmath18 fermi surface disappears , and again the fermi surface multiplicity reduces to two , but in this case one around @xmath30 and another around @xmath25 .
such a disappearance of the @xmath18 hole fermi surface was first noticed in the band calculation of ca@xmath31al@xmath32o@xmath33fe@xmath31as@xmath31 @xcite by miyake _
_ in refs.@xcite .
this material indeed has very small bond angle of about 102 degrees . and
@xmath34 orbital characters at ( 0,0 ) is indicated.,width=302 ] fig.[fig3 ] explains schematically the band structure / fermi surface variation against the bond angle reduction@xcite . as the bond angle is reduced , the @xmath35 band below the fermi level around @xmath25 rises up , while another @xmath35 portion of the band above the fermi level around @xmath30 comes down .
this band deformation can be understood from fig.[fig1](c ) , where the @xmath35 portion of the band for the original lafeaso is extracted . in the tightbinding picture ,
the energy difference between the wave vectors @xmath30 and @xmath25 is roughly equal to @xmath36 , where @xmath37 is the nearest neighbor hopping of the @xmath35 orbital . as the bond angle is reduced , the contribution to @xmath37 from the fe@xmath38as@xmath38fe path decreases , while that from the direct fe@xmath38 fe path increases .
the two contributions have opposite signs , so that the reduction of the bond angle results in a decrease of @xmath37@xcite . when the bottom of the upper @xmath35 portion sinks below the @xmath34 bands around @xmath30 , a band structure reconstruction takes place , and now the two bands that are degenerate at @xmath30 repel with each other ( one going up , the other going down ) as the wave vector moves away from @xmath30 ( fig.[fig3](c ) ) .
in this situation , the band below these two bands has @xmath35 character near @xmath30 , and changes its character to @xmath34 as the wave number increases .
therefore , just before the inner hole ( @xmath18 ) fermi surface disappears , it has strong @xmath35 character . .
the black ( red ) portions indicate the bands with strong @xmath35 @xmath39 orbital character.,width=302 ]
having understood the fermi surface variation against the bond angle , we now consider how this should affect the gap function of the superconductivity mediated by spin fluctuations .
the pairing interaction mediated by spin fluctuations is repulsive , so the superconducting gap has an tendency of changing its sign between the initial and final wave vectors of the pair scattering .
another important point is that the spin fluctuations develop at wave vectors that bridge the portions of the fermi surface having the same orbital character@xcite . in the presence of the @xmath35 originated @xmath17 and @xmath34 originated @xmath18 and @xmath19 fermi surfaces ( case ( b ) in fig.[fig3 ] ) ,
it is known that the @xmath17-@xmath29 interaction among portions having @xmath35 character and the @xmath28-@xmath29 interactions among portions with @xmath34 character dominate , and the superconducting gap is fully open on all the fermi surfaces while changing its sign as @xmath40 , @xmath41 , @xmath40 along @xmath28 , @xmath29 , @xmath17 fermi surfaces .
this is the fully gapped @xmath4 gap@xcite . on the other hand , in the absence of the @xmath17 fermi surface ( case ( a ) in fig.[fig3 ] ) , the pairing interaction between the @xmath35 portions of the @xmath29 fermi surfaces and that between the @xmath34 portions of the @xmath28 and @xmath29 fermi surfaces results in a frustration in the sign of the gap function .
this situation was studied in several previous papers@xcite .
this can result in either nodal @xmath4-wave or @xmath42-wave pairings , where the nodes of the gap go into the @xmath29 fermi surface in the former , and @xmath28 in the latter .
a schematic figure summarizing the above is shown in fig.[fig4]@xcite . as was shown in ref.@xcite
, this frustration effect degrades @xmath3 of the superconductivity , so that the lattice structure acts as a switch between high @xmath3 nodeless and low @xmath3 nodal superconductivity .
wave , the nodal @xmath4 wave and @xmath42 wave gap are schematically shown .
the solid red ( dashed blue ) curves represent positive ( negative ) sign of the gap .
the arrows indicate the dominating nesting vectors.,width=377 ]
here we discuss another situation where the frustration arises in the sign of the superconducting gap function . as mentioned in section[fsvsangle ]
, the @xmath35 orbital character strongly mixes into the @xmath18 fermi surface just before the fermi surface vanishes as the bond angle is reduced .
in this situation , there are now @xmath35 components on @xmath18 , @xmath29 , and @xmath17 ( if present ) fermi surfaces .
since these fermi surfaces interact with repulsive pairing interactions , once again , a frustration arises in the sign of the superconducting gap ( fig.[fig5](a ) ) .
in addition to this , there can also be some @xmath34 component remaining in the @xmath18 fermi surface , and this portion tends to change the sign from the @xmath29 fermi surfaces ( fig.[fig5](b ) ) so this can be another factor for the frustration . to actually see this frustration effect ,
here we consider hypothetical lattice structures of ca@xmath32al@xmath31o@xmath33fe@xmath31as@xmath31@xcite , where we vary the bond angle from 108 to 109 degrees , and contrust a two dimensional five orbital model in the unfolded brillouin zone . in this angle regime ,
the @xmath18 fermi surface is barely present , and it is indeed composed of mixed @xmath35 and @xmath34 orbital components .
we apply fluctuation exchange approximation to this model@xcite , and obtain the eigenfunction ( gap function ) of the linearized eliashberg equation as was done in ref.@xcite . in fig.[fig6 ] , we show the gap function for the two angles for 10 percent electron doping and temperature @xmath43ev .
it can be seen that the magnitude of the gap on the @xmath18 fermi surface is very small , and its sign actually changes as the bond angle is varied , reflecting the frustration . and ( b ) @xmath34 portions of the fermi surface in the case where the inner hole fermi surface ( @xmath18 ) is barely present . in this case
, @xmath18 is a mixture of @xmath35 and @xmath34.,width=302 ] al@xmath31o@xmath33fe@xmath31as@xmath31 .
the bond angle @xmath28 is varied to 108 or 109 degrees , while the bond length is fixed to the original value.,width=453 ]
when the systems exhibit some three dimensionality , the above mentioned variation of the fermi surface configuration against the bond angle depends on @xmath2 .
this is shown schematically in fig.[fig7 ] . in systems with moderate three dimensionality ,
the orbital character change of the @xmath18 fermi surface and its disappearance first starts at @xmath44 ( around z point ) as the bond angle is reduced , and ends at @xmath16 ( @xmath45 point ) .
namely , the @xmath18 fermi surface becomes three dimensional before it disappears completely . in this case , the above mentioned frustration effect should be present around the top and the bottom portions of the three dimensional fermi surface . in fact ,
the above mentioned three dimensionality of the band structure is rather common for the iron - based superconductors .
this has been discussed in detail in ref.@xcite . in fig.[fig8 ] , we show the band structure of various iron - based superconductors for @xmath16 and @xmath44 planes ( in the original folded brillouin zone ) calculated by using the wien2k package@xcite .
one can see that , except for lafeaso , the two bands degenerate at z point @xmath46 repel with each other as the wave vector moves away from z in the @xmath44 plane .
this means that configurations ( c ) or ( d ) in fig.[fig3 ] are realized for a certain range of @xmath2 . on the other hand , in fese , lifeas and bafe@xmath31as@xmath31 the band structure near the @xmath16 plane takes configuration ( b ) , so that in these materials , the band structure near the fermi level is indeed three dimensional .
conversely , the 1111 systems can be considered as somewhat exceptional in that , although the three dimensional @xmath35 band does exist , it does not come down too rapidly before the @xmath35 originated @xmath17 fermi surface appears .
in fact , ndfeaso , one of the materials having the highest @xmath3 , seems to have the least frustration from the above mentioned viewpoint .
namely , as seen in the band structure shown in fig.[fig9 ] in the unfolded brillouin zone@xcite , the @xmath17 fermi surface is present but the three dimensional @xmath35 band still lies above the @xmath34 bands for all @xmath2 . on the other hand , from the comparison to the band structure of lafeaso ( fig.[fig1](b ) ) , it can be seen that as a trade - off for the appearance of the @xmath17 fermi surface around @xmath25 , the three dimensional @xmath35 band along @xmath47-@xmath46 has certainly come down very close to the @xmath34 band , and for smaller bond angle or higher pnictogen position , the band reconstruction and thus the frustration starts to take place .
therefore , from the present viewpoint , the `` sweet spot '' for high @xmath3 is restricted to a narrow window , which may explain the experimental observations@xcite . .
the `` sweet spot '' is restricted to the regime where the @xmath35 originated @xmath17 fermi surface is effective and the @xmath18 fermi surface has @xmath34 character for all @xmath2 .
, width=377 ] and @xmath44 planes are shown . the red solid circles ( or elipse )
indicates the portion where the two bands degenerate at wave vector @xmath48 repel with each other as the wave number increases , while the dashed blue cirles are the portions where the two bands degenerate at ( 0,0 ) both form a hole fermi surface . for lafeaso ,
a typical example for 1111 systems , the three dimensional band along @xmath45-z having @xmath35 character ( green dashed elipse ) stays above the degenerate @xmath34 bands ( red solid).,width=377 ] fermi surface is present around the wave vector @xmath25 , and ( ii ) the three dimensional @xmath35 band from @xmath47 to @xmath46 does not intersect the @xmath34 bands for all @xmath2.,width=377 ]
as described in the present paper , the multiplicity of the fermi surfaces and even their orbital characters change as the lattice structure is varied .
because of the presence of multiple fermi surfaces and the repulsive pairing interaction mediated by the spin fluctuations , frustration can arise in the sign of the gap on the disconnected fermi surfaces . this can result in nodal structures in the gap function , and should generally work destructively against superconductivity .
conversely , from the present viewpoint , the optimal situation for the spin fluctuation mediated superconductivity is when the @xmath35 @xmath17 fermi surface is effective and the @xmath18 fermi surface has @xmath34 orbital character for _ all _ @xmath2 . as far as the first principles band calculations are concerned , this situation is realized in limited materials like ndfeaso and smfeaso , where the highest @xmath3 among the iron based superconductors is observed experimentally . from this viewpoint
, further band structure calculation studies may give useful information for obtaining new related superconductors with higher @xmath3 .
we acknowledge c.h .
lee , t. miyake , o.k .
andersen , h. mukuda , h. kinouchi , and y. kitaoka for illuminating discussions .
99 y.kamihara , t.watanabe , m.hirano , and h. hosono , j. am .
chem . soc . * 130 * , 3296 ( 2008 ) .
i.i . mazin , d.j .
singh , m.d .
johannes , and m.h .
du , phys .
* 101 * , 057003 ( 2008 ) .
k.kuroki , s.onari , r.arita , h.usui , y.tanaka , h.kontani , and h.aoki , phys .
* 101 * , 087004 ( 2008 ) .
k. kuroki and r. arita , phys .
b * 64 * , 024501 ( 2001 ) .
k. kuroki , t. kimura and r. arita , phys .
b * 66 * , 184508 ( 2002 ) .
h. usui and k. kuroki , phys .
b * 84 * , 024505 ( 2011 ) .
s. baroni , a. dal corso , s. de gironcoli , p. giannozzi , c. cavazzoni , g. ballabio , s. scandolo , g. chiarotti , p. focher , a. pasquarello , k. laasonen , a. trave , r. car , n. marzari and a. kokalj , http://www.quantum - espresso.org/. n. marzari and d. vanderbilt , phys .
b * 56 * , 12847 ( 1997 ) ; i. souza , n. marzari and d. vanderbilt , phys . rev .
b * 65 * , 035109 ( 2001 ) .
the wannier functions are generated by the code developed by a. a. mostofi , j. r. yates , n. marzari , i. souza and d. vanderbilt , ( http://www.wannier.org/ ) .
d.j.singh and m.h .
du , phys .
* 100 * , 237003 ( 2008 ) .
v. vildosola , l. pourovskii , r. arita , s. biermann , and a. georges , phys . rev .
b * 78 * , 064518 ( 2008 ) .
s. lebegue , z.p .
yin , and w.e .
pickett , new j. phys . * 11 * , 025004 ( 2009 ) .
o.k . andersen and l. boeri , annalen der physik , * 1 * 8 ( 2011 ) .
p.m. shirage , k. kihou , c.h .
lee , h. kito , h. eisaki , and a. iyo , applied physics letters * 97 * 172506 ( 2010 ) .
t. miyake , t. kosugi , s. ishibashi , and k. terakura , j. phys .
. jpn . * 79 * , 123713 ( 2010 ) .
t. kosugi , t. miyake , s. ishibashi , and k. terakura , j. phys .
. jpn . * 81 * , 014701 ( 2012 ) .
k. kuroki , h. usui , s. onari , r. arita , and h. aoki , phys .
b * 79 * , 224511 ( 2009 ) .
f. wang , h. zhai , and d .- h .
lee , phys .
b * 81 * , 184512 ( 2010 ) .
r. thomale , c. platt , w. hanke , and b.a .
bernevig , phys .
lett . * 106 * , 187003 ( 2011 ) .
s. graser , t. a. maier , p. j. hirschfeld , and d. j. scalapino , new j. phys . * 11 * , 025016 ( 2009 ) .
bickers , d.j .
scalapino , and s.r .
white , phys .
lett . * 62 * , 961 ( 1989 ) .
p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , j. luitz in wien2k , an augmented plane wave + local orbitals program for calculating crystal properties , ( karlheinz schwarz/ techn .
universit@xmath49 t wien , wien , austria , 2001 ) .
lee , a.iyo , h. eisaki , h. kito , m.t .
fernandez - diaz , t. ito , k. kihou , h. matsuhata , m. braden , and k. yamada , j. phys .
soc . jpn.*77 * , 083704 ( 2008 ) . y. mizuguchi and y. takano , j. phys .
. jpn . * 79 * , 102001 ( 2010 ) and references therein .
|
in the present paper , we describe how the band structure and the fermi surface of the iron - based superconductors vary as the fe - as - fe bond angle changes .
we discuss how these fermi surface configurations affect the superconductivity mediated by spin fluctuations , and show that in several situations , frustration in the sign of the gap function arises due to the repulsive pairing interactions that requires sign change of the order parameter .
such a frustration can result in nodes or very small gaps , and generally works destructively against superconductivity .
conversely , we propose that the optimal condition for superconductivity is realized for the fermi surface configuration that gives the least frustration while maximizing the fermi surface multiplicity .
this is realized when there are three hole fermi surfaces , where two of them have @xmath0 orbital character and one has @xmath1 _ for all @xmath2 _ in the three dimensional brillouin zone .
looking at the band structures of various iron - based superconductors , the occurrence of such a `` sweet spot '' situation is limited to a narrow window .
|
eta carinae has a peculiar habit of settling into a photometric trend and spectroscopic state long enough for observers to grow accustomed to it , and then abruptly changing .
dramatic transitions occurred around 1840 , 1890 , and 1945 , see @xcite and refs . therein ; and there are hints that another may have begun in the late 1990 s . for almost 50 years after the 1940 - 1950 change , the star plus ejecta brightened at an average rate of 0.025 visual magnitude per year , largely due to expansion of the dusty homunculus nebula which had been ejected around 1840 . essentially a bipolar reflection or scattering nebula , the homunculus dominated @xmath0 car s total brightness throughout the 20th century .
after 1997 , however , observations with the hubble space telescope ( hst ) revealed that _ the central star _ was now brightening far more rapidly @xcite . despite brief declines during the 1998.0 and 2003.5 spectroscopic events , its average rate from 1998 to 2006 was about 0.15 mag / yr . this was a recent development , since earlier speckle data @xcite show that the star had not brightened that quickly in the fifteen years before 1997 .
ground - based photometry of the homunculus after 1997 showed an increase that was less dramatic , but which exceeded any of the fluctuations seen between 1955 and 1995 .
most likely the circumstellar extinction began to decrease rapidly in the mid-1990 s ; perhaps the rate of dust formation near the star had decreased , or dust was being destroyed , or both .
either case requires some undiagnosed change in the dense stellar wind @xcite .
thus , suspecting that another `` change of state '' was underway , we were surprised when ground - based photometry of @xmath0 car plus the homunculus showed a 0.3-magnitude _ decrease _ in brightness between august 2006 and march 2007 @xcite .
, note that the decline may have begun in the season that is unfavorable for observing @xmath0 car .
it may have started as early as august or as late as december of 2006 . ]
the change occurred at all bvri wavelengths and initially resembled the behavior seen during a spectroscopic event ( fig .
[ photom ] ) .
this was the largest decline in @xmath0 car s groundbased light curve event in at least the past 50 years , see @xcite , @xcite , and refs . therein .
the integrated photometry of the homunculus and central star vary in parallel @xcite . therefore ,
if this change proved to be a reversal of the previous several years behavior , then one might expect the central star to have faded by more than half a magnitude in 20062007 , largely invalidating our conjectures about the future trend @xcite .
thus it was important to obtain fresh hst photometry of the central star .
fortunately this was already a goal of a small existing program outlined in section 2 below .
as usual in @xmath0 car research , the results differ from any scenario proposed beforehand .
the photometric decline soon leveled off , but our hst photometry strongly suggests that it was _ not _ merely a temporary reversal of the previous brightening trend ( 3 ) . suspecting that the photometric reversal might signal unforeseen spectroscopic changes in the stellar wind
, we obtained data with the gemini multi - object spectrometer ( gmos ) in 2007 which can be compared with earlier data from the hst s space telescope imaging spectrograph ( stis ) in 19982004 , and with vlt / uves spectra from 2006 prior to the photometric decline .
the spectroscopy revealed only subtle changes halfway through @xmath0 car s 5.5-year spectroscopic cycle , possibly related to the latitude - dependent morphology of the stellar wind .
any interpretation of @xmath0 car s recent behavior must take into account its 5.54-year spectroscopic cycle ; for general information see @xcite , @xcite , and many references cited therein .
most authors agree that the 1998.0 , 2003.5 , and 2009.0 `` spectroscopic events '' probably occurred at periastron passages of a hypothetical companion star in a highly eccentric orbit . at times
more than 7 months from such an event , the two stars are 15 to 30 au apart seemingly too distant for the secondary star to significantly influence the appearance of the primary wind in the observations discussed here . although @xcite have suggested that even at apastron the ionizing radiation from the secondary may not be negligible .
at least _ in most scenarios proposed so far , _
one expects the 20062007 changes to signify structural effects in the primary wind more than wind wind interactions . in the next section
we describe the photometric and spectroscopic observations including a description of the differences between ground - based and hst / stis spectra and the complications that must be accounted for when comparing them . in [ three ] we discuss the observed variations in @xmath0 car in 20062007 , mid way in its 5.5 yr spectroscopic cycle , and compare them with spectra from the same phase in the previous cycle ( 20002002 ) .
the last section summarizes the observations with a brief discussion of their possible origin .
normal ground - based images of @xmath0 car do not consistently separate the star from the inner parts of the homunculus nebula , which has structure on all scales from 0.1 to 10 . until recently the homunculus appeared brighter than the star , though the ejecta - to - star brightness ratio has evolved substantially in the past 20 years @xcite .
precise ground - based photometry generally includes the entire homunculus plus the star .
hst images , however , have good enough spatial resolution to show the brightness of the star itself with only minor contamination by the ejecta located at @xmath1 .
in recent years we have obtained a number of images of this type , to monitor the central star s brightness and detect changes in the surrounding ejecta .
we obtained hubble space telescope acs / hrc images of @xmath0 car in the f250w and f330w filters on 2006 august 4 ( mjd 53951.1 ) , near peak brightness , and 2007 january 20 ( mjd 54120.4 ) , during the unexpected decline .
the central wavelengths of these broad - band filters are near 250 and 330 nm respectively .
all of the images were reduced , measured , and calibrated as described in @xcite . for brightness measurements we used a weighted virtual aperture centered on the star , with diameter 0.3 ( about 10.7 acs / hrc pixels ) .
the aperture weighting function was @xmath2 with @xmath3 = 0.15 .
= 0.3 , which was really the diameter @xmath4 .
@xcite stated this detail correctly . ]
see also @xcite for aperture size effects in acs images of @xmath0 car .
shortly after our january 2007 observations the acs electronics failed .
we then used the older hst / wfpc2 camera on 2007 august 23 ( mjd 54335.1 ) and 2008 february 14 ( mjd 54510.8 ) , with filters f255w and f336w which are comparable to the acs / hrc filters .
these images were reduced using the standard stsci data reduction pipeline . a correction factor of 0.9915
was applied to the raw counts to compensate for geometric pixel distortion @xcite , and the photflam keyword induced a pipeline conversion from raw counts to flux values .
we measured the brightness of the central star in the wfpc2 data , using the same 0.3 weighted virtual aperture ( @xmath5 6.6 pc2 pixels ) as for the acs / hrc data .
we estimated the aperture correction based on two 12th - magnitude stars in the images , well separated from the homunculus , by comparing their fluxes in the virtual aperture to those in a standard 1 measurement aperture corrected to infinite size @xcite .
the resulting geometrical transmission factors are listed in table [ caltab ] . to assess the effect of a known `` red leak '' in the wfpc2 f336w filter transmission curve , we convolved an hst / stis spectrum of @xmath0 car with the response function for the planetary camera with this filter .
the flux at @xmath6 nm contributed only 0.94% of the counts , negligible for our purposes .
the stmag photometric system is calibrated for direct comparison of fluxes measured with similar filters in different instruments @xcite .
we tested the calibration by convolving the instrument and filter response functions with hst / stis spectra of @xmath0 car and comparing the results for the acs / hrc f250w and f330w filters to those for the wfpc2 f255w and f336w filters .
the computed values agreed to four significant digits , giving us confidence that comparisons between our acs / hrc and wfpc2 results are valid .
our experience with the large existing hst data set for @xmath0 car , 19912009 , has never contradicted this assessment to a serious degree , despite the spectral forest of emission lines and strong uv absorption .
the acs / hrc and wfpc2 photometry are listed in table [ phottab ] , and plotted in figure [ photom ] together with our previous measurements @xcite .
we obtained ground - based slit spectroscopy of @xmath0 car with the gemini multi - object spectrograph ( gmos ) on the gemini south telescope in late june and early july 2007 , just after the decrease in brightness .
the parameters of these observations and subsequent gmos observations in february and july 2008 are listed in table [ spectab ] .
the parameters of the spectrograph are given in table [ instab ] .
we prepared 2-d spectrograms with the standard gmos data reduction pipeline in the gemini iraf package and extracted 1-d spectra via a routine developed earlier for use with hst / stis @xcite ( figure [ fullspec ] ) . at each wavelength
our software integrates the counts along a line perpendicular to the dispersion , weighted by a mesa - shaped function centered on the local spectral trace .
for these data we used a mesa function with base - width = 11 pixels and top - width = 7 pixels , about 0.8 and 0.5 respectively .
the seeing was roughly 1.0 , so each gmos spectrum discussed in 3 below represents a region about 1 across .
hst / stis spectra of @xmath0 car had been obtained in october 2001 and january 2002 , at approximately the same phase in its 5.54-year spectroscopic cycle @xcite .
we reduced them using a modified version of the goddard calstis reduction pipeline described in @xcite .
the spectra are listed in table [ spectab ] and the instrument parameters are given in table [ instab ] .
the january 2002 data most closely match the spectroscopic phase of july 2007 .
of course the spatial resolution of gemini , limited by atmospheric seeing , is greatly inferior to hst / stis and the inner ejecta are unavoidably mixed with the spectrum of the star .
in particular our gmos spectra include the `` weigelt knots '' 0.3 northwest of the star .
fortunately , by 2007 the star had become substantially brighter than the nearby ejecta unlike the case 10 or 20 years earlier @xcite .
the 1-d stis spectra used in this paper were prepared with an extraction width of 0.20 , which excludes most of the narrow emission from the inner ejecta ; see important comments in subsection [ two.three ] .
moreover , the slow - moving inner ejecta produce narrow emission lines which are distinguishable from the broad stellar wind lines ; typical widths are of the order of 20 and 400 km s@xmath7 respectively . in the wavelength range of interest , the spectral resolution of gmos is roughly 75 km s@xmath7 while stis provided about 40 km s@xmath7 .
the narrow lines are therefore more blurred in the gemini data while the broad stellar wind features and their p - cygni absorption components are well resolved by both instruments .
we have smoothed the stis data plotted in section [ three ] to match the resolution of gmos .
h@xmath8 emission is so bright in @xmath0 car that it saturates the detector pixels in even the shortest gmos exposures centered on the star .
we attempted to use a narrower slit , offset from the star by 0.6 to sample the wings of the star s p.s.f .
this technique produced an unsaturated h@xmath8 profile , but it is probably too contaminated by surrounding ejecta to be trusted ; see subsection [ two.three ] .
unfortunately @xmath0 car was not observed with gemini / gmos in 20052006 before the photometric fading .
the eso science archive does have spectra obtained with the vlt / uves instrument in 2006 car from 2002 to 2005 are available at http://etacar.umn.edu/. the uves spectral sequence is also described by @xcite , but they did not note the potential problems with the later uves spectra . ] .
we reduced the 2006 spectra with the standard uves pipeline available from eso .
the spectra were extracted using a mesa function 3 by 2 pixels wide , about 0.75@xmath9 by 0.5@xmath9 .
the seeing was 0.8 to 0.9@xmath9 so this extraction corresponds to about 1@xmath9 on the sky .
none of the 2006 spectra are concurrent with the decrease .
the closest in time is from june 2006 but internal clues suggest that the slit was not centered on the star .
the spectra do not closely resemble earlier uves observations ( 2002 - 2004 ) which were definitely centered on the star ; the stellar - wind features are weaker and the narrow emission lines are much weaker than what one expects to see in a ground - based spectrum .
the published position and acquisition image for the june 2006 spectrum also suggest that the slit might be offset from the star .
the june 2006 uves spectrum thus appears unsuitable for our purposes here .
the narrow emission lines are also weaker than expected in the two spectra from 2005 and the april 2006 spectrum and weaker than in the 2007 gmos spectra .
this may also be due to pointing differences with respect to the earlier uves spectra and the gmos spectra .
although the april 2006 spectrum is referenced in the later discussion , it is used here with reservations for this reason . provided the extraction size is less than 0.3
, an hst spectrum of @xmath0 car shows the stellar wind with only slight contamination .
ground - based spectroscopy , on the other hand , with seeing of the order of 1 , also includes the spectrum reflected by dust at @xmath10 0.15 to 1 , plus narrow emission lines created in that region .
since reflection usually has little effect on the strengths of spectral features relative to the continuum , one might expect a ground - based spectrum of @xmath0 car ( i.e. , central object plus reflected spectrum plus narrow - line emission ) to closely resemble an hst spectrum , supplemented by the narrow lines .
in fact , however , the earliest hst spectra showed that _ this is not the case _
ground - based data show paradoxically weaker balmer emission lines .
[ stisuves ] shows how conspicuous this effect is .
here we use vlt / uves to exemplify a modern ground - based instrument , because @xmath0 car was observed with stis and uves on 2003 february 12 and 14 respectively , only 1.2 days apart .
we show the h@xmath11 feature because it plays a leading role in 3.2 .
h@xmath11 emission in the uves data has equivalent width 26 compared to 33 in the stis spectrum ( measured the same way in both data sets , with no correction for narrow lines ) .
old hst / fos results and various ground - based data suggest that this difference is real , not merely an instrumental peculiarity of either stis or uves @xcite
. why does the difference occur ?
wavelength - dependent balmer absorption within the reflecting dust - gas mixture seems implausible , because each stellar - wind emission line is much broader than the velocity dispersion in the relevant material .
a more appealing explanation is that the spectrum `` seen by '' the dust intrinsically differs from what we see directly in other words , the stellar wind spectrum depends on the direction from which it is viewed .
@xcite explored this idea via observations farther out in the homunculus nebula , and concluded that @xmath0 car s wind is strongly latitude - dependent . note that the dusty `` weigelt knots , '' which contribute strongly to any ground - based spectrum of the central object , are thought to be near the equatorial plane while our direct viewpoint is around 45@xmath12 latitude e.g. davidson et al ( 2001 ) .
this explanation is not entirely satisfying .
it implies that the equivalent widths of balmer lines are substantially smaller at equatorial latitudes , whereas @xcite and @xcite suggested the same for _ polar _ directions .
apparently hst s high - spatial - resolution direct view of the stellar wind gives larger equivalent widths than we find in the reflected spectrum anywhere in the ejecta .
( this statement probably applies to broad emission features in general , not just balmer lines . ) furthermore , the uves vs. stis comparison mentioned above seems to require a surprisingly large latitude variation in the h@xmath11 equivalent width , almost a factor of two .
thus we fear that some other , more subtle process may be altering the reflected spectra .
this problem merits further study .
independent of what causes the discrepancies , for this paper the practical implication is that _
one must be very careful when comparing ground - based spectra of @xmath0 car to the hst / stis data .
_ in particular we can not directly compare the equivalent widths .
moreover , the offset h@xmath8 observations mentioned in 2.2 are not useful here because they are most likely affected by reflection and emission in their spatial locales . in
4 below we compare recent gemini / gmos observations to earlier hst / stis results , because neither data set spans two spectroscopic cycles ; nor is any other existing data set suitable for this problem . in those comparisons
, we exercise the necessary caution and caveats .
in recent years relatively little attention has been given to @xmath0 car s behavior between spectroscopic events .
at such times the companion star must be located 1535 au from the primary .
mid - cycle changes at uv - to - red wavelengths are therefore significant for at least one of two reasons : ( 1 ) proposed explanations for the 5.54-year cycle generally do not predict irregular or short - term effects more than a few months from an event . if they do occur and are related to the cycle , then some new factor must be taken into account . ( 2 )
if , on the other hand , such effects are not related to the cycle , then they may represent phenomena in the primary wind .
observations in 19992002 and 20052007 therefore merit attention .
here we are concentrating on 20062007 , supplemented by a few data from 20002002 for comparison purposes . to avoid misunderstandings , note that continuous _
systematic _ changes occur in mid - cycle , not directly related to the primary star . @xcite
have shown that some quasi - nebular spectral lines emitted close to our line of sight to the star gradually brighten to a maximum near mid - cycle and then fade . @xcite earlier found a similar result for one line of this type in an unresolved set of emission regions . the spectral features in question ( [ ] , [ ] , [ ] , etc . )
probably indicate photoionization by the hot secondary star , and their behavior in the mehner et al .
analysis may represent the cyclic variation of column density between us and the orbiting secondary star .
the most popular class of orbit - and - wind models can probably be adjusted to match this effect .
fluctuations described below , however , differ in several respects .
they directly concern the spectrum of the primary stellar wind rather than nearby ejecta ; they do nt have a smooth obvious correlation with phase in the 5.5-year cycle ; and each of them occurred in only one of two observed cycles .
the hst images confirm that the central star became fainter during 20062007 ( fig .
[ photom ] ) , but not as much as the ground - based photometry had led us to expect . from 1997 to 2006
the star s apparent magnitude had brightened three times faster than the surrounding homunculus , but _
its relative fading in 2006 was only about the same as for the homunculus .
_ whatever the explanation is , this detail indicates that the change in 20062007 was not a simple reversal of the 19982006 trend
. moreover , the decline stopped and the upward brightening trend resumed after the 2009 spectroscopic event , leaving more or less intact our earlier speculations about secular changes @xcite .
the f330w filter showed a much steeper decline than f250w .
this is interesting because the former samples balmer continuum emission plus other continuum , whereas f250w includes many strong absorption features . during a spectroscopic event such as 1998.0 or 2003.5 ,
these absorption lines strengthen so much that the star becomes faint in the 230290 nm wavelength range @xcite .
this is why fig .
[ photom ] shows a substantial temporary decrease in f250w during the 2003.5 event , with scarcely any corresponding effect in f330w .
evidently the 2006 - 2007 fading was quite different .
the same figure suggests two possibilities : ( 1 ) the _ increased _ f250w
/ f330w flux ratio in 2007 does not favor an explanation based on increased extinction by circumstellar dust .
( 2 ) perhaps the balmer continuum emission intrinsically weakened in 2006 .
the ground - based b , v , r , and i magnitudes all faded by about @xmath13 in 20062007 @xcite .
while @xmath0 car s b and r values are strongly influenced by the very bright h@xmath14 and h@xmath8 emission lines , v almost entirely measures continuum brightness .
unfortunately the central star could not be measured with hst / wfpc2 at these wavelengths , because every wfpc2 filter redward of 400 nm either saturates the central pixels in the shortest allowed exposure time , or else is affected by some particular emission line .
motivated by the visual decline , its duration , and the peculiar uv photometry from hst / wfpc2 , we obtained the gemini / gmos data described in
[ specsec ] to learn whether the spectrum had changed during the brightness decrease . unfortunately no truly suitable earlier ground - based spectra are available for comparison particularly none from 2005 or 2006 . instead , therefore , we used stis observations made in october 2001 and january 2002 , at approximately the same phase of @xmath0 car s 5.54-year spectroscopic cycle as the gmos data .
there is no _ a priori _ reason to think that the 2006 photometric change was related to the cycle , but on the other hand it would be imprudent to use observations at much different phases as comparisons . if , as usually assumed , the cycle is modulated by a hot companion star in an eccentric orbit , then fig .
[ orbitfig ] shows the approximate orbital location of the secondary star when our spectra were obtained .
is qualitatively valid for nearly all models that have been proposed .
see @xcite and refs . therein . ] in this part of the orbit the two stars are more than 20 au apart and their motion is slow .
the most conspicuous spectral features in 2007 ( fig .
[ fullspec ] ) closely resembled the 20012002 data .
consider for instance the brightest balmer lines . since h@xmath8 is overexposed in the gmos data and h@xmath15
is confused with other features in this object , in fig .
[ balmer ] we show h@xmath14 and h@xmath11 .
the measured equivalent widths for h@xmath11 turn out to be like those mentioned in 2.3 : 26 in the 2007 gmos spectrum and 33 in the 20012002 stis data . unlike the earlier examples ,
however , these values have been corrected for weak superimposed narrow lines .
the internal uncertainties chiefly systematic rather than statistical are probably less than @xmath16 2 .
car is hard to define and weaker additional emission lines contribute to the fluxes , these results depend on the measurement protocol .
we measured `` continuum '' fluxes near 4080 and 4160 and we integrated the net h@xmath11 flux from 4082 to 4116 ( vac ) .
the correction for unrelated emission on the long - wavelength wing is small , about @xmath17 in the gmos data and less for stis . in the last stis data in early 2004
the equivalent width for h@xmath11 was only about 29 , but lingering effects of the 2003.5 spectroscopic event may have reduced it below `` normal '' @xcite .
regarding earlier epochs , the first hst spectra gave an e.w .
of 35 for h@xmath11 in 1991 @xcite , or probably about 33 if we could apply the corrections mentioned above . ]
if we make allowances for the effect described in 2.3 , evidently h@xmath11 was essentially as strong in 2007 ( relative to the continuum ) as it had been 5.5 years earlier . in the february and
july 2008 gmos spectra are 25 each and 26 in the april 2006 uves spectrum . ]
the h@xmath14 equivalent width was about 163 , 175 , and 156 , respectively , in the 2001 and 2002 stis data and the 2007 gmos data ; this line may be less accurate than h@xmath11 because its very large line / continuum flux ratio is disadvantageous for the measurements .
but the h@xmath11 profile does show one potentially significant difference : weak p - cygni absorption was present near @xmath18 km s@xmath7 in 20012002 and in the april 2006 uves spectrum , but not in 2007 .
this is not merely a result of different spectral resolutions , because the stis data in fig .
[ balmer ] have been smoothed to match the gmos resolution .
the difference in spatial resolution is probably not responsible either , since the stis vs. uves comparison in fig .
[ stisuves ] shows at most a very weak p - cygni feature in the ground - based data rather than stis .
( for reasons why that might occur , see @xcite along with 2.3 above . ) the gmos spectrum and h@xmath11 profile from february 2008 is essentially identical to the 2007 spectrum , while its july 2008 spectrum shows the reappearance of weak p cygni absorption in h@xmath11 prior to the onset of the 2009 event ( figure [ gmoshdel ] ) .
although the spectroscopic analysis by @xcite pertained to the hst / stis spectrum only one year after the 1998.0 event , they showed that the hydrogen emission originates predominantly in the outer wind of the primary star .
altogether , then , the p - cygni difference is a real effect , meaning that the column density of excited hydrogen atoms along our line of sight was less after the photometric decline .
a parallel , but more dramatic , difference occurred in some of the helium features .
helium lines whose lower levels are 1s2s @xmath19s , 1s2s @xmath20s , or 1s2p @xmath20p@xmath21 show prominent p - cygni absorption in @xmath0 car s mid - cycle spectrum ; good examples are the triplet features at @xmath224026,4472,4713 .
qualitatively , one expects such absorption to occur in a dense photoionized he@xmath23 zone where recombination strongly populates the metastable 1s2s levels @xcite . such a zone is thought to exist in those parts of @xmath0 car s primary wind that are closest to the hot secondary star before encountering the wind - wind shocks ; see section 6 of @xcite and refs . therein . if fig .
[ orbitfig ] is more or less valid , that zone was situated between us and the primary wind in 2001 - 2002 and 2007 ; so p - cygni absorption is unsurprising . for our purposes here , the interesting point is that _ the helium p - cygni velocities differed between those two cycles . _
[ hebalmer ] shows @xmath244027 . in 20012002
its p - cygni feature was centered at @xmath25 km s@xmath7 like that of h@xmath11 , but in 2007 the absorption had broadened and shifted by about 100 km s@xmath7 to smaller ( i.e. , less negative ) average velocities .
this most likely indicates a quantitative change in the primary wind structure .
similar effects occurred in the other helium lines .
the p cygni absorption is also stronger in this line than the corresponding emission .
we also note that the he i profile has continued to shift to even lower velocities in the february 2008 gmos spectrum , but it has shifted bluewards , back to @xmath26 km s@xmath7 , in the july 2008 spectrum ( fig [ heallgmos ] ) .
interestingly , the @xmath244027 p cygni profile in the april 2006 uves high resolution spectrum appears to have two absorption components centered at velocities of @xmath27 and @xmath28 km s@xmath7 ( fig [ hegmosuves ] ) .
it is uncertain however , if the uves slit is centered on the star or intercepting the same scattered and reflected light from the surrounding ejecta as the gmos spectra .
we plan to model the behavior of the he i lines during @xmath0 car s spectroscopic cycle , but this is beyond the scope of this paper .
by now it is clear that @xmath0 car is not a simple 5.54-year clock - like mechanism .
hst / stis data show a partially different set of changes in the previous mid - cycle interval , 20002001 @xcite .
for example , p cygni absorption in h@xmath8 , usually absent in a mid - cycle spectrum of the star @xcite , showed a marked increase in absorption between march 2000 , when there was no detectable p cygni absorption , and october 2000 .
its appearance was brief since it was gone by january 2001 .
unfortunately the october 2000 stis data only covered h@xmath8
so we could not compare the profiles of any of the other balmer features or he i triplet lines . at the same time , there was a significant increase in the strength of the narrow -140 km s@xmath7 absorption feature in h@xmath8 .
the lower spectroscopic resolution of gmos makes it difficult to determine if the narrow @xmath29 km s@xmath7 feature was present in 2007 .
the variable emission feature near 6307 reaches its maximum strength mid - cycle , in 20012002 and again in 2005 @xcite . in the vlt / uves spectra from 2005 it had an equivalent width more than twice that previously observed .
it is also present in the april 2006 uves spectrum and june 2007 gmos spectrum at 0.2 0.3 equivalent width , equal to its typical strength reported 19992002 between spectroscopic events .
it is slightly stronger in the april 2006 spectrum relative to the june 2007 spectrum .
we note that an increase in the uv flux from the central star occurs in 20012002 and 20052006 and corresponds with the maximum strength of the 6307 line .
there was at least one episode of mid - cycle activity in each of the last two spectroscopic cycles .
each episode was different and each occurred at a different phase relative to the spectroscopic events . in the next section
we summarize the mid - cycle variability and discuss probable variations in the primary star s wind that would account for these changes .
the mid - cycle variability in 2006 - 2007 described in 3 occurred when the secondary star was at its greatest distance from the primary and most likely reveals basic fluctuations of @xmath0 car s primary stellar wind , independent of the companion star and the 5.5-year cycle .
this is the _
simplest _ view for reasons outlined below , and it makes sense in relation to several known or probable facts . for the first time in 160 years , the central star now strongly and directly affects ground - based photometry .
the homunculus nebula dominated the total apparent brightness throughout the 20th century , see refs . in @xcite ; but since 1998 the star has been brightening rapidly and now accounts for half or more of the integrated light .
this development may explain why ground - based magnitudes ( i.e. , homunculus plus star ) have recently been fluctuating with greater amplitude than in the 19531995 record . see for example , the recent light curve at http://etacar.fcaglp.unlp.edu.ar .
apparently our direct view of the star shows stronger variability than does the average of many viewpoints reflected in the homunculus , though other factors should not be ignored .
since @xmath0 car s spectroscopic events have attracted attention in recent years , there is a natural tendency to ascribe observed variability to the 5.5-year cycle , the wind - wind interaction , and the influence of the companion star .
but in this case there are good arguments against that response .
according to nearly all proposed models for the orbit , in 20062007 the wind - wind interaction should have been minimal because the companion star was then located far from the primary . during a 3-year interval around apastron ,
the separation is more than 20 au ( fig .
[ orbitfig ] ) .
this implies the following points for conventional models : 1 .
tidal effects are weaker than those at periastron by a factor of order 500
if the primary mass loss rate and line - of - sight wind speed are roughly @xmath30 yr@xmath7 and wind speeds along the line of sight typically 500600 km s@xmath7 , then the optical depth for thomson scattering at @xmath31 au is small , of the order of 0.1 .
optical depths for continuum absorption are expected to be even smaller .
a detailed model is beyond the scope of this paper , but this assertion is based on two generalities : standard processes give @xmath32 at the likely densities and temperatures , and one can also judge the matter from the sizes of photospheres that produce suitable continua .
the visual and near - uv photosphere in the primary wind can not be very much larger than the circle for the primary in fig .
[ orbitfig ] , @xmath33 au ; otherwise its emergent continuum would be too cool .
therefore the optical depth for continuum absorption must be quite small at @xmath34 au . in conventional models
the broad - emission - line spectrum originates mostly within @xmath35 au @xcite , where the secondary star and wind - wind interaction can have at most only a minor indirect influence .
the radiation density is too high for appreciable circumstellar dust in this region .
likely grain formation distances are 150300 au @xcite . @xcite
have suggested that dust can form close to the star , but also note that more than 170 days from periastron dust can form only at large distances . in any case
as explained earlier dust , formation can not explain the photometric decline .
5 . along relevant photon paths ( see point 6 below ) , optical depths through the fast , hot , low - density secondary wind are even smaller than those in the primary wind .
the radiation density is too high for appreciable circumstellar dust in this region .
likely grain formation distances are 150300 au @xcite .
when viewing fig . [ orbitfig ] , it is important to bear in mind the 45 inclination of the orbit plane . at 2008.54 , for instance , our line of sight to the primary star did not pass close to the secondary star even though the figure might give that impression .
between 2005.5 and 2008.0 our projected view of the system would have changed slowly and continuously at least in conventional models .
the same statement applies to expected line - of - sight column densities .
8 . at visual and near - uv wavelengths
, the shocked region between the two winds should be practically transparent for absorption and line formation .
its high temperatures , @xmath36 k , greatly weaken free - free and other absorption opacities at these wavelengths .
figures depicting shock simulations such as @xcite , @xcite , and @xcite do not depict the visual appearance ; indeed the shocked regions are expected to emit less than 0.01% of @xmath0 car s
total red - to - near - uv light . 9 .
flow simulations show dramatic spiral patterns in the shocked structure around periastron @xcite . during
most of the orbit , however , the secondary star moves at speeds less than 50 km s@xmath7 , much slower than either wind ; so the flow structure then approximates an axial symmetry around the the star star axis , with only a modest tilt angle and negligible spirality .
near periastron the colliding wind structure tends to become unstable and chaotic @xcite , but this effect is far less serious at larger distances and lower densities . 11 .
the hot secondary star is thought to ionize the pre - shock zones in the primary wind @xcite .
this probably affects the p cyg features described in 3 , but it does nt explain why the column densities and speeds should differ between two cycles . at most visual and near - uv wavelengths ,
continuum emission from this photoionized zone should be about two orders of magnitude weaker than that of the primary wind . in the balmer continuum
it might account for almost 10% of the total , but this contribution should be quite steady , and insensitive to local gas densities etc . in summary , the above points plus the persistence of the spectroscopic changes supports variations in the primary s wind as opposed to a wind - wind interaction interpretation of the mid - cycle irregularities .
we can not absolutely rule out an explanation of that type , but the obvious alternative is far simpler
i.e. , that the observations indicate fluctuations in the primary wind .
similar arguments apply to unconventional models wherein the secondary star plays the leading role , e.g. see soker ( 2007 ) and kashi and soker ( 2009a ) .
the decline in apparent brightness in 20062007 was temporary .
the flux at all wavelengths subsequently leveled off and hst / wfpc2 photometry after the 2009 event ( to be discussed in a forthcoming paper on the 2009 event ) shows that the central star resumed its decades long increase in brightness .
furthermore the multi - wavelength visual photometry , and the near - uv flux ratio ( fig .
[ photom ] ) show that increased extinction by dust can not explain the decline .
allowable explanations for the observed decrease in brightness are thus limited .
one feasible possibility is a decrease in the balmer continuum emission which could be due a change in the stellar wind , assuming a simple dependence of the continuum on the wind density .
the decrease in the h@xmath11 p cygni absorption between 2001 and 2007 suggests a decrease in density of the primary star s wind between the two cycles at about the same phase in the orbit . if the density @xmath37 of an opaque wind decreases , then its photosphere tends to move inward . at constant luminosity ( generally assumed for @xmath0 car )
, this raises the characteristic temperature t@xmath38 of the emergent radiation .
since t@xmath38 is well above 10000@xmath39 k , in this case , the brightness therefore tends to decrease at wavelengths greater than 300 nm as the radiation shifts further in the uv .
this is consistent with the observed photometric decrease in 20062007 .
crude estimates suggest that a density decrease on the order of 20% would suffice .
the change in the he i profiles also show a corresponding decrease in the wind speed .
the primary s wind thus became both slower and less dense mid - cycle .
the balmer and he i profiles show little change between the gmos mid 2007 spectrum and the spectrum from early 2008 .
the primary s wind thus stayed in this slower , less dense state up until the time of the spectrum from mid 2008 , just prior to the onset of the 2009 event .
a possible explanation for these mid - cycle changes in the stellar wind may be related to @xmath0 car s aspherical wind , confirmed with near - infrared interferometry @xcite .
our line of sight to @xmath0 car is at an angle midway between its fast , dense polar outflow and its slower , lower density wind in the equatorial region @xcite .
a small shift in @xmath0 car s latitude - dependent wind could thus account for the observed spectroscopic changes with our line of sight intercepting more of the lower latitude wind from mid 2006 to mid 2008 .
the he i double p cygni absorption profile in the 2006 uves spectrum is intriguing in this connection , but with doubts about the slit position , its interpretation is uncertain .
@xcite have suggested that small changes in the primary s wind can explain the short term variations observed in the x - ray light curve @xcite at apastron and propose that larger variations may have been responsible for the much shorter x - ray minimum during the recent 2009 event @xcite .
however we note that there were no large fluctuations in the x - ray flux during the time of the photometric decline and our observed spectral changes from @xmath40 late 2006 to at least early 2008 , although the observed 20% change in the wind velocity may result in only a 7% decrease in the x - ray flux @xcite .
we also noted mid - cycle changes in 2000 - 2002 although they were somewhat different and did not occur at the same phase in the orbit .
these small mid - cycle fluctuations may be part of a longer term secular trend perhaps related to the star s rapid brightening in the past decade as it recovers from the giant eruption .
most recent work on @xmath0 car has focused on its spectroscopic events , but mid - cycle observations merit more attention
. mid - cycle variability can provide important information on the primary star s intrinsic instability that the spectroscopic events do not .
we thank jean - rene roy , deputy director and head of science operations of the gemini observatory , for granting us director s discretionary time on short notice to obtain spectra of @xmath0 car in 2007 .
we also thank the staff and observers of the gemini south observatory in la serena for their help and support in putting together a successful observing program .
we are also acknowledge use of the aavso international database .
altamore , a. , baratta , g. b. , cassatella , a. , rossi , l. , & viotti , r. 1986 , new insights in astrophysics : eight years of uv astronomy with iue , 303 biretta , j. a. , et al .
1996 , wfpc2 instrument handbook , version 4.0 ( baltimore : stsci ) cassatella , a. , giangrande , a. , & viotti , r. 1979 , , 71 , l9 corcoran , m. f. , 2005 , , 129 , 2018 damineli , a. , conti , p.s . , & lopes , d.f .
1997 , new astronomy , 2 , 107 damineli , a. , hillier , d. j. , corcoran , m. f. , et al .
2008 , , 386 , 2330 davidson , k. , ebbets , d. , weigelt , g. , humphreys , r.m .
, hajian , a.r . ,
walborn , n.r . , & rosa , m. 1995 , , 109 , 1784 davidson , k. 1997 , new astronomy , 2 , 387 davidson , k. , & humphreys , r.m .
1997 , , 35 , 1 davidson , k. , et al . 1999a , , 118 , 1777 davidson , k. , ishibashi , k. , gull , t.r . , & humphreys , r.m .
1999b , eta carinae at the millenium , asp conf .
179 ( ed .
morse , r.m .
humphreys , & a. damineli ) , 227 davidson , k. , smith , n. , gull , t. r. , ishibashi , k. , & hillier , d. j. 2001 , , 121 , 1569 davidson , k. 2004 , stsci newsletter , spring 2004 , 1 davidson , k. , et al . 2005 , , 129 , 900 discala , g. , & jones , r. w. 2008 , observations from the aavso international database , private communication .
feinstein , a. , & marraco , h. g. 1974 , , 30 , 271 feinstein , a. 1967 , the observatory , 87 , 287 fernandez - lajus , e. , gamen , r. , schwartz , m. , salerno , n. , llinares , c. , farina , c. , amorn , r. , & niemela , v. 2003 , information bulletin on variable stars , 5477 , 1 fernndez - lajs , e. , faria , c. , torres , a. f. , schwartz , m. a. , salerno , n. , caldern , j. p. , von essen , c. , calcaferro , l. m. , giudici , f. , llinares , c. , & niemela , v. , 2009 , , 493 , 1093 gull , t. r. , et al .
2009 , , 396 , 1308 hillier , d.j . , & allen , d.a .
1992 , , 262 , 153 hillier , d.j . , davidson , k. , ishibashi , k. , & gull , t. r. 2001 , , 553 , 837 hillier , d.j . ,
gull , t. , nielsen , k. , et al .
2006 , , , 642 , 1098 hofmann , k .- h . , & weigelt , g. 1988 , , 203 , l21 holtzman , j. a. , et al .
1995 , , 107 , 156 humphreys , r. m. , & stanek , k. ( eds . ) 2005 , asp conf .
332 , the fate of the most massive stars humphreys , r. m. , davidson , k. , & koppelman , m. 2008 , , 135 , 1249 ishibashi , k. , eta carinae and other mysterious stars , asp conf .
ser . 242 ( ed . t.r .
gull , s. johansson , & k. davidson ) , 53 kashi , a. & soker , n. 2008a , 390 , 1751 kashi , a. & soker , n. 2008b new astronomy , 13 , 569 kashi , a. & soker , n. 2009a new astronomy , 14 , 11 kashi , a. & soker , n. 2009b , 701 , l59 koornneef , j. , bohlin , r. , buser , r. , horne , k. , & turhshek , d. 1986 , highlights astron .
, 7 , 833 martin , j. c. 2005 , the fate of the most massive stars , asp conf . ser . 332 , 111 martin , j. c. , & koppelman , m. d. 2004 , , 127 , 2352 martin , j. c. , davidson , k. , humphreys , r. m. , hillier , d. j. , & ishibashi , k. 2006a , , 640 , 474 martin , j. c. , davidson , k. , & koppelman , m. d. 2006b , , 132 , 2717 martin , j. c. , davidson , k. , hamann , f. , stahl , o. , & weis , k. 2006c , , 118 , 697 mattei , a. , & foster , g. 1998 , international amateur - professional photoelectric photometry communications , 72 , 53 mehner , a. , davidson , k. , ferland , g. j. , & humphreys , r. m. 2010 , , in press nielsen , k. e. , kober , g. vieira , weis , k. , gull , t. r. , stahl , o. , & bomans , d. j. 2009 , 181 , 473 oconnell , d. j. k. 1956 , vistas in astronomy , 2 , 1165 okazaki , a.t . , owocki , s.p . , russell , c.m.p . , & corcoran , m.f . 2008 , , 388 , l39 osterbrock , d.e . , & ferland , g.j .
2006 , _ astrophysics of gaseous nebulae and active galactic nebulae _ parkin , e. r. , pittard , j. m. , corcoran , m. f. , hamaguchi , k. & stevens , i. r. 2009 , , 394 , 1758 pian , e. , campana , s. , chincarini , g. , corcoran , m. f. , hamaguchi , k. , gull , t. , mazzali , p. a. , thoene , c. c. , morris , d. , & gehrels , n. 2009 , arxiv:0908.2819 pittard , j.m . , & corcoran , m.f .
2002 , , 383 , 636 sirianni , m. , et al .
2005 , , 117 , 1049 smith , n. , davidson , k. , gull , t. r. , ishibashi , k. , & hillier , d. j. 2003 , , 586 , 432 smith , n. et al , 2004 , , 605 , 405 soker , n. 2007 , , 661 , 490 sterken , c. , freyhammer , l. , arentoft , t. , & van genderen , a. m. 1999 , , 346 , l33 van genderen , a. m. , sterken , c. , de groot , m. , & burki , g. 1999 , , 343 , 847 van genderen , a. m. & sterken , c , 2004 , , 423 , l1 viotti , r. , rossi , l. , cassatella , a. , altamore , a. , & baratta , g. b. 1989 , , 71 , 983 weigelt , g. , & ebersberger , j. 1986 , , 163 , l5 weigelt , g. , kraus , t. driebe , t. et al 2007 , , 464 , 87 whitelock , p. a. , feast , m. w. , marang , f. , & breedt , e. 2004 , , 352 , 447 zanella , r. , wolf , b. , & stahl , o. 1984 , , 137 , 79 lllrrrr + + j9p603onq&54120.426&2007.055&1.2&1.189&6.212&[email protected] + j9p603opq&54120.426&2007.055&1.2&1.159&6.239 & + j9p603orq&54120.430&2007.055&1.2&1.174&6.225 & + j9p603otq&54120.430&2007.055&1.2&1.172&6.228 & + + + + ua140109m&54335.188&2007.643&0.6&1.094&6.302&6.302 + ua140209m&54510.797&2008.124&0.6&1.064&6.333&6.333 + + + + j9p603obq&54120.410&2007.055&0.4&1.336&6.085&[email protected] + j9p603odq&54120.410&2007.055&0.4&1.338&6.084 & + j9p603ofq&54120.414&2007.055&0.4&1.329&6.091 & + j9p603ohq&54120.414&2007.055&0.4&1.334&6.087 & + + + + ua140101m&54335.121&2007.643&0.1&1.273&6.138&[email protected] + ua140102m&54335.125&2007.643&0.1&1.312&6.105 & + ua140103m&54335.121&2007.643&0.1&1.253&6.155 & + ua140104m&54335.125&2007.643&0.1&1.254&6.154 & + ua140201m&54510.770&2008.124&0.1&1.154&6.245&[email protected] + ua140202m&54510.773&2008.124&0.1&1.135&6.263 & + ua140203m&54510.773&2008.124&0.1&1.151&6.247 & + ua140204m&54510.777&2008.124&0.1&1.105&6.292 & + lrrr + + s20070616s0045&2007.45&b1200&5.5 + s20070616s0046&2007.45&b1200&5.5 + s20070618s0011&2007.45&b1200&5.5 + s20070618s0012&2007.45&b1200&5.5 + s20070630s0026&2007.49&r831&3.5 + s20070630s0035&2007.49&r831&1.5 + s20070630s0040&2007.49&r831&4.5 + s20080211s0087&2008.11&b1200&5.5 + s20080211s0088&2008.11&b1200&5.5 + s20080718s0068&2008.54&b1200&5.5 + s20080718s0069&2008.54&b1200&5.5 + + + + o6ex03010&2001.75&g750 m & 6.0 + o6ex03020&2001.75&g750 m & 0.6 + o6ex03060&2001.75&g430 m & 36.0 + o6ex030e0&2001.75&g430 m & 36.0 + o6ex030f0&2001.75&g430 m & 3.0 + o6ex02010&2002.05&g750 m & 6.0 + o6ex02020&2002.05&g750 m & 0.4 + o6ex02030&2002.05&g430 m & 18.0 + o6ex020u0&2002.05&g430 m & 8.0 + o6ex020v0&2002.05&g430 m & 36.0 + + + + uves.2006 - 04 - 09&2006.27&cd#2&60.0 + t02:41:56.828 + car .
the curve marked `` aavso '' is a smoothed representation of ground - based v - magnitudes for the homunculus nebula plus the star @xcite .
other data shown here include only the central star or rather its opaque wind observed with hst in the uv .
crosses ( + ) show photometry synthesized from stis / ccd spectra near @xmath41 330 and 250 nm @xcite .
boxes ( @xmath42 ) were measured in acs / hrc images ( @xcite and this paper ) .
asterisks ( @xmath43 ) are photometry from wfpc2 images .
these hst points are stmag values @xcite .
an arrow marks the time when ground - based photometry indicated an unusual decrease in brightness .
the vertical dotted line indicates the time of the gmos spectrum discussed in this paper . ]
profile in @xmath0 car , observed with both hst / stis and vlt / uves at practically the same time in february 2003 .
the stis tracing samples a diameter less than 0.2 , while the uves spectrum represents a seeing - limited region about 1 across .
both data sets have been normalized so their continuum fluxes match near @xmath44 and 4160 , and the bottom edge of the figure corresponds to zero flux .
the spectral resolution of the uves tracing has been degraded to match that of stis , roughly 40 km s@xmath7 . ]
4027 triplet line .
as in figure [ balmer ] , the june 2007 gmos spectrum is the solid black line .
the solid gray and dashed black lines are the stis spectra from october 2001 and january 2002 respectively smoothed to the same resolution as the gmos spectrum . ] 4027 triplet line in the gmos spectra from july 2007 ( solid black line ) , february 2008 ( solid gray line ) , and july 2008 ( dashed black line ) .
note the blueward shift of the p - cygni absorption in the july 2008 spectrum . ]
4027 triplet line in the gmos spectra from july 2007 ( solid black line ) and the uves spectrum from april 2006 ( solid gray line )
. the uves data has been binned by a factor of five to smooth out noise but it is still plotted with five times greater resolution than the gmos spectra .
note that the p - cygni absorption in the 2006 uves spectra is asymmetric and has two velocity components . ]
|
in late 2006 , ground - based photometry of @xmath0 car plus the homunculus showed an unexpected decrease in its integrated apparent brightness , an apparent reversal of its long - term brightening .
subsequent hst / wfpc2 photometry of the central star in the near - uv showed that this was not a simple reversal .
this multi - wavelength photometry did not support increased extinction by dust as the explanation for the decrease in brightness . a spectrum obtained with gmos on the gemini - south telescope , revealed subtle changes mid - way in @xmath0 car
s 5.5 yr spectroscopic cycle when compared with hst / stis spectra at the same phase in the cycle . at mid - cycle
the secondary star is 2030 au from the primary .
we suggest that the spectroscopic changes are consistent with fluctuations in the density and velocity of the primary star s wind , unrelated to the 5.5 yr cycle but possibly related to its latitude - dependent morphology .
we also discuss subtle effects that must be taken into account when comparing ground - based and hst / stis spectra .
|
it has been known long that old stellar populations exist in the dwarf irregular galaxies and dwarf elliptical galaxies and the @xmath0-band magnitude of the tip of these old populations ( red giant branch ) has been often used to estimate their distances ( lee et al .
however , little is known about the size and morphological structure of the halos of these old stellar populations in the dwarf galaxies .
a study of the morphological structure of the halos in dwarf galaxies may provide some clues to understanding the origin of the dwarf irregular galaxies and the dwarf elliptical galaxies .
we have carried out a project to investigate the halos in dwarf irregular galaxies using the megaprime camera at the cfht . here
we present a case study of ngc 6822 which is a famous dwarf irregular galaxy in the local group .
ngc 6822 is known to have a huge rotation hi disk ( de blok & walter 2000 ) and a wide distribution of carbon stars ( letarte et al .
it is an well isolated galaxy belonging neither to our galaxy nor to m31 .
ngc 6822 is located at the distance of 500 kpc , and one arcmin at this distance corresponds to 145.4 pc .
we display in figure 1 the @xmath1 color - magnitude diagram of three regions at @xmath2 , @xmath3 , and @xmath4 where @xmath5 is the angular distance from the center of ngc 6822 .
this is part of the data covering a 1 deg @xmath6 1 deg field including ngc 6822 ( corresponding to a field of 8.7 kpc @xmath6 8.7 kpc ) .
figure 1 shows several distinguishable stellar populations in ngc 6822 as marked by the boxes : blue main sequence stars and blue supergiants ( ms ) , red supergiants ( rsg ) , old red giant branch stars ( rgb ) , asymptotic giant stars ( agb ) , and young red giants ( rg ) .
a broad vertical structure seen at @xmath7 represent foreground stars .
young stars are mostly seen at @xmath8 , while the old rgb stars are abundant even at @xmath9 .
this old population may belong to the halo of ngc 6822 .
we have used the stars inside each box for the following analysis to investigate the morphological structure of each population .
( top panel ) , @xmath3 ( middle panel ) , and @xmath4 ( bottom panel ) where @xmath5 is the angular distance from the center of ngc 6822 .
thick solid line boxes represent the regions selected for the analysis of the different stellar populations . ]
we have derived the surface number density maps for the ms , young rg and old rgb in the field of 1 deg @xmath6 1 deg to investigate their spatial distribution , displaying them in figure 2 in comparison with the greyscale map of the digitized poss map of ngc 6822 .
figure 2 shows several striking features .
first , the structure of the old rgb ( halo ) is dramatically different from the main body ( @xmath10 ) of ngc 6822 seen in the poss map .
it is surprisingly much larger than the poss image , exceeding one squaredegree field .
its inner region is elongated horizontally , while its outer region is elongated along the ne - sw direction .
its inner region looks elliptical , very similar to the structure of dwarf elliptical galaxies .
however , its outer part shows some distorted structure in the ne - sw end indicating some tidal interaction .
the existence of this tidal feature is intriguing , because ngc 6822 has no known companion galaxies in the nearby distance .
second , the spatial distribution of the young ms in the central region is similar to the shape of the main body of ngc 6822 seen in the poss map , but it is much more extended and elongated along the nw - se direction .
this structure is consistent with the hi structure ( de blok & walter 2000 ) .
third , the spatial distribution of the young rg looks round and similar to that of the central part of the old rgb distribution , but its center is offset from the center of the rgb distribution .
in addition , it shows stronger central concentration than that of the old rgb .
we have also investigated the radial variation of the surface number density for the rgb , rg , ms , agb and carbon stars ( letarte et al . 2002 ) as plotted in figure 3 .
figure 3 shows that the young populations ( ms and rg ) are more centrally concentrated than the older populations ( rgb , agb and carbon stars ) .
through the deep wide - field imaging we have found that ngc 6822 has a giant halo made of old red giants , extending much farther than the optical body and hi disk of this galaxy .
the structure of this stellar halo looks similar to those of dwarf elliptical galaxies so that we conclude that the halos of dwarf irregular galaxies may share the same origin with those of the dwarf elliptical galaxies . ngc 6822 also is known to have an old globular cluster at as far as 13 kpc from the center , supporting that its halo is really huge ( hwang et al .
. the existence of the giant halos may be common among dwarf irregular galaxies , waiting to be discovered .
|
we present a discovery of a giant stellar halo in ngc 6822 , a dwarf irregular galaxy in the local group .
this halo is mostly made of old red giants , showing striking features : 1 ) it is several times larger than the main body of the galaxy seen in the optical images , and 2 ) it is elongated in the direction almost perpendicular to the hi disk of ngc 6822 .
the structure of this stellar halo looks similar to the shape of dwarf elliptical galaxies , indicating that the halos of dwarf irregular galaxies share the same origin with those of the dwarf elliptical galaxies .
|
discovering new phases of matter is a primary objective of physics .
the fractionalized spin liquid in two spatial dimensions ( @xmath3 ) has provided a popular candidate framework for models describing the exotic properties observed in many strongly correlated electronic materials , including frustrated quantum magnets and high@xmath4 superconductors @xcite . the spin
liquid state is characterized by spinon excitations carrying unit charge under a compact u(1 ) gauge field . however
, polyakov has argued @xcite that a pure compact u(1 ) gauge theory is always confining at zero temperature for @xmath3 , confinement between test particles with opposite charges being produced by the proliferation of instanton tunneling events .
by contrast , the case for confinement by instanton proliferation in spin systems is rather more involved @xcite , and the situation at a critical point offers additional possibilities .
these considerations underlie a recent discussion of the deconfined quantum critical point @xcite as the foundation for the emergence of spin liquid phases with fractional excitations .
however , while this set of elegant ideas appears most plausible , it is not yet clear from any analytical or numerical calculations that there exists a microscopic hamiltonian , and particularly one with only simple electronic interactions , exemplifying this class of physical phenomena .
quasiparticle fractionalization has been demonstrated rigorously in one dimension ( 1d ) and in the quantum hall effect . for frustrated quantum spin systems ,
recent progress was made in the context of a two dimensional model with high degeneracy in the ground
state manifold which exhibits ( partial ) dimensional reduction @xcite and consequent spinon deconfinement @xcite .
we will expand upon and systematize these ideas to provide a further demonstration of a class of physical systems displaying fractional excitations . to this end , we begin with the natural question of whether a dilute gas of deconfined spinons can be stabilized in dimensions @xmath5 in a finite region of the phase diagram .
this latter requirement is crucial for observing deconfined spinon excitations in real systems .
we will provide an affirmative answer by considering a highly frustrated @xmath6 model on the @xmath3 ( checkerboard ) and @xmath7 pyrochlore lattices .
we demonstrate that for @xmath1 the spinon excitations are deconfined at the klein critical point , whose nature we will explain in detail . as expected from above , this zero temperature deconfined phase is unstable under any realistic perturbation characterized by an energy scale @xmath0 .
we argue that in the regime @xmath8 , where @xmath9 is the characteristic energy scale of the model , a dilute coulomb gas of spinons is stabilized in the temperature range @xmath2 .
this result raises the possibility of observing spinons in real higher dimensional systems . the checkerboard and pyrochlore lattices ( fig .
[ plattice ] ) which form the focus of our analysis have long been known to possess a geometry which is amenable to very large ground
state degeneracies .
the pyrochlore lattice is composed of corner sharing tetrahedra , arranged in a cubic ( fcc ) structure , such that each vertex is common to two 4site units ( see panel ( b ) of fig . [ plattice ] ) .
this geometry is rather common in transition metal and rare earth oxide systems , specifically in the pyrochlore ( the origin of the name ) and spinel structures .
the pyrochlore systems @xcite have chemical formula a@xmath10b@xmath10o@xmath11 , where both a and b may be taken from a wide range of metal ions , particularly rare earths , and both occur on a pyrochlore lattice .
spinel systems @xcite have chemical formula ab@xmath10o@xmath12 , where it is more typical for the a and b ions to be transition metals and only the b sites form a pyrochlore structure . individual compounds in these classes are usually strongly insulating and of fixed ionic valences , and the presence of non
magnetic ions on one of the sublattices ( the a sites for the spinel ) leads to a pyrochlore lattice of interacting spins on the other .
it was noted at a very early stage @xcite that the connectivity of such a structure is not dissimilar to the arrangement found in water ice , a point which we will both explain and exploit extensively below . the checkerboard lattice can be considered as a 2d version of the pyrochlore structure , in the sense that it also is composed of corner sharing tetrahedral units , these being represented by the crossed plaquettes ( the terminology we adopt henceforth ) in panel ( a ) of fig . [ plattice ] .
it is the relatively low coordination number and tetrahedral connectivity of the checkerboard and pyrochlore lattices which permits a proliferation of ground states .
some of the essential considerations are presented for classical pyrochlore antiferromagnets in ref .
@xcite and for quantum pyrochlore antiferromagnets in ref .
while a significant body of recent work has led to a better understanding of ice like phases in pyrochlore ferromagnets @xcite , the situation for both classical and quantum antiferromagnets remains far from satisfactory , to the extent that certain results conflict @xcite .
here we provide a further contribution to the body of knowledge concerning quantum pyrochlore antiferromagnets .
we will demonstrate further that the highly unconventional properties listed above arise within simple , nearest neighbor , bilinear and biquadratic spin models on the 2d and 3d pyrochlore lattices .
these models are derived directly from half filled , nearest
neighbor hubbard models , _
i.e. _ from models with coulomb interactions of the shortest range , and are thus fully realistic for electronic systems . the key features giving rise to exotic behavior are ( i ) an exponential number of degenerate ground states ; ( ii ) exact critical behavior with an emergent divergent correlation length , accompanied by the absence of long ranged order ; ( iii ) fractionalization of excited spin states into spinons which propagate freely in space , implying a separation of spin and charge upon doping these systems ( we will show that the emergent fractionalized excitations have well defined quantum numbers which may be described as residing at the ends of strings obeying simple rules ) ; ( iv ) an effective dimensional reduction occurring precisely at the critical point ; ( v ) topological order , in a sense which we will define explicitly below .
all of these features lead in combination to a distinct type of critical point whose static and dynamic properties we will characterize in detail , illustrating in addition their connection to a multitude of ideas and concepts emerging in the recent literature in the context of deconfinement and criticality . in sec .
ii we introduce the type of spin model displaying a klein point , and discuss its foundation in the electronic hamiltonian of real , correlated insulating materials . in sec .
iii we begin our analysis of the model at the klein critical point and at zero temperature , focusing on the characteristic properties here and illustrating their origin in the extensive ground state degeneracy . in sec .
iv we remain at the klein point but consider finite temperatures , to derive the effective entropic interaction of spinon excitations and thus demonstrate their thermally driven deconfinement . in secs .
v and vi we restore some of the perturbing effects present in the realistic hamiltonian of sec .
ii . working at zero temperature , in sec .
v we introduce and apply a specifically constructed variational procedure based on the ( non orthogonal ) dimer coverings of the ground state manifold , to demonstrate the selection of specific valence
bond orderings as the ground states of the perturbed system . in sec .
vi we extend our considerations to finite temperatures and present one example of an exactly critical regime .
these calculations indicate explicitly both the confinement of spinons at low temperatures and their liberation at higher temperatures , thus reinforcing the results of secs .
iii and iv .
section vii contains a discussion and conclusion .
the spin model upon which we focus is a @xmath6 hamiltonian on the checkerboard and pyrochlore lattices , @xmath13 where @xmath14 denotes all pairs of sites in the tetrahedron @xmath15 and @xmath16 is a scalar product of the two spins . on the checkerboard lattice @xmath17
denotes each crossed plaquette , by which is meant those in fig .
[ plattice](a ) with cross coupling interactions , and the first term is equivalent to both nearest and next neighbor heisenberg interactions of strength @xmath9 .
this model was considered on the square lattice ( all plaquettes crossed ) by two of the present authors @xcite , and we will contrast the two @xmath3 systems below . the heisenberg interaction on the tetrahedral units ( we will also refer to the crossed plaquettes of the checkerboard lattice as tetrahedra ) of the @xmath3 and @xmath7 pyrochlore lattices is strongly frustrated , and the symmetric four spin interaction pushes the model to maximal frustration at @xmath18 @xcite .
a straightforward but important observation is that @xmath19 may be recast in the form @xmath20 with @xmath21 the net spin of the tetrahedral unit @xmath22 , @xmath23 @xmath24 and @xmath25 @xcite .
this hamiltonian ( [ esh ] ) can be obtained from the hubbard model on the pyrochlore lattice , @xmath26 where @xmath27 is the number operator for electrons of spin @xmath28 .
this is perhaps the simplest possible description of interacting electrons with only nearest
neighbor hopping @xmath29 and a local coulomb repulsion @xmath30 .
systems which are half filled and have strong on site interactions are localized insulators where , with the exception of virtual processes , the kinetic energy gain from hopping is sacrificed to avoid the coulomb cost of double occupancy .
the effective hamiltonian obtained from the virtual hopping processes at lowest ( second ) order in the small parameter @xmath31 is a nearest
neighbor heisenberg model of the type contained in the first term of eq .
( [ symham ] ) , with @xmath32 and electronic spin @xmath33 ( @xmath34 ) . a fourth order strong coupling expansion of eq .
( [ eeh ] ) returns interaction terms bilinear in spins on the same and on different tetrahedra and biquadratic in spins on the same tetrahedron @xcite . as shown in detailed studies performed for plaquettes of the square lattice in the context of planar cuprates @xcite , and verified by fits to magnon dispersion relations for the same systems @xcite ,
the latter interactions are stronger than the former by approximately one order of magnitude . specifically , a far larger number of intra tetrahedron processes , whose sum has the symmetric form contained in the second term of eq .
( [ esh ] ) , contributes to the effective fourth order hamiltonian than do processes of the next neighbor heisenberg type .
thus by retaining only the intra
tetrahedron processes to fourth order in @xmath29 one obtains the spin hamiltonian of eq .
( [ symham ] ) with @xcite @xmath35 the general , su(2)invariant spin hamiltonian arising from a realistic electronic model may thus be considered as an intra
plaquette interacting system of exactly the form given in eq .
( [ symham ] ) , with only very weak perturbations from inter tetrahedron terms . to quantify this statement , in the half filled system eqs .
( [ symham ] ) and ( [ eeh ] ) are related by @xmath36 where @xmath37 is the effective spin hamiltonian obtained from a fourth order strong coupling expansion of @xmath38 .
the sum is over pairs of sites @xmath39 separated by two bonds and which do not belong to the same tetrahedron .
the exchange coupling @xmath40 is given by @xmath41 it is evident from eqs .
( [ j12 ] ) and ( [ j3 ] ) that for small @xmath31 the inter plaquette corrections ( @xmath42 ) originating from the hubbard hamiltonian are indeed one order of magnitude smaller than the intra plaquette terms of @xmath19 ( [ symham ] ) .
a particle - hole transformation leaves the spectrum of @xmath38 invariant at half - filling .
since the net effect of this transformation is to change the sign of the hopping , @xmath37 can not contain any term of odd order in @xmath29 . in sec .
iii we will analyze the properties of eq .
( [ symham ] ) at the klein point , @xmath43 @xcite or @xmath44 , where the semi positive definite hamiltonian of eq .
( [ symham ] ) attains its minimal value ( when @xmath45 or 1 on every tetrahedron ) .
the klein point is quasi - exactly solvable .
[ its ground states may be determined exactly . ] following the observations in the preceding paragraph , in secs .
v and vi we will expand our treatment to include perturbations around the klein point , which are of two types : the stronger , represented by nearest
neighbor heisenberg terms , correspond to deviations of eq .
( [ esh ] ) from the point @xmath46 ; the weaker , represented by next
neighbor heisenberg terms ( and , for the checkerboard , cyclic terms on the uncrossed plaquettes ) , correspond to fourth order inter
plaquette interactions .
specific @xmath6 materials can be expected at least in the 3d pyrochlore structure , such as that formed by the magnetic ions at both a@xmath47 and b@xmath48 ( @xmath49 ) sites in the stereotypical pyrochlore systems a@xmath10b@xmath10o@xmath11 , and on the b sites in a@xmath47b@xmath50o@xmath12 ( @xmath51 ) spinels . however , despite the considerable choice of magnetic and non magnetic ionic species afforded by these two structural classes , to date we are not aware of the successful synthesis of any such compounds which maintain the full cubic lattice symmetry .
equation ( [ symham ] ) suggests an even simpler way of expressing the hamiltonian , @xmath52 with @xmath53 a quartic polynomial in the total spin of the four sites forming a given tetrahedral unit of the pyrochlore lattice , @xmath54 . for clarity
we provide a brief review of the possible two and four particle states of @xmath6 spins . for a system of four coupled spins , the total spin @xmath55 of any tetrahedron obeys @xmath56 . the 16state spin space is decomposed into two singlet states ( @xmath57 ) , three triplets @xmath58 , and one quintet ( @xmath59 ) , @xmath60 in this explicit decomposition the right hand side labels the disjoint net spin ( @xmath55 ) sectors while the left
hand side encodes the 16dimensional space spanned by the direct product of the four @xmath61 particles .
the sum of any two nearest
neighbor spins on the lattice @xmath62 takes only the two values @xmath63 .
if any two spins within a given tetrahedron are in a singlet state ( @xmath64 ) , then the total spin of this tetrahedron can not exceed unity , @xmath65 .
for the parameter choice @xmath46 ( @xmath66 ) , @xmath19 ( [ symham ] ) can be rewritten , up to irrelevant constants , as a klein hamiltonian @xcite , @xmath67 where @xmath68 is the projection operator onto the subspace of net spin @xmath69 .
next we note , following ref .
@xcite , that because the number of lattice sites is double the number of tetrahedra , @xmath70 , the set of ground states with one singlet per tetrahedron can be mapped onto the set of hard
core dimer coverings of the lattice .
thus any dimer covering of the lattice with precisely one dimer per tetrahedron is a ground state of @xmath71 .
further , any state @xmath72 which is a superposition of dimer coverings each of which has one dimer per tetrahedron , @xmath73 ( the dimer coverings are labeled by @xmath74 ) , is also a ground state of @xmath71 .
one of the simplest ground states is afforded by the dimer coverings depicted in fig .
[ pr ] . while it is clear that any state of the form of eq .
( [ dimer ] ) is a ground state , a proof that all ground states are of this form is far less obvious .
it may nevertheless be shown rigorously @xcite that for klein models on the pyrochlore lattices all ground states are indeed of the type specified in eq .
( [ dimer ] ) . as a first qualitative consequence of this result
, it was noted in ref .
@xcite for the analogous point in the square
lattice hamiltonian that satisfying this constraint leads to a significant degeneracy because dimer singlets may be rearranged along diagonal lines of the lattice with no energy cost .
the ground
state entropy scales with the perimeter of the system , and pairs of single
spin ( spinon ) excitations on the diagonal lines have no binding energy for any separation .
thus the effective dimensional reduction from 2d to 1d leads to the presence of deconfined spinons whose propagation is essentially unidimensional . on the pyrochlore lattice
the open structure of crossed plaquettes creates a less constrained system , and the number of dimer configurations in the ground
state manifold is strongly enhanced , the still more massive degeneracy implying similarly exotic physics in this case .
we proceed quantitatively by computing the ground
state entropy in sec .
in addition to the simple coverings shown in fig .
[ pr ] , a far richer variety of states exists .
the mapping onto the spin ice problem mentioned briefly above provides a useful classification of these ground states .
we stress that the states with one dimer per tetrahedral unit defined by eq .
( [ dimer ] ) map exactly to the spin ice problem @xcite where , motivated by the structure of h atoms in solid water @xcite , two of the sites of any elementary unit ( a tetrahedron ) are associated with an ingoing arrow pointing towards the center of the tetrahedron and two sites lie on arrows pointing outwards .
we label the two sites belonging to a singlet dimer @xmath75 in a given tetrahedron @xmath22 by two incoming arrows from sites @xmath76 and @xmath77 to the center of the unit @xmath78 . in this fashion
it is clear that the system is mapped to a set of continuous directed lines such that each tetrahedral unit has exactly two incoming and two outgoing arrows relative to its center .
an example of this mapping is illustrated in fig .
[ pr ] . a longer but ultimately equivalent version of this mapping , enabling a study of correlations and degeneracies , was reported in ref .
the mapping is one to one : any spin ice configuration determines a unique singlet covering state with one dimer per tetrahedral unit and vice versa .
the conservation of incoming and outgoing arrows in the spin
ice representation is the foundation for the divergence
free condition we will employ below . throughout this work we focus on systems with periodic boundary conditions ( pbcs ) . a system with open boundary conditions ( obcs ) would possess a number of additional states not captured by the six vertex mapping , the entropy of which scales with its surface . because the number of pyrochlore spin ice states is bounded from below @xcite by the pauling limit @xcite , the number of ground states ( or , more precisely , of non orthogonal dimer coverings ) is given by @xmath79 with @xmath80 the number of vertices of the pyrochlore lattice and @xmath81 the number of ground state singlet dimer coverings .
we emphasize that this degeneracy is exponential in the system volume , yielding an extensive entropy with a simple , analytical expression for the lower bound , the pauling entropy @xmath82 .
the `` exact '' entropy of the ice problem is known from series expansion methods @xcite , which yield the numerical result @xmath83 similarly , for the checkerboard lattice the system is reduced to the exact spin
ice model @xcite on the square lattice formed by the centers of the crossed plaquettes , leading to an exact singlet dimer covering degeneracy @xcite of @xmath84 with @xmath80 the number of vertices of the checkerboard lattice , whence @xmath85 . in this derivation
we have assumed that the singlet dimer coverings , although non orthogonal , are linearly independent @xcite .
there are additional `` defect '' states in the ground
state manifold , in which two singlet dimers may occupy a single tetrahedron , or unpaired spins may be present , without altering the total energy @xcite .
however , because the entropy of these states scales with the system surface they constitute a set of measure zero in comparison with the extensive set of dimer coverings , and will be neglected in what follows . to examine the structure of general ground states of the model ( [ esh ] ) at the klein point ,
we invoke the line representation of the six vertex model @xcite depicted in fig .
[ rules ] . in this convention
, every line is composed of links whose arrows point to the right , and there is clear one to one mapping between the six vertex and the line representations .
an essential property of the lines is their `` chiral '' nature : they always represent a motion only to the right ( fig .
[ rules ] ) .
we will show in detail below ( fig .
[ lightcone ] ) that this chirality is not merely an artificial feature of the line representation , but encodes the physical restrictions on the possible spinon paths in the background of the allowed dimer states . finding
the elementary process connecting a given dimer covering with other ground states is straightforward in the line representation .
[ fig : pr1](c ) shows the line representation of the ground state depicted in fig .
the simplest process in the dimer basis corresponds in the line representation to flipping over one of the line corners , as depicted in figs .
[ fig : pr1](c ) and ( d ) .
note that there is an uncrossed plaquette ( in the original lattice ) associated with each such corner flip .
returning to the dimer representation , such a process corresponds to a cyclic rotation of the four dimers connected to the uncrossed plaquette [ fig .
[ fig : pr1](b ) ] .
a similar process around a hexagon whose edges belong to six different tetrahedra is obtained for the pyrochlore lattice . for clarity of explanation
we comment that in figs .
[ fig : pr1](c ) and ( d ) , and later in this section , the six vertex ( thin lines with arrows ) and line representations ( thick ) are superposed to aid in illustrating their equivalence , but either alone is a complete representation of the system in terms of independent sets of quantum numbers .
the mapping to the line representation shows that the low energy sector of a hamiltonian at the klein point can be mapped into a string model .
it is clear that the total number of lines is a conserved quantity under any local physical process , where by `` local '' we mean that the process changes only a finite number of ( spin ) degrees of freedom .
the dimer coverings may therefore be classified according to this line quantum number . as an example , in fig .
[ fig : vac ] we show that the uniform or `` ferroelectric '' @xcite dimer covering corresponds to the vacuum of lines .
number conservation implies that this state can not be connected with any other ground state by a local process .
the other `` ferroelectric '' dimer state ( of opposite polarization ) which is obtained by spatial inversion corresponds in the line representation to the state which is full of lines , and by the same argument is not connected by a local process to any other states in the ground manifold .
spatial inversion corresponds in the line representation to the line antiline transformation ( line conjugation ) .
not unexpectedly , the exact solution of the six vertex model @xcite reveals that the ground states of greatest importance ( by which is meant those becoming increasingly sharp in the thermodynamic limit ) are those containing , on a checkerboard lattice of size @xmath86 , precisely @xmath87 lines .
this sector has the highest statistical ( entropic ) weight by virtue of the many ground states it possesses .
analogously , we will see in sec .
v that when the system is perturbed away from the klein point at @xmath1 the state stabilized by the perturbation is generally one with @xmath87 lines @xcite . to expand upon the previous statements , we stress that all of the dimer coverings in the ground state manifold are individually eigenstates of the hamiltonian
thus there are no quantum fluctuations at the klein point , and a zero temperature transition through this point , for example as a function of @xmath88 , is a discontinuous , first
order quantum phase transition @xcite .
the klein point is a distinct type of critical point which is essentially classical in nature with critical thermal fluctuations at all temperatures @xmath89 .
it has long been known @xcite that the square lattice ice model formed by the centers of the crossed plaquettes in the checkerboard lattice exhibits critical power
law correlations , which have been investigated extensively in refs .
@xcite . elementary entropic arguments combined with the ice condition
lead to a simple dipolar analogy suggesting that the dimer correlations in @xmath90 dimensions exhibit an @xmath91 decay , where @xmath92 is the separation between dimers @xcite . specifically , denoting by @xmath93 the first two six vertex configurations of fig .
[ rules ] , and similarly by @xmath94 and @xmath95 the other two pairs , the line representation applied for the pyrochlore lattice yields @xcite @xmath96 \label{dipole_eqn}\ ] ] for large separation @xmath97 , with @xmath98 denoting the spatial components of @xmath99 and @xmath100 the unit vector in the direction of @xmath92 ; @xmath101 is a constant .
the correlation function for the checkerboard lattice has a similar @xmath102 decay . when interpreted as flux lines , the six
vertex arrows ( fig .
[ rules ] ) adhere at every vertex to a strict condition of no divergence .
this correspondence , along with the entropic arguments , underlies the dipolar correlations .
the zero flux condition at every vertex leads to the long ranged correlations ( in fact with infinite correlation length ) of the dimers .
the ground
state manifold at @xmath103 corresponds to the high temperature limit of the ice model , in which every six vertex configuration is realized with equal amplitude . as such @xcite ,
the system is in the `` disordered '' critical phase described earlier where , in particular , the expectation value of any local six vertex order parameter vanishes @xcite . on the checkerboard ( pyrochlore ) lattice the different sets of states in the manifold remain characterized by the net number of vertical lines on each horizontal row ( plane ) of lattice sites ( or conversely ) : this number is conserved and corresponds to a topologically invariant `` flux '' of lines . because lines of arrows can not terminate in the six vertex representation the flux can not change from one plane to the next in the pyrochlore @xcite .
we remark for clarity that for the checkerboard lattice the `` horizontal '' and `` vertical '' directions specified above for the dual lattice are rotated by 45@xmath104 relative to the horizontal and vertical directions of the original lattice .
although the system does not possess local order it displays this elementary topological order characterized by the topologically invariant flux @xcite .
the model discussed here forms a high dimensional realization of wen s string nets @xcite . on the checkerboard lattice , conservation of the line quantum number
means that if @xmath105 is the number of vertical lines intercepting a horizontal span of fixed length at position @xmath106 on the @xmath107 axis , the non local correlation function for two strings at @xmath108 and @xmath109 , which is given by @xmath110 with @xmath111 arbitrary , is maximal , @xmath112 . because in general @xmath113 , for unrestricted values of @xmath114 and @xmath115 , here one has simply @xmath116 .
similar results apply for the pyrochlore lattice , where @xmath117 denotes the number of lines intersecting the plane @xmath118 .
equation ( [ topoeq ] ) presents a high dimensional analog of the well known fixed value of string correlation functions in @xmath119 spin chains at the aklt point @xcite . at large separations
these uniform non local correlations dwarf the local dimer
dimer correlations , which retain an algebraic decay ( as @xmath120 ) . as a direct consequence ( below )
, the system exhibits in general a long ranged dimer order away from the klein critical point .
the energy cost of exciting one of the singlets of a given dimer configuration is proportional to @xmath9 ; as depicted in fig .
[ spinon ] , this energy is not altered by changes in the separation of the two @xmath6 objects forming the triplet state , which may thus be regarded as two deconfined spinon excitations . single
spinon propagation can be described as a constrained random walk because the allowed paths may not increase the number of tetrahedra without a singlet dimer , but in contrast to the situation in ref .
@xcite the paths are @xmath90dimensional rather than being constrained to a one dimensional path . in this context
it is instructive to consider the dimensionality of the allowed spinon paths for the checkerboard and pyrochlore klein models , to examine its relation to the dimensionality @xmath121 of the symmetry group emerging in the ground state sector @xcite . here
@xmath121 is the dimensionality of the minimal non empty set of spins influenced by these symmetry operations . assuming that the ground state sector does not transform trivially under these symmetry operations , the entropy @xmath122 scales according to @xmath123 , where @xmath124 .
the klein model on the pyrochlore lattice exhibits a zero
dimensional @xmath125 ) symmetry , meaning that there exist local operations which link different ground states ( _ e.g. _ the operation depicted in fig .
[ fig : pr1 ] ) .
the presence of these symmetry operators enables spinon excitations to propagate with no energy cost in @xmath126 dimensions .
thus spinons on the pyrochlore and checkerboard lattices propagate freely in regions whose size scales with the volume of the entire lattice , @xmath127 . for the square lattice model of ref .
@xcite , @xmath3 while @xmath128 , as a consequence of which the spinons can propagate only in @xmath129 dimensional regions ( lines ) .
this analysis provides a more explicit definition of the terms `` complete '' ( pyrochlore and checkerboard ) and `` partial dimensional reduction '' ( square lattice ) , of their origin in terms of system entropy ( scaling respectively with the volume and with the perimeter ) , and of their consequences for spinon dynamics . as noted earlier , perturbation
theoretic calculations , including those illustrating the effects of spinons , are somewhat involved in the dimer basis due to the non orthogonality of the dimer states . in the remainder of this section we comment only briefly and qualitatively on possible spinon exchange paths connecting different dimer states of the low energy sector ; a detailed discussion of one systematic approach to quantitative calculations in the dimer basis is deferred to sec .
v. while we have used the words `` free propagation '' of spinons to indicate that their motion occurs in all spatial dimensions , spinon dynamics on the pyrochlore and checkerboard lattices at the klein point remain constrained by the effective ice rules obeyed by the dimer configurations in the ground state manifold .
these rules , upon which we remark further below , become most evident by considering the line representation of the six vertex model ( fig . [ rules ] ) . in the presence of a single dimer excitation ,
the allowed lines specify the possible spinon motion . against the background of the uniform `` ferroelectric '' state ( corresponding to the line vacuum , fig .
[ fig : vac ] ) , where all horizontal and vertical arrows point in the same directions , the spinon lines which may be inserted are all chiral .
the spinons are restricted to propagate only within a cone defined by two spatial quadrants ( fig .
[ lightcone ] ) , and no closed spinon paths are possible in such states .
the same is true for the opposing `` ferroelectric '' state , the state full of lines , in which the string created between the two spinons appears as an antiline .
as suggested by the simple example of the previous paragraph , the spinon motion can be determined from the line rules .
the trace left by two spinons which are created on a common tetrahedron and move away from each other corresponds , in the line representation , to a string which changes the state of the underlying link : the string creates a line segment if the link was empty and annihilates the corresponding segment if it is already present . following the rules of fig .
[ rules ] , if the line segment created by the string crosses a preexisting line , the resulting state on the corresponding tetrahedron is that indicated as 2 in fig .
[ rules ] .
once again we show for clarity in figs .
[ spinon](c ) and ( d ) the line representations of the dimer / spinon states depicted in figs .
[ spinon](a ) and ( b ) for the checkerboard lattice . in the general case of
non uniform states with a finite density of lines , the spinon strings are no longer constrained to be chiral ( fig .
[ spinon ] ) . when the string of a propagating spinon collides with a six vertex line representing the dimer configuration , the lines ( and hence the spinon path ) are effectively reflected . .
disruption of the dimer background dictates that the line or string created between the propagating spinons can move only in the direction indicated by the black arrows .
this implies that the spinons are restricted to move within a cone defined by the horizontal and vertical directions . in the line representation ,
the position of each spinon is indicated with a small dotted circle.,width=340 ]
as noted above , the dipolar correlations characterizing the deconfined klein critical point are driven by classical fluctuations , whence the most appropriate designation is as a classical @xmath130 critical point . away from @xmath1 ,
the critical thermal fluctuations of the dimer coverings produce an entropic spinon spinon interaction .
this interaction may be determined by replacing each spinon with a static monomer inserted in the dimer covering , reducing the problem to the classical statistical mechanics of close packed dimer coverings @xcite .
this reduction makes use of the fact that the singlet coverings are linearly independent @xcite .
the critical dipolar fluctuations of the underlying dimer field produce an effective `` coulomb '' interaction between the spinons @xcite , which in the pyrochlore lattices under consideration arises from the local conservation law of zero divergence .
we will demonstrate the existence of the coulomb phase for the checkerboard lattice ; a similar analysis may be applied for the 3d pyrochlore case .
the derivation of the effective coulomb interaction begins by assigning a charge to each spinon , for which it is convenient to expand the six vertex representation in order to include states containing two spinons .
it is important to note that spinons are created in pairs ( singlet triplet excitation ) , and each pair generates a single crossed plaquette of higher energy , because is possesses no singlets , which will be called a `` defect '' .
when the spinons propagate away from each other they are also separated from the defect ( fig .
[ spinon ] ) , and preservation of the ice
rules requires that the number of defects does not increase
. we will show below that the effective charge @xmath131 of the defect is opposite in sign to the effective charge @xmath132 of the spinons , with magnitude @xmath133 .
[ superficially , this overall neutrality of three quasiparticles forms a geometric analog of the quark content of the neutron , whose basic quark structure is @xmath134 with the charges of the @xmath135 and @xmath90 quarks being respectively @xmath136 and @xmath137 . ] to represent states with two spinons , we will include the four additional vertex configurations depicted in fig .
[ spinvert ] .
note that these configurations have non zero divergence , in contrast to the six vertex states ( fig .
[ rules ] ) used to represent the dimer coverings ( which are the vacuum of charge ) . according to the previous rules , the defect is represented by four outgoing arrows [ fig .
[ spinvert](a ) ] . for the checkerboard geometry
it is convenient to introduce the two nel sublattices a and b of the underlying square lattice ; a number of equivalent definitions may be found for a two sublattice decomposition of the 3d pyrochlore .
a one to one mapping between states containing spinons and the vertex configurations may be found by restricting the spinons to one of the sublattices ( for example a ) .
a complete classification of all spinon configurations may be obtained by introducing an additional ising like quantum number which specifies on which of the two sublattices our plaquettes are located .
such an additional tabulation is not necessary for the examination of the long distance interaction .
we may rely instead only on the consequences of the vertex representation illustrated in fig .
[ spinvert ] for plaquettes of one sublattice , which are sufficient for determining the asymptotic ( long distance ) form of the interspinon interaction .
the fact that spinons may propagate from one sublattice to the other due to the existence of diagonal bonds in dimer coverings of the checkerboard lattice is nevertheless crucial .
this is not the case for quantum dimer models on bipartite lattices , whose dimers always link the two different sublattices .
we remind the reader that on a bipartite lattice each spinon may be assigned a unique effective charge given by the sublattice parity of the spinon location ( usually + 1 on one sublattice , and -1 on the other ) .
this simple assignment is invalid on non bipartite lattices such as the pyrochlore , where the additional range of possible spinon configurations necessitates an essential extension of the definition of the effective spinon charge .
because the spinons occupy sites of the original lattice , their position will be denoted by one arrow on the dual lattice .
specifically , we take this as the ingoing arrow coming from the lower left corner of a plaquette with only one outgoing arrow [ fig .
[ spinvert](b ) ] .
if one considers vertex states such that for each vertex with maximal divergence [ fig .
[ spinvert](a ) ] there are two with only one outgoing arrow [ fig .
[ spinvert](b ) ] , the mapping between the two representations is one to one ( under the condition that all the spinons occupy the same sublattice ) . the coulomb interaction and the effective charge of the spinons emerge from the standard procedure used to map the ice problem onto a compact u(1 ) gauge theory @xcite . here
we follow the derivation and notation of ref .
@xcite to describe this procedure and derive the effective charges of defects and spinons .
the local zero divergence condition defining the dimer coverings on the checkerboard and pyrochlore lattices implies that these states can be considered as the physical hilbert space of a compact @xmath138 gauge theory .
the local @xmath138 gauge invariance becomes more explicit if one defines pseudospin1/2 operators @xmath139 on each site @xmath76 of the lattice .
the dual lattice formed by the centers of the tetrahedrons ( vertices ) can be divided into two sublattices @xmath140 and @xmath141 .
the site coordinate @xmath76 of the original lattice denotes the links of the dual lattice .
@xmath142 if the arrow on link @xmath76 points from the @xmath140 to the @xmath141 sublattice , @xmath143 if the arrow points from @xmath141 to @xmath140 . the zero divergence condition can be expressed in terms of these variables as @xmath144 and the operators @xmath145 are the local u(1 ) symmetry transformations associated with this conservation law .
we note that the conservation law is a property of the hilbert space and is therefore present for any theory defined on this space .
an equivalent constrained low
energy hilbert space was derived for an xxz model on a pyrochlore lattice in ref .
@xcite .
we present in some detail the consequences of the zero divergence condition for the checkerboard lattice ( @xmath3 ) , for which the notation is slightly more compact ; the results we obtain are unchanged for the pyrochlore lattice ( @xmath7 ) .
the @xmath138 gauge structure of the hilbert space indicates that one may gain further insight by identifying the variables playing the role of the electric field and of the vector potential in electrodynamics .
for this purpose , it is convenient to introduce quantum rotor variables and the coordinate @xmath146 to span the dual lattice ( the lattice of centers of the tetrahedrons ) .
the primitive vectors @xmath147 and @xmath148 define the two perpendicular directions on the dual lattice . on each link of this lattice
we define an angular momentum variable @xmath149 , where @xmath150 indicates the oriented link from @xmath92 to a neighboring dual lattice site @xmath151 .
the state @xmath152 ( @xmath153 ) corresponds to an arrow parallel ( anti parallel ) to the relative vector @xmath154 . with this definition , @xmath155 changes sign under inversion : @xmath156 ( @xmath157 indicates the oriented link from @xmath92 to @xmath158 ) . for convenience
we introduce the variable @xmath159 , which takes the integer values 0 or 1 .
we employ the discrete gradient @xmath160 to reexpress the ice rules as an explicit zero divergence condition on the vector field @xmath161 , @xmath162 equation ( [ gauss ] ) corresponds to gauss law for electrodynamics in the absence of external charges , @xmath163 .
vertex configurations violating the ice rules , such as those depicted in fig .
[ spinvert ] , should then carry an `` effective charge '' @xmath164 . here
@xmath165 for the configurations of fig .
[ spinvert](a ) ( defects ) and @xmath166 for the three configurations of fig .
[ spinvert](b ) ( spinons ) : the spinons carry a positive unit charge while the defects carry two units of negative charge .
the `` electric field '' is then defined as the operator @xmath167 with eigenstates @xmath168 and eigenvalues @xmath169 . the canonically conjugate operator @xmath170 satisfying @xmath171= i \delta_{{\bf r},{\bf r ' } } \delta_{\mu,\nu}\ ] ] plays the role of the vector potential .
because in this section we focus only on the high temperature ( @xmath172 ) fixed point whose properties are determined solely by the gauge structure of the hilbert space , we will not discuss any particular quantum gauge theory .
the above derivation is used only to identify which variables play the role of the electric field and the charge .
the klein point at @xmath130 corresponds to the infinite temperature fixed point .
such a point is characterized by a deconfined two component ( spinons and defects ) coulomb gas .
the spinons exist only as gapped excitations ( of gap @xmath173 ) , and the spinon
spinon interaction is then logarithmic , @xmath174 , for @xmath175 and a power law , @xmath176 , for @xmath7 .
the spinon
defect interaction has the opposite sign ( attractive ) and its absolute value is two times greater .
although a spinon defect pair is critically confined in the @xmath3 case by the logarithmic interaction , the prefactor @xmath177 is generally sufficiently small that there is a divergence in the expectation value of @xmath178 , which implies a divergent `` dielectric constant '' @xcite . at finite temperatures
there is a finite concentration @xmath179 [ @xmath180 of thermally excited spinons in equilibrium , and the effective spinon spinon or spinon defect interaction is screened , decaying exponentially for distances @xmath181 .
however , the critical behavior remains observable for @xmath182 . for general @xmath183 and @xmath184 ,
the energy difference between the singlet and triplet states on each tetrahedron is much smaller than the gap to the quintet state ( @xmath185 ) .
@xcite it is clear that for @xmath186 the degeneracy between the singlet and triplet states is effectively restored at temperatures @xmath187 such that @xmath188 in this case , only the manifold of singlet dimer coverings found as the ground states at @xmath103 is accessible .
thus the effect of a finite temperature on a system in the proximity of a klein point is the effective restoration of extensive configurational entropy , critical dipolar correlations , topological order , and spinon deconfinement .
we comment that the appearance of deconfinement at high temperatures , also in systems far more general than those considered here , is hardly surprising .
we repeat well known facts : various ordered phases generally melt via the appearance of topological defects ; these defects become deconfined at the melting temperature ; the origin of deconfinement is an energy
entropy balance ; entropic effects modify the effective interactions between the defects , leading to a vanishing interaction at the onset of deconfinement .
a careful treatment of entropic effects often involves detailed contour counting arguments @xcite . the effective force between topological defects in a wide range of systems , including vortices in superconductors and dislocations and disclinations in elastic media ,
is universally expected to be algebraic , as may be seen directly from duality arguments @xcite . in the current context
the topological defects correspond to monomers or spinons in the dimer coverings and the coulomb field lines to the many possible trajectories for the motion of a defect .
pure coulomb interactions vanish at large separations in dimensions @xmath189 , suggesting the possibility of high dimensional fractionalization .
the peculiar aspect of the pyrochlore systems under consideration is that the critical point separating the low temperature confined state from the high temperature deconfined phase may be driven to @xmath190 by optimizing the frustration in @xmath19 .
this is the singular characteristic of the klein point : the exponentially large ground
state sector is defined solely by the local zero divergence constraint .
the system is then driven only by entropic fluctuations , which in combination with the local constraint give rise to the classical critical behavior .
in this section we examine explicitly the possible ground states away from the klein point . to this end , we begin by developing a systematic procedure for evaluating `` matrix elements '' , @xmath191 , within the non orthogonal basis of singlet dimer coverings which constitute the low energy sector in the vicinity of the klein point . we note that because of the non orthogonality a standard ( bra operator ket ) expectation value computed between two states in the manifold is not a true matrix element , and is referred to henceforth as a `` bracket '' .
we will illustrate the calculational procedure and results for the checkerboard lattice , and state only that their generalization to the pyrochlore is straightforward ( if lengthy ) .
the evaluation procedure consists of elementary rules which are applied to the graph associated with the bracket to be evaluated , and it is a generalization of the standard method used to compute the overlap between two singlet dimer coverings on bipartite lattices @xcite . we first extend this method to compute the overlap between non
bipartite singlet coverings , because on the pyrochlore system the dimers are not necessarily formed from sites on different sublattices . secondly and more importantly , we provide straightforward rules for computing brackets of spin product operators of heisenberg type , @xmath192 . brackets of products of heisenberg spin operators , as in the second term of eq .
( [ esh ] ) , can be evaluated by a simple extension of these rules . in this section
, we will consider a perturbation which is proportional to the first term of eq .
( [ esh ] ) , a heisenberg interaction between any pair of sites in the same crossed plaquette @xmath193 such that after the perturbation @xmath194 .
for the klein model on the square lattice it is transparent that all states within the ground
state basis remain degenerate to first order in the perturbation @xmath195 .
this statement follows from two observations : the diagonal brackets , @xmath196 , are the same for any dimer state @xmath197 ; the off diagonal terms , @xmath198 , in particular for @xmath199 .
thus the degeneracy is lifted only to second order in @xmath195 on the square lattice .
the situation is completely different for the same model on the pyrochlore and checkerboard lattices : because in these cases the ground
state degeneracy is extensive at the klein point , two different ground states can be connected by local operators . as a consequence the off
diagonal brackets do not in general vanish , and the degeneracy is lifted to first order in the perturbation .
the usual approach to this problem is to neglect the non orthogonality of the singlet dimer coverings and propose a minimal quantum dimer model which includes only processes of the rokhsar
kivelson ( rk ) type @xcite .
we note , however , that this procedure does not lead to the rk model in the pyrochlore geometries under consideration because , although the hamiltonian is the same ( a u(1 ) gauge magnet @xcite ) , the zero divergence or gauge condition which defines the `` physical '' hilbert spaces is different . in what follows
we depart from this conventional treatment , and show that it is indeed incorrect for the case of interest .
the non orthogonality of the dimer basis introduces subtleties not directly evident from the conventional approach , as a consequence of which we will show that the valence bond crystal ( vbc ) orderings chosen by tunneling events differ from the ones expected for a theory which includes only processes of the rk type . in terms of the arrows ( the six vertex representation ) shown in fig .
[ afes ] , the antiferroelectric state becomes a staggered flux phase in which each square plaquette of the dual lattice has a well defined chirality ( all of its arrows circulate in the same way ) and the direction of circulation alternates from clockwise to counterclockwise . with this starting state one
may invert the sign of the current ( direction of the arrows ) around any plaquette .
this is the elementary process , or corner flip in the line representation , depicted in fig .
[ fig : pr1](b ) .
the antiferroelectric state allows for the largest number of these elementary processes , and each can be interpreted as the creation of a localized defect on an uncrossed plaquette of the antiferroelectric background .
the tunneling between states with defects in different positions lowers the energy of the system .
while the creation of one of these defects is indeed the elementary rk process normally invoked as the most relevant one in these cases @xcite , we will show that its amplitude is exactly equal to zero for the perturbation @xmath195 .
to first order in the perturbation the state of lowest energy is the linear combination of singlet dimer coverings which minimizes @xmath195 .
up to an irrelevant constant , the hamiltonian @xmath195 can be rewritten in the more convenient form @xmath200 where @xmath201 is the permutation operator which interchanges the spins on sites @xmath76 and @xmath77 of plaquette @xmath17 .
the diagonal brackets are manifestly the same for all singlet dimer coverings .
specifically , on the checkerboard lattice and for any system with pbcs , @xmath202 with @xmath80 the number of lattice sites and @xmath203 denoting any dimer covering satisfying the klein constraint ( all crossed plaquettes contain one dimer ) . as noted in sec .
iii , systems with obcs possess additional ground states containing crossed plaquettes with two dimers or ( one or two ) spinons , and these would have a different diagonal energy but are not thermodynamically relevant .
we focus henceforth on ground states satisfying pbcs .
the degeneracy of dimer coverings is lifted only by superposing different allowed coverings with finite amplitudes , _
i.e. _ by considering states of the form @xmath204 with the normalization condition @xmath205 which results from superposing two different dimer coverings [ ( a ) and ( b ) ] .
these dimer coverings correspond to an antiferroelectric background with a single elementary process operating on different ( nearest neighbor ) uncrossed plaquettes.,width=340 ] we proceed by listing a complete set of rules for the evaluation of brackets in the dimer basis . these general , purely topological rules extend early and pioneering ideas concerning resonating valence bond states , and initial results obtained by using these concepts for square lattices @xcite .
the normalization of @xmath206 and computation of @xmath207 require evaluating the overlaps @xmath208 and brackets of the form @xmath209 , with @xmath203 and @xmath210 any states in the dimer basis . to compute the overlap
we first assign a given orientation to each singlet dimer , indicated with arrows pointing upwards for vertical dimers and to the right for diagonal and horizontal bonds ( fig .
[ graph ] ) .
[ we remark that `` vertical '' and `` horizontal '' refer here to the real lattice , and not to the rotated lattice used in the discussion of the six vertex and line representations in sec .
iii . ] with this convention , a superposition of dimer configurations corresponding to @xmath203 and @xmath211 , with the differing bonds marked as in fig .
[ graph](c ) ( bonds which do not differ give trivial loops of length two , introducing factors of unity ) , forms a graph @xmath212 . in general
, this graph may consist of more than one disconnected loop @xmath213 .
let @xmath214 be the length of the loop @xmath213 ( the number of singlet dimers forming the loop ) , @xmath215 be the number of disconnected loops in graph @xmath212 , and @xmath216 be the number of arrows in the graph which circulate clockwise in their loop . with these definitions
the overlap @xmath217 may be expressed as @xmath218 where @xmath219 is the total length of the graph .
the determination of the non trivial brackets @xmath220 proceeds by examining how the length and dimer orientation of the graph @xmath212 are affected when the permutation operator @xmath221 is applied to the state @xmath222 .
the resulting graph will be denoted as @xmath223 .
the application of @xmath221 may alter the loop geometry in only a restricted set of simple ways , of which we next provide a systematic list . 1
. [ l1 ] if @xmath224 and they belong to the same loop @xmath225 , there are two possibilities : 1 .
[ l11 ] when the distance ( number of dimers ) on the loop between sites @xmath76 and @xmath77 is an odd number , the effect of @xmath221 is to cut the loop into two segments and to reconnect them while exchanging the ends of one of the segments [ figs .
[ pijtypei](a ) and ( b ) ] .
the length of the loop under consideration is unaltered in this process .
the only effective modification is the reversal of an odd number of arrows , causing a sign change relative to @xmath217 , @xmath226 2 .
[ l12 ] + when the distance between sites @xmath76 and @xmath77 is even , @xmath221 divides the loop @xmath225 into two smaller loops , @xmath227 and @xmath228 , preserving the total length , @xmath229 [ figs .
[ pijtypei](c ) and ( d ) ] .
the circulation of the arrows is unaltered , and thus from eq.([gab ] ) the net change relative to graph @xmath212 is a supplementary factor of 2 due to the increase in the number of loops , @xmath230 , whence @xmath231 2 .
[ l2 ] if only one of the two sites ( @xmath76 or @xmath77 ) belongs to graph @xmath212 , the effect of @xmath221 is to increase the length of the graph by two links , @xmath232 .
there is no change of sign because the arrows of the two additional bonds circulate in opposite directions relative to their point of contact . from eq .
( [ gab ] ) the net result relative to graph @xmath212 is a factor of @xmath233 due to the length increase , and so @xmath234 3 . [ l3 ] if neither @xmath76 nor @xmath77 belong to graph @xmath212 , there are two possible cases : 1 .
[ l31 ] if there is a singlet bond connecting the sites @xmath76 and @xmath77 , @xmath235 2 .
[ l32 ] if the sites @xmath76 and @xmath77 are not connected by a singlet bond , @xmath234 4 .
[ l4 ] finally , if @xmath224 but the sites belong to different loops @xmath236 and @xmath237 , the effect of @xmath221 is to fuse both loops into a single loop @xmath238 . because the total length , @xmath239 , and the sign are preserved under this operation , the net result is a factor @xmath233 which arises from the decrease by one in the number of loops , @xmath240 , leading to @xmath234 on graph @xmath212 ( fig . [ graph ] ) when sites @xmath76 and @xmath77 belong to the same loop of @xmath212 and their separation is odd [ case ( a ) @xmath241 ( b ) ] or even [ case ( c ) @xmath241 ( d)].,width=340 ] these rules allow a very straightforward derivation of eq .
( [ uniform ] ) : because @xmath242 is the null graph , the effect of @xmath221 corresponds to case [ [ l3 ] ] . for any dimer covering
there are @xmath243 bonds of type [ [ l31 ] ] and @xmath244 bonds of type [ [ l32 ] ] , and adding the two contributions yields @xmath245 from @xmath246 [ see eq .
( [ gab ] ) ] we recover eq .
( [ uniform ] ) . with the same rules one may compute the energy of the most general state within the klein
point ground state manifold whose expression is given by eq .
( [ superposition ] ) , @xmath247 .
\label{general_energy}\ ] ] here @xmath248 , where @xmath249 ( @xmath250 ) is the number of pairs @xmath251 which are of type [ [ l11 ] ] ( [ [ l12 ] ] ) relative to the graph @xmath252 , while @xmath253 is the total number of pairs @xmath251 of type [ i ] .
the reason why only cases of type [ i ] need be considered in eq .
( [ general_energy ] ) is that terms of types [ [ l2 ] ] , [ [ l32 ] ] , and [ [ l4 ] ] all give the same contribution , @xmath254 .
resulting from the superposition of two dimer coverings which differ minimally .
the physical process connecting these is the rk type or corner
flip process represented in fig .
[ fig : pr1].,width=340 ] next we apply the rules of subsec .
vb to the checkerboard lattice .
equation ( [ general_energy ] ) allows one to compute the energy of any variational state .
its second term corresponds to the energy change @xmath255 due to the linear combination of different singlet dimer coverings .
the magnitude of this change therefore reflects the degree of `` mixing '' between different dimer coverings induced by @xmath195 . to develop an intuitive understanding we begin by considering the linear combination of two dimer coverings , a natural choice being to superpose two coverings whose
dimer configurations differ minimally ( such as the two states depicted in fig .
[ fig : pr1 ] ) , _ i.e. _ by the rk or corner flip process discussed above .
we remind the reader that the six vertex representation of this process entails a reversal of the arrows on a single plaquette of the dual lattice with a well defined chirality [ figs . [
rk](a ) and ( b ) ] .
if @xmath256 and @xmath257 are any such pair of dimer coverings which differ minimally , their overlap is @xmath258 because @xmath259 and @xmath260 [ fig .
[ rk](c ) ] . to obtain the energy of the normalized state @xmath261
one replaces the coefficients of eq .
( [ general_energy ] ) with the values dictated by the above rules .
the number of pairs @xmath262 with both sites on the graph is @xmath263 , composed of 8 pairs of type [ [ l11 ] ] and 4 of type [ [ l12 ] ] , as a result of which @xmath264 . replacing these quantities in eq .
( [ general_energy ] ) gives @xmath265 the fact that the energy is unchanged ( @xmath266 ) by taking the linear combination ( [ n - d ] ) indicates that the elementary rk type process is not generated by @xmath195 .
we note that all pairs of sites @xmath267 in the only loop of the graph @xmath268 [ fig .
[ rk](c ) ] are separated by a distance on the loop no greater than 2 ; loops satisfying this condition will be defined as _
simple_. as an example , the loop depicted in fig .
[ graph](c ) is not simple because it contains site pairs @xmath251 separated by five and six bonds . in general
, @xmath255 vanishes for any pair of dimer configurations , @xmath269 and @xmath270 , such that all loops in the graph @xmath212 are simple .
this property follows immediately from eq .
( [ general_energy ] ) : if all loops are simple one has that @xmath271 , implying @xmath272 and @xmath273 . resulting from the superposition of two states connected by inverting the arrows of two next neighbor plaquettes of the dual lattice which have opposing , well defined chiralities .
the two plaquettes are highlighted in panels ( a ) and ( b).,width=340 ] because @xmath195 does not connect two minimally different dimer configurations [ @xmath274 , it is necessary to seek the minimal relevant process , meaning that with the minimum value of @xmath275 , which is generated by @xmath195 .
the next possible value is @xmath276 ( no graphs exist with @xmath277 ) , an example of such a physical process being that connecting the states ( a ) and ( b ) in fig .
[ graph ] . in the six vertex representation
this process is permitted when all the arrows of a rectangle formed by two adjacent plaquettes of the dual lattice have the same chirality ; this rectangle is highlighted in figs .
[ graph](a ) and ( b ) .
the process consists of inverting the orientation of the six arrows , or the flux of the rectangle . for any pair of dimer coverings @xmath278 and
@xmath279 which differ by this process , @xmath280 and for the normalized linear combination @xmath281 the energy change is @xmath282 the minimum energy is obtained for @xmath283 , which together with the normalization condition gives @xmath284 for @xmath285 and @xmath286 for @xmath287 .
the overlap is @xmath288 , giving @xmath289 , as expected for the mixing of two orthogonal states with the same diagonal energy .
this remains true for any pair of states connected by @xmath195 because @xmath290 is an upper bound for the absolute value of their overlap . in the next process to consider , the arrow directions on the perimeters of two next neighbor plaquettes of the dual lattice are inverted if both plaquettes have opposing , well defined chiralities , as indicated in the examples of figs .
[ graph2](a ) and ( b ) . in this case
we denote the corresponding dimer coverings by @xmath291 and @xmath292 , and the length of the associated graph is @xmath293 .
the overlap is @xmath294 and the energy shift due to the perturbation for the linear combination @xmath295 is @xmath296 with @xmath297 for latexmath:[$\delta j = \pm the other graph @xmath212 of length @xmath299 corresponds to a process which is analogous to the previous case but with next neighbor plaquettes of the dual lattice having the same well defined chirality ( fig .
[ twoloops ] ) .
the energy change @xmath255 associated with this process is exactly that obtained for @xmath300 .
any other process generated by @xmath195 has an associated graph of length @xmath301 , and because the amplitude of these processes falls in proportion to @xmath302 ( @xmath303 is the number of pairs @xmath251 on the same loop separated by a distance greater than 2 ) , these may safely be cut off beyond a certain length @xmath304 . resulting from the superposition of two states connected by inverting the arrows of two next neighbor plaquettes of the dual lattice which have the same well defined chirality .
the two plaquettes are highlighted in panels ( a ) and ( b).,width=340 ] in order to propose a wave function @xmath206 which takes advantage of the three processes giving non
zero contributions for @xmath305 , we introduce the bosonic operators @xmath306 and @xmath307 . to define these operators we classify the uncrossed plaquettes of the checkerboard lattice according to the sublattice @xmath101 or @xmath308 ( sec .
iv ) of the dual square lattice [ fig .
[ sublattices](a ) ] . in the six vertex representation of the antiferroelectric state ( fig . [ afes ] ) , these two sublattices have opposite chirality , corresponding to a staggered flux phase .
if the uncrossed plaquette @xmath309 has a well defined chirality , the operator @xmath306 raises the flux from negative to positive if @xmath310 and lowers it from positive to negative if @xmath311 , _ i.e. _
@xmath306 changes the direction of arrow circulation from counterclockwise to clockwise if @xmath312 and conversely if @xmath311 . for states of no
well defined chirality on plaquette @xmath309 , and when the flux is positive ( negative ) for @xmath310 ( @xmath308 ) , @xmath306 becomes the null operator , @xmath313 .
minimizing the energy of a variational state is equivalent to maximizing the number of allowed tunneling processes with a negative sign .
the appropriate dimer background for this optimization is provided by the antiferroelectric state @xmath314 of fig .
[ afes ] ; only for this state can the elementary process of fig . [ rk ]
be applied to every uncrossed plaquette . moreover ,
the three processes which give finite contributions to @xmath255 for @xmath315 can be applied to any pair of nearest neighbor ( fig .
[ graph ] ) and next neighbor ( figs .
[ graph2 ] and [ twoloops ] ) plaquettes of the dual lattice when the antiferroelectric background is suitably populated with the local defects ( bosons ) created by @xmath316 . for positive @xmath317 ,
these processes lead to a uniform energy gain ( @xmath318 ) when the bosons are created with opposite phases on the sublattices @xmath101 and @xmath308 [ fig .
[ sublattices](a ) ] . the considerations for @xmath285 mirror those for positive @xmath317 with a change of the decomposition in two sublattices : for negative @xmath317 the uniform energy gain
occurs when the bosons are created with opposite phases on the sublattices @xmath319 and @xmath320 [ fig .
[ sublattices](b ) ] . employing the definitions and observations introduced earlier in this section , the simplest wave functions for approximating the ground state in the presence of @xmath195 are @xmath321 where @xmath322 correspond to @xmath323 .
the coefficients @xmath135 and @xmath324 are determined from the normalization of the corresponding wave functions [ eq . ( [ norm ] ) ] .
[ sublattices ] illustrates the sublattice decompositions of the checkerboard geometry for the two cases : @xmath325 for @xmath326 ( @xmath327 ) and @xmath328 for @xmath285 ( @xmath329 ) . a similar decomposition into plaquettes @xmath330 ,
@xmath331 is obtained by a @xmath332 rotation of fig .
[ sublattices](b ) .
the wave functions @xmath333 and @xmath334 represent linear combinations including all possible two plaquette resonance processes , the phase relations of which are specified by fig .
[ sublattices ] .
although these are superpositions of many states from the manifold of dimer coverings , the number of participating coverings remains a very small fraction of the total , and the selected linear combinations are vbc states . in the pseudospin @xmath335 space generated by the two possible states of each uncrossed plaquette of @xmath314 , @xmath327 and @xmath334 are analogous to the classical , magnetically ordered nel [ @xmath336 and collinear [ @xmath337 or @xmath338 states polarized in the @xmath339direction . in this space
@xmath314 corresponds to one of the fully polarized ferromagnetic states ( such as @xmath340 for all @xmath309 ) .
however , @xmath327 and @xmath334 are not exactly the classical nel and collinear states because two nearest neighbor uncrossed plaquettes can not be `` flipped '' independently of each other , _ i.e. _ @xmath341 if @xmath309 and @xmath342 are nearest neighbor uncrossed plaquettes . in the pseudospin representation the two plaquettes can not be simultaneously in the @xmath343 state , as a result of which these states are projected out of the superpositions corresponding to the classical nel and collinear states to obtain @xmath327 and @xmath334 .
we stress that the symmetries of @xmath344 and @xmath334 for @xmath323 are quite different .
( crosses ) and @xmath308 ( circles ) , which dictates the form of the variational wave function ( [ var_form ] ) for @xmath345 ; ( b ) decomposition into sublattices @xmath319 ( crosses ) and @xmath320 ( circles ) which yields the form of the low energy wave function for @xmath285.,width=340 ] the energies of the wave functions @xmath327 and @xmath346 obtained from eq .
( [ general_energy ] ) are @xmath347 the three processes represented in figs .
[ graph ] , [ graph2 ] , and [ twoloops ] give an extensive negative contribution to @xmath348 .
the lack of frustration among these different processes indicates that @xmath327 and @xmath334 are representative of the vbc orderings which are stabilized respectively for @xmath345 and @xmath285 .
we draw particular attention to the fact that in these valence bond orderings the four dimers around each uncrossed plaquette resonate between the two configurations shown in fig .
the relative phase is positive on one sublattice ( @xmath101 for @xmath345 and @xmath319 for @xmath285 ) and negative for the other ( @xmath308 for @xmath326 and @xmath320 for @xmath285 ) .
note that the rk process would lead to a different ground state in which this relative phase is the same for all the plaquettes of the dual lattice . in the pseudospin basis
this corresponds to the ferromagnetic state [ @xmath349 with polarization in the @xmath339direction for @xmath350 and the @xmath351direction for @xmath352 . with regard to possible four spin perturbation interactions , we remark that the dominant type of such a perturbation would be that contained in the second term of eq .
( [ esh ] ) . however , the effect of altering @xmath353 is trivially identical to the perturbation of eq .
( [ heisenberg_perturbation ] ) , and the results derived throughout this section may be applied directly to eq .
( [ esh ] ) with a sign inversion of @xmath354 relative to @xmath317 .
we now summarize the extension of our results to the pyrochlore lattice . here
any given dimer covering leads to precisely the same , uniform energy shift ( [ uniform ] ) when a supplementary nearest
neighbor heisenberg interaction is inserted . when superpositions of dimer coverings are considered , the elementary graph ( or analog of the rk process ) circumscribes a hexagon formed by six neighboring tetrahedra .
this elementary hexagonal loop replaces the sides of the uncrossed plaquettes on the checkerboard lattice . as for the checkerboard ,
the graph corresponding to the elementary process is simple , implying that this process is not generated by the perturbation @xmath195 .
once again it is necessary to consider larger loops to find the valence bond ordering stabilized by the perturbation .
as stated above , the local zero divergence constraint defining the low energy subspace around the klein point implies that the low energy theory induced by a general perturbation is always a compact u(1 ) gauge theory . following the argument of polyakov , the phase stabilized by a general perturbation is always confining at @xmath1 . for the spin models we consider ,
this confining phase consists of a vbc configuration .
the perturbation @xmath195 treated in this subsection is a particular example of this general observation , and also perhaps the most physically relevant .
the effect of a general perturbation in the space of interaction parameters is illustrated in the schematic phase diagram of fig .
[ pdiag ] .
we conclude this section with a summary : we have introduced a loop based approach to account for the physical processes in short ranged quantum dimer systems with only local interactions , which is powerful and general .
we have used this procedure to compute energies and variational wave functions in the non orthogonal basis of dimer coverings on the checkerboard lattice .
these considerations include all of the physically relevant perturbations of the klein hamiltonian , as discussed in sec .
ii , either explicitly or by direct analogy .
we conclude that the general case is one in which vbc phases , which are gapped and confined , are stable away from the klein point .
at issue for the physical properties of a system is rather less the behavior at @xmath1 than that at finite temperature . in the preceding section we devised a systematic , purely topological , diagrammatic method to compute the energies of all variational wave functions in the dimer basis .
this procedure is exact for the degenerate manifold of basis states , including most importantly when this is non orthogonal , and as shown in sec .
v presents a tractable problem at zero temperature . at finite temperatures
the need to evaluate brackets between all possible states in the hilbert space presents a more serious problem . in this section
we discuss the extent to which progress is possible in this regime , by first illustrating how to establish the existence of a finite temperature region of critical behavior for general pertubations , and then consider a simple , anisotropic perturbation to ( a ) demonstrate explicitly the existence of this critical region and ( b ) derive its approximate phase boundaries . in the remainder of this section
we set @xmath355 and for rigor assume implicitly that @xmath356 .
@xmath9 is the gap scale separating the ground state energies from the lowest excited states .
the presence of a spectral gap in a klein model was shown in @xcite .
consider a general perturbation @xmath357 which augments the klein point hamiltonian to give @xmath358 eq .
( [ heisenberg_perturbation ] ) provides one specific example .
we employ the notation @xmath0 ( sec .
iv ) to denote the small energy scale of @xmath357 . in what follows , we prove that unless the ground states of @xmath359 are ordered , in the sense that the different ground states are distinguishable by local order parameters , then a region of finite temperature criticality must occur if the perturbation is not linear in an order parameter .
we begin by stating that the system displays critical correlations ( i ) when @xmath360 at any finite temperature and ( ii ) at @xmath361 .
we also note that ( iii ) any local order parameter vanishes for any fixed @xmath0 at sufficiently high temperatures @xmath362 .
finally we assume ( iv ) that at sufficiently high temperature the two point correlator of an order parameter decreases monotonically in temperature at long enough distances . while ( i ) , from sec .
iv , and ( iii ) are self
evident , the demonstration of statement ( ii ) represents the core of the proof : because critical correlations appear in the high temperature limit , ( iii ) and ( iv ) imply that correlations at any finite but high enough temperature can not be weaker than algebraic .
thus the system exhibits critical correlations at sufficiently high but finite temperature .
the derivation of statement ( ii ) proceeds from the definition of the relevant order parameter . by analogy with the classical dimer polarization operators presented in sec .
iiic , we define the corresponding quantum operators . for a given plaquette a formed by the four sites @xmath363 appearing in counterclockwise order ,
the polarization is given by @xmath364 .
\end{aligned}\ ] ] the dimer
dimer correlation function between two crossed plaquettes @xmath101 and @xmath308 is then given , under the assumption that all states in the klein manifold are linearly independent @xcite , and at finite temperatures @xmath365 , by the expectation value @xmath366 } { tr \big[e^{-\beta h}\big ] } \nonumber \\
\lim_{t/\epsilon \to \infty } \chi_{ab } & = & \frac{1}{n_{g } } \sum_{|\psi \rangle } \langle \psi| \vec{p}_{a } \cdot \vec{p}_{b } |\psi \rangle \nonumber \\ &
\propto & ~ |r_{ab}|^{-d } , \label{explain_ortho}\end{aligned}\ ] ] with @xmath367 the separation between the centers of the plaquettes @xmath101 and @xmath308 , @xmath90 the spatial dimensionality , @xmath81 the number of states in the ground state manifold ( sec .
iiia ) and the sum taken over a complete set of orthonormal states @xmath368 . at the klein point ,
the correlation function @xmath369 is algebraic , as shown in sec .
the essential property underlying the physics of the finite temperature derivation is that the value of @xmath369 is the same for all temperatures when the magnitude @xmath0 of the external perturbation vanishes .
this is a consequence of the fact that all states within klein
point manifold carry equal weight , whence @xmath369 is given by the last two lines of eq .
( [ explain_ortho ] ) and the system is manifestly critical . as the klein
point limit is approached for @xmath370 , this average becomes equivalent to a sum over all states spanning the klein
point hilbert space and its value is precisely the same as for the @xmath190 case when @xmath0 is taken to zero . here
the probability of any state @xmath206 as given by the finite temperature density matrix is @xmath371/z = 1/n_{g}$ ] . for @xmath370
the correlation function @xmath369 then takes the form specified by eq .
( [ explain_ortho ] ) , decaying algebraically to @xmath372 as stated in ( ii ) above .
although it is not relevant to the above derivation , we remark briefly that topological rules similar to those of sec .
v may be devised for the computation of the dimer dimer correlation function .
as a consequence of the two scalar products appearing in each term of @xmath369 , two permutation operations can be performed on the loop formed by two states ( as opposed to the single operation discussed in sec .
v ) . a similar set of topological rules for the computation of @xmath369 will be provided elsewhere . the use of statements ( ii ) and ( iii ) for arbitrary pertubations
is assisted by considering the form of general order parameters .
let @xmath373 denote the ground states of @xmath359 and @xmath374 the order parameter by which they are distinguished . in a system which displays local order , @xmath375 is an operator which depends on a finite number of fields in a sphere ( or disk ) of finite radius @xmath376 which surrounds a given point @xmath92 ) .
a general algorithm for the construction of the order parameter by examining reduced density matrices is given in ref .
@xcite . at @xmath1 within any ground state , @xmath377 for separations @xmath378 .
following eq .
( [ explain_ortho ] ) , @xmath379 where @xmath380 denotes a set of pure dimer states and @xmath381 a complete set of orthonormal states spanning the klein point basis .
equation ( [ average_explicit ] ) makes use of the overcompleteness relation @xmath382 with @xmath383 the unit operator in the klein
point subspace [ and @xmath384 the projection operator onto the subspace of net spin @xmath69 ( sec .
ii ) ] ; eq . ( [ kp ] ) expresses the fact that the dimer states completely exhaust the klein
point subspace , as proven in ref .
@xcite . as emlpoyed in @xcite ,
all klein model ground states can be expressed as a sum of projection operators on these ground states which decompose them into ( young tableaux ) sectors which are well defined under the permutation of pair sites .
the presence of a singlet dimer between two sites enforces antisymmetry amongst the pertinent lattice sites . as evident in expressing the ground states in terms of the irreducible representations of the permutation group , the most general operations allowed within this basis amount to permutations .
@xcite as the most general permutation can be written as a product of pairwise permutations and as @xmath385 , where @xmath221 is the operator permuting sites @xmath76 and @xmath77 , the most general permutation is a functional of scalar spin products .
for an operator @xmath375 which is a functional of spin products of the form @xmath386 , the correlation function @xmath387 is equivalent to a correlation function between dimer products . from the analogy to a dipolar system ( sec .
iii ) , correlation functions containing more scalar spin products decay algebraically according to the strength of the coulomb interaction between more complicated dipole configurations .
moments of higher order than dipolar are important for those dimers which are close to each other but far from all other dimers .
all correlation functions thus decay algebraically with the dimer dimer separation . the analysis based on eq .
( [ explain_ortho ] ) may now be extended to more general correlation functions which involve the order
parameter fields @xmath375 .
returning to statement ( iii ) , as in our earlier considerations for the polarization @xmath388 this is based on the result that any finite temperature expectation value @xmath389 will decrease monotonically with increasing temperature , and asserts that when the expectation value of the order parameter vanishes , @xmath390 , the corresponding energy scale sets the transition temperature @xmath391 [ ( iii ) ] .
this quite universal property is reviewed briefly in ref .
we reiterate [ ( ii ) ] the fact that in the high temperature limit , where correlations are weakest , the correlation function @xmath392 retains a decay which is only algebraic in the separation @xmath393 . from these results
it is clear that there is a finite temperature region surrounding the `` klein line '' @xmath394 , and specified explicitly by @xmath362 , where critical correlations appear .
we conclude the general discussion by noting that for a perturbation which is linear in a general order parameter , @xmath395 , it is possible that the two point correlation function takes a finite value , @xmath396 as @xmath397 , at any finite @xmath187 .
this type of perturbation may then yield correlations which are algebraic only in the limit @xmath398 @xcite . in the next subsection we consider a particular perturbation
whose effect can be approximated by the anisotropic six vertex model to show that the dimer correlations may also remain critical up to a threshold value of the perturbation . in this subsection
we use a specific anisotropic perturbation to investigate the nature of the phase diagram .
we will illustrate explicitly that the phase diagram contains a finite temperature critical region , exemplifying the general concepts of the preceding subsection , and then derive its approximate phase boundaries .
we consider a perturbation of the form @xmath399 where @xmath400 denotes all diagonal pairs on the same crossed plaquette .
this type of interaction emulates the weakening ( for @xmath401 ) of exchange interactions between neighboring diagonal sites on a checkerboard plaquette relative to those of horizontal or vertical spin pairs , or a lifting ( to tetragonal ) of the cubic symmetry in a pyrochlore tetrahedron . focusing for specificity on the checkerboard geometry , for @xmath402
it is clear that in the regime @xmath403 , the effect of @xmath404 is to favor certain dimer , or equivalently six vertex , configurations over others , and in particular that for @xmath405 the diagonal contribution , @xmath406 , favors singlet formation on the diagonal bonds of all plaquettes . because off
diagonal brackets are significantly smaller than diagonal contributions , these may be neglected .
the favored ground states in this case are precisely the two antiferroelectric states .
similarly , for @xmath401 the single dimer on each crossed plaquette lies preferentially on the vertical or horizontal bonds .
because only the diagonal contributions @xmath407 are considered , the ground state degeneracy for @xmath405 remains exponentially large : the system supports that subset of all allowed ground states which corresponds to the ground state manifold of the klein model on the square lattice , the size of which scales with @xmath408 , where @xmath80 is the number of lattice sites @xcite .
we remind the reader that the states in this restricted set of dimer coverings , in which precisely one dimer is present on every square
lattice plaquette , are strictly orthogonal to each other in the thermodynamic limit .
the general case has a very much larger manifold of ground states , the classification of which proceeds by the line representation presented in sec .
the illustration that the phase diagram in the presence of this perturbation contains a finite temperature critical regime is based on the exact limits examined in subsec .
( [ finite ] ) and constitutes an explicit implementation of this discussion .
we begin by noting , as before , that when @xmath409 the system remains at the klein point for all temperatures , displaying critical correlations , and the analysis of sec .
iv is applicable .
the susceptibilities @xmath410 are a direct probe of dimer correlations : at @xmath409 the correlation function @xmath369 is critical for all @xmath89 . with reference to the considerations below eq .
( [ h_d ] ) , for @xmath405 , @xmath411 increases in magnitude relative to its klein point value for diagonal dimers only .
thus , _ ( i ) _ the decay of @xmath412 with distance is no faster than algebraic .
however , no vbc order is possible at high temperatures @xmath413 , and thus _
( ii ) _ @xmath414 as the plaquette separation @xmath415 .
conditions _ ( i ) _ and _ ( ii ) _ taken together imply that for @xmath413 the system exhibits algebraic correlations .
similar considerations apply when @xmath401 , where @xmath416 $ ] increases relative to its klein point value , corresponding to correlations of horizontal or vertical dimers .
we thus conclude that for @xmath417 the system is critical .
these statements are completely general , and provide one explanation for why it is that the six vertex model always has critical , and only critical , correlations at sufficiently high temperatures .
@xcite we now derive the approximate forms of the phase boundaries of the critical region .
this calculation is not exact for the reasons stated at the start of this section , and is included to provide a qualitative indication of the nature of the phase diagram .
we begin with an explicit statement of the approximations employed in this derivation . _
( i ) _ in contrast to sec .
v , for this finite temperature derivation the non orthogonality of the different dimer states is neglected . at the klein point ( @xmath409 )
all dimer configurations carry equal weight and their orthogonality is irrelevant , a fact exploited in sec .
v. all of the properties of the system are a consequence of the entropic considerations presented in sec .
iv . however , away from the klein point the configurations no longer have equal probability ( in the example of eq .
( [ h_d ] ) , states with diagonal dimer configurations are favored when @xmath405 , and suppressed when @xmath401 ) .
_ ( ii ) _ off diagonal contributions in the full hamiltonian , _
i.e. _ terms of the form @xmath418 with @xmath419 , are neglected . when mapped to the six vertex representation , provided that the conditions @xmath420 are satisfied , the anisotropic perturbation of eq .
( [ h_d ] ) is reflected in a change in the relative weights of the six vertex configurations ( which at low temperatures are the only states with non vanishing weights ) . at @xmath1 , for @xmath405 these nonuniform weights are represented by vanishing energies for four of the vertex types while the other two contribute an equal and finite energy , and conversely for @xmath401 . .
neglecting both non orthogonality and off diagonal contributions , an antiferroelectric valence bond crystal ( afe vbc ) ordering is stabilized below @xmath421 for @xmath422 .
the system is critical at all points within the region shaded in light grey . in the absence of off
diagonal processes , it remains critical to zero temperature for @xmath401 ; the qualitative effect of these processes is to remove the low temperature critical behavior by stabilizing a vbc ordering in the region shaded dark grey . while the phase boundaries shown above are approximate , the existence of a finite temperature critical region is not ( sec .
via).,width=415 ] with reference to fig .
[ rules ] , we choose the energies of the six vertex configurations displayed to be @xmath423 and @xmath424 in the presence of the perturbation .
the correlations and thermodynamics of this model are best analyzed through the fugacities @xmath425 = \exp[- \beta \epsilon_{2 } ] , ~b = \exp [ -\beta \epsilon_{3 } ] = \exp [ -\beta \epsilon_{4}]$ ] , @xmath426 = \exp[-\beta \epsilon_{6}]$ ] , where @xmath427 is the inverse temperature . the fugacities determine a parameter @xmath428 whose value determines in turn the phase in which the six vertex model lies , @xmath429 following lieb s original solution of the six vertex model @xcite , all models with @xmath430 are in the disordered phase . such small@xmath431 systems , including the ice model relevant for the isotropic checkerboard lattice where @xmath432 , are in the disordered phase and exhibit power law correlations .
by contrast , for @xmath433 the system adopts an ordered configuration where all arrows have a definite chirality ( such as up and to the right or down and left ) . in the system with perturbation @xmath434 ,
@xmath435 while @xmath436 $ ] , and one finds that if @xmath437 then @xmath430 , indicating that the system is in its disordered phase . in this case
we find power
law correlations among dimer pairs and the system is trivially critical for all non zero temperatures when @xmath438 , or critical only at sufficiently high temperatures [ whose lower bound is given by eq .
( [ t_6v ] ) ] whenever @xmath405 .
when @xmath1 , one finds that for all @xmath401 the system lies precisely on the boundary between the ordered and the critical phase @xmath439 .
equation ( [ t_6v ] ) defines a finite region within the @xmath440 plane ( fig .
[ jd ] ) in which the system is critical .
the above results are applicable under the conditions that all relevant states lie within the klein basis manifold , _ i.e. _ @xmath441 , that the effects of non orthogonality are neglected , and that only the diagonal corrections induced by @xmath442 are considered .
although these are small , off diagonal contributions of the type latexmath:[$\langle \psi_a| h_d in removing the `` marginally critical '' @xmath1 correlations obtained for @xmath401 , and stabilize once again a particular valence bond ordering pattern of the types considered in detail in sec .
v. in this situation the marginally critical line may be moved to finite temperatures , separating the low@xmath187 ordered phase from a finite@xmath187 critical regime . in summary , it is possible to demonstrate rigorously that specific types of perturbation around the klein point lead to an exotic , exactly critical regime .
quantitative calculations of the properites of this regime are not exact , but approximations of the kind applied here may be used for the qualitative determination of quantities such as the boundaries of this phase .
the pyrochlore ( a@xmath10b@xmath10o@xmath11 ) and spinel ( ab@xmath10o@xmath12 ) structures are very common among magnetic oxides , occurring for a wide variety of transition and rare earth metals as both the a and b ions and thereby offering a range of valence states and magnetic moments @xmath444 in a pyrochlore lattice of interacting spins . because the orbital contribution to the magnetic moment of transition metal ions in the 3@xmath90 series is quenched by the crystal
field splitting , the spin
orbit coupling has only a perturbative effect and the relevant physical spin models are effectively su(2)invariant . while the link to a physical pyrochlore or spinel system satisfying all the necessary criteria remains to be found ( sec .
ii ) , we have shown here that for these structures a class of su(2)invariant @xmath6 models of this type , whose local interactions emerge directly from the simplest hubbard hamiltonian , gives rise to spinon excitations which propagate in the full lattice . at the root of the exotic behavior exhibited by this type of system
is the massive ( extensive ) degeneracy of the ground
state manifold in the vicinity of the klein critical point .
we have gained physical insight into the multiple essential aspects of this problem by exploiting exact mappings between the spin system under consideration and other physical systems @xcite .
in particular , because the ground state sector of the @xmath6 klein models on the pyrochlore lattice obeys ice rules of the type discussed in sec .
iii , it is possible to exploit analogies with six vertex models @xcite , string gases and u(1 ) gauge magnets to make a number of powerful qualitative and quantitative statements .
the local constraint of zero divergence imposed by the ice rules implies that the low energy sector is described by an effective u(1 ) gauge theory in the neighborhood of the klein point .
this theory is , however , not necessarily the minimal u(1 ) gauge magnet , whose terms involve only the smallest loops , which is usually invoked in these cases @xcite . the variational approach which we introduce and employ in sec .
v indicates that in some cases , as a result of the non orthogonality of the singlet dimer coverings , processes involving larger loops are relevant for realistic physical perturbations .
although the spinons are deconfined excitations only at a single point ( the klein point ) of the quantum phase diagram ( @xmath1 ) , this deconfinement exists over a finite region of the phase diagram due to thermal fluctuations which give rise to a dilute plasma phase of spinons .
the spinon spinon correlations in this region are dominated by the @xmath445 klein critical point , whose critical fluctuations are of entropic origin .
the primary characteristic of the klein point is the stabilization of a low energy manifold of states satisfying a local constraint of zero divergence , or equivalently the ice rules @xcite , and the removal of any low energy scale . as a consequence of both conditions , the system becomes critical due to entropic fluctuations induced even at infinitesimal temperatures .
these features are directly relevant for the potential observation of spinons in two or three dimensional systems . in the case of the pyrochlore lattice models considered here
, the deconfined phase is expected to appear near a transition between two valence bond crystals .
similar considerations were applied to the @xmath6 model studied in ref .
@xcite , the difference in that case being that the spinons propagate along one dimensional paths because the dimensional reduction at the corresponding klein point is incomplete .
it is also interesting to note that an exact deconfined quantum critical point ( qcp ) @xcite can be induced by adding a zeeman term to the klein hamiltonian of eq .
( [ ho ] ) .
the qcp is induced at the critical magnetic field @xmath446 which closes the singlet
triplet spin gap .
the excited two spinon deconfined state becomes the ground state at @xmath446 . for higher values of the field ( @xmath447 ) ,
the concentration of spinons becomes finite and the degeneracy between the underlying singlet dimer coverings is lifted .
consequently , spinon spinon confinement is restored and the resulting triplet pairs are expected to condense , giving rise to an xy type antiferromagnet ( xy af ) in the plane perpendicular to the applied field . in the presence of a magnetic field ,
the system at the klein point displays a true deconfined qcp between a line of classical critical points in the region @xmath448 and a magnetically ordered region at @xmath447 with two coexisting order parameters .
a schematic phase diagram illustrating the finite extent of the deconfined regime is shown in fig .
[ pdiagb ] . very different and much more conventional behavior
is obtained away from the klein point , where the ground state degeneracy is lifted .
here the region @xmath448 has a vbc ground state with gapped magnon excitations , separated by a conventional qcp from an ordered phase with spin wave excitations .
the ideas and concepts connected with deconfinement at quantum critical points have formed an extensive recent literature best summarized in ref . @xcite . as noted in sec .
i , many of these studies depart from effective u(1 ) gauge theoretical treatments rather than microscopic models , and it is not yet clear that specific systems exist which realize the desired deconfinement properties . the klein point in certain pyrochlore lattice models was considered briefly in ref .
@xcite , although these authors did not dwell on either the detailed physical properties of such a point or on their origin .
we have found that the mapping used in this study to a quantum dimer model with only rk interactions is not in fact justified for a heisenberg antiferromagnet in a pyrochlore geometry .
the scenario of the `` constrained entropic critical point '' introduced recently in ref .
@xcite contains certain parallels to the physics of the klein point with regard to high temperature constraints .
however , this is said to be a `` top down '' construction based on different energy scales , some of which are present in a number of the effective models considered , and does not contain specific properties emerging from a microscopic hamiltonian for all temperature regimes .
we conclude this discussion with a speculation . the exponential ground
state degeneracy [ eqs .
( [ pauling.],[elliot . ] ) ] and resulting extensive configurational entropy which we have found suggest , but certainly do not mandate , that glassy spin dynamics may occur naturally in pyrochlores .
we stress that these systems are uniform and disorder free but frustrated , whence this suggestion would reinforce the widely held belief that frustration rather than disorder is the fundamental requirement for the dynamical properties of structural glasses .
the basic premise of the vogel fulcher ( vf ) form of glassy dynamics is that the relaxation times appear to diverge at a temperature which correlates well with the intercept ( @xmath449 ) of the extrapolated entropy of the super cooled liquid to that of the solid . according to a prevalent line of reasoning @xcite ,
as the system is cooled the configurational entropy first becomes extensive at @xmath450 , the onset temperature of multiple local free energy minima ; at a lower temperature , @xmath449 , these minima become stable and the configurational entropy vanishes .
if an exponentially large number of metastable states is found for @xmath451 , one may invoke the analysis of entropic droplets @xcite to obtain the characteristic free energy barrier height @xmath452 linearizing the extensive configurational entropy , @xmath453 leads to vf dynamics @xcite with relaxation times @xmath454 .
\label{vft}\ ] ] here @xmath449 a temperature scale close to the kauzmann temperature @xcite at which an `` ideal glass transition '' would occur , and where the extrapolated entropy of the assumed liquid undergoes a crisis .
we stress that this very general derivation of a systematic definition for glassy behavior , by consideration of free energy barriers @xcite , does not require that the different ground states be linked only by infinite length processes .
( however , if only the spatially longest processes were operational this would indeed lead to very robust , exponentially slow quantum dynamics , as first noted by chamon and coworkers @xcite . ) for the pyrochlore system we have shown that there exist finite
length tunneling processes linking different ground states within the highly degenerate low
energy sector , and thus suggesting an obvious candidate system for this type of glass .
we comment that the physical processes indicated here , namely of glassy spin dynamics arising from frustration in a periodic lattice , are similar to early ideas of the `` topological spin glass '' of ref .
empirically , glassy dynamics are strongly indicated in a number of frustrated magnetic systems , of which we quote only a selective list : evidence of phases with certain glassy characteristics has been reported in some pyrochlores , including y@xmath455mo@xmath455o@xmath456 @xcite , in the stacked kagome layer system srcr@xmath457ga@xmath458o@xmath459 @xcite , and in the kagome bilayer compound ba@xmath455sn@xmath455zncr@xmath460ga@xmath461o@xmath462 @xcite .
further indications for glassy behavior are found in some quasi triangular antiferromagnets , including niga@xmath455s@xmath463 @xcite . in conclusion , we have performed a detailed analysis of a physically motivated quantum spin model with only near neighbor interactions on the two and three dimensional pyrochlore lattices .
this reveals a wealth of exotic behavior which can be traced to the extensive degeneracy of the system at the klein point and thus to a complete dimensional reduction .
the complete absence of quantum fluctuations in the ground
state manifold at the klein point leads to a new type of classical critical point , with exactly known critical behavior , which we stress emerges from a highly frustrated , microscopic quantum spin model .
the critical correlations at the klein point are driven by entropic ( thermal ) fluctuations , leading to an effective plasma phase with coulombic interactions .
these are not confining , whence the elementary excitations are deconfined spinons which propagate freely in all directions ( a further manifestation of complete dimensional reduction ) .
this high dimensional fractionalization would also be manifest as a spin charge separation in the dilute limit of added holes if the carrier hopping were much smaller than the magnetic energy scale @xmath9 .
physically relevant perturbations away from the klein point lead in general to confined phases of static valence bond order with spin gaps to @xmath119 excitations .
however , at finite temperatures in the vicinity of the klein point , the classical criticality is dominant and deconfinement persists over a finite region of the phase diagram .
this type of deconfined behavior goes well beyond the simple , thermal decoupling of a system ( to obtain quasi
one dimensional behavior ) in that the classical critical point is driven to @xmath130 : the microscopic origin of deconfinement lies in the proximity of the system to a klein point , and spinon propagation remains @xmath90dimensional .
we have obtained our conclusions , and as a result a rather complete picture of a distinctive paradigm for classical and quantum criticality , from a number of rigorous techniques .
the dimer coverings of the checkerboard and pyrochlore lattices may be treated , through their connection with the ice rules , by both the six vertex mapping and a line representation , which exploit the underlying topological order of the physical system to classify the states of the manifold .
we have developed a loop representation of the physical processes in the dimer basis which has very general applicability to calculations involving the non orthogonal dimer states ; this gives a clear and intuitive picture of all contributing local processes , reflected in the sizes of corresponding loops , and a straightforward but rigorous set of rules systematizing their computation .
we thank b. kumar for helpful discussions .
we also wish to thank f. pollmann and k. shtengel for their enthusiastic interest and for encouraging us to emphasize the absence of the rokhsar - kivelson type processes in our bare systems .
this work was sponsored by the u.s .
doe under contract no .
w7405eng36 and by pict 0306343 of the anpcyt of argentina .
two months after the initial appearance of our work @xcite , pollmann and coworkers reported , in a series of preprints @xcite , on a spinless fermi model which possesses several of the qualitative features which we found here for spin systems on the pyrochlore and checkerboard lattices .
we provide a brief summary of selected contributions to the pyrochlore problem .
the case of an ising antiferromagnet on a pyrochlore lattice was investigated by anderson @xcite , who noted the similarity to the spin ice problem .
analytical treatments of this model @xcite have yielded the precise form of the correlation functions , which have been confirmed by numerical studies @xcite .
the classical antiferromagnet has been investigated by the large@xmath80 technique @xcite , and quantum heisenberg spins on the pyrochlore structure have been addressed in a number of studies @xcite by perturbative treatments based on the ising result with sufficiently large transverse terms .
these approaches lead to the rich physics of ring exchange interactions @xcite , as well as to a rather elegant charge model @xcite , and suggest the strong possibility of fractionalized excitations . while some approaches @xcite indicate the absence of spin order in the quantum pyrochlore antiferromagnet , numerical studies suggest specific singlet ordering patterns for the ground states of certain @xmath6 antiferromagnetic models @xcite .
specific results for simple , microscopic , isotropic , nearest
neighbor quantum spin hamiltonians are lacking . contradicting the support for fractionalization , the large@xmath444 technique suggests for the classical heisenberg antiferromagnet on the pyrochlore and checkerboard lattices @xcite that magnetic order
may be established at low temperatures , driven by an order by disorder @xcite effect where tunneling between states lifts the classical ground
state degeneracy and stabilizes order .
recent studies @xcite of systems similar to the effective spin models of ref .
@xcite have provided rigorous demonstrations of this effect for the classical case ( _ i.e. _ with thermal rather than quantum fluctuations ) .
j. villain , solid state comm .
* 10 * , 967 ( 1972 ) ; f. h. stillinger and m. a. cotter , j. chem . phys . * 58 * , 2532 ( 1973 ) ; r. w. youngblood and j. d. axe , phys .
b * 23 * , 232 ( 1981 ) ; l. b. ioffe and a. i. larkin , phys .
b * 40 * , 6941 ( 1989 ) ; d. a. huse .
w. krauth , r. moessner , and s. l. sondhi , phys .
lett . * 91 * , 167004 ( 2003 ) ; c. l. henley , phys . rev .
b * 71 * , 014424 ( 2005 ) .
antiferromagnetic nearest neighbor interactions on the pyrochlore lattice have a strong , intrinsic geometrical frustration . for both classical and quantum spins
it is impossible to bring all the individual exchange interactions close to their minimal value simultaneously .
the competition of individual heisenberg interactions is exacerbated by the inclusion of the symmetric four spin interaction .
the extent of the frustration can be characterized quite generally by the ground
state degeneracy , and this is maximized at the point @xmath465 .
a discussion of the nature of the level crossing occurring at this point is contained in sec . v.
the reader may wish to verify the mapping between eqs .
( [ esh ] ) and ( [ symham ] ) by application of the standard @xmath6 identities @xmath466 the first relation follows from the anticommutation of different spin components at a single lattice site , @xmath467 and the second follows , among other derivations , from the identity @xmath385 , where @xmath221 is the operator permuting sites @xmath76 and @xmath77 . in the case of the checkerboard lattice
there is also a biquadratic spin term which involves the four sites of each uncrossed plaquette .
these processes are not included in the hamiltonian of eq .
( [ symham ] ) .
this lies outside the range of validity of @xmath468 as an effective low energy hamiltonian for @xmath38 ( @xmath469 for @xmath470 to @xmath471 ) .
however , the klein condition @xmath469 can be obtained for smaller values of @xmath31 , such that @xmath468 becomes a valid description of the low energy spectrum , if a ferromagnetic nearest
neighbor direct exchange interaction is added to the hubbard model .
such a direct exchange cotribution is always present in real systems .
d. j. klein , j. phys .
a : math . gen . *
15 * , 661 ( 1982 ) . for completeness
, we remark that klein s original work examined the total spin formed by sites which surround a given lattice site .
this is slightly different than the case that we examine in our work . here
, the projection operators act on the total spin of all sites compromising a basic building block such as the tetrahedron .
such a similar extended idea was employed in @xcite .
l. onsager and m. dupuis , in _
rendiconti della scuola internationale di fisica ( enrico fermi ) x coroso _ , societa italiana di fisica , bologna ( 1960 ) ; b. simon , _ the statistical mechanics of lattice gases _ , volume i princeton university press ( 1993 ) , esp .
two arbitrary states are in general non orthogonal , but their overlap is exponentially small in the length of closed loops obtained by superposing them .
linear independence of singlet dimer coverings has been proven for other lattices in ref .
@xcite .
we assume here that the diagonal elements of the perturbative term , @xmath472 , do not break the degeneracy ; this would be the case when , for example , @xmath472 is a heisenberg hamiltonian of the type given by the first term of @xmath19 .
in contrast to the mappings between quantum spin and dimer models , orthogonality between different dimer coverings is not a requirement in this case because the effective spinon
spinon interaction has an entropic origin and the entropy is determined solely by the number of linearly independent states . the energy of eq .
( [ symham ] ) when evaluated for a single tetrahedron at @xmath473 is @xmath474 for the singlet , triplet , and quintet states respectively . for @xmath475 ,
the energy difference between the singlet and triplet states is very much smaller than the gap to the quintet state .
g. rumer , gttinger nachr . 337
( 1932 ) ; s. liang , b. doucot , and p. w. anderson , phys .
lett . * 61 * , 365 ( 1988 ) ; b. sutherland , phys .
b * 37 * , 3786 ( 1988 ) ; m. kohmoto and y. shapir , phys . rev .
b * 37 * , 9439 ( 1988 ) ; r. mcweeny , _ valence bond theory and chemical structure ( studies in physical and theoretical chemistry , vol .
64 ) _ , eds .
d. j. klein and n. trinajstic ( elsevier , amsterdam , 1990 ) p. 13 ; m. havilio , phys .
b * 54 * , 11929 ( 1996 ) . a separate algorithm for the computation of energies and correlation functions in valence
bond states was developed shortly after this work by k. s. d. beach and a. w. sandvik , nucl .
b * 750 * , 142 ( 2006 ) .
the approach of these authors does not apply the topological rules exploited here , but offers insight into a number of other related points .
an introductory exposition to young tableaux is available in many texts ( e.g. @xcite , @xcite ) .
each of the klein model ground states is a superposition of a product of singlets@xcite . in the young tableaux notation , these ground states therefore lie exclusively in the two row sector in which the length of both rows is the same .
the top row labels in the young tableaux denotes sites in which up spins are antisymmetrized ( to form singlet dimers ) with down state spins of the belonging to site indices in the row just below it .
consider now any of these ground states ( @xmath476 ) . in order for an operator @xmath477 acting on these states ( @xmath478 ) to lie in the same young tableaux sector and allow for finite matrix elements @xmath479 in the klein ground state basis
, we must have that @xmath477 merely permutes the site indices on the top and bottom rows
. the operator @xmath477 can , of course , also do nothing at all ( and act as the identity ) ; this is the case for a uniform rotation operator which leaves the klein spin state invariant .
[ after all , any klein basis spin state is a global singlet ( @xmath480 ) as it is a superposition of a product of singlet dimers . ]
thus , the most general operator @xmath477 which has non - vanishing matrix elements in the klein ground state basis has components which are the a superposition of the most general permutation operations ( including the trivial permutation- the identity operator ) .
we re - emphasize that this follows as ( i ) the different young tableaux sectors are orthogonal to one another and that ( ii ) within the young tableaux pertinent to the klein basis sector , all sites on the top row of the tableaux lie in the up spin state while all sites in the bottom row ( whose length is equal to that of the top row ) are indexed by a down spin state .
no operators @xmath477 other than those involving a sum of permutations can link states in this young tableaux sector . in the absence of external symmetry breaking fields ,
the competition of energy and entropy is resolved universally in favor of disorder at sufficiently high temperatures ( symmetry restoration ) .
we summarize those parts of the physical origin of this phenomemon most relevant to the current considerations .
general disordering events ( similar to those in sec .
iv ) have positive energy penalties ( @xmath481 ) and entropy gains ( @xmath482 ) . at sufficiently high temperatures
@xmath187 , the free energy change due to the increase in disordering processes , @xmath483 , becomes negative : defects proliferate and order is destroyed . for general , discrete order parameters in a system of finite range interactions ,
the scaling of @xmath481 with defect ( domain wall ) size is no faster than that of @xmath482 , whence domain wall deconfinement leads to a lower free energy at sufficiently high @xmath187 . for continuous order parameters there
exist further , lower energy , deformations in addition to domain walls , which are already favorable .
thus there is always a temperature @xmath391 beyond which the free energy is lowered monotonically by contributions from disorder increasing processes . as a consequence , for @xmath484 the expectation value @xmath390 .
the development of long range order in @xmath485 [ @xmath486 at all finite @xmath187 ] in this case is not in conflict with finite temperature singularities in the free energy which may lead to algebraic correlations in quantities other than @xmath485 .
one example is afforded by the classical six vertex model in any symmetry breaking field , which is analogous to the parameter @xmath0 in @xmath472 . in this model
the field may cause all six states to have different energies , but criticality appears nevertheless at finite fields and temperatures [ see for example fig .
8.6 in ref .
@xcite ] .
we should remark that this finite @xmath187 critical behavior is distinct from the exact deconfinement present at @xmath488 albeit very similar to it . at @xmath489 , the entropic contribution @xmath490 to
the free energy leads to an effective ( marginally confining ) logarithmic interaction between monomers at the ice point .
however , as discussed in sec.([thermalklein ] ) , this logarithmic interaction is screened ( by virtue of a finite defect concentration at @xmath489 ) at sufficiently large distances .
t. r. kirkpatrick and p. g. wolynes , phys .
b * 35 * , 3072 ( 1987 ) ; * 36 * , 8552 ( 1987 ) ; t. r. kirkpatrick and p. g. wolynes , phys . rev . a * 40 * , 1045 ( 1989 ) ; t. r. kirkpatrick and d. thirumalai , phys . rev . lett . * 58 * , 2091 ( 1987 ) .
f. pollmann , p. fulde , e. runge , phys . rev .
b 73 , 125121 ( 2006 ) [ cond - mat/0604122 ] ; f. pollmann , p. fulde , europhys .
lett . , 75 ( 1 ) , pp . 133 - 138 ( 2006 ) [ cond - mat/0604666 ] ; f. pollmann , j. j. betouras , k. shtengel , p. fulde , cond - mat/0607202 ( 2006 ) ; f. pollmann , j. j. betouras , e. runge , p. fulde , cond - mat/0609122 ( 2006 )
|
the ground states of klein type spin models on the pyrochlore and checkerboard lattice are spanned by the set of singlet dimer coverings , and thus possess an extensive ground state degeneracy . among the many exotic consequences
is the presence of deconfined fractional excitations ( spinons ) which propagate through the entire system . while a realistic electronic model on the pyrochlore lattice is close to the klein point , this point is in fact inherently unstable because any perturbation @xmath0 restores spinon confinement at @xmath1 .
we demonstrate that deconfinement is recovered in the finite temperature region @xmath2 , where the deconfined phase can be characterized as a dilute coulomb gas of thermally excited spinons .
we investigate the zero temperature phase diagram away from the klein point by means of a variational approach based on the singlet dimer coverings of the pyrochlore lattices and taking into account their non orthogonality .
we find that in these systems , nearest neighbor exchange interactions do not lead to rokhsar - kivelson type processes .
|
it is well known that the all - particle spectrum of primary cosmic rays follows a power law of dj / de @xmath0 e@xmath1 , but steepens at energies around 4@xmath210@xmath3 ev where the power index @xmath4 changes sharply from @xmath52.7 to @xmath53.1 @xcite .
such structure of the all - particle energy spectrum is called the knee " , which is considered to be closely related to the origin , acceleration and propagation mechanism of cosmic rays . in order to explain the existence of the knee ,
many hypotheses and mechanisms @xcite have been proposed .
although all these approaches can well describe the knee structure , there are much discrepancies in the prediction of the individual components at the knee region .
thus , it is critical to measure the primary chemical composition or mass group at energies 50 - 10000 tev , especially , to measure the primary energy spectra of individual component and determine a break energy of the spectral index for individual component .
direct cosmic - ray measurements on board balloons or satellites are the best way to study the chemical composition , while the maximum energy they can cover is up to 10@xmath6 ev / nucleon at most due to limited detection area or exposure time .
we may have no choice but to rely on ground - based air - shower ( as ) measurements to study the primary chemical composition around the knee .
the study of cosmic - ray composition around the knee was done by a hybrid experiment of the emulsion chambers ( ecs ) , the burst detectors ( bds ) and the as array ( tibet - ii ) , where ecs and bds of total area 80 m@xmath7 were set up near the center of the as array and operated for three years @xcite . the threshold energy of ecs capable of analyzing the fine structure of as cores is about 1 tev , so that it is not difficult to separate the as events induced by light - component of protons and helium nuclei , while the energy range of primary particles is limited to be above @xmath5200 tev for protons and @xmath5400 tev for helium nuclei @xcite .
this experiment suggests that the flux of light component is less than @xmath530% of the total , resulting in that the knee is dominated by nuclei heavier than helium @xcite .
a demerit of this experiment is that there are few statistics of the high - energy core events due to the high detection threshold energy of the experiment as mentioned above . to improve this problem ,
a new air - shower core detector named yac ( yangbajing air shower core detector ) has been developed and improved so as to meet our requirements .
one important improvement is to lower the detection threshold energy of primary particles to several times 10 tev , about one order of magnitude smaller than the previous experiment . with this improvement
, the energy spectra of individual components measured by yac will overlap with those of direct measurements , which may help us to examine the knee of light component , such as proton knee " or helium knee " .
another important improvement of yac is its ability to count the number of shower particles passing through each detector in a wide dynamic ranging from 1 to 10@xmath8 particles , making it possible to observe the primary cosmic rays in the energy range from @xmath510 tev to @xmath510 pev . until now ,
we have constructed and operated yac - i as a prototype of full - scale yac comprising 400 core detectors . yac - i is a small array consisting of 16 core detectors which were placed near the center of the tibet - iii as array as shown in fig .
[ figure-01 ] , while being able to observe a lot of as - core events in the energy around @xmath9 ev by the operation of a few months .
as the primary composition around this energy region is fairly well known by the direct observations @xcite , the data from yac - i may be used to test the interaction models such as sibyll2.1 , qgsjetii-04 and epos - lhc being currently used in the monte carlo ( mc ) simulations . in this paper , we discuss the performance and sensitivity of yac for observing light - component spectrum of primary particles through mc simulations based on the yac - i experiment .
each , which cover the area of about 10 m@xmath7 .
each detector of yac - i consists of a lead plate with the thickness of 3.5 cm ( 6.3 radiation lengths ) , a supporting iron plate with the thickness of 0.9 cm ( 0.5 radiation lengths ) and a plastic scintillator with the thickness of 1 cm.,scaledwidth=80.0% ] the yac - i array consists of 16 core detectors which are placed on a 4@xmath10 square grid covering the area of about 10 m@xmath7 as shown in fig . [ figure-01 ] and has been operating since may 2009 , together with the tibet - iii as array .
each core detector of yac - i comprises a plastic scintillator with the size of 40 cm @xmath2 50 cm and a lead plate with the thickness of 3.5 cm ( 6.3 radiation lengths ) being put on the scintillator .
the lead plate is used to select as particles and cores capable of having sufficient energy to create cascade showers in the lead plate and pass through the scintillator .
the plastic scintillator in the core detector is divided into 10 pieces of the width of 4 cm and the scintillation lights are collected through wavelength shifting ( wls ) fibers as shown in fig .
[ figure-02 ] .
such design ensures the geometrical uniformity of detector response within 5 % .
the details about the hardware of yac detector is described in @xcite .
two photomultiplier tubes ( pmts ) of high - gain ( hamamatsu : r4125 ) and low - gain ( hamamatsu : r5325 ) are equipped to cover a wide dynamic range from 1 mip(minimum ionization particle ) to 10@xmath8 mips as seen in fig .
[ figure-02 ] . the corresponding linearity and saturation effect of pmt and scintillator were examined by use of cosmic - ray muons and electron beams provided by the beam facility of bepcii ( beijing electron positron collider , ihep , china ) @xcite .
the stability of the pmt gain was checked and corrected using cosmic - ray muons . in this experiment ,
the yac - i array works to observe air - shower cores and their accompanying ass are observed simultaneously with the tibet - iii as array .
the tibet - iii provides information on the arrival time and direction of each air shower and its as size corresponding to the energy of primary particle @xcite .
when a yac - i event is triggered , its accompanying as is simultaneously recorded by the tibet - iii array and the matching between yac - i and tibet - iii events is made by their arrival time stamps recorded by a gps clock .
we have carried out a full monte carlo ( mc ) simulation on the development of air showers in the atmosphere using the simulation code corsika @xcite ( version 7.3500 ) .
three hadronic interaction models , including sibyll2.1 @xcite , epos - lhc ( v3400 ) @xcite and qgsjetii-04 @xcite , are used to generate the air - shower events in the atmosphere . for the primary cosmic rays
, we examined three composition models , namely , he - poor " , he - rich " and gaisser - fit " models , in order to evaluate the systematic errors attributable to primary composition models .
the he - poor " model is based on the hd ( heavy dominant ) model mentioned in the paper @xcite , and it is slightly revised to match with the new all - particle energy spectrum @xcite , to be the new he - poor " model . the he - rich " model is the model b ( a lightly harder spectrum than the previous on by taking account of the nonlinear effects ) " mentioned in the paper @xcite .
the gaisser - fit " model is the three - population " model mentioned in the paper @xcite .
the proton spectra of the former two models are fitted to the direct measurements at the low energy and consistent with the spectrum obtained from the tibet - ec experiment at the high energy .
the he spectrum of he - poor model coincides with the results from runjob , but the he spectrum of he - rich model coincides with the results from jacee , atic2 and cream .
the gaisser - fit model fits to a higher he model ( almost same as our he - rich model ) at the low energy range and to the kascade - qgsjet data at high energy range in which light components ( p and he ) dominate in the chemical composition . in all models mentioned above
, each component is summed up so as to match with the all - particle spectrum with a sharp knee , which was obtained with the tibet - iii as array @xcite .
table [ table-1 ] is a summary of the fractions of the components ( p , he , medium and fe ) of the three composition models in given energy regions for three primary models .
the energy spectra of individual components ( or mass groups ) for three primary models are shown in fig .
[ figure-03 ] .
it is seen that all the individual components of the three models in the low energy range ( less than 100 tev ) are in good agreement with direct measurements while differ significantly at higher energy .
the all - particle spectra of three models , however , coincide with each other and reproduce the sharp knee structure as well .
@llccc composition & & @xmath11 ev & @xmath12 ev & @xmath13 ev + models & & ( % ) & ( % ) & ( % ) + & p & 31.5 & 22.7 & 9.6 + & he & 22.4 & 18.8 & 9.7 + & medium & 26.6 & 26.6 & 26.1 + & fe & 19.5 & 31.9 & 54.6 + & p & 31.1 & 26.3 & 10.0 + & he & 25.1 & 28.7 & 17.5 + & medium & 32.4 & 34.4 & 50.3 + & fe & 11.3 & 10.6 & 22.2 + & p & 32.8 & 29.0 & 19.6 + & he & 34.4 & 37.4 & 37.4 + & medium & 20.1 & 20.4 & 25.3 + & fe & 12.7 & 13.2 & 17.7 + in this simulation , primary cosmic rays at the top of the atmosphere within the zenith angles smaller than 60 degrees are thrown into the atmosphere isotropically and the minimum energy of primary cosmic rays is set to 40 tev . all shower particles in the atmosphere are traced down to the minimum energy of 1 mev .
the as events generated are randomly dropped onto the area of 32.84 m @xmath2 32.14 m , which is a 15 m wider in each side of the yac - i array .
this dropping area is large enough to collect the as events more than 99.5% under our core - event selection conditions ( see below in the text ) .
observation of the mc events is made with the same method as that of the experiment .
the detector responses to shower particles falling on the detectors of ( yac - i+tibet - iii ) array are calculated using the geant4 @xcite ( version 9.5 ) , where the detector performance , trigger efficiency and effective area are adequately taken into account based on the experimental conditions .
the number of charged particles passing through the scintillator is defined as the pmt output ( charge ) divided by that of the single - particle peak .
the single - particle peak is determined by a probe calibration @xcite using cosmic rays , typically muons .
the value of single - particle peak is measured as 1.98 mev for yac - i detectors and 6.28 mev for the tibet - iii detectors .
these values are used in this mc simulation .
the main purpose of this work is to check the sensitivity of yac to observe the light component of primary cosmic rays as well as to evaluate the systematic errors by adopting different primary composition models and interaction models mentioned above . for this
, we selected five combinations of interaction models and primary composition models .
four combinations of sibyll2.1+he - rich , sibyll2.1+he - poor , epos - lhc+he - poor and epos - lhc+gaisser - fit are to check the sensitivity of yac to the light component and uncertainties due to the adoption of the different composition models .
other three combinations of sibyll2.1+he - poor , qgsjetii-04+he - poor and epos - lhc+he - poor are to check the interaction models and also uncertainties under the same primary composition model .
it has value to point out here that there is no serious difference among the current interaction models on the particle production in the forward region and proton - air inelastic cross sections in our concerned energy region from 10 tev to @xmath14 tev since all the models are well tuned using recent accelerator data including lhc , while there are big differences among primary composition models because of a lack of direct observation data at energies above @xmath5200 tev .
the number of air - shower events generated for each model is 7.40@xmath210@xmath15 , 6.57@xmath210@xmath15 , 4.67@xmath210@xmath15 , 6.25@xmath210@xmath15 and 5.18@xmath210@xmath15 , respectively , as shown in table [ table-2 ] .
the analysis of these mc events was made by the same method used in the experiment .
information on the size @xmath16 and arrival direction of each air shower event hitting both yac - i and tibet - iii arrays can be easily obtained from the mc events observed with the tibet - iii as array simultaneously .
details of its analysis are found in the paper @xcite . from the yac - i array ,
we can obtain the following five quantities reflecting the characteristic of as cores : ( 1 ) @xmath17 , the number of `` fired '' detectors with @xmath18@xmath19 200 , where @xmath20 is the number of particles ( burst size ) observed by each core - detector ; ( 2 ) @xmath21@xmath18 , the total sum of @xmath20 of fired detector ; ( 3 ) @xmath18@xmath22 , the maximum @xmath18 among the fired detectors ; ( 4 ) @xmath23 , the mean lateral spread defined as @xmath23 = @xmath21@xmath24/(@xmath25 - 1 ) ; ( 5 ) @xmath26 , the mean energy - flow spread defined as @xmath27 , where @xmath28 denotes the number of particles observed in @xmath29-th fired detector and @xmath30 represents the lateral distance from the burst center ( @xmath31 ) , where ( @xmath31 ) = @xmath32 .
it is confirmed that the five quantities mentioned above are basic and enough to separate the light component ( p+he ) from others .
a use of ann - method @xcite may further improve the quality of separation .
@lcc & primaries & core events + & ( e @xmath19 40 tev ) & ( mode energy : @xmath5200 tev ) + sibyll2.1+he - rich & 7.40@xmath210@xmath15 & 64331 + sibyll2.1+he - poor & 6.57@xmath210@xmath15 & 47580 + qgsjetii-04+he - poor & 4.67@xmath210@xmath15 & 31928 + epos - lhc+he - poor & 6.25@xmath210@xmath15 & 42137 + epos - lhc+gaisser - fit & 5.18@xmath210@xmath15 & 49390 + in order to obtain the light - component spectrum using the data from both arrays of yac - i and tibet - iii , we select the high - energy core events by imposing the conditions of @xmath33 , @xmath34 , @xmath35 and @xmath36 .
the mode energy of primary particles producing such high energy core events is then estimated to be about 200 tev .
the statistics of the data - sets selected based on the five models are listed in table [ table-2 ] . shown in fig . [ figure-04 ] is the effective @xmath37 of yac - i array to observe the as - core events satisfying the event select conditions , where @xmath38 denotes the detection area and @xmath39 the solid angle .
the effective @xmath37 depends weakly on the model used , but its difference is found to be smaller than 25% in our concerned energy range . of yac
- i array to observe the light - component for various models used in mc . ]
we check the sensitivity of yac array to the interaction models and primary cosmic - ray models using the high - energy core events selected under the conditions discussed in the previous section .
it is well known that the absolute intensity of the total burst sizes depends sensitively on the increase of cross sections , inelasticity , and also on the primary cosmic - ray composition .
shown in fig .
[ figure-05]-(a ) is the integral total burst - size spectrum ( @xmath40 ( sumnb ) ) obtained by the respective mc model for comparison .
the @xmath40 spectra obtained by five mc models are compared each other by taking the flux ratio to that by the sibyll2.1+he - rich model in fig .
[ figure-05]-(b ) .
it is seen that the epos - lhc+gaisser - fit model gives the highest flux in all @xmath40 region . according to our mc simulation ,
the observed as cores in the size region of @xmath40 = 2@xmath41 - 4@xmath42 are produced mostly by the light component ( p+he ) with its primary energies of several times 10@xmath6 - 10@xmath3 ev .
the fraction of light component in the primary of this energy region is about 66% for gaisser - fit model while about 55% for the he - rich model and 42% for the he - poor model as seen in table [ table-1 ] .
it should be , however , noted that about 70% of the observed high - energy core events are originated by the light component , that is , the contribution from other nuclei is fairly small .
( sumnb ) spectra obtained from five mc models , ( b ) the intensity ratios of @xmath40 to that obtained by the sibyll2.1+he - rich model , ( c ) the mean energy - flow lateral spreads @xmath26 in the respective energy interval for five mc models.,title="fig : " ] ( sumnb ) spectra obtained from five mc models , ( b ) the intensity ratios of @xmath40 to that obtained by the sibyll2.1+he - rich model , ( c ) the mean energy - flow lateral spreads @xmath26 in the respective energy interval for five mc models.,title="fig : " ] ( sumnb ) spectra obtained from five mc models , ( b ) the intensity ratios of @xmath40 to that obtained by the sibyll2.1+he - rich model , ( c ) the mean energy - flow lateral spreads @xmath26 in the respective energy interval for five mc models.,title="fig : " ] here , we discuss the uncertainties due to different interaction models in reference to the spectrum of high - energy core events . before entering in discussion
, we should first remind that the production rate of high - energy core events is most sensitive to the energy per nucleon of primary particles .
thus , the mean energy per nucleon of helium nuclei is 1/4 of protons when compared at the same primary energy and also the interaction mean free path of helium nuclei in the air is about a half of that of protons . for the he - poor model , it is seen that the flux of protons is slightly higher than that of helium nuclei or almost same in the energy region over about 300 tev and also their power indices are almost same as learned from fig .
[ figure-03 ] and table [ table-1 ] .
if we combine this with above discussion , it may be allowed to ignore the contribution from helium nuclei to the core events observed in this primary model , that is , it is regarded as those produced by protons . under this assumption , it may be noticed from fig .
[ figure-05]-(b ) that the flux values by sibyll2.1 , qgsjetii-04 and epos - lhc using the same primary model of he - poor match well with an error of smaller than 10% , which may be attributed to uncertainties due to the interaction models used .
a deviation of qgsjetii-04 in the core - size region above about 2@xmath42 may be due to a low statistics of the events , that is , within a statistical fluctuation ( at most 2@xmath43 level ) .
on the other hand , when we arrange the light component flux of protons and helium nuclei in descending order in the @xmath44 -@xmath14 tev region , it becomes as gaisser - fit @xmath45 he - rich @xmath45 he - poor .
it is then confirmed that the flux of core events is in the same order as the primary light - component flux and also the shape of each primary spectrum is well reflected in the corresponding core - event spectrum , as seen in fig .
[ figure-05]-(b ) .
a typical example is seen in the case of gaisser - fit primary model .
this means that high energy core - event observation with yac is very sensitive to the primary light - component spectrum . a correlation between @xmath26 and @xmath40 is shown in fig . [ figure-05]-(c ) .
this figure tells us that the gaisser - fit primary model gives smaller lateral spread than others , while the qgsjetii-04+he - poor model gives larger spread than others . as protons with the long interaction
mean free path can penetrate deep in the atmosphere and produce as cores near the observation level , resulting in giving smaller lateral spread .
the gaisser - fit primary is light - component dominant as seen in fig .
[ figure-03 ] , so that the core spread by this model should be smaller than others . when the primary model is fixed as he - poor , the mean spread of qgsjetii-04 is slightly larger than that of sibyll2.1 .
this may slightly depend on the interaction model since the energy spectrum of secondary particles in the very forward region ( feynman @xmath46 0.1 - 0.3 ) produced at collision in the sibyll2.1 model is harder than that of qgsjetii-04 , that is , the former contains slightly larger number of very high - energy secondaries than the latter .
the difference of the intensity and lateral spread between both models could be attributed to the number of very high energy particles as those penetrate deep in the atmosphere .
it should be noted that the lateral spread of as - cores is mostly caused by coulomb scattering of shower electrons and positrons in the atmosphere , not by transverse momentum of secondaries produced at collisions except within the depth of 1 - 2 radiation lengths from the interaction point in the atmosphere . in connection with this ,
the energy loss of as cores generated by qgsjetii-04 may be faster than those by sibyll2.1 , resulting in giving lower as - core intensity . hence , a precise as experiment like the ( yac + tibet - iii as ) array will be able to examine the interaction models to some extent . in this analysis
, we use the ann technique to separate the light - component from others .
this method is shown to be quite effective for such purpose as confirmed by our previous works @xcite . in this ann analysis , we use the following seven quantities : ( 1 ) @xmath25 , ( 2 ) @xmath21@xmath20 , ( 3 ) @xmath47 , ( 4 ) @xmath23 , ( 5 ) @xmath26 , ( 6 ) @xmath16 and ( 7 ) @xmath48 ( zenith angle ) . these are input to the ann with 35 hidden nodes and 1 output unit . to train the ann in separating light - component ( p+he ) from other nuclei , the input patterns for light - component and others are set to 0 and 1 , respectively .
we then define a critical value of @xmath49 to calculate the corresponding purity and selection efficiency of the selected ( p+he)-like events .
= 0.4 . ]
@lcccc & primaries & core events & + & ( p+he)/all(% ) & ( p+he)/all(% ) & purity ( % ) & efficiency ( % ) + sibyll2.1+he - rich & 56.2 & @xmath50 & @xmath51 & @xmath52 + sibyll2.1+he - poor & 46.8 & @xmath53 & @xmath54 & @xmath55 + qgsjetii-04+he - poor & 46.8 & @xmath56 & @xmath57 & @xmath58 + epos - lhc+he - poor & 46.8 & @xmath59 & @xmath57 & @xmath60 + epos - lhc+gaisser - fit & 67.0 & @xmath61 & @xmath62 & @xmath63 + figure [ figure-06 ] shows the ann output distribution trained using the epos - lhc+he - poor model . as seen in this figure , the events with @xmath64 could be regarded as the ( p+he)-like events , and the average selection purity and efficiency over whole energy range of ( p+he)-like events are 95% and 76% , respectively .
table [ table-3 ] is a summary of the ratios of the ( p+he)/all before and after ann training analysis based on the five mc models over the whole energy range .
thanks to the performance of the yac - i array , the ratio of the ( p+he)/all before ann training ( core events ) has already reached @xmath570% by the core - event selection conditions ( @xmath587% for gaisser - fit model , as seen in table [ table-3 ] ) . with the ann training , the purity of selected ( p+he)-like events is further increased up to @xmath595% as learned from table [ table-3 ] .
this high quality data set is used for reconstructing the light - component primary spectrum .
table [ table-3 ] teaches us that after ann training the difference of the selection purity and efficiency is within @xmath56% among three hadronic interaction models , and @xmath58% among three primary composition models .
overall uncertainties due to ann training for the mc models are then estimated to be about 10% . using the ann trained by the epos - lhc+he - poor events , we select the ( p+he ) like events from all the observed events and
also obtain the selection purity and efficiency .
the primary energy @xmath65 of each selected events is then estimated using the as size @xmath16 obtained by the tibet - iii .
the relation between air shower size @xmath16 and primary energy @xmath65 is expressed as @xmath66 where the parameters of @xmath67 and @xmath68 are estimated from as events generated by the mc for the tibet - iii as array , while @xmath67 and @xmath68 depending on the zenith angle of air showers .
details of this procedure is described in the paper @xcite . shown in fig .
[ figure-07 ] is the correlation between @xmath65 and @xmath16 of the ( p+he)-like events selected by the ann method .
the solid line denotes the best fit curve for this correlation , and the parameter values of @xmath69 and @xmath70 are then obtained for air showers with @xmath71 .
the energy resolution is also estimated as about 25% at energies around 200 tev . ) and the estimated shower size ( @xmath16 ) of ( proton+helium)-like events based on epos - lhc+he - poor model with @xmath72 .
solid line shows the fitting result of @xmath65 = 0.794@xmath2 @xmath73 gev . ]
we also checked a dependence of the correlation of @xmath16 and @xmath65 on the interaction and primary composition model , and obtained the similar relationship in the other models .
we then found that there is less than 10% difference for the determination of the primary energy by use of different interaction and primary composition models .
figure [ figure-08 ] shows the estimated primary energy spectrum of the light - component ( p+he ) in comparison with the assumed primary spectrum .
it is seen that the estimated energy spectrum well reproduces the assumed one within about 10% errors .
almost same results are obtained for other mc models . in this work ,
we discuss the systematic errors coming from the models used in the mc simulation for deriving the primary ( p+he ) spectrum using the ( yac - i+tibet - iii ) array .
the systematic errors caused by each step of the analysis procedures are investigated including the dependence of the mc data on the interaction models , the primary composition models and the algorithms for the primary mass identification .
we summarize these systematic errors as follows : 1 ) errors due to the observation efficiency ( @xmath37 of ( p+he ) ) depending on the interaction and primary composition models are found to be smaller than 25% ; 2 ) errors due to the selection of ( p+he)-like events with ann training are about 10% , in which the model dependence on the purity and efficiency is totally included ; 3 ) errors due to the estimation of primary energy with use of the conversion from @xmath16 to @xmath65 are 10% , in which the dependence on both interaction and primary composition models are taken into account ; 4 ) errors due to the reconstruction procedures of the primary light - component spectrum from the observed core events are estimated to be smaller than 10% .
the total systematic errors are then estimated to be about 30% as the square root of quadratic sum of those four systematic errors , which may be somewhat overestimated because of a little correlation among four error estimation parts .
in this paper , we have carried out a full mc simulation to examine the capability of measuring the energy spectrum of primary light - component ( p+he ) of cosmic rays at the knee energies using the ( yac - i+tibet - iii ) array .
the models used in this mc simulation are sibyll2.1 , epos - lhc ( v3400 ) and qgsjetii-04 for the interaction and he - poor , he - rich and gaisser - fit for the primary cosmic - ray composition .
the corsika code was used to generate as events in the atmosphere and the geant4 code was used to treat the shower particles entering in the detectors .
the air - shower core events observed with the yac - i array were analyzed to select those induced by the light component using the ann technique . in this paper , we focused on the sensitivity of the yac - i array to observe the light component of cosmic rays around the knee and discussed the systematic errors coming from the the models used in the mc indispensable for obtaining the result which should be independent on the model as possible .
it is shown that the yac+tibet as array is powerful to study the primary cosmic - ray chemical composition , in particular , to obtain the energy spectrum of light - component ( p+he ) of cosmic rays at the knee energies .
a full - scale yac consisting of 400 core detectors covering more than 5000 m@xmath7 could be operated together with the tibet - iii array in the very near future .
the authors would like to express their thanks to the members of the tibet as@xmath4 collaboration for the fruitful discussion .
this work is supported by the grants from the national natural science foundation of china ( y11122005b , y31136005c and y0293900tf ) and the chinese academy of sciences ( h9291450s3 ) and the key laboratory of particle astrophysics , institute of high energy physics , cas . the knowledge innovation fund ( h95451d0u2 and h8515530u1 ) of ihep ,
china also provide support to this study .
|
a new air - shower core - detector array ( yac : yangbajing air - shower core - detector array ) has been developed to measure the primary cosmic - ray composition at the knee " energies in tibet , china , focusing mainly on the light components . the prototype experiment ( yac - i ) consisting of 16 detectors has been constructed and operated at yangbajing ( 4300 m a.s.l . ) in tibet since may 2009 .
yac - i is installed in the tibet - iii as array and operates together . in this paper
, we performed a monte carlo simulation to check the sensitivity of yac - i+tibet - iii array to the cosmic - ray light component of cosmic rays around the knee energies , taking account of the observation conditions of actual yac - i+tibet - iii array .
the selection of light component from others was made by use of an artificial neural network ( ann ) .
the simulation shows that the light - component spectrum estimated by our methods can well reproduce the input ones within 10% error , and there will be about 30% systematic errors mostly induced by the primary and interaction models used .
it is found that the full - scale yac and the tibet - iii array is powerful to study the cosmic - ray composition , in particular , to obtain the energy spectra of protons and helium nuclei around the knee energies .
_ keywords _ : cosmic rays , hadronic interaction , knee , composition , energy spectrum
|
bouncing cosmologies and more specifically the matter bounce scenario ( see @xcite for a review of this kind of cosmologies ) is the most promising alternative to the inflationary paradigm .
there are many ways to obtain cosmologies without the big bang singularity : for example violating the null energy condition in general relativity by incorporating new forms of matter such as phantom @xcite or quintom fields @xcite , galileons @xcite or phantom condensates @xcite , or by adding terms to einstein - hilbert action @xcite , but the simplest one is to go beyond general relativity and consider holonomy corrected loop quantum cosmology ( lqc ) , where a big bounce replaces the big bang singularity @xcite .
in fact , other future singularities such as type i ( big rip ) and type iii ( big freeze ) are also forbbiden in holonomy corrected lqc @xcite .
on the other hand , it is well known that a matter domination period in the contracting phase is dual to the de sitter regime in the expanding phase @xcite , which provides a flat power spectrum of cosmological perturbations in the matter bounce scenario . moreover ,
an abrupt ekpyrotic phase transition is needed , in the contracting phase , in order to solve the famous belinsky - khalatnikov - lifshitz ( bkl ) instability : the effective energy density of primordial anisotropy scales as @xmath1 in the contracting phase @xcite and , more important , to produce enough particles which will be responsible for thermalizing the universe in the expanding phase @xcite .
this new scenario named matter - ekpyrotic scenario was introduced in @xcite , being developed within the two - field model in @xcite , while in @xcite there is a numerical discussion on the primordial anisotropy issue ( see @xcite for a review ) .
it was treated in the context of loop quantum cosmology in @xcite , showing that it could be compatible with the new experimental data @xcite provided by planck s team ( see also @xcite ) .
the main goal of the present work is to study mathematically , from the viewpoint of the dynamical systems theory , the matter - ekpyrotic scenario in the context of holonomy corrected lqc .
more precisely , we study its background dynamics and the evolution of cosmological scalar perturbations in this scenario , showing that it leads to a nearly flat power spectrum of perturbations , whose spectral index and running match correctly with the recent experimental data .
moreover , we will also point out that the phase transition , in the contracting epoch , from matter domination to the ekpyrotic phase , is the responsible for the production of enough particles whose energy density , in the expanding phase , will be dominant , leading to a reheating temperature compatible with the nucleosynthesis bounds .
the work is organized as follows : the dynamical study of the matter - ekpyrotic scenario in general relativity is performed in section ii . in section iii
we study the dynamics of the background , demonstrating that in loop quantum cosmology the matter - ekpyrotic scenario has two different regimes : at very early times , the universe in the contracting phase is matter dominated and evolves to an ekpyrotic regime , and after the bounce , we show that the universe leaves the ekpyrotic phase and enters in a kination ( or deflationary ) regime @xcite .
section iv is devoted to the calculation of the power spectrum of scalar perturbations in this scenario , showing that the power spectrum is proportional to the square of the value of the hubble parameter at the transition time .
the mechanism to reheat the universe is studied in section v , where we show that the phase transition from the matter to the ekpyrotic regime is essential to produce enough particles to reheat the universe . in last section ,
we build a simple potential in this scenario that leads to theoretical values of the spectral index and its running that fit well with the last observational data provided by planck s team .
the units used throughout the paper are @xmath2 and , thus , the reduced planck s mass is @xmath3 .
in this section , we consider the flat friedmann - lematre - robertson - walker ( flrw ) metric in the context of general relativity .
then , the friedmann , conservation and raychauduri equations are respectively given by @xmath4 where @xmath5 is the hubble parameter , @xmath6 the energy density and @xmath7 the pressure .
to achieve the matter - ekpyrotic scenario we will deal with a single scalar field , namely @xmath8 .
then , the energy density and pressure are @xmath9 being @xmath10 the potential .
our first goal is to mimick , with a single scalar field @xmath8 , a fluid with equation of state ( eos ) @xmath11 , i.e. , we want to obtain a potential @xmath10 which provides a solution of the conservation equation that leads to the same background as a barotropic fluid with eos @xmath11 , where @xmath12 .
the first step is to write the eos @xmath11 as follows @xmath13 which means that for @xmath14 the potential must be positive , negative for @xmath15 , and vanishes for @xmath16 . on the other hand , for a fluid with eos @xmath11 ,
the friedmann and conservation equations of general relativity could be easily solved , giving as a result the background @xmath17 secondly , using the raychauduri equation @xmath18 one has @xmath19 , and if we only consider increasing functions , i.e. , @xmath20 we will obtain @xmath21 where @xmath22 ( resp .
@xmath23 ) refers to the contracting ( resp . expanding )
in fact , @xmath24 is defined for @xmath25 , and @xmath26 in @xmath27 .
finally , to reconstruct the potential we use the relation @xmath11 and the raychaudhuri equation , to obtain @xmath28 using , a simple calculation leads to @xmath29 and , by writing @xmath30 , the analytical solution becomes @xmath31 once we have obtained the potential for a single scalar field , the dynamical equations ( friedmann and conservation ) are @xmath32 where @xmath33 .
the conservation equation is a second order differential equation , meaning that there are infinitely many solutions that depict different universes .
for example , if one deals in the expanding phase , i.e. we take @xmath34 , then for the potential @xmath35 , the analytical solution @xmath36 with @xmath37 , defines a universe with eos @xmath11 all the time . however , the other solutions do not define a universe with this eos all the time .
for this reason , what we want to study is the behaviour of the analytical solution , i.e. , we want to know whether the orbit in the plane @xmath38 defined by the solution ( [ 8 ] ) will be an attractor or a repeller . to do that , first of all we study the dynamics in the contracting phase .
performing the change of variable @xcite @xmath39 the conservation equation will become @xmath40 that could be written as a first order one - dimensional differential equation : @xmath41 where @xmath42.\end{aligned}\ ] ] to perform the dynamical study we have to differentiate between @xmath43 cases : 1 .
@xmath16 . in this case
the differential equation becomes : @xmath44,\end{aligned}\ ] ] which means that in the half plane @xmath45 , one has @xmath46 and the general solution is given by @xmath47 + that is , there are two kind of orbits in the plane @xmath38 @xmath48 and @xmath49 + the first kind corresponds to a stiff fluid because @xmath50 leads to @xmath51 . and
the second kind are fixed points that correspond to @xmath52 .
+ on the other hand , in the plane @xmath53 , it is easier to study directly the conservation equation @xmath54 obtaining @xmath55 + what is important is that all the orbits @xmath56 with @xmath57 are stable and all the phase space is foliated by them ( see figure [ retratgrw1 ] ) .
+ plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath16 .
the orbits of the families @xmath50 ( blue , upper semiplane ) and @xmath59 ( blue , lower semiplane ) , separated by the equilibrium points @xmath60 ( black dots ) fill the plane.,title="fig : " ] plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath16 . the orbits of the families @xmath50 ( blue , upper semiplane ) and @xmath59 ( blue , lower semiplane ) , separated by the equilibrium points @xmath60 ( black dots ) fill the plane.,title="fig : " ] 2 .
the system has one fixed point , namely @xmath61 which is a global repeller because @xmath62 , and for @xmath63 , @xmath64 is negative , which means that in cosmic time the orbits move away from @xmath65 .
+ this solution corresponds to @xmath66 whose orbit depicts , all the time , in the contracting phase , a universe with eos @xmath11 .
+ note that the fact that solution is a repeller means that all the orbits depict , at early times , a matter dominated universe .
moreover , since time goes forward , the solutions separate from , meaning that the backgrounds depicted by them cease to be matter dominated .
the phase portrait is drawn in figure [ retratgrw0 ] .
+ plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath67 and @xmath68 .
the analytical orbit @xmath69 of eq .
( black ) is a repeller.,title="fig : " ] plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath67 and @xmath68 .
the analytical orbit @xmath69 of eq .
( black ) is a repeller.,title="fig : " ] + in this scenario , a fundamental quantity is the effective eos parameter , namely @xmath70 which could be used to indicate the evolution of the eos for each orbit of the dynamical system . in figure [ weffgrw0 ] , we can see that all orbits tend asymptotically at the beginning of the contracting phase to @xmath71 . + for orbits represented in the phase portrait in figure [ retratgrw0 ] . ]
the system is only defined for @xmath72 . moreover ,
the system has three fixed points , namely @xmath73 + @xmath74 is an attractor in the half plane @xmath75 because @xmath76 and @xmath77 are repellers because these solutions correspond to @xmath52 , and in the contracting phase @xmath5 decreases as time goes forward , that is , the orbits moves away from @xmath77 .
+ the first solution corresponds to @xmath78 whose orbit depicts , all the time , in the contracting phase , a universe with eos @xmath11 .
the other ones satisfy @xmath52 .
since the orbit @xmath79 is an attractor in the half plane @xmath80 , this means that all the other orbits in the half plane @xmath80 depict , at late time , a universe with eos @xmath11 .
plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath81 and @xmath82 .
the analytical orbits of eq .
define the orbits @xmath69 ( black ) , @xmath83 ( red ) , @xmath84 ( green ) .
nearby orbits ( blue ) show how @xmath69 is an attractor for the orbits between it and @xmath84 ( corresponding to the region where @xmath85 in phase space ) , while @xmath86 are repellers.,title="fig : " ] plane ( left ) and in the @xmath58 plane ( right ) in the case @xmath81 and @xmath82 .
the analytical orbits of eq .
define the orbits @xmath69 ( black ) , @xmath83 ( red ) , @xmath84 ( green ) .
nearby orbits ( blue ) show how @xmath69 is an attractor for the orbits between it and @xmath84 ( corresponding to the region where @xmath85 in phase space ) , while @xmath86 are repellers.,title="fig : " ] note that in figure [ retratgrwgt1 ] , orbits approaching either of the fixed points @xmath87 reach such value ( corresponding to @xmath88 ) in a finite cosmic time .
since near @xmath87 the dynamical system behaves as @xmath89 , it is reached in a finite @xmath8 time .
given that @xmath8 is finite , @xmath90 is finite as well and , hence , in a finite cosmic backward time the contracting phase starts after a bounce at @xmath88 that has enabled the system to leave the expanding phase with no singularity .
in figure [ weffgrwgt1 ] we show the evolution of @xmath91 . we see that for those orbits approaching to the analytical solution in the plane @xmath58 , @xmath91 tends to @xmath92 , while for those in which @xmath93 diverge
, @xmath91 tends to @xmath94 . for the orbits represented in the phase portrait in figure [ retratgrwgt1 ] . ]
the same analysis shows that , in the expanding phase , for @xmath16 there are solutions that correspond to a stiff fluid , and others with @xmath52 .
for @xmath14 the solution that depicts a fluid with eos @xmath11 is an attractor . for @xmath15 , the solution that depicts
a fluid with eos @xmath11 is a repeller , and the one with @xmath52 an attractor .
once we have performed this previous study , it is interesting to consider the matter - ekpyrotic scenario given , in the contracting ( resp .
expanding ) phase , by a potential ( @xmath15 ) with the shape @xmath95 with @xmath96 and @xmath97 . in figure
[ fig : superposgr ] we show phase portraits , in contracting phase , for the superposed potential with @xmath98 and @xmath99 .
the colouring code is that the black , red , green solutions are those given by the initial condition of the corresponding critical orbits in the case @xmath100 .
moreover , in figure [ fig : weffsuperposgr ] we have drawn the parameter @xmath91 for black orbits in each superposition with @xmath99 .
plane for the superposed potential in the case @xmath101 and @xmath98 , with @xmath99 respectively.,title="fig : " ] plane for the superposed potential in the case @xmath101 and @xmath98 , with @xmath99 respectively.,title="fig : " ] plane for the superposed potential in the case @xmath101 and @xmath98 , with @xmath99 respectively.,title="fig : " ] plane for the superposed potential in the case @xmath101 and @xmath98 , with @xmath99 respectively.,title="fig : " ] for the distinguished orbits indicated in black in figure [ fig : superposgr ] , again with @xmath99 respectively .
one can see that for all the orbits @xmath91 starts at @xmath102 and ends at @xmath92.,title="fig : " ] for the distinguished orbits indicated in black in figure [ fig : superposgr ] , again with @xmath99 respectively .
one can see that for all the orbits @xmath91 starts at @xmath102 and ends at @xmath92.,title="fig : " ] for the distinguished orbits indicated in black in figure [ fig : superposgr ] , again with @xmath99 respectively .
one can see that for all the orbits @xmath91 starts at @xmath102 and ends at @xmath92.,title="fig : " ] for the distinguished orbits indicated in black in figure [ fig : superposgr ] , again with @xmath99 respectively .
one can see that for all the orbits @xmath91 starts at @xmath102 and ends at @xmath92.,title="fig : " ] from our previous study it seems clear that in the contracting phase , at early times , all the orbits will depict a matter dominated universe , and after leaving this behaviour they will evolve , at late times , to an ekpyrotic one . on the contrary , in the expanding phase , at early times , orbits will depict a universe in the ekpyrotic phase that eventually evolves to a matter dominated one .
in loop quantum cosmology ( lqc ) , where holonomy corrections are introduced as a consequence of the discrete nature of the space - time , the modified friedmann equation depicts an ellipse in the plane @xmath103 given by @xcite @xmath104 where @xmath105 is the so - called critical energy density whose value is approximately 252 @xcite .
once again , we want to mimick by a scalar field a fluid with eos @xmath11 .
solving the friedmann and consevation equations in lqc for this kind of fluid one has the solution @xcite @xmath106 then , the equation @xmath11 becomes @xmath107 whose solution is given by @xcite @xmath108 to reconstruct the potential we use the formula @xmath109 and we isolate @xmath110 as a function of @xmath8 , obtaining @xcite @xmath111 where we have chosen @xmath112 .
we can see , as in the case of general relativity , for @xmath16 the potential vanishes , for @xmath14 @xmath113 is positive , and negative for @xmath15 . as in general relativity
, we will differentiate between @xmath43 cases so as to study the following dynamical system : @xmath114 where @xmath115 .
1 . @xmath16 . in this case
the differential equation becomes : + @xmath116 + obtaining for @xmath117 orbits in the plane @xmath118 of the form + @xmath119 + where @xmath120 , and for @xmath85 + @xmath121 + where @xmath122 .
+ we observe that all orbits are stable and all the phase space is foliated by them . + in order to study the remaining two cases , it is suitable to perform the following change of variables , motivated by : @xmath123 +
hence , equation of conservation becomes + @xmath124 + where @xmath115 and @xmath125 .
2 . @xmath14 .
the dynamical system is defined for @xmath126 and has two fixed points : + @xmath127 + in figure [ retratlqcw0 ] , we have represented the phase portrait of the dynamical system , such that blue orbits stand for the analytical ones ( fixed points @xmath128 , corresponding to equation ) , red colour is for the contracting phase and green colour is for the expanding phase .
the coloured zone in the phase portrait is the forbidden region , in which @xmath129 .
note that the bounce from the contracting to the expanding phase happens in @xmath130 curve .
we see that there are two types of orbits : those that cross axis @xmath131 and those that cross axis @xmath132 .
therefore , it can be clearly concluded that the analytical orbit is a repeller in the contracting phase and an attractor in the contracting phase , as one can also verify with the effective eos parameter in figure [ wefflqcw0 ] .
+ plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath67 . the analytical solution has been outlined in blue.,title="fig : " ] plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath67 .
the analytical solution has been outlined in blue.,title="fig : " ] + .
the orbit that crosses @xmath133 is the one which passes through @xmath134 .
both tend asymptotically at @xmath135 and @xmath136 at @xmath92 . ]
the dynamical system is defined for @xmath137 and has four fixed points : + @xmath138 + as already happened with general relativity for the @xmath15 case , those orbits approaching @xmath139 ( corresponding to @xmath88 ) reach it at a finite time because of the same argument that was used in general relativity , since holonomic corrections can be discarded near @xmath88 .
therefore , in these cases we will see a bounce at @xmath88 .
in the phase portrait in figure [ retratlqcwgt1 ] ( with the same colour notation as in the former case ) , we have seen that all orbits different from the analytical one ( @xmath128 , corresponding to ) perform two cycles in the @xmath103 ellipse , with two bounces at @xmath130 and one bounce at @xmath88 .
the process is the following : * contracting phase 1 : at @xmath136 , @xmath88 for infinite values of @xmath140 and @xmath93 and at a finite time @xmath141 the orbit reaches @xmath130 , occurring a bounce .
* expanding phase 1 : after the first bounce , the orbit enters into the expanding phase reaching either @xmath142 or @xmath143 ( corresponding to @xmath88 ) at a finite time @xmath144 .
there , another bounce occurs . *
contracting phase 2 : after the second bounce , the orbit starts the second cycle in the ellipse @xmath103 and reaches @xmath130 at a finite time @xmath145 ( third bounce ) .
* expanding phase 3 : after this last bounce , the orbit enters into the expanding phase and tends asymptotically for @xmath135 to @xmath88 for infinite values of both @xmath140 and @xmath93 .
+ therefore , we infere that the analytical solutions @xmath128 are attractors in the contracting phase and repellers in the expanding phase , while the solutions @xmath139 are repellers in the contracting phase and attractors in the expanding phase . with regards to the behaviour of @xmath91 , represented in [ wefflqcwgt1 ]
, it is seen that both at @xmath136 and @xmath135 , @xmath146 , whereas it diverges at the bounce at @xmath88 .
+ plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 . the analytical solution has been outlined in blue.,title="fig : " ] plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 .
the analytical solution has been outlined in blue.,title="fig : " ] + . in the first cicle
, @xmath91 starts being @xmath94 at very early times and diverges at the end of the cicle . in the second cicle , in the contracting phase , @xmath91 approaches to @xmath92 near the bounce and when the orbit enters in the expanding phase the effective eos parameter goes asymptotically to @xmath94 . ] from these previous results , to obtain a simple version of the matter - ekpyrotic bouncing scenario in lqc , we can consider the potential @xmath147 where @xmath15 , @xmath96 and @xmath97 and the abrupt phase transition , needed in order to produce enough particles to reheat the universe , occurs at @xmath148 .
for this potential some orbits will lead to an effective eos being zero at early times , approximately @xmath92 just before the bounce , and finally approaching asymptotically to @xmath94 in the expanding phase .
these orbits could be obtained by taking as an initial condition a point @xmath149 corresponding to an orbit in the second cycle of the figure [ retratlqcwgt1 ] .
thus , with these initial conditions , integrating the conservation equation , one will set up , for the potential , orbits with an effective eos parameter satisfying the required properties .
in fact , in last section we will obtain a potential which leads to a viable matter - ekpyrotic bounce scenario , i.e. , a scenario which provides theoretical values of the spectral parameters that match at @xmath150 confidence level with the recent observational data .
some final remarks are in order : 1 .
as we have already explained in the introduction , holonomy corrections provide a big bounce .
then , the conservation equation @xmath151 has to be studied in the contracting and expanding phase .
firstly , the contracting phase , i.e. with @xmath152 , takes place and , when the universe bounces at @xmath52 , it enters into the expanding one and , thus , its dynamics will now be given by the conservation equation with @xmath34 .
2 . note also that , in loop quantum cosmology , when one considers @xmath91 one has to use the last formula of , i.e. , @xmath153 , because the intermediate one only holds in general relativity ( when holonomy corrections can be disregarded ) .
3 . in @xcite
, the authors obtain an ekpyrotic phase with the following potential : + @xmath154 + when @xmath155 , this potential is symmetric and , by using the change of variables @xmath156 , the new potential becomes @xmath157 which has the asymptotic behaviour as our former potential for @xmath158 .
+ if @xmath159 , the bounce at @xmath131 ( @xmath160 ) takes place at @xmath71 .
therefore , in this case , all orbits have two cycles except the one bouncing at @xmath71 , which has a single cycle , as we can see in figure [ cairebot ] .
however , this last orbit does not correspond to the analytical solution since the value of @xmath91 does not remain constant , but approaches asymptotically to 1 for both @xmath135 and @xmath136 , as we can see in figure [ cairebotweff ] .
+ plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 and @xmath161 .
the orbit with the bounce at @xmath71 has been outlined in blue.,title="fig : " ] plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 and @xmath161 .
the orbit with the bounce at @xmath71 has been outlined in blue.,title="fig : " ] + . ] + nevertheless , when a sufficiently low @xmath113 is used , orbits do not reach @xmath88 in a finite time and , hence , they have a single cycle , as we can see in figures [ cai ] and [ caiweff ] .
+ plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 and @xmath162.,title="fig : " ] plane ( left ) and in the @xmath58 plane for the lqc system with potential for @xmath81 and @xmath162.,title="fig : " ] + . ]
in this section we will calculate explicitly the power spectrum of cosmological perturbations in the matter - ekpyrotic bounce scenario . in loop quantum cosmology the corresponding mukhanov - sasaki equation @xcite , in fourier space , is given by @xcite @xmath163 where the velocity of sound is given by @xmath164 , @xmath165 , being @xmath166 the derivative with respect the conformal time and @xmath167 the conformal hubble parameter . during the matter - domination period , the scale factor in conformal time
could be chosen as follows @xcite @xmath168 where @xmath169 , and @xmath170 is the transition time . then , since at early times the energy density is smaller than the critical one , holonomy corrections could be disregarded @xmath171 and the bunch - davies vacuum will be given by the modes @xcite @xmath172=\frac{e^{-ik(\eta-\eta_1)}}{\sqrt{2k}}\left(1-\frac{i}{k(\eta-\eta_1 ) } \right).\end{aligned}\ ] ] the modes will eventually leave the hubble radius , and thus , for modes well outside of the hubble radius @xmath173 , one has @xmath174 on the other hand , for these modes we can use the long - wavelenght approximation @xmath175 where in a matter dominated universe at early times , @xmath176 is given by @xmath177 , because the holonomy corrections could be disregarded .
a simple calculation leads to @xmath178 matching both expressions , we obtain the following values of @xmath179 and @xmath180 @xmath181 and thus , the curvature fluctuation in co - moving coordinates is given by @xmath182 note that , if @xmath176 is continuous , at the transition time @xmath183 , the formula holds for every time , and we can calculate the power spectrum as @xmath184 where , when @xmath185 the asymptotic form of @xmath186 is given by , i.e. , at very early times @xmath187 .
this is essential in order to perform numerical calculations , because once an orbit has been chosen , the value of @xmath188 is given by formula , the value of @xmath189 is obtained from the relation @xmath190 , the asymptotic condition at early times @xmath187 and the value of @xmath191 via the relation @xmath192 . as an example , if one disregards the phase transition , i.e. , if one only considers a matter domination during all the background evolution , and one uses the orbit given by the solution ( [ 28 ] ) with @xmath67 , formula leads to the result @xmath193 ( see @xcite and @xcite for a derivation of the result ) , where @xmath194 is the planck s energy density , which is incompatible with the experimental data @xmath195 @xcite , because in lqc the value of the critical density is given by @xmath196 @xcite .
this means that , to be viable , the matter bounce scenario must be implemented , for example , with an ekpyrotic phase transition . in this way we can introduce a transition
however , when it is too abrupt , formula does not hold . to be more precise
, we consider the simple example @xcite where the scale factor is given , in terms of the cosmic time , by @xmath197 where @xmath198 , being @xmath199 the value of the hubble parameter at the time of the phase transition and @xmath200 is the time when the universe enters in a deflationary or kination regime . so that the hubble parameter be continuous at @xmath201 one has to choose @xmath202 where we have used that @xmath203 .
in @xcite the authors heuristically obtained that @xmath204 .
then , since the observational values of the power spectrum states @xmath205 , one will deduce that @xmath206 . to reproduce mathematically this result we consider , for @xmath207 , the solution in the long wave - lenght approximation @xmath208 then matching at @xmath183 one obtains @xmath209 that is , @xmath210 because for modes well outside to the hubble radius one has @xmath211 .
first we are interested in the evolution of @xmath212 during the ekpyrotic regime .
for this reason we will calculate @xmath213 where we have used that , during the ekpyrotic phase , @xmath214 performing the change of variable @xmath215 one obtains @xmath216.\end{aligned}\ ] ] the integrals that appear in this expression are of the same order as @xmath217 @xmath218 because in this model @xmath219 . finally , using that @xmath220 , we will get @xmath221 and , approximately , we have @xmath222 during the deflationary phase , the curvature fluctuation evolves as @xmath223 thus , performing the matching at @xmath224 one obtains @xmath225
the deflationary regime ends when the universe becomes reheated and enters in a radiation dominated phase . for this reason we have to calculate @xmath226 using the formulas @xmath227 and @xmath228 of @xcite @xmath229 and the fact that @xmath230 and @xmath231 , one obtains @xmath232 where @xmath233 is the reheating temperature .
now , taking into account that in lqc @xmath234 and that the nucleosynthesis bounds impose @xmath235 , one has @xmath236 if we impose the first term on the right hand side to be greater than the second one , one will have the constraint @xmath237 now , with this assumption one will have @xmath238 when the universe enters in a radiation dominated phase one will have @xmath239 where the holonomy corrections could be disregarded and then , @xmath240 with @xmath241 . matching at @xmath242 one will get @xmath243 a simple integration shows that , at very late times @xmath244\cong \frac{c_2(k)}{6a_r^3h_r}. \end{aligned}\ ] ] this last quantity is equal to @xmath245 where we have used that during the deflationary phase one has @xmath246 then , we conclude that , if @xmath199 satisfies the conditions @xmath247 , one will have @xmath248 , which means that the power spectrum is given by @xmath249 and thus , to match with observational data one has to choose @xmath250 .
finally , the condition @xmath247 is fulfilled for @xmath251 .
as we have already explained , when one deals with a matter dominated universe ( without the ekpyrotic phase transition ) , the power spectrum of cosmological perturbations is of the order @xmath105 , and since lqc provides a value around planck s density , its value is in contradiction with the current observations because it is too high .
this is another reason to implement an ekpyrotic phase in the matter bounce scenario .
we will consider a heavy massive quantum field , namely @xmath252 , conformally coupled with gravity , and first of all we consider the model ( [ cai - wilson ] ) proposed by @xcite .
this model has been studied in @xcite , where it has been shown that the energy density of the produced particles is given by @xmath253 during the ekpyrotic phase , the background evolves as @xmath254 , then , since @xmath15 , during the contracting phase @xmath255 increases faster than @xmath256 , and thus , the universe continues being driven by the background . however , after the bounce , the universe enters in a kination ( @xmath16 ) regime and the background evolves as @xmath257 , where @xmath258 is the minimum value of the scale factor . since in the expanding phase @xmath6 decreases as @xmath1 and @xmath259 as @xmath260 , the energy density of the created particles will eventually start to dominate , and at that moment , namely @xmath261 , when both energy densities will be of the same order , the universe will become reheated with a temperature @xmath262 . to calculate this temperature we use the identity @xmath263 to obtain @xmath264 where @xmath265 . and , thus , the reheating temperature is given by @xmath266 taking into account , as we have shown in section @xmath267 , that for @xmath268 one has @xmath269 , one will obtain in figure [ fig : tvsw1 ] the dependence of @xmath270 as a function of @xmath92 . using formula ( [ xxx ] ) . ] consequently , since nucleosynthesis bounds constrain the reheating temperature to be between @xmath271 and @xmath272 , from figure [fig : tvsw1 ] we can see that the matter - ekpyrotic model only holds when @xmath92 is of the order @xmath273
. however , the reheating temperature is very high and is in the borderline . then , if one wants a lower reheating temperature , one has to consider a phase transition less abrupt than the one considered above . to do this
, we will consider the model defined in next section , where in the phase transition the second derivative of the hubble parameter is discontinuous .
assuming that the mass of the quantum field is very high , for example is of the order of the reduced planck mass , we could approximate the vacuum modes by its first order wkb solution , obtaining that the @xmath274 bogoliubov coefficient is given by @xcite @xmath275 the energy density of the produced particles is given by @xmath276 on the other hand , after the phase transition the energy density of the background evolves as @xmath277 , and after the bounce , since the universe enters in a kination domination , it evolves as @xmath257 , where the value of the scale factor at the bouncing time is given by formula . in the contracting phase ,
the energy density is always dominant , but in the expanding one it will eventually become subdominant .
when both energy densities are of the same order the universe becomes reheated , this happens when @xmath278 where the constant @xmath279 has been introduced in .
then , the reheating temperature will be @xmath280 where @xmath281 . and , after reheating , the universe is dominated by a thermalized relativistic plasma and matches with the standard hot friedmann universe .
finally , from formula , if one takes , for instance , @xmath282 , one will obtain the picture drawn in figure [ fig : tvsw2 ] , which shows that the reheating temperatures belong to the range between @xmath271 and @xmath272 for values of @xmath92 up to order @xmath283 . using formula . ]
two final remarks are in order : 1 .
when one considers very light particles non - conformally coupled with gravity , the energy density of the produced particles is @xmath284 @xcite , leading to the reheating temperature @xmath285 which , as we can see in figure [ fig : tvsw3 ] , is always greater that @xmath286 , i.e. , the matter - ekpyrotic bounce scenario only supports the creation of massless particles when the reheating temperature of the universe is very high . + using the formula ( [ light ] ) . ] 2 .
the difference between the matter - ekpyrotic scenario and the current models in quintessence inflation is that in the later one the phase transition occurs after the end of inflation and , thus , the value of the hubble parameter is of the order of @xmath287 @xcite , leading to a reheating temperature of the order of @xmath288 , which is less than the one obtained in matter - ekpyrotic lqc scenario .
in the matter - ekpyrotic bounce scenario , for modes that leave the hubble radius at very early times , when the universe is matter dominated , the spectral index and its running are given by @xcite @xmath289 it is clear that a pure matter - ekpyrotic bouncing scenario leads to @xmath290 then , to match with the current observational data , one needs to improve this scenario introducing a quasi - matter domination period at very early times . to do this ,
first of all one has to consider the raychaudhuri equation in lqc ( see for instance @xcite ) @xmath291 which for the linear eos @xmath11 becomes @xmath292 then , in the ekpyrotic phase one has to simply choose @xmath15 , and for quasi - matter domination one could choose @xmath293 ( we have written @xmath294 for convenience , but in principle @xmath295 can be positive or negative ) .
however , at the phase transition , the derivative of the hubble parameter is discontinuous and , as we have seen in the previous section , this kind of models could be ruled out by the nucleosynthesis bounds due to its high reheating temperature .
for this reason we need to match these dynamics in a continuous way , and a simple way to do this is presented in the following model @xmath296 note that in the contracting phase , near @xmath297 , since holonomy corrections could be disregarded , these dynamics could be written as follows @xmath298 moreover , the effective equation of state ( eos ) parameter , namely @xmath299 , is given by @xmath300 then , we can see that when @xmath301 one has @xmath302 , that is , the universe is quasi - matter dominated .
on the other hand , the potential that leads to these dynamics could be obtained approximately using the reconstruction method , as follows : when holonomy corrections could be disregarded , raychaudhuri equation becomes @xmath303 , and then , @xmath304 , obtaining for our model , when @xmath305 @xmath306 where we have used formula @xmath307 of @xcite @xmath308 conversely , @xmath309 since , when holonomy corrections are disregarded , the potential is given by @xmath310 , one has approximately for @xmath311 @xmath312 on the contrary , as we have seen in section @xmath43 , in the ekpyrotic phase the potential is given by @xcite @xmath313 and , since for the potential one has @xmath314 , assuming the continuity of the potential at the transition phase one gets @xmath315 therefore , if one takes into account the approximation @xmath316 one will have @xmath317 summing up , the potential we propose in the matter - ekpyrotic bounce scenario is @xmath318 note that when @xmath319 the potential satisfies @xmath320 , that is , the universe is matter dominated at early times , and for @xmath321 the universe is in the ekpyrotic regime . to perform the calculations , we choose , for example , as a pivot scale @xmath322 where subindex @xmath102 means the current time , and we choose , as usual , @xmath323 @xcite . to calculate the value of @xmath324
we need the current value of the scale factor which could be calculated as follows : for the sake of simplicity we take @xmath325 .
then , from equation , the value of the scale factor at the beginning of the radiation era will satisfy @xmath326 .
now , we use the relation @xcite @xmath327 to obtain @xmath328 , where we have used that the current temperature of the universe is @xmath329 , and that for @xmath16 the reheating temperature is @xmath330 .
consequently , @xmath331 and , thus since @xmath332 , one concludes that @xmath333 . on the other hand ,
let @xmath334 be the time when the pivot scale leaves the hubble radius during the quasi - matter domination in the contracting phase .
then one will have @xmath335 , which means @xmath336 because the pivot scale leaves the hubble radius , in the contracting phase , before the phase transition .
then , we can conclude that @xmath337 , and thus @xmath338 and @xmath339 planck13 observational data at @xmath340 c.l .
give the following results : @xmath341 and @xmath342 .
then , to match our theoretical model , with the @xmath94-marginalized domain for the spectral index at @xmath150 c.l .
, one has to choose the effective eos parameter @xmath343 satisfying @xmath344 on the other hand , since @xmath345 we can see that the theoretical value of the running enters in the @xmath94- marginalized domain for the running at @xmath150 c.l .
a final important remark is in order : as pointed out in @xcite , in the matter - ekpyrotic scenario , dealing with the analytical solution , the power spectrum of tensor and scalar perturbations are of the same order .
this means that the ratio of tensor to scalar perturbations , namely @xmath346 , does not satisfy the constraint @xmath347 .
there are several ways to address this problem , one of them is to consider at @xmath348cdm universe at early times @xcite because , due to the small velocity of sound , this increases the power spectrum of scalar perturbations and conserves the power spectrum of tensor ones , that is , it suppresses the tensor / scalar ratio .
however , one has to realize that these calculations are performed for analytical solutions , but when one deals with a scalar field there are infinitely many solutions , then for some of these solutions , as has been shown numerically in @xcite for the simple case of a matter dominated universe , some of them suppress the ratio of tensor to scalar perturbations satisfying the bound @xmath347 provided by the joint analysis data from bicep2/keck array and planck @xcite .
thus , in the present case some numerical calculations might support this fact .
this is , of course , a point that deserves future investigation .
in the present work we have shown , with all the mathematical details , that the matter - ekpyrotic bounce scenario in loop quantum cosmology , whose background has a matter domination regime in the contracting phase at very early times followed by a phase transition to an ekpyrotic epoch up to the bounce where the universe enters in a kination phase , is a promising alternative to the inflationary paradigm , because it leads to theoretical values of some spectral parameters - the power spectrum of scalar perturbations , the spectral index and its running- that match at @xmath150 confidence level with the current observational data . moreover , the phase transition , produced in the contracting phase , is able to produce enough particles -very massive conformally coupled with gravity or massless non conformally coupled- to reheat the universe , in the expanding phase , at temperatures compatible with the bounds coming from nucleosynthesis , and thus , matching with the hot friedmann universe .
99 m. novello and s.e .
perez bergliaffa , phys .
rept . * 463 * , 127 ( 2008 ) [ arxiv:0802.1634 ] .
brandenberger , ( 2012 ) [ arxiv:1206.4196 ] .
brandenberger , int . j. mod
ser * 01 * , 67 ( 2008 ) [ arxiv:0902.4731 ] .
brandenberger , aip conf . proc . * 1268 * , 3 ( 2010 ) [ arxiv:1003.1745 ]
brandenberger , pos ( icfi 2010 ) * 001 * , ( 2010 ) [ arxiv:1103.2271 ] .
a. ashtekar , t. pawlowski and p. singh , phys .
. lett . * 96 * 141301 , ( 2006 ) [ arxiv:0602086 ] .
+ a. ashtekar , t. pawlowski and p. singh , phys . rev . *
d74 * , 084003 ( 2006 ) [ arxiv:0607039 ] .
+ a. corichi and p. singh , phys.rev . *
d78 * 024034 , ( 2008 ) [ arxiv:0805.0136 ] .
+ t. zhu , a. wang , k. kirsten , g. cleaver and q. sheng , [ arxiv:1607.06329 ] .
t. cailleteau a. barrau , j. grain and f. vidotto , phys . rev . *
d86 * 087301 , ( 2012 ) [ arxiv:1206.6736 ] .
+ t. cailleteau , j. mielzczarek , a. barrau and j. grain , class .
* 29 * , 095010 ( 2012 ) [ arxiv:1111.3535 ] .
|
we will perform a detailed study of the matter - ekpyrotic bouncing scenario in loop quantum cosmology using the methods of the dynamical systems theory .
we will show that when the background is driven by a single scalar field , at very late times , in the contracting phase , all orbits depict a matter dominated universe , which evolves to an ekpyrotic phase .
after the bounce the universe enters in the expanding phase , where the orbits leave the ekpyrotic regime going to a kination ( also named deflationary ) regime .
moreover , this scenario supports the production of heavy massive particles conformally coupled with gravity , which reheats the universe at temperatures compatible with the nucleosynthesis bounds and also the production of massless particles non - conformally coupled with gravity leading to very high reheating temperatures but ensuring the nucleosynthesis success . dealing with cosmological perturbations
, these background dynamics produce a nearly scale invariant power spectrum for the modes that leave the hubble radius , in the contracting phase , when the universe is quasi - matter dominated , whose spectral index and corresponding running is compatible with the recent experimental data obtained by planck s team .
@xmath0departament de matemtica aplicada , universitat politcnica de catalunya + diagonal 647 , 08028 barcelona , spain + * pacs numbers : *
|
cryogenic detectors are currently the preferred technology for astronomical observations over most of the electromagnetic spectrum , notably in the far infrared through millimeter ( 0.13 mm ) @xcite , x - ray @xcite , and gamma - ray @xcite wavelength ranges . in the important ultraviolet , optical , and near infrared ( 0.15 @xmath0 m ) wavelength range a variety of detector technologies based on semiconductors , backed by large investment from both consumer and military customers , has resulted in detectors for astronomy with large formats , high quantum efficiency , and low readout noise . however , these detectors are fundamentally limited by the band gap of the semiconductor ( 1.1 ev for silicon ) and thermal noise sources from their high ( @xmath1100 k ) operating temperatures @xcite .
cryogenic detectors , with operating temperatures on the order of 100 mk , allow the use of superconductors with gap parameters over 1000 times lower than typical semiconductors .
this difference allows new capabilities . a superconducting detector can count single photons with no false counts while determining the energy ( to several percent or better ) and arrival time ( to a microsecond ) of the photon .
it can also have much broader wavelength coverage since the photon energy is always much greater than the gap energy .
while a ccd is limited to about 0.31 @xmath0 m , the new arrays described here are sensitive from 0.1 @xmath0 m in the uv to greater than 5 @xmath0 m in the mid - ir , enabling observations at infrared wavelengths vital to understanding the high redshift universe .
this approach has been pursued in the past with two technologies , superconducting tunnel junctions ( stjs ) @xcite and transition edge sensors ( tess ) @xcite . while both of these technologies produced functional detectors , they are limited to single pixels or small arrays due to the lack of a credible strategy for wiring and multiplexing large numbers of detectors , although recently there have been proposals for larger tes multiplexers @xcite .
microwave kinetic inductance detectors , or mkids@xcite , are an alternative cryogenic detector technology that has proven important for millimeter wave astrophysics@xcite due to their sensitivity and the ease with which they can be multiplexed into large arrays .
mkids use frequency domain multiplexing @xcite that allows thousands of pixels to be read out over a single microwave cable .
while the largest stj array is 120 pixels @xcite and the largest optical tes array is 36 pixels @xcite , the mkid arrays described below are 1024 pixels , with a clear path to megapixel arrays .
the ability to easily reach large formats is the primary advantage of mkid arrays . in this paper
we describe the first photon - counting , energy - resolving ultraviolet , optical , and near infrared mkid focal plane array .
these optical lumped element ( ole ) mkid arrays have significant advantages over semiconductor detectors like charge coupled devices ( ccds ) @xcite .
they can count individual photons with essentially no false counts and determine the energy and arrival time of every photon with good quantum efficiency . their physical pixel size and maximum count rate
is well matched with large telescopes .
these capabilities enable powerful new astrophysical instruments usable from the ground and space .
mkids work on the principle that incident photons change the surface impedance of a superconductor through the kinetic inductance effect @xcite .
the kinetic inductance effect occurs because energy can be stored in the supercurrent of a superconductor .
reversing the direction of the supercurrent requires extracting the kinetic energy stored in the supercurrent , which yields an extra inductance .
this change can be accurately measured by placing this superconducting inductor in a lithographed resonator .
a microwave probe signal is tuned near the resonant frequency of the resonator , and any photons which are absorbed in the inductor will imprint their signature as changes in phase and amplitude of the probe signal .
since the quality factor @xmath2 of the resonators is high and their microwave transmission off resonance is nearly perfect , multiplexing can be accomplished by tuning each pixel to a different resonant frequency with lithography during device fabrication .
this is accomplished by changing the total length of the inductor with a `` trombone section '' , resulting in a lower inductance and therefore a higher resonant frequency .
a comb of probe signals can be sent into the device , and room temperature electronics can recover the changes in amplitude and phase without significant cross talk @xcite , as shown in figure [ fig : detcartoon ] .
are absorbed in a superconducting film , producing a number of excitations , called quasiparticles .
( b ) to sensitively measure these quasiparticles , the film is placed in a high frequency planar resonant circuit . the amplitude ( c ) and phase ( d ) of a microwave excitation signal sent through the resonator .
the change in the surface impedance of the film following a photon absorption event pushes the resonance to lower frequency and changes its amplitude .
if the detector ( resonator ) is excited with a constant on - resonance microwave signal , the energy of the absorbed photon can be determined by measuring the degree of phase and amplitude shift .
right : the top panel shows the results the equivalent circuit of multiplexed mkids , and the bottom panel shows microwave transmission data from actual mkids with very accurate frequency spacing . ]
mkids are extremely versatile , as most resonators with a superconductor as the inductor will function as a mkid .
we have decided to pursue a lumped element resonator design @xcite , shown in figure [ fig : ole ] .
the resonator itself consists of a 20 nm thick sub - stoichiometric titanium nitride ( tin@xmath3 ) film @xcite , with the nitrogen content tuned with @xmath4 such that the superconducting transition temperature @xmath5 is about 800 mk . due to the long penetration depth of these films ( @xmath11000 nm )
the surface inductance is an extremely high 90 ph / square , allowing a very compact resonator fitting in a 100@xmath6100 @xmath0 m square . due to bandwidth limitations of our electronics
we use two feedlines to read out the array , each serving 512 resonators .
the resonators are designed to be separated by 2 mhz within a 45 ghz band .
m pitch , with slot widths inside the resonator of 0.5 @xmath0 m .
right : a sem of a ole mkid pixel .
the microwave feedline runs down the middle , with ground straps shorting the finite ground planes together .
an l - shaped piece of niobium is connected to the center strip and enables strong coupling of the resonator to the feedline .
the resonant frequency is adjusted by changing the length of a `` trombone section '' .
the tapering is visible as the slow increase in leg width with increasing distance from the feedline . ] to avoid crosstalk between pixels the inductors are made with a double meander design that allows the electric field from the charge in each meander leg to be precisely cancelled by the adjacent leg @xcite .
the array is designed so that resonators close together in resonant frequency are physically far apart . to improve the quantum efficiency of the device a 100 @xmath0 m pitch circular microlens array
is used to focus the incoming light on the inductor , since photons hitting the capacitor or wiring will not be detected or will appear as photon events with an energy significantly below their true energy .
the circular microlenses used in these measurements limits the effective fill factor to 67%
. an improved lens with square lens elements could increase the fill factor above 95% . in order to achieve high energy resolution
, the ole mkid must have a uniform response to photons hitting anywhere inside the spot produced by the microlens , which is expected to have a wavelength dependent diameter of around 15 @xmath0 m .
the responsivity of an ole mkid depends on the current density in the meander leg at the location the photon is absorbed . since the capacitance of our resonator is small the current density changes by nearly a factor of two over the length of the inductor .
diffusion does not even out the quasiparticle distribution since the quasiparticle diffusion length in tin is expected to be short ( on the order of 10 @xmath0m)@xcite . in order to normalize the response we taper the width of each leg to give a uniform current density in the last eight legs ( the microlens target ) based on electromagnetic simulations of the current density using the sonnet software package , as shown in figure [ fig : taper ] .
the widths of the legs vary from 2.54 @xmath0 m , while the slots are 0.5 @xmath0 m .
early prototypes used coplanar waveguide ( cpw ) or coplanar slotline ( cps ) feedlines to bring the probe signals to and from the resonators .
these designs exhibited extreme variations in coupling quality factor ( @xmath7 , the strength of the resonator s coupling to the feedline ) and resonant frequency , presumably due to undesired modes being excited on the feedline at discontinuities such as resonator couplers . in order to suppress these modes we switched to a finite groundplane cpw ( fcpw ) feedline with regular straps that connect the ground planes together to suppress the undesired slotline mode .
the feedlines are made out of niobium since it is difficult to make 50 ohm feedlines with tin due to its high surface inductance .
an extended niobium groundplane was also added to reduce crosstalk between the two feedlines . in order to achieve strong coupling ( low @xmath7 )
a l - shaped extension of the center strip was created .
our ole mkid array is fabricated on an high resistivity ( 1020 kilo - ohm cm ) si @xmath8 wafer to reduce two - level system ( tls ) noise @xcite .
wafers have the native oxide removed immediately before tin film growth by dipping them in a buffered oxide etch ( boe ) .
the tin film is deposited by reactive sputtering from a 99.99% purity ti target in a mixture of ultra - high purity nitrogen and argon .
depositions are done at room temperature at @xmath9 pa background pressure .
the tin film was deposited at a rate 37 nm / min using roughly 12% n@xmath10 in ar by flow at 0.266 pa total pressure .
conditions are tuned to provide a sub - stoichiometric composition with slight compressive stress ( @xmath1100 mpa ) , tc@xmath1800 mk , and resistivity @xmath1100 @xmath11 cm .
layers are patterned using a canon stepping projection aligner .
the tin is reactive ion etched ( rie ) with a chlorine containing gas mixture in an inductively coupled plasma system ( icp ) using a photoresist mask .
the wafer surface is solvent cleaned after etch step and given a mild o@xmath10 plasma clean .
tin resonators are protected by a blanket deposit of 80 nm rf - bias sputtered sio@xmath10 . this protection layer is next patterned and etched away from transmission line areas . a coplanar waveguide feedline is fabricated in three steps .
the center line is sputter deposited nb which is patterned by lift - off .
an interlayer dielectric ( ild ) consisting of 120 nm of sio@xmath10 is then deposited and patterned .
finally , the surface is cleaned and patterned again for lift - off of a 160 nm nb film used for the ground plane metal .
sio@xmath10 protecting the tin resonators must be removed in the final step of the process for reduced tls noise .
a stencil of positive resist is patterned over the cpw structure to protect the ild .
the tin is exposed by dipping the wafer in boe for 90 seconds .
photoresist is applied to protect the wafer for dicing into chips and not removed until the device is mounted in a sample box .
one concern during fabrication was how accurately the tapered shape designed into the resonators would be reproduced in the final devices , since the taper is a small fraction of the line width and not visible in a standard lab optical microscope . to measure this ,
devices from three separate wafer runs were measured with an scanning electron microscope ( sem ) .
these measurements showed that while there was an overall random error of @xmath1100 nm in the width of the slots , the widths of the legs were consistent with each other to @xmath125 nm .
extensive simulations show that in an ideal design with these realistic lithographic errors non - uniform current density will not limit our energy resolution until r = e/@xmath12e exceeds 80 .
the device was mounted in a gold - plated copper sample box and wire bonded to duroid transition boards .
the sample box was inserted into a mkid testbed based on a dilution refrigerator and cooled to approximately 100 mk .
a weinreb microwave hemt amplifier with a noise temperature of approximately 4 kelvin is used to amplify the signal .
the testbed allows collimated light fed from an external fiber to illuminate the array while up to 512 resonators are probed with room temperature electronics .
the electronics consist of an anritsu signal generators , marki iq mixers , and a national instruments analog to digitial converter .
more details on the electronics and cryogenics can be found in @xcite .
a mercury argon light source with narrow band filters allows illumination with monochromatic light . .
the fall time of the pulse of 50 @xmath0s limits the maximum count to around 2000 counts / pixel / second .
the top right panel shows a histogram of the optimally estimated pulse height based on the detection of @xmath150,000 254 nm photons .
the solid red curve is the fit of the sum of two gaussian to the histogram , showing an energy resolution r=16 , with a slightly broader shoulder extending to lower energies .
the low energy shoulder is likely due to photons that miss the circular microlens and are absorbed in less sensitive areas of the resonator and photons that hit the substrate between the legs of the inductive meander .
the bottom panel shows the fourier transform of the average pulse template in black ( arbitrary scale ) as well as the measured phase ( blue dotted line ) and amplitude ( red dashed line ) noise . ] the resonators were illuminated with 254 nm hg line photons , and their response was recorded as shown in figure [ fig : r ] . after processing with a wiener optimal filter an energy resolution r = e/@xmath12e=16 was measured , with @xmath12e being the fwhm . the expected energy resolution based on the measured noise power spectrum and average pulse template was also r=16 , showing good agreement with the actual measurements . as shown in the bottom panel of figure [ fig : r ] ,
the noise consists of both white amplifier noise and pink tls noise exclusively in the phase direction .
most of the signal from the photons comes at frequencies above @xmath13 hz . at these frequencies the phase and amplitude noise are nearly identical , indicating that the dominant noise source is the hemt amplifier
the white noise level of -83 dbc / hz at the device readout power of -103 dbm is consistent with an amplifier noise temperature of 4 kelvin assuming 3 db of loss between the mkid and the amplifier .
reduction of the amplifier noise temperature or an increase in the maximum usable readout power should immediately improve the energy resolution .
in a device with a fixed response ( degrees of phase shift per ev of photon energy ) the energy resolution scales linearly with photon energy , except for a region between 350 and 700 nm where a significant fraction of photons pass through the metal and are absorbed in the silicon substrate .
these substrate events occur when a photon is absorbed close to the silicon / tin interface , causing a significant fraction of the phonons ( @xmath1470% ) created in the substrate to diffuse into the tin and break cooper pairs .
these substrate events can give quite large signals , and will be discussed in depth in a future paper .
simple device changes , such as making the tin film thicker or using a transparent substrate such as sapphire , should eliminate these unwanted substrate events .
significant improvements can be made to increase the energy resolution , as the theoretical energy resolution set by the creation statistics of the quasiparticles created during downconversion is @xmath15 , where @xmath16 is the efficiency of creating quasiparticles @xcite , @xmath17 is the energy of the incident photon , @xmath18 is the gap energy of the superconducting absorber , and @xmath19 is the fano factor @xcite .
this works out to r=150 at 5 ev for an operating temperature of 100 mk .
an operating temperature of 15 mk could allow a theoretical maximum energy resolution of r=400 at 5 ev , although it is likely other noise sources , like two level system noise @xcite , will become more important as future development increases the energy resolution .
figure [ fig : meas ] shows data from a typical 32x32 pixel device . in this device , 85% of the resonators were usable .
simple number counts showed that at least 95% of the resonators were present , but variations in thickness and t@xmath20 of the tin film caused some resonators to have similar resonant frequencies and overlap in frequency .
future improvement to the uniformity of the tin and a more robust algorithm for placement of resonators should significantly decrease the number of overlapping resonators .
, was approximately @xmath21 .
the bottom right panel is the frequency spacing in mhz between each resonator and its nearest neighbor .
most of the missing resonators are too close together in frequency ( @xmath22 khz , noted with a dashed line ) , resulting in only one resonator being included in the plot . ] the quantum efficiency of the device was estimated by depositing a 40 nm tin film on sapphire and measuring the reflection and transmission of the film as a function of wavelength , then subtracting these quantities from unity to give the fraction of photons absorbed in the superconducting film , as shown in the top right panel of figure [ fig : meas ] .
this was done with a varian cary 5000 spectrometer with accessories for absolute transmission and absolute specular reflectance .
this figure slightly overestimates the quantum efficiency of the final device because it does not account for losses in the microlens array , or the @xmath110% correction due to photons below 0.5 @xmath0 m slipping through the slots in the inductor .
the quantum efficiency is extremely good in the uv , but declines to about 30% at 1 @xmath0 m .
further development will likely result in significant increases in quantum efficiency .
improvements in the fabrication of tin films will lead to increased uniformity and fabrication yield .
the array size can be grown nearly arbitrarily by adding more resonators per feedline , likely to a maximum of around 10,000 resonators in a 1020 ghz band , and then by adding more microwave feedlines . the primary challenges to growing the array are in the digitization bandwidth and processing power of the room temperature electronics ,
not the device fabrication .
ole mkid arrays have now been proven in the lab , and the first astronomical test at the palomar 200 inch telescope has been conducted with data taken on a variety of astronomical objects . these arrays will bring extreme performance improvements to some of the most exciting areas of astrophysics , such as coronagraphic planet finding @xcite , transient and time variable sources @xcite , and high redshift galaxy evolution @xcite , and will likely also find application in other fields such as quantum optics @xcite and biological imaging @xcite . this material is based upon work supported by the national aeronautics and space administration under grant nnx09ad54 g , issued through the science mission directorate , jet propulsion lab s research & technology development program , and a grant from the w.m .
keck institute for space studies .
part of the research was carried out at the jet propulsion laboratory , california institute of technology , under a contract with the national aeronautics and space administration .
the authors would like to thank rick leduc , jonas zmuidzinas , sunil golwala , david moore , peter day , and omid noroozian for useful insights .
|
microwave kinetic inductance detectors , or mkids , have proven to be a powerful cryogenic detector technology due to their sensitivity and the ease with which they can be multiplexed into large arrays .
a mkid is an energy sensor based on a photon - variable superconducting inductance in a lithographed microresonator , and is capable of functioning as a photon detector across the electromagnetic spectrum as well as a particle detector .
here we describe the first successful effort to create a photon - counting , energy - resolving ultraviolet , optical , and near infrared mkid focal plane array . these new optical lumped element ( ole ) mkid arrays have significant advantages over semiconductor detectors like charge coupled devices ( ccds ) .
they can count individual photons with essentially no false counts and determine the energy and arrival time of every photon with good quantum efficiency . their physical pixel size and maximum count rate
is well matched with large telescopes .
these capabilities enable powerful new astrophysical instruments usable from the ground and space .
mkids could eventually supplant semiconductor detectors for most astronomical instrumentation , and will be useful for other disciplines such as quantum optics and biological imaging . 10 d. bintley , m. j. macintosh , w. s. holland , p. friberg , c. walther , d. atkinson , d. kelly , x. gao , p. a. r. ade , w. grainger , j. house , l. moncelsi , m. i. hollister , a. woodcraft , c. dunare , w. parkes , a. j. walton , k. d. irwin , g. c. hilton , m. niemack , c. d. reintsema , m. amiri , b. burger , m. halpern , m. hasselfield , j. hill , j. b. kycia , c. g. a. mugford , and l. persaud , `` characterising the scuba-2 superconducting bolometer arrays , '' proc .
spie * 7741 * , 2 ( 2010 ) .
m. d. niemack , y. zhao , e. wollack , r. thornton , e. r. switzer , d. s. swetz , s. t. staggs , l. page , o. stryzak , h. moseley , t. a. marriage , m. limon , j. m. lau , j. klein , m. kaul , n. jarosik , k. d. irwin , a. d. hincks , g. c. hilton , m. halpern , j. w. fowler , r. p. fisher , r. dnner , w. b. doriese , s. r. dicker , m. j. devlin , j. chervenak , b. burger , e. s. battistelli , j. appel , m. amiri , c. allen , and a. m. aboobaker , `` a kilopixel array of tes bolometers for act : development , testing , and first light , '' j. low temp .
phys .
* 151 * , 690 ( 2008 ) .
j. e. carlstrom , p. a. r. ade , k. a. aird , b. a. benson , l. e. bleem , s. busetti , c. l. chang , e. chauvin , h .-
m .
cho , t. m. crawford , a. t. crites , m. a. dobbs , n. w. halverson , s. heimsath , w. l. holzapfel , j. d. hrubes , m. joy , r. keisler , t. m. lanting , a. t. lee , e. m. leitch , j. leong , w. lu , m. lueker , d. luong - van , j. j. mcmahon , j. mehl , s. s. meyer , j. j. mohr , t. e. montroy , s. padin , t. plagge , c. pryke , j. e. ruhl , k. k. schaffer , d. schwan , e. shirokoff , h. g. spieler , z. staniszewski , a. a. stark , c. tucker , k. vanderlinde , j. d. vieira , and r. williamson , `` the 10 meter south pole telescope , '' pasp * 123 * , 568 ( 2011 ) .
r. l. kelley , s. r. bandler , w. b. doriese , y. ezoe , r. fujimoto , l. gottardi , r. den hartog , j .- w .
den herder , h. hoevers , k. irwin , y. ishisaki , c. a. kilbourne , p. de korte , j. van der kuur , k. mitsuda , t. ohashi , l. piro , f. s. porter , k. sato , k. shinozaki , p. shirron , s. j. smith , y. takei , p. whitehouse , and n. y. yamasaki , `` the x - ray microcalorimeter spectrometer for the international x - ray observatory , '' aip conf
. proc . * 1185 * , 757 ( 2009 ) .
w. b. doriese , j. n. ullom , j. a. beall , w. d. duncan , l. ferreira , g. c. hilton , r. d. horansky , k. d. irwin , j. a. b. mates , c. d. reintsema , l. r. vale , y. xu , b. l. zink , m. w. rabin , a. s. hoover , c. r. rudy , and d. t. vo , `` 14-pixel , multiplexed array of gamma - ray microcalorimeters with 47 ev energy resolution at 103 kev , '' appl .
. lett . * 90 * , 3508 ( 2007 ) .
m. d. eisaman , j. fan , a. migdall , and s. v. polyakov , `` single - photon sources and detectors , '' rev .
sci .
inst .
* 82 * , 071101 ( 2011 ) .
d. d. e. martin , p. verhoeve , a. peacock , a. g. kozorezov , j. k. wigmore , h. rogalla , and r. venn , `` resolution limitation due to phonon losses in superconducting tunnel junctions , '' appl .
phys .
lett . * 88 * , 3510 ( 2006 ) .
r. a. hijmering , p. verhoeve , d. d. e. martin , i. jerjen , a. g. kozorezov , and r. venn , `` direct position resolution measurement with droids at optical wavelengths , '' j. low temp . phys . * 151 * , 298 ( 2008 ) .
r. w. romani , a. j. miller , b. cabrera , s. w. nam , and j. m. martinis , `` phase - resolved crab studies with a cryogenic transition - edge sensor spectrophotometer , '' ap .
j. * 563 * , 221 ( 2001 ) .
j. burney , t. j. bay , j. barral , p. l. brink , b. cabrera , j. p. castle , a. j. miller , s. nam , d. rosenberg , r. w. romani , and a. tomada , `` transition - edge sensor arrays for uv - optical - ir astrophysics , '' nucl .
instrum . meth .
a * 559 * , 525 ( 2006 ) .
m. d. niemack , j. beyer , h. m. cho , w. b. doriese , g. c. hilton , k. d. irwin , c. d. reintsema , d. r. schmidt , j. n. ullom , and l. r. vale , `` code - division squid multiplexing , '' appl .
. lett . * 96 * , 3509 ( 2010 ) .
p. day , h. leduc , b. mazin , a. vayonakis , and j. zmuidzinas , `` a superconducting detector suitable for use in large arrays , '' nature * 425 * , 817821 ( 2003 ) .
j. a. schlaerth , n. g. czakon , p. k. day , t. p. downes , r. duan , j. gao , j. glenn , s. r. golwala , m. i. hollister , h. g. leduc , b. a. mazin , p. r. maloney , o. noroozian , h. t. nguyen , j. sayers , s. siegel , j. e. vaillancourt , a. vayonakis , p. r. wilson , and j. zmuidzinas , `` mkid multicolor array status and results from democam , '' proc .
spie * 7741 * , 4 ( 2010 ) .
( c ) 2010 : american institute of physics .
m. roesch , a. bideaud , a. benoit , a. cruciani , f. x. dsert , s. doyle , s. leclercq , f. mattiocco , k. f. schuster , l. swenson , and a. monfardini , `` characterization of lumped element kinetic inductance detectors for mm - wave detection , '' proc .
spie * 7741 * , 16 ( 2010 ) .
b. mazin , p. day , k. irwin , c. reintsema , and j. zmuidzinas , `` digital readouts for large microwave low - temperature detector arrays , '' nucl .
instrum . meth .
a * 559 * , 799801 ( 2006 ) .
p. verhoeve , d. d. e. martin , r. a. hijmering , j. verveer , a. van dordrecht , g. sirbi , t. oosterbroek , a. peacock , `` s - cam 3 : optical astronomy with a stj - based imaging spectrophotometer , '' nucl .
instrum . meth .
a * 559 * , 598 ( 2006 ) .
g. e. smith , `` the invention and early history of the ccd , '' app .
. lett . * 109 * , 102421 ( 2011 ) .
d. c. mattis and j. bardeen , `` theory of the anomalous skin effect in normal and superconducting metals , '' phys .
rev . * 111 * , 412417 ( 1958 ) .
s. doyle , p. mauskopf , j. naylon , a. porch , and c. duncombe , `` lumped element kinetic inductance detectors , '' j. low temp .
phys . * 151 * , 530536 ( 2008 ) .
h. g. leduc , b. bumble , p. k. day , b. h. eom , j. gao , s. golwala , b. a. mazin , s. mchugh , a. merrill , d. c. moore , o. noroozian , a. d. turner , and j. zmuidzinas , `` titanium nitride films for ultrasensitive microresonator detectors , '' appl .
. lett . * 97 * , 102509 ( 2010 ) .
o. noroozian , p. day , b. h. eom , h. leduc , and j. zmuidzinas , `` crosstalk reduction for superconducting microwave resonator arrays , '' ieee trans .
microw .
theory tech , submitted ( 2011 ) .
j. gao , m. daal , a. vayonakis , s. kumar , j. zmuidzinas , b. sadoulet , b. a. mazin , p. k. day , and h. g. leduc , `` experimental evidence for a surface distribution of two - level systems in superconducting lithographed microwave resonators , '' appl .
. lett . * 92 * , 152505 ( 2008 ) .
b. a. mazin , `` microwave kinetic inductance detectors , '' , ph.d .
thesis , california institute of technology ( 2004 ) .
a. g. kozorezov , j. k. wigmore , d. martin , p. verhoeve , and a. peacock , `` electron energy down - conversion in thin superconducting films , '' phys .
rev .
b * 75 * , 094513 ( 2007 ) .
u. fano , `` ionization yield of radiations 2 : the fluctuations of the number of ions , '' phys .
rev . * 72 * , 2629 ( 1947 ) .
j. r. crepp , l. pueyo , d. brenner , b. r. oppenheimer , n. zimmerman , s. hinkley , i. parry , d. king , g. vasisht , c. beichman , l. hillenbrand , r. dekany , m. shao , r. burruss , l. c. roberts , a. bouchez , j. roberts , and r. soummer , `` speckle suppression with the project 1640 integral field spectrograph , '' ap . j. * 729 * , 132 ( 2011 ) . r. j. bouwens , g. d. illingworth , m. franx , and h. ford , `` z 7 - 10 galaxies in the hudf and goods fields : uv luminosity functions , '' ap .
j. * 686 * , 230 ( 2008 ) .
l. ma , s. nam , h. xu , b. baek , t. chang , o. slattery , a. mink , and x. tang , `` 1310 nm differential - phase - shift qkd system using superconducting single - photon detectors , '' new j. phys . * 11 * , 045020 ( 2009 ) .
i. tinoco and r. l. gonzalez , `` biological mechanisms , one molecule at a time , '' gene dev . , * 25 * , 12051231 ( 2011 ) .
|
quasielastic studies are traditionally a good source of information about nuclear and nucleon structure . the main tool has been usually represented by electron scattering experiments , since in this case the elementary probe - nucleon reaction mechanism is regarded to be under better control .
however , not all the possible nuclear response functions can be accessed in this way and , furthermore , the extraction from the data of the response functions that enter electron scattering can be model dependent .
hadronic probes are an alternative source of quasielastic data . among them
it is particularly interesting the case of @xmath0-nucleus scattering , since , for kaon laboratory momenta below 800 mev / c , the elementary @xmath0-nucleon ( @xmath4 ) interaction is much weaker than other hadron - nucleon interactions .
in fact , a weaker elementary interaction allows the projectile to penetrate deeper inside the nucleus , thus probing regions of higher density and , consequently , being more sensitive to collective phenomena .
such an experiment has been performed at bnl , where quasielastic @xmath0-nucleus double - differential cross sections have been measured for kaons with a laboratory momentum of 705 mev / c , using d , c , ca and pb as targets @xcite .
data have been taken for scattering at 24@xmath5 , 34@xmath5 and 43@xmath5 , corresponding to approximately fixed momentum transfers of 290 , 390 and 480 mev / c . in the case of c ,
the data have been taken for all the momenta , whereas for ca and pb they are available at 290 and 480 mev / c . in refs .
@xcite the data have been analyzed using various relativistic models to describe the nuclear dynamics ; on the other hand , the distortion of the incoming kaons has been accounted for employing a very simple model , based on the effective number of nucleons participating in the reaction , @xmath6 ( see sec .
[ subsec : hp ] ) .
the `` experimental '' values found for @xmath6 are @xmath7 higher than the ones calculated in the glauber theory .
this finding is rather puzzling , also in view of the fact that multiple scattering theory underestimates the nuclear elastic scattering data @xcite , a fact which has been interpreted as a signal of an enhancement of the in - medium @xmath4 cross section . as also noted in ref .
@xcite , the quasielastic cross section is , in general , proportional to the elementary single differential @xmath4 cross section and to the effective number of participating nucleons , the latter depending , in turn , on the total @xmath4 cross section ( see sec . [ subsec : hp ] for details ) . at the energy of the experiment ,
@xmath4 scattering is nearly isotropic , so that the differential and total cross sections are proportional ; however , an increase in the @xmath4 cross section reduces @xmath6 ( and viceversa ) , leaving a quasielastic cross section little dependent on the input @xmath4 amplitudes .
thus , the quasifree experiment on the one side does not shed light on the elastic scattering data and , on the other , it seems to pose another puzzling problem . in this paper
we would like to reanalyze the @xmath0-nucleus quasielastic data , using a more realistic model for the reaction mechanism , already applied to ( @xmath8 ) and ( @xmath9 ) quasifree scattering @xcite , based on a consistent implementation of glauber theory for quasifree scattering .
the nuclear dynamics has been treated through a finite nucleus continuum random phase approximation ( rpa ) calculation , also accounting for the coupling of the particle - hole ( ph ) states to higher order configurations ( spreading width ) .
the dynamics is non - relativistic , but the correct relativistic kinematics has been used .
we will show that the puzzling experimental outcome for @xmath6 is very likely an artifact of the naive fitting procedure employed in the analysis of the data and that a realistic ( and _ without free parameters _ ) theoretical framework has no difficulties in giving a rather accurate description of the experimental cross sections . in sec .
[ sec : tm ] we introduce the theoretical methods employed in the calculations : here , one can find a brief description of the quasielastic nuclear responses in finite nuclei ; of the rpa using density - dependent interactions ; of the specific model for the ph effective potential that we have used ; and of the response functions to hadronic probes up to two - step processes in the glauber framework . in sec .
[ sec : results ] , we display the results of our calculations and in the last section we discuss their implications for the interpretation of the data .
the nuclear response function to an external probe , transferring momentum @xmath10 and energy @xmath11 , is proportional to the imaginary part of the polarization propagator@xcite : @xmath12 the latter reads @xmath13 \ ,
\label{eq : pi}\end{aligned}\ ] ] where \{@xmath14 } is a complete set of nuclear eigenstates of energy @xmath15 , @xmath16 the second quantized expression of the vertex operator and @xmath17 labels the spin - isospin channel . in the following , for the purpose of illustration , we shall deal explicitly only with the scalar - isoscalar channel , which is , by the way , the dominant one in @xmath4 scattering at the energies of concern to us , where @xmath18 : the case of the spin modes , which is slightly complicated by the spin algebra , is treated in detail in ref .
@xcite .
the angular part of @xmath19 ( we now drop , for simplicity , the channel label @xmath17 ) can be handled through a multipole decomposition that reads @xmath20 so that @xmath21 in a mean field ( shell model ) framework , one has for the particle - hole ( ph ) polarization propagator : @xmath22 { q^{jj0}_{ph}}^*(q ' ) \ , \label{eq : pj0}\ ] ] where @xmath23 ( @xmath24 ) labels a complete set of single particle ( hole ) quantum numbers and @xmath25^{1/2 } { \cal i}_{\ell , ph}(q)\left ( \begin{array}{ccc } \ell_p&\ell_h&\ell\\ 0&0&0 \end{array } \right ) \ .
\label{eq : qjl}\ ] ] in ( [ eq : qjl ] ) , @xmath26 is the standard @xmath27 recoupling coefficient and @xmath28 @xmath29 being the radial particle ( hole ) wave function and @xmath30 the associated eigenvalues .
they are obtained by solving the schroedinger equation with the woods - saxon potential @xmath31 ^ 2\frac{w_{so}}{a r } \frac{e^{(r - r)/a}}{\left[1+e^{(r - r)/a}\right]^2}\ , \bbox{\ell}\cdot\bbox{\sigma}\ ] ] ( neglecting , for simplicity , the coulomb term ) , where @xmath32 is the pion mass and the following set of parameters has been employed : @xmath33 note that the sum appearing in eq .
( [ eq : pj0 ] ) should be understood as a sum over the discrete part of the spectrum and an integration over the continuum one .
indeed , contrary to the widespread procedure of calculating the polarization propagator in coordinate space and then fourier transform to momentum space , we have directly evaluated eq .
( [ eq : pj0 ] ) in the latter . besides being fast and reliable
, this procedure allows also for an important extension to the shell model polarization propagator , namely the inclusion of the _ spreading width _ of the ph states .
this can be accomplished by adding in eq .
( [ eq : pj0 ] ) a complex ph self - energy , i. e. through the following substitution : @xmath34 although @xmath35 could in principle be calculated , we shall actually employ a phenomenological parameterization , writing @xmath36 with @xmath37 ( 1+c_{ts } ) \nonumber\\ \\
\delta_{ph}(\omega ) & = & [ \delta_p(\hbar\omega+\epsilon_h)+\delta_h(\epsilon_p-\hbar\omega ) ] ( 1+c_{ts } ) \ . \nonumber\end{aligned}\ ] ] the arguments of the functions @xmath38 and @xmath39 have been inferred from the analysis of the second order particle and hole self - energy contributions to @xmath35 .
the coefficients @xmath40 represent the corrections , in each isospin @xmath41spin @xmath42 channel , due to the ph interference diagrams : these have been estimated @xcite to contribute around 5% in @xmath43 and @xmath44 in @xmath45 , @xmath46 and @xmath47 , @xmath48 ; given the smooth dependence of the quasielastic response on the ph self - energy and the small contribution of these modes to the @xmath0-nucleus cross section , these corrections are not important and we shall set for simplicity @xmath49 in these channels . concerning the channel @xmath50 , the estimate of ref .
@xcite gives @xmath51 , resulting in a complete cancellation of the ph spreading width .
this outcome is valid in nuclear matter and in the limit of very large angular momenta , which is presumably good in the quasielastic peak ( qep ) region .
however , at low transferred momenta the region of small energy transfers of the quasielastic response is dominated by resonances associated to specific ( low ) angular momenta . in that case it is a better approximation to retain the full width and , accordingly , we shall choose @xmath52 $ ] with @xmath53 mev : then , resonances in the @xmath54 mev range get a width of a few mev , whereas for @xmath55 mev there is essentially no spreading .
finally , the @xmath38 are chosen according to the parametrization of ref .
@xcite , namely @xmath56 symmetrical with respect to the fermi energy ( @xmath57 ) , which gives a reasonable fit of the particle widths for medium - heavy nuclei , using @xmath58 mev , @xmath59 mev and @xmath60 mev @xcite .
the corresponding real parts are obtained through a once - subtracted dispersion relation@xcite . a final remark , concerning relativistic kinematics effects on the response functions , is in order .
we shall be concerned with momentum transfers up to 500 mev / c , where purely kinematical relativistic effects are starting to be sizable . at @xmath61 mev
/ c the relativistic position of the qep is @xmath62 mev below the non - relativistic one , the effect being even larger for energies on the right hand side of the qep . in ref .
@xcite , it had been shown that the non - relativistic fermi gas can be mapped into the relativistic one through the simple prescription @xmath63 the validity of ( [ eq : relativ ] ) in a finite nucleus shell - model calculation has been checked in ref .
the main effect of properly accounting for the relativistic dispersion relation turns out to be a moderate shrink of the response functions at the right of the qep ( see sec .
[ sec : results ] ) .
the polarization propagator introduced in the previous subsection is used in this paper as input in a continuum rpa calculation of the response functions . for a central density - independent interaction one
has to solve , for each multipole @xmath64 , the following integral equation ( see ref .
@xcite for the complications introduced by the tensor interaction ) : @xmath65 @xmath66 being the fourier transform of the two - body potential
. however , realistic effective nuclear interactions are in general density - dependent , especially in the scalar channels .
for a general interaction potential @xmath67 , one can introduce the double fourier transform @xmath68 then , instead of eq .
( [ eq : rpa ] ) , one has to solve the following rpa equations : @xmath69 having defined the kernel @xmath70 clearly , in the case @xmath71 one has @xmath72 : the kernel ( [ eq : kernel ] ) is then reduced to @xmath73 and one gets back to eq .
( [ eq : rpa ] ) . from eq .
( [ eq : rpa - dd ] ) it is apparent that the solution of the rpa equations with density - dependent forces does not pose any additional technical problem , apart from the input kernel , whose calculation is more involved . as discussed in detail in the next subsection , in the calculations of sec .
[ sec : results ] we have been using a parameterization of the effective nuclear interaction linear in the density .
in general , in any ( non - tensor ) channel one has @xmath74 with @xmath75 , @xmath76 and @xmath77 , @xmath78 representing the nuclear density , here approximated by the standard fermi distribution . in momentum space ( [ eq : vdd ] ) reads @xmath79 and its @xmath64th component in the angular momentum expansion ( [ eq : vddj ] ) turns out to be @xmath80 where @xmath81 @xmath82 representing the ordinary legendre polinomials .
two strategies are possible in order to determine the effective ph interaction in the nuclear medium : one can either directly fix an effective potential by fitting some phenomenological properties or start with a bare nucleon - nucleon interaction and calculate the related @xmath83-matrix . parameterizations of the ph interaction based upon the first procedure are generally only available at very low momentum transfers ( in terms of migdal - landau parameters ) ; since we are probing relatively high momenta , we have resorted to use a @xmath83-matrix and we have chosen the calculation of ref .
@xcite , which , in our view , has the following appealing features : it is based upon a realistic boson exchange potential ; ( nonlocal ) exchange contributions are included in the effective interaction , which is conveniently parameterized in terms of yukawa functions ; a parameterization of the density dependence is also provided .
the potential is given using the standard representation in spin and isospin ( no spin - orbit contribution will be considered in the following ) : @xmath84 where @xmath85 , @xmath86 ( @xmath87 , @xmath88 being the relative momenta in the initial and final state , respectively ) and the coefficients are density and momentum dependent .
however , before utilizing the nuclear matter interaction of ref .
@xcite in a finite nucleus calculation of quasielastic responses , a few issues have to be addressed .
\a ) the density dependence of the @xmath83-matrix is given in terms of density dependent coupling constants , which is not very useful for our purposes .
furthermore , the parameterization is fitted for @xmath89 , @xmath90 being the fermi momentum , which spans a range of densities down to roughly 1/3 of the central density : extrapolation of their parameterization to lower densities gives unreasonable results .
thus , we have chosen to employ the linear @xmath91 dependence of eq .
( [ eq : vdd ] ) [ or ( [ eq : vddq ] ) ] , which is known to provide a reasonable parameterization ( see , e. g. , ref .
one can see in fig .
[ fig : veff - kf ] a comparison of the two parameterizations for the @xmath90 dependence of the effective interaction .
it should be noted that most of the contribution to the quasielastic responses comes from densities where the two descriptions differ by a few per cent .
\b ) in order to obtain , at a fixed density , a local interaction , one can use the relation between @xmath10 , @xmath92 and @xmath93 , i. e. , @xmath94 , substituting @xmath93 with a suitably chosen average value , @xmath95 ; then , the only independent momentum is @xmath10 . the authors of ref .
@xcite were interested in a potential for nuclear structure calculations : hence , they put the momenta of the two nucleons in the initial state on the fermi surface and averaged over the relative angle , getting @xmath96 .
clearly , in this case one has the constraint @xmath97 . on the other hand ,
we are interested in the ph interaction in the quasielastic region : one nucleon in the initial state is below the fermi sea , while the other can be well above it .
a look at fig .
[ fig : veff - diag ] shows that @xmath87 is defined , in terms of the particle and hole momenta , as @xmath98 .
thus , at fixed @xmath99 one should average @xmath93 over @xmath100 and @xmath101 , getting @xmath102 .
now @xmath93 is growing with @xmath10 , so that there are no longer constraints on @xmath10 , and the exchange momentum turns out to be constant , @xmath103 .
one can see in fig .
[ fig : veff - nontensor ] the resulting interaction in the non tensor channels .
\c ) in the case of the tensor channels things are simpler , since in the parameterization of ref .
@xcite there is no explicit density dependence ( fig .
[ fig : veff - tensor ] ) .
the coefficients of the exchange tensor operator , @xmath104 and @xmath105 , display a very mild density dependence , induced by @xmath92 , which is completely negligible .
the only catch concerns the treatment of @xmath106 : assuming that @xmath99 and @xmath107 are orthogonal , with some algebra one can show that @xmath108 .
note , however , that @xmath109 of the quasielastic cross section for @xmath0 scattering is due to the scalar - isoscalar channel .
the formalism introduced above is not enough when one is dealing with strongly interacting probes , in which case a framework for the reaction mechanism must be provided . for this purpose
, we have chosen the glauber approach@xcite , including up to two - step inelastic processes .
simple treatments within the glauber theory usually amount to including the effects due to rescattering in an effective number of nucleons participating in the reaction , thus effectively renormalizing the response functions defined above . in the case of @xmath0 quasielastic scattering
, this approach has been followed in refs .
@xcite .
however , a consistent treatment within glauber theory leads one to the definition of _ surface response functions_@xcite . a detailed derivation , in particular for the spin - isospin channel ,
can be found in ref .
@xcite . here , we briefly sketch the relevant points , again using the scalar - isoscalar channel as example .
a surface polarization propagator can be obtained from ( [ eq : pi ] ) by substituting the vertex operator @xmath110 , which describes the probe - nucleon coupling , with @xmath111 @xmath112 and @xmath113 being bidimensional vectors in the plane orthogonal to the direction of motion of the projectile , @xmath114 the elementary probe - nucleon amplitudes , @xmath115 the effective total probe - nucleon cross section and @xmath116 @xmath117 being the nuclear density . with these definitions ,
the effective number of participating nucleons is given as @xmath118 @xmath115 depends on the energy and we shall take the average over the total cross sections at the incoming and outgoing projectile energies : for @xmath0 scattering at @xmath119 mev / c , the variation of @xmath115 is rather weak , going roughly from 14 mb to 13 mb for energy losses up to 250 mev .
one then finds , for the @xmath64-th multipole , @xmath120 \pi_{j(\alpha)}(\lambda,\lambda';\omega ) \nonumber\\ & & -2\frac{1}{|f_{\alpha}(q)|^2}\int_0^\infty d\lambda\,\lambda\,\text{re}[f^*_{\alpha}(q)\ , f^{\phantom{*}}_{\alpha}(\lambda)\ , h^{(0)}_j(\lambda;q)]\pi_{j(\alpha)}(q,\lambda;\omega ) \ , \end{aligned}\ ] ] having set @xmath121 where @xmath122 j_m(\lambda b)\,j_m(qb)\ ] ] and @xmath123 with the previous definitions , the one - step surface response functions are given as @xmath124 and the double differential cross section turns out to be @xmath125 to be compared with the effective number approximation , which reads @xmath126 multiple scattering inelastic contributions can , in principle , be incorporated along the same lines . since in the kinematic regions of interest they turn out to be much smaller than the one - step terms , it is not worth going through a very complex formalism and one can safely stick to the effective number approximation . in this case ,
the two - step contribution is proportional to the convolution of two one - step response functions@xcite . by summing over all the possible channels and using free one - step response functions
, one can define the two - step response as @xmath127 where @xmath128 is the momentum of the projectile , @xmath129 the @xmath130 amplitude in the @xmath131 channel and @xmath132 is proportional to the effective number of pairs participating in the double scattering . the full quasielastic @xmath0-nucleus cross section discussed in sec .
[ sec : results ] is given by the sum of one- and two - step terms : @xmath133 \ .\ ] ] we conclude this section with a few remarks about the elementary @xmath134 amplitudes employed in the calculations .
the amplitude for elastic @xmath134 scattering can be written as @xmath135 , with @xmath136 , @xmath137 being a unit vector normal to the scattering plane . in fig .
[ fig : amplitudes ] one can see the amplitudes as a function of the momentum transfer for @xmath119 mev / c@xcite : the dominant channel is clearly the scalar - isoscalar one .
some care should be taken in the choice of the reference frame where the @xmath134 amplitudes are evaluated . in order to be able
to factorize the two - body amplitude out of the nuclear response functions , an `` optimal '' choice of the reference frame has to be done@xcite : the optimal momentum of the struck nucleon turns out to be @xmath138 such as to reduce to zero at the qep and to the breit frame value at @xmath139 .
the most relevant consequence of this choice of frame is that the elementary projectile - nucleon amplitudes should be evaluated at an effective laboratory kinetic energy defined by @xmath140 @xmath141 and @xmath128 being the projectile mass and momentum , respectively .
this introduces a dependence on @xmath11 of the amplitudes entering in the evaluation of the quasielastic cross section , which can be important if the former are strongly energy dependent : at the kinematics of relevance to us , the effective @xmath0 momentum varies roughly in the range @xmath142 mev / c , where the amplitudes change sufficiently slowly to make the effects not at all dramatic .
let us start by briefly discussing the relativistic kinematical effects , introduced at the end of sec .
[ subsec : qnr ] . in fig .
[ fig : rel ] one can see the rpa scalar - isoscalar response of @xmath143c to @xmath0 probes in the non - relativistic case compared to the response function where the relativistic dispersion relation has been accounted for . as expected , the effect is negligible at the lowest momentum transfer ( @xmath1 mev / c ) , whereas at higher momenta ( @xmath144 mev / c ) it produces a moderate shrink of the response at the right of the qep , compensated by an enhancement ( @xmath145 ) at the peak position .
next , we would like to discuss the uncertainties connected to the choice of the effective ph interaction for quasielastic calculations .
the knowledge of the latter suffers , in general , from many shortcomings : when it is theoretically calculated , as the one we employ here ( see sec . [ subsec : eff - int ] ) , sources of uncertainties arise because of the specific many - body scheme which has been adopted , from relativistic effects ( see ref .
@xcite for an effective interaction along the lines of ref .
@xcite , but in a relativistic context ) and from the approximations made in the actual calculations or in extrapolating to the ph kinematical regime ( see sec .
[ subsec : eff - int ] ) ; when the effective interaction is just fitted to phenomenological properties , it can suffer from ambiguities in the parameterization ( many sets of parameters reproducing the same body of data ) and from the limited range of momenta which is covered ( see , for instance , ref .
@xcite for a phenomenological density - dependent interaction expressed in terms of migdal - landau parameters at @xmath146 ) .
on the other hand , one can of course reverse the argument and use the quasielastic studies to gain insight into the effective ph interaction at momentum transfers of a few hundreds mev .
one obvious difficulty in this case is related to the fact that it is not always possible to disentangle the various spin - isospin channels : the only reaction for which this has been achieved is the ( @xmath147 ) charge - exchange one , where the separated isovector spin - longitudinal and spin - transverse responses have been extracted @xcite .
however , the strong interaction of protons with nuclei constrains that reaction to the low density peripheral region of the nucleus , making it little sensitive to rpa effects ; moreover , while the spin - longitudinal channel can be well described , the transverse one shows much more strength than expected @xcite . in this
regard , the @xmath0-nucleus reaction is much more promising for two reasons : first of all , the small @xmath134 cross section makes the kaon enter inside the nucleus much more deeply than protons or pions , in turn implying stronger collective effects ; secondly , as seen in fig .
[ fig : amplitudes ] , the scalar - isoscalar channel is largely dominant , making this reaction a `` quasipure '' probe of the @xmath50 mode . whatever be the attitude towards this problem , we believe it is useful to gain some feeling about the sensitivity of the collective effects to the input effective potential . for this purpose , we display in fig . [
fig : rpa ] , for the four spin - isospin channels , the surface rpa response functions at @xmath144 mev / c , calculated with the ph interaction discussed in sec .
[ subsec : eff - int ] , comparing them to the rpa responses calculated with the same interaction scaled by @xmath148 .
this amount for the rescaling is not an estimate of the theoretical or phenomenological uncertainty of the ph interaction ; however , concerning the scalar - isoscalar channel , there are reasons to believe that the @xmath83-matrix estimate of ref .
@xcite gives too much attraction : as discussed by the authors , the extracted @xmath149 landau parameter makes nuclear matter unstable , a shortcoming that can be cured by higher order contributions channel has been shown to be reduced by nearly a factor @xmath150 at @xmath151 ; however , in the range of momenta of interest to us , the relativistic and non - relativistic @xmath83-matrices appear to be comparable . ] .
thus , while the case with the interaction increased by @xmath152 is taken mainly for illustrative purposes , the case with a weaker interaction is somewhat more realistic . by inspecting fig .
[ fig : rpa ] one can note the following : \i ) the dependence of the responses on the potential is somehow desensitized by the rpa chain : for the @xmath50 mode the variation of the response function at the left of the qep is typically about @xmath153 , apart from the very low energy region ( @xmath154 mev ) , where it grows up to @xmath155 ; for the other modes , still at the left of the qep , the variation is always much smaller , say @xmath156 ; \ii ) at the right of the qep the effect of correlations is much smaller , being limited to a few per cent for the @xmath50 mode and to practically nothing for the other modes ; \iii ) the relatively weak sensitivity of the isovector and spin channels to variations of the input potential makes the small ( @xmath157 ) contamination from these modes in the @xmath0-nucleus cross section rather stable with respect to uncertainties due to correlations , enhancing the argument in favour of this reaction as a probe of the scalar - isoscalar channel . the results if fig .
[ fig : rpa ] are given for @xmath144 mev / c : similar considerations apply also to the lowest momenta . in fig .
[ fig : c12 ] one can see the data for the quasielastic @xmath0-c cross section @xcite compared to our calculations . in the theoretical cross sections , besides the case with free response functions ( dotted ) , there are included the nuclear responses corresponding to the interaction of sec .
[ subsec : eff - int ] ( solid ) and to a reduction of @xmath152 of the same interaction ( dashed ) ; the curves include the two - step contribution , which is also shown separately ( dot - dashed ) . from inspection of the figure
, one sees that the strength of the cross sections on the right hand side of the qep is well reproduced by both the correlated and the free responses ( apart from the very high energy tail ) at all the momentum transfers ; however , at the lowest momenta the data show a clear distortion of the typical quasielastic shape , which is not compatible with the uncorrelated cross sections .
the experimental energy resolution is not sufficient to cut out the elastic scattering contribution ; moreover , low - lying discrete nuclear excitations are much affected by the detail of the nuclear model ( such as shell model parameters and spreading width ) .
if , for these reasons , one excludes the very low energy tail ( say , @xmath158 mev ) it appears that the model with a weaker ph interaction gives a remarkably good description of the data at @xmath1 and @xmath2 mev / c . at the highest momentum , the elastic contamination and the discrete excitations
have been washed out ; the interaction in both the correlated models is sufficiently weak to cause little distortion of the quasifree shape , but the low energy part of the spectrum seems again to favour the model with the weaker ph interaction .
it is worth noticing the smallness of the two - step contribution , which is at most a few per cent of the one - step term at @xmath144 mev / c : this can be contrasted , for instance , to the isovector spin - transverse response to 500 mev protons , where one finds , at @xmath159 mev / c , a sizeable @xmath153 contribution from two - step processes @xcite .
protons of 500 mev of kinetic energy have a squared momentum which is roughly twice the one of 700 mev / c kaons ; furthermore , the two - step factor @xmath160 of eq .
( [ eq : r2 ] ) and ( [ eq : d2 ] ) is larger by roughly a factor two for the kaon reaction , because of the smaller probe - nucleon total cross section @xmath115 , but this is compensated by the factor two arising from the two possible orderings of the charge - exchange reaction ( see formula ( 8) of ref .
hence , apart from a factor 2 in the coefficient , the relative weight of the two - step term in the two reactions has to be driven by the elementary amplitudes entering ( [ eq : r2 ] ) ( the response functions being in all the cases the free ones ) , that is by the factor @xmath161 in the analogous expression for the ( @xmath9 ) reaction , @xmath99 being the external momentum and @xmath162 the integration variable , whereas @xmath163 and @xmath164 are the isovector spin - transverse and the scalar - isoscalar @xmath165 amplitudes , respectively .
the different size of the two - step contributions is indeed determined by the different strength and the different momentum behaviour of the above amplitudes : in the kaon case , the scalar - isoscalar amplitude has little momentum dependence , so that @xmath166 is roughly between one and two , while the remaining squared amplitude is on average @xmath167 mb ( in the laboratory ) ; on the other hand , for the ( @xmath9 ) reaction at @xmath159 mev / c , the ratio @xmath168 is typically in the range @xmath169 , whereas the remaining amplitude gives contributions mainly in the range from 10 to 22 mb . in qualitative words , we can say that because the @xmath165 cross section is relatively forward peaked , unlike the @xmath134 one , a double scattering in which the momentum transfer @xmath10 is shared by the two nucleons is favored in the @xmath165 case with respect to the @xmath134 one . finally , we display also the available data for ca at @xmath1 and @xmath3 mev / c ( fig .
[ fig : ca40 ] ) . to these figures
one can apply the same considerations made for the carbon data , although the preference for the weaker ph effective potential is even more clear now , since in heavier nuclei collective effects tend to be stronger .
indeed , the sensitivity of the response functions to variations in the input potential is a little more pronounced in @xmath170ca than in @xmath143c , being about @xmath171 for a @xmath152 change of the potential in the @xmath50 channel , to be compared to the @xmath153 sensitivity found in @xmath143c ( see the discussion of fig .
[ fig : rpa ] ) . also the two - step contribution is ,
as expected , larger in @xmath170ca than in @xmath143c .
from the discussion in the previous section , it appears that the model presented here gives a good description of the quasielastic @xmath0 data , the main uncertainty being related to the strength of the effective ph interaction in the scalar - isoscalar channel .
a good description of the data was achieved by using the interaction of ref .
@xcite quenched by @xmath152 .
it was shown that the use of this interaction with the full strength led to results in clear contradiction with the data .
indeed , there are theoretical indications that the strength of this interaction in the scalar - isoscalar channel should be reduced . on the other hand , from figs .
[ fig : c12 ] and [ fig : ca40 ] one can see that the @xmath0 data can be used to constrain the strength of the ph potential for the @xmath50 mode .
there are , however , two issues that need to be commented upon , namely the reported high experimental values for the effective number of nucleons @xmath6 @xcite ( see the introduction ) and the comparison with the available calculations for this reaction that employ relativistic dynamical models @xcite .
concerning the first point , it might appear curious that the `` experimental '' value for @xmath6 quoted in refs .
@xcite be @xmath7 higher than the one calculated in the glauber model , since we have seen that the latter gives a nice estimate of the quasielastic cross sections .
a source of error may of course be given by the use of the effective number approximation ( [ eq : neff - approx ] ) ; however , we believe that the main reason for the overestimate of @xmath6 lies in the way followed to extract the effective number from the data @xcite . indeed , @xmath6 has been obtained by integrating the quasielastic data over the whole range of transferred energies : in order to do this , the data have been fitted with gaussian distributions and , although a subtraction of the elastic scattering and low energy excited levels contributions has been attempted , _ no subtraction for any high energy background has been included_. from figs .
[ fig : c12 ] and [ fig : ca40 ] it appears that the very high energy tail of the cross sections is underestimated by the calculations , both for the free and the correlated models ( remember , from the discussion of sec .
[ sec : results ] , that while rpa correlations sizably affect the low energy side of the qep , they have little influence on the high energy part ) .
the data actually seem to show some structure beyond the ph response region : in the case of quasifree electron scattering , the strength in this region is attributed to effects beyond rpa ( such as meson - exchange currents or 2p2h excitations ) .
note also that a simple , `` straight line '' , estimate of the background brings the value of @xmath6 in close agreement with the glauber estimate @xcite .
meson - exchange currents effects tied to the interaction of the kaon with the nuclear pion cloud have been evaluated in refs .
@xcite and found to lead to small corrections to the @xmath0-nucleus total cross section .
the consideration of contact terms using chiral lagrangians makes the corrections even smaller @xcite , so that this source of corrections can not account for the strength at large values of @xmath11 . on the other hand ,
the tail of the @xmath172 excitation reaches this region , since the @xmath172 acquires a finite width , even below the pion production threshold , due to the @xmath173 reaction in the nucleus .
hence , the channel of kaon induced @xmath172 excitation in nuclei should become a target of both theoretical and experimental investigation in order to further clarify these issues . concerning the calculations with relativistic models of refs .
@xcite , it is obviously difficult to make a comparison with our results , since in those papers the rpa response functions ( calculated in a variety of models ) have been multiplied by the `` experimental '' @xmath6 , so that their strength is just fitted to the data ( of course , the use of the calculated value for @xmath6 would result in a general underestimate of the cross sections ) .
however , it should be noticed that the relativistic rpa isoscalar response is _ quenched _ , in contrast to our non - relativistic rpa response , which is mainly shifted to the left and enhanced at very low energies .
then , it is clear that this quenching has to be compensated by a higher @xmath6 . also to be noticed is the fact that the relativistic rpa calculations are able to describe ( albeit through a fit of the total cross section ) only the data at high momentum transfers . at @xmath1
mev / c , the large enhancement in the cross section visible at moderate energy transfers ( @xmath174 ) is clearly not predicted , even in a finite nucleus calculation @xcite , in contrast to our non - relativistic results . of course , before drawing more firm conclusions
one should test the model also against the ( @xmath175 ) data and calculations in this direction are in progress . in this connection , it is interesting to note that a recent reanalysis of the ( @xmath175 ) world data @xcite seems to rule out the long - standing problem of the missing strength in the charge channel , which had been interpreted in terms of a quenching of the charge response .
we would like to acknowledge r. j. peterson for sending us detailed information on the experiments .
this work has been supported by the program of human capital and mobility of the eu , contract n. chrx
ct 930323 , and by cicyt , contract n. aen 961719 .
r. d. smith , in _ proceedings of the international conference on spin observables of nuclear probes _ ,
telluride , colorado , 1988 , edited by c.j .
horowitz , c.d .
goodman and g. walker , ( plenum , new york , 1989),p . 15 .
|
quasielastic @xmath0-nucleus scattering data at @xmath1 , @xmath2 and @xmath3 mev / c are analyzed in a finite nucleus continuum random phase approximation framework , using a density - dependent particle - hole interaction .
the reaction mechanism is consistently treated according to glauber theory , keeping up to two - step inelastic processes .
a good description of the data is achieved , also providing a useful constraint on the strength of the effective particle - hole interaction in the scalar - isoscalar channel at intermediate momentum transfers .
we find no evidence for the increase in the effective number of nucleons participating in the reaction which has been reported in the literature .
|
in the spring of 1909 evershed published a remarkable discovery of strong horizontal mass flows in sunspots penumbra , the outer part of sunspots characterized by filamentary magnetic field structures @xcite .
the flows , with typical speed of 14 km / s , start at the boundary between the umbra and penumbra and expand radially , accelerating with distance and suddenly stopping at the outer sunspot boundary .
the evershed effect may play a significant role in the formation , stability and dynamics of sunspots and is considered as one of the fundamental process in solar physics .
this phenomenon caused significant interest and detailed observational and theoretical investigations , but the understanding of the physical mechanism is still missing ( a recent review is published by @xcite ) . high - resolution observations from large ground - based telescopes and the hinode space mission revealed a complicated filamentary structure of these flows @xcite and their non - stationary dynamics in the form of quasiperiodic
evershed clouds " @xcite . in some cases
, the flows showed a large - scale coherent behavior across several flow channels @xcite , and also provided evidence of a wave - like behavior @xcite .
theories of the evershed effect can be divided in two categories , describing it as channel flows in magnetic flux tubes @xcite or as elongated magnetoconvective rolls @xcite .
recent numerical simulations @xcite successfully modeled the filamentary magnetic structure of sunspot penumbra and horizontal outflows , thus providing a strong support to the magnetoconvective nature of the evershed effect @xcite . in this paper , we present a study of solar magnetoconvection in the presence of inclined magnetic field , based on the realistic radiative mhd simulations , and link the evershed effect to the phenomenon of traveling magnetoconvection waves .
the convective waves is a very interesting mhd phenomenon @xcite , which , in fact , has been previously suggested as a reason of the evershed flows @xcite , but did not receive further development .
our study provides a basis for explaining the evershed effect as a result of traveling magnetoconvection waves in highly inclined magnetic field of sunspot penumbra .
the phenomenon of traveling magnetoconvection waves is considered also in the dynamics of the earth s core @xcite , and may happen in various astrophysical objects , such as magnetic stars , accretion disks , compact objects and active galactic nuclei .
thus , detailed observational and theoretical studies of this phenomenon are of great interest .
we use a three - dimensional non - linear radiative - magnetohydrodynamics code developed for simulating the upper solar convection zone and lower atmosphere @xcite .
this code takes into account several physical phenomena : compressible fluid flow in a highly stratified medium , 3d multi - group radiative energy transfer between the fluid elements , a real - gas equation of state , ionization and excitation of all abundant species , and magnetic effects .
a important feature of this code is implementation of various subgrid scale turbulence models . in this paper
, we adopted the most widely used smagorinsky model @xcite in the compressible formulation @xcite .
the turbulent electrical conductivity is calculated by using the extension of the smagorinsky model to the mhd case @xcite .
we simulate the upper layer of the convection zone , extending from 5 mm below the visible surface to 0.5 mm above the surface .
the horizontal size varied from 6.4 mm @xmath0 6.4 mm to 25 mm @xmath0 25 mm . the computational grid step size varied from 25 to 100 km .
the results of this paper are obtained using @xmath1 grid with the step size of 50 km ( except fig . 5 , which is obtained using @xmath2 grid ) .
we have verified that the results do not change in computations with finer grids ( except of the development of smaller scale turbulent motions ) .
the initial uniform magnetic field is imposed on a snapshot of the preexisting hydrodynamic convection @xcite .
the initial field strength , @xmath3 , varies from 0 to 2000 gauss , and the inclination angle , @xmath4 , varies from 0 to 90 degrees .
the lateral boundary conditions are periodic , and the top and bottom boundary conditions maintain the total magnetic flux and the mean inclination .
this formulation allows us to carry out a series of controlled numerical experiments and investigate how the structure and dynamics of solar turbulent convection depend on the magnetic field properties in regimes close to the observed in sunspot penumbra , and elucidate the physical mechanism of the evershed effect .
outside magnetic field regions the solar convection forms granular cells of a typical size of 1 - 2 mm and lifetime of about 10 min . in the presence of magnetic field the structure of convection strongly depend on the field strength and inclination .
when the magnetic field is vertical the granules become smaller @xcite , and their overturn time is shorter resulting in generation of high - frequency turbulence and acoustic waves ( `` halos '' ) @xcite . in the presence of an inclined magnetic field , such as observed in sunspot penumbra , the granular cells becomes naturally elongated in the direction of the field because magnetic field restricts motions across the magnetic field lines .
but the most interesting effect is that the inclined field changes the nature of solar convection . instead of a stationary overturning convection pattern
the simulation reveal traveling convection waves , which become more apparent and stronger for higher field strengths and inclination .
this convection develops long narrow structures of velocity , thermodynamic parameters and magnetic field , resembling the filamentary structure and motions in the penumbra of sunspots .
these structures are illustrated in figures 1 and 2 .
figure 1 shows a 3d slice of our computational domain with a sample of magnetic field lines ( red curves ) , velocity field ( black arrows ) and a volume rendering of the temperature structures ( blue - red color scale ) .
the initial 1000 gauss magnetic field is oriented in the @xmath5-plane and inclined by 85 degrees to the @xmath6-axis , so that the @xmath7-vector is the positive in the @xmath8-direction . evidently the strongest motions occur in the direction of the field inclination in narrow structures , with upflows and downflows at the initial and end point of these structures .
the magnetic field lines change in accord with these elongated motions , rising up at the initial points and declining at the end points , thus giving an impression of rising and falling loop - like motions .
the temperature is typically higher at the start points and lower at the end points .
the typical vertical velocity around these points is about 1 km / s , but the horizontal velocity between them in the positive @xmath8-direction reaches 46 km / s .
most of the horizontal mass flow occur in these relatively narrow patches , which strongly resemble evershed clouds " , discovered in observations @xcite .
significantly weaker flows in the opposite direction are also observed .
these are often originate at the initial upflow points .
vertical cuts through the flow field ( e.g. left @xmath9-plane in fig .
1 ) reveal associated vortex - type motions below the surface .
when the background field is strong , the horizontal flow patches become quite narrow , with the width of 0.5 mm or less .
the magnetic field variations also becomes more filamentary .
this is illustrated in fig . 2 , which shows the surface structure of the horizontal flows and the @xmath10 component for two different initial magnetic field strengths , 1000 and 1200 g. the simulations show strong interaction between the plasma flows and magnetic field .
the magnetic field controls the general direction of the flows , but in the strong flow patches , the magnetic field is pushed aside and has a reduced magnetic field strength
. this may give impression , sometimes , reported from observations that the flows occur in magnetic field gaps " .
nevertheless , the plasma flows remain magnetized . the filamentary magnetic structures and flows
are strongly coupled .
the most interesting feature of the simulations , which , we argue , is a key for understanding the evershed effect , is the traveling wave pattern of magnetoconvection in the presence of strongly inclined magnetic field .
the simulations show that the velocity patches and magnetic field perturbations migrate in the direction of the field inclination .
vertical cuts in @xmath5-planes show rapidly moving inclined convective cells ( a snapshot is illustrated in fig .
this process is best seen in the movies ] , and also in the time - distance slices of the surface @xmath11 velocity component along the @xmath8-axis .
figure 4 shows an example of these slices for the initial magnetic field , @xmath12 g , and the inclination angle of @xmath13 . in this case
the convective velocity reaches @xmath14 km / s , and a pattern of convection waves traveling in the direction of the field inclination with a speed of 1 - 2 km / s can be identified .
the general picture is that the overturning convection motions are swept by the traveling waves .
this interaction amplifies the flows in the direction of the waves .
this is accompanied by weaker plasma motions in the opposite direction .
in fact , the initial points of the convective upflows often move in the opposite direction .
this may explain the puzzling discrepancy between the outward flow direction and the apparent motion of penumbra grains " .
it is intriguing that the traveling convection pattern shows variations with a characteristic time of 20 - 50 min , resembling the quasi - periodic behavior noticed in the observations @xcite . by increasing the computational domain up to 25 mm
we have checked that the quasi - periodicity does not depend on the size of the domain and , thus , is not due to the periodic boundary condition .
this is an intrinsic property of the inclined field magnetoconvection , but understanding of this phenomenon require further investigation . as we have pointed out the high - speed ( 4 - 6 km / s ) horizontal flows occur in localized patches corresponding to the
evershed clouds " .
an averaged over time and space velocity is smaller , about 12 km / s .
the flows are concentrated in a shallow subsurface layer less than 1 mm deep ( fig .
the velocity peaks about 100200 km below the surface .
this also corresponds to the observations showing that the velocity of the evershed flows increases with depth .
the averaged velocity does not change much with the magnetic field strength in the range of 10001500 g , but it strongly depends on the inclination angle ( fig .
the mean horizontal flow is much weaker for small inclination angles .
the radiative mhd simulations of solar magnetoconvection in regions of inclined magnetic field qualitatively and quantitatively describe many observed features of the evershed effect in sunspots .
the results indicate that the principal physical mechanism of the evershed flows is the traveling wave nature of the magnetoconvection .
the traveling waves have been extensively studied in idealized situations @xcite , and it has been suggested that they play significant role in sunspot flows @xcite .
our simulations model this phenomenon in the realistic solar conditions , and show that indeed many details correspond to the observations , thus providing a basis for explaining the evershed effect . in particular , the simulations show that the high - speed flows reaching 4 - 6 km / s occur in the direction of the field inclination in narrow , 23 mm long patches , which tend to appear quasi - periodically on the time - scale of 15 - 40 min .
these patches correspond to the so - called evershed clouds " @xcite and represent the main component of the evershed flows .
these horizontal flows originate from convective upflows of hotter plasma , like in ordinary convection , but are channeled by the magnetic field and amplified by the traveling convective waves .
the whole process is highly non - linear and stochastic with high - speed patches appearing randomly , but the simulations also show large - scale organization patterns across the simulation domain , which seem to be associated with the traveling waves .
these patterns are evident in the simulation movies .
some observations showed a signature of coherence in appearance of the evershed clouds @xcite but this has not been fully established @xcite .
the simulations suggest that a large - scale coherence may be a fundamental property of the traveling wave phenomenon , and certainly encourage further observational studies . of course , in real sunspots the magnetic field stricture is highly inhomogeneous , and this may affect the large - scale appearance .
this must be investigated in future simulations . in the past several models
were suggested to explain the evershed effect .
interestingly , some features of these models can be found in our simulations .
one of the first models describes the penumbra filaments as convective rolls along the direction of magnetic field @xcite , first suggesting the convective nature of the evershed effect .
the apparent observed wave - like behavior inspired attempts to explain the evershed effect as magnetoacoustic or magnetogravity waves @xcite .
our model specifies that these waves are convective in nature .
the rising and falling thin - flux tube model @xcite was suggested to describe the discrepancy between the apparent motion of penumbra features and the main evershed flows .
our simulations explain this naturally , and also reveal upward and downward loop - like motions of magnetic field lines synchronized with the high - velocity patches .
the siphon model @xcite suggested that the flow is driven by the pressure difference between the initial and end points , and indeed , in the simulations the gas pressure in the initial points is higher than at the end points .
the recent numerical simulations of the sunspot structure @xcite led to the suggestion that the evershed effect is caused by the overturning convection @xcite , but the flow speed was not sufficiently high . our simulation show that the high - speed matching the observations is achieved if the magnetic field is strong , 1000 - 1500 g , and highly inclined , when the magnetoconvection has properties of traveling waves .
thus , it seems that the mhd simulations provide a unified description of the models and the key observed features , and , perhaps , lead to the understanding of the 100-year old discovery .
bunte , m. , darconza , g. , & solanki , s. k. 1993 , , 274 , 478 cabrera solana , d. , bellot rubio , l. r. , beck , c. , & del toro iniesta , j. c. 2007 , , 475 , 1067 cabrera solana , d. , bellot rubio , l. r. , borrero , j. m. , & del toro iniesta , j. c. 2008 , , 477 , 273 danielson , r. e. 1961 , , 134 , 289 evershed , j. 1909 , , 69 , 454 georgakilas , a. a. , & christopoulou , e. b. 2003 , , 584 , 509 germano , m. , piomelli , u. , moin , p. , & cabot , w. h. 1991 , physics of fluids , 3 , 1760 heinemann , t. , nordlund , . , scharmer , g. b. , & spruit , h. c. 2007 , , 669 , 1390 hurlburt , n. e. , matthews , p. c. , & proctor , m. r. e. 1996 , , 457 , 933 hurlburt , n. e.,matthews , p. c. , & rucklidge , a. m. 2000 , , 192 , 109 ichimoto , k. , et al . 2007a , science , 318 , 1597 ichimoto , k. , et al . 2007b , , 59 , 593 jacoutot , l.,kosovichev , a. g. , wray , a. , & mansour , n. n. 2008a , , 684 , l51 jacoutot , l. , kosovichev , a. g. , wray , a. a. , & mansour , n. n. 2008b , , 682 , 1386 maltby , p. , & eriksen , g. 1967 , , 2 , 249 meyer , f. , & schmidt , h. u. 1968 , mitteilungen der astronomischen gesellschaft hamburg , 25 , 194 moin , p. , squires , k. , cabot , w. & lee , s. 1991 , phys .
fluids a , 3 , 11 , 2746 montesinos , b. , & thomas , j. h. 1997 , , 390 , 485 rempel , m. , schssler , m. , & knlker , m. 2009 , , 691 , 640 rimmele , t. r. 1994 , , 290 , 972 rimmele , t. r. 1995 , , 298 , 260 scharmer , g. b. , nordlund , .
, & heinemann , t. 2008 , , 677 , l149 schlichenmaier , r. , jahn , k. , & schmidt , h. u. 1998 , , 493 , l121 shine , r. a. , title , a. m. , tarbell , t. d. , smith , k. , frank , z. a. , & scharmer , g. 1994 , , 430 , 413 smagorinsky , j. , 1963 , monthly weather review , 93 , 99 stein , r. f. , & nordlund , .
2002 , solmag 2002 .
proceedings of the magnetic coupling of the solar atmosphere euroconference , 505 , 83 theobald , m. l. , fox , p. a. , sofia , s. 1994 , phys .
plasmas , 1 , 9 tritschler , a. 2009 , structure and dynamics of sunspots .
arxiv e - prints arxiv:0903.1300 walker , m. r. , & barenghi , c. f. 1999 , physics of the earth and planetary interiors , 111 , 35 weiss , n. o. 1991 , geophysical and astrophysical fluid dynamics , 62 , 229 zhang , k. 1999 , physics of the earth and planetary interiors , 111 , 93 , is 1000 g , the inclination angle , @xmath4 , is 85 degrees with respect to the vertical .
red curves show magnetic field lines ; the color 3d structures show the temperature variations relative to the mean vertical stratification ( dark green and blue correspond to lower temperature ) ; the arrows show the velocity field .
the horizontal size of the box is 6.4 mm , the depth is about 2.5 mm . ] , in the direction of the field inclination , and arrows show the horizontal velocity at the solar surface for different initial magnetic field strength , @xmath3 : a ) 1000 g , b ) 1200 g , and the inclination angle , @xmath15 . ]
|
discovered in 1909 the evershed effect represents strong mass outflows in sunspot penumbra , where the magnetic field of sunspots is filamentary and almost horizontal . these flows play important role in sunspots and have been studied in detail using large ground - based and space telescopes , but the basic understanding of its mechanism is still missing .
we present results of realistic numerical simulations of the sun s subsurface dynamics , and argue that the key mechanism of this effect is in non - linear magnetoconvection that has properties of traveling waves in the presence of strong , highly inclined magnetic field .
the simulations reproduce many observed features of the evershed effect , including the high - speed evershed clouds " , the filamentary structure of the flows , and the non - stationary quasi - periodic behavior .
the results provide a synergy of previous theoretical models and lead to an interesting prediction of a large - scale organization of the outflows .
|
the experimental success of quantum electrodynamics ( qed ) lies in the domain of small fields where observations impressively match the theoretical calculations based on perturbation theory in the fine structure constant @xmath9 . in calculations involving bound states of a nucleus of charge @xmath0 the fine structure @xmath10
constant additionally appears in the combination @xmath11 . even though @xmath10 is small
, @xmath11 may not be , so that perturbative analysis can fail when @xmath12 .
one of the most profound physical effects predicted to take place in this regime is the instability of the ground state ( the vacuum ) against creation of electron - positron pairs , resulting in a screening space charge of electrons with positrons leaving physical picture @xcite .
experimental study of this effect has not been possible , as stable nuclei with @xmath13 have not been created , and attempts to look for positron production in a temporarily created overcritical system of slowly colliding uranium nuclei have not been successful @xcite .
the goal of this paper is to demonstrate that in the condensed matter setting the corresponding problems are the impurity states in narrow - band gap semiconductors ( nbgs ) @xcite and weyl semimetals ( ws ) @xcite .
the advantage of these systems is that the charges and fields required to see the ground state instability are modest and readily achievable . some of the effects may have already been seen @xcite without fully appreciating what they represent .
the outline of this paper is as follows : in section ii the phenomenon of critical charge is first explained heuristically ( iia ) followed by more precise semiclassical argument ( iib ) that relates the effect to that of quantum - mechanical fall to the center @xcite .
then the critical charge problem for the coulomb potential modified at small distances is analyzed via dimensional analysis ( iic ) which in section iii is employed to demonstrate the feasibility of observation of its condensed matter analog , instability with respect to creation of electron - hole pairs in semiconductors . in section
iv thomas - fermi ( tf ) theory of screening by space charge is derived and its deficiencies and a modification are discussed .
one of the by - products of the analysis is the conclusion that the observable charge of an arbitrary overcritical source in the ws case is always given by the reciprocal of the inverse fine structure constant for the material . in sections v and vi the tf theory
is solved in two steps - uniformly charged half - space @xmath14 spherically symmetric charge distribution - so that the existence of several physically different regimes of screening can be appreciated , and to establish a relationship with previous analysis @xcite .
the tf analysis leads to the conclusion that there is total screening in the ws case , through an argument that is parallel to landau s `` zero charge effect '' in qed @xcite .
this would have readily observable consequences .
however , we will argue in section vii that this claim of complete screening is not right , so that there after all can be an observable charge .
the dirac equation for an electron in vacuum in the field of a point charge @xmath0 ( the dirac - kepler problem ) becomes invalid for @xmath15 @xcite .
this feature can be heuristically understood by starting with classical expression for the energy of an electron of mass @xmath16 and momentum @xmath17 in the field of charge @xmath0 @xmath18 and trying to estimate the ground - state energy . since the electron position can not be determined to better than @xmath19 divided by the uncertainty of momentum , @xmath17 and @xmath20 entering eq.([cl_energy ] )
may be regarded as the typical momentum and size of the quantum state , respectively .
then the state energy can be estimated as @xmath21 where @xmath22 measures the nuclear charge in units of the reciprocal of the fine structure constant @xmath23 ; both these `` natural '' units ( lower case letters ) and the usual units for charge ( upper case letters ) will be used throughout this paper . minimizing with respect to the free parameter @xmath17 we find @xmath24 where @xmath25 is the electron compton wavelength that sets
the scale for the uncertainty of measurement of the electron position and @xmath26 is the classical electron radius .
the lowest ( ground - state ) energy is then @xmath27 while reproducing the well - known ground - state properties of a hydrogen - like atom in the non - relativistic @xmath28 limit ( [ gs_properties ] ) , as well as ( coincidentally ) matching the exact expression for the ground - state energy ( [ gs_energy ] ) based on the analysis of the dirac equation @xcite , these arguments also predict that the minimum of ( [ energy_vs_p ] ) only exists for @xmath29 ( @xmath30 ) .
as @xmath31 , the ground state becomes sharply localized ( @xmath32 ) , the typical electron momentum diverges ( @xmath33 ) , and the ground - state energy vanishes ( @xmath34 ) .
the conclusions ( [ gs_properties])-([gs_energy ] ) become meaningless for @xmath35 ; specifically the ground state energy is predicted to become imaginary .
the counterintuitive independence of @xmath36 of the electron mass @xmath16 can be explained via dimensional analysis : the dirac - kepler problem is fully specified by the dimensionless parameter @xmath22 , and by the electron compton wavelength @xmath25 in ( [ gs_properties ] ) .
if there exists a critical value of the charge @xmath36 , it can only be a function of the remaining independent dimensionless parameters of the problem .
however having only one length scale @xmath25 available makes it impossible to use it in a dimensionless combination .
therefore @xmath36 can not depend on @xmath25 , and thus on the electron mass @xmath16 ; the only possible outcome is @xmath37 .
these observations imply that the @xmath38 anomaly of the dirac - kepler problem persists in the weyl - kepler problem ( @xmath39 ) , where the estimate ( [ energy_vs_p ] ) becomes @xmath40 as a result , a charged weyl fermion placed in the field of a point charge with @xmath29 is always delocalized ( @xmath41 , @xmath42 , and @xmath43 ) ; the spectrum is not discrete ( i.e. no bound states ) . on the other hand , a sufficiently attractive charge
@xmath38 leads to a sharply localized ground state ( @xmath44 , @xmath45 , and @xmath46 ) .
the @xmath38 instability in the dirac - kepler problem can be identified with a strong field limit of the schwinger effect @xcite : the creation of electron - positron pairs in vacuum in a uniform electric field .
the phenomenon is characterized by the schwinger typical electric field @xmath47 for which the work to separate the constituents of the electron - positron pair over the length scale of the compton wavelength is equal to the rest energy of the pair : @xmath48 : @xmath49 for @xmath50 the pairs are created by tunneling with the vacuum being in a metastable state while for @xmath51 the vacuum is absolutely unstable with respect to pair creation . for the coulomb problem the instability sets in when the electric field of the nucleus a compton wavelength away from its center , @xmath52 , reaches the order of magnitude of the schwinger field ( [ schwinger ] ) , thus predicting @xmath37 . in view of its mass independence
, the prediction @xmath53 also applies to the weyl - kepler problem .
the @xmath54 anomaly of the dirac equation is related to the `` fall to the center '' effect of quantum mechanics @xcite . for a classical electron of energy @xmath55 and angular momentum
@xmath56 moving in a central field @xmath57 the equation of conservation of energy can be written as @xmath58 where @xmath59 is the square of total momentum and @xmath60 is the radial component of the momentum .
the particle can reach the origin ( `` fall to the center '' ) if @xcite @xmath61 specifically for zero angular momentum @xmath56 the fall to the center occurs for any attractive potential decreasing faster than @xmath62 as @xmath63 . in quantum mechanics
@xmath64 has to be replaced by the eigenvalues of the square of the angular momentum operator which will be chosen in the semi - classical langer form @xmath65 @xcite so that both the effects of zero - point motion ( @xmath66 ) and angular momentum ( @xmath67 ) are accounted for .
the smallest value of @xmath64 is @xmath68 which implies that the fall to the center can occur for @xmath62 potentials that are more attractive than the critical potential satisfying @xcite @xmath69 when this condition is met there is no lower bound on the spectrum . for relativistic classical particle moving in a central field ,
energy and momentum are related by @xmath70 . for bound states we have @xmath71 , where beyond the lower limit the system is unstable against pair creation . at the lower limit
the range of motion can be found by solving for the radial momentum to obtain a form analogous to eq.([nr_energy_conservation ] ) : @xmath72 if @xmath57 is diverging as @xmath63 , the @xmath73 term dominates , and for attractive @xmath57 the origin can be reached if @xcite @xmath74 classically the fall to the center for the @xmath75 state is possible for a potential that at @xmath63 is more attractive than a @xmath76 potential .
the quantum case is different : substituting in ( [ r_confition_to_fall ] ) minimal @xmath77 we infer that the fall to the center is possible for @xmath76 potentials more attractive than the critical potential @xmath78 comparing with the coulomb field @xmath79 , we deduce @xmath80 which is the correct critical charge of the kepler problem for a spinless particle @xcite .
it is half the value found for the dirac particle ( the `` missing '' half is due to the electron spin ) .
the dirac case can not be fully understood semi - classically but an insight can be gained by observing that the relativistic case with @xmath81 is equivalent to a non - relativistic problem with the effective potential ( compare eqs.([nr_energy_conservation ] ) and ( [ r_energy_conservation ] ) ) @xmath82 and zero total energy .
we now see that for the kepler problem , @xmath83 , the particle is _ repelled _ at large distances .
the same effective potential is obtained from the klein - gordon equation transformed into the schrdinger form @xcite .
the dirac equation can be also brought into a schrdinger form with an effective potential at @xmath81 resembling eq.([eff_potential ] ) but also exhibiting extra terms attributed to the electron spin @xcite .
their role can be ( approximately ) summarized in a form similar to eq.([eff_potential ] ) with different amplitude of the @xmath62 term .
specifically for the dirac - kepler problem we have @xcite @xmath84 we see that the fall to the center occurs for @xmath54 and then the particle is confined to the central region of radius @xmath85 as is the case of the fall to the center problem , proper treatment of the instability of the dirac equation requires accounting for finite radius @xmath86 of atomic nucleus that modifies the @xmath76 attraction at short distances and removes the difficulty for @xmath38 @xcite . for a nucleus with @xmath87 ( @xmath88 )
the ground - state energy reaches the boundary of the lower continuum , @xmath89 @xcite . past that point
the total energy of the production of an electron - positron pair becomes negative and the vacuum becomes unstable with respect to pair creation ; the positron repelled by the nucleus escapes to infinity while the electron remains near the nucleus @xcite .
many of the conclusions of previous analysis of the critical charge problem that accounted for finite radius @xmath86 of atomic nucleus @xcite can be reproduced by a combination of dimensional analysis and simple physical arguments .
indeed now we have a problem fully specified by independent dimensionless combinations @xmath3 , @xmath10 , and @xmath90 .
then if there exists a critical value @xmath4 , it can only be a function of @xmath10 and @xmath90 .
the electrostatic potential inside the nucleus has the form : @xmath91 where @xmath92 and @xmath93 is finite .
then the parameters @xmath3 and @xmath10 appear together in the @xmath94 combination .
therefore @xmath95 or @xmath96 where @xmath97 is a function that depends on the shape of the charge distribution within the nucleus . the properties of @xmath97 can be inferred from the following arguments : \(i ) for @xmath98 one has @xmath99 which implies the small argument behavior @xmath100 .
this additionally means that @xmath99 for @xmath39 for arbitrary @xmath86 ( the weyl - kepler problem with cutoff ) .
\(ii ) in the classical @xmath101 limit the planck s constant must drop out of eq.([dim_analysis ] ) .
this translates into the large argument behavior @xmath102 with the consequence @xmath103 .
this is indeed what is expected on physical grounds : the vacuum becomes unstable when the electron potential energy at the center of the nucleus @xmath104 reaches the boundary of the lower energy continuum @xmath105 .
this argument applied to the uniformly charged ball model of the nucleus predicts @xmath106 . on the other hand , @xmath107 for the constant potential ball model of the nucleus . for ordinary heavy nuclei the nuclear size @xmath86 depends on @xmath3 according to the fermi formula @xmath108 the critical charge @xmath36
is found by simultaneous solution of eqs.([dim_analysis_zeta ] ) and ( [ nucl_matter ] ) for @xmath109 : @xmath110 the electron compton wavelength is known to be much larger than the nuclear size which means the argument of the function @xmath111 in eq.([fermi_model_electron ] ) is much smaller than unity . in this limit
the distinction between different models of nuclear charge distribution disappears thus explaining the nearly model - independent value of @xmath36 close to @xmath112 . to add credibility to dimensional analysis we now show that the latter easily solves the problem of instability of the muon vacuum .
indeed , the muon compton wavelength has the same order of magnitude as the nuclear size since the muon is more than @xmath113 times heavier than the electron .
the solution to the problem is then described by eq.([dim_analysis ] ) with the electron mass @xmath16 replaced by the muon mass @xmath114 .
we are now in the large argument limit , @xmath102 , which determines the critical @xmath3 for the muon to be @xmath115 these conclusions agree with existing analysis of the problem @xcite .
in fact we observe that approximating the true @xmath97 dependence for _ all _ @xmath116 by its @xmath117 limit @xcite @xmath118 suffices to quickly estimate the critical charge in the practically relevant case of the uniform density model of the nucleus . in the @xmath119 limit , where this approximation is expected to work poorly
, the combination of eqs.([fermi_model_electron ] ) and ( [ choice_of_f ] ) predicts @xmath120 which is close to the accepted value of @xmath121 .
inspection of previous results @xcite shows that , except for a narrow vicinity of @xmath122 , the function @xmath97 is basically a straight line of the right slope with a larger - than - unity offset .
the significance of the linear approximation ( [ choice_of_f ] ) is that it allows us to reliably estimate the critical charge in the condensed matter setting .
qed s predictions of screening by space charge can be tested in performable experiments involving condensed matter systems , both presently available and those that will become available in the near future .
our primary example is that of nbgs whose physics is known to mimic qed @xcite .
excitation of an electron - hole pair is analogous to the creation of an electron - positron pair in qed , with the band gap representing the combined rest energy of the particles .
creation of the electron - hole pairs in the presence of a uniform electric field takes place via zener tunneling @xcite which is analogous to the schwinger effect @xcite .
our contention is that moderately charged impurity regions in semiconductors can trigger a space charge around them that parallels the effects that would occur in qed for unrealistically large @xmath123 .
the idea that the @xmath124 anomaly of the original dirac - kepler problem may have observable condensed matter implications is due to keldysh @xcite . in his study of the impurity states in semiconductors
keldysh noted that the effective mass approximation @xcite , while successful in describing shallow impurity states , fails to explain deep states whose binding energy is comparable with the band gap .
such states are formed near multi - charged impurity centers , vacancies etc . and they can not be associated with either conduction or valence bands .
the experiment presented another puzzle : some highly charged impurities acted as very efficient recombination centers that managed to trap both electrons and holes but an explanation why that was the case was lacking .
keldysh argued that experimental findings can be explained in a two - band approximation ( well obeyed in nbgs of the @xmath125 type ) where the low - energy electron ( hole ) dispersion law is relativistic @xcite @xmath126 here the upper and lower signs correspond to the conduction and valence bands , respectively , @xmath127 is the energy band gap that parallels twice the rest energy of a particle of mass @xmath128 , and @xmath129 is the velocity of a high - momentum particle analogous to the speed of light @xmath130 . compared to their vacuum electron - positron counterparts , electrons and holes in nbgs have two orders of magnitude smaller effective mass ( @xmath131 ) and limiting velocity @xmath129 nearly three orders of magnitude smaller than the speed of light ( @xmath132 ) . as a result
their band gap @xmath133 is seven orders of magnitude smaller than the rest energy of the electron - positron pair @xcite . due to these parameter values
the analog of large field qed effects are readily realizable in nbgs . with this in mind , the determination of the impurity states reduces to solving the dirac equation for a particle of mass @xmath128 in the field of a charge @xmath0 screened by the dielectric constant @xmath134 of the semiconductor . in view of the peculiarity of the dirac - kepler problem ( now @xmath135 ) , keldysh argued that for @xmath136 the impurity states are given by the known solution to the dirac equation @xcite while the @xmath38 case with `` collapsed '' ground state describes a recombination center .
an expression for the nbgs critical charge can be written in a form that parallels eq.([dim_analysis_zeta ] ) : @xmath137 where now @xmath86 is the radius of the impurity region , @xmath138 is the semiconductor analog of the electron compton wavelength and @xmath139 is the semiconductor counterpart of the classical electron radius ( defined as band electron s delocalization size at which its potential self - energy @xmath140 matches its rest energy @xmath141 .
we note that since both @xmath142 and @xmath143 significantly exceed the lattice spacing , macroscopic theory of impurity states ignoring the lattice structure of the material suffices .
the density of nuclear matter is known to have the order of magnitude set by the classical electron radius @xmath144 ( see eq.([nucl_matter ] ) ) .
it will be made clear shortly , that the large field effects in nbgs become prominent at impurity charge densities set by the nbgs electron radius @xmath145 .
therefore the relationship between the radius @xmath86 and the charge @xmath3 of a uniformly charged region will be chosen as @xmath146 that parallels its nuclear physics counterpart ( [ nucl_matter ] ) ; the numerical factor corresponds to the charge density @xmath147 to be justified below .
the nbgs critical charge can be determined by solving the equation @xmath148 that is nearly identical to its qed counterpart ( [ fermi_model_electron ] ) . because the value of the limiting velocity @xmath129 is known
, the semiconductor equivalent of the fine structure constant is @xmath149 , an order of magnitude larger than its qed counterpart ( we employed @xmath150 @xcite ) . with this value of @xmath10 and choosing the function @xmath97 in the simple form ( [ choice_of_f ] ) , the solution to eq.([sc_unif_den_model ] ) is @xmath151 which implies @xmath152 .
the corresponding critical cluster size is @xmath153 according to eq.([sc_matter ] ) . surely @xmath154 impurity clusters with sizes in excess of several nanometers are more common objects than @xmath123 nuclei .
in addition to making it possible to study the regime of large effective fine structure constant , condensed matter systems also offer possibilities that can not be realized in qed .
indeed , over forty years ago abrikosov and beneslavski @xcite predicted the existence of ws having points in the brillouin zone where the valence and conduction bands meet with a dispersion law that is linear in the momentum .
this is the @xmath155 case of eq.([dispersion ] ) .
the low - energy excitations in ws ( realizing massless versions of qed ) are described by the weyl equation .
we already know that the critical charge for the weyl - kepler problem is @xmath99 .
eq.([m=0_energy ] ) additionally implies lack of the discrete spectrum for @xmath156 ; for @xmath38 , a space charge of weyl fermions is present in the ground state .
such substances are likely to be realized in doped silver chalcogenides @xmath157 and @xmath158 @xcite , pyrochlore iridates @xmath159 ( where @xmath160 is yttrium or a lanthanide ) @xcite , and in topological insulator multilayer structures @xcite .
the zero energy gap of ws implies that in a _ uniform _ electric field the creation of a space charge of weyl fermions is spontaneous .
the dielectric constant of ws is of order @xmath161 with @xmath162 @xcite , thus leading , like in the nbgs case , to @xmath163 independent of the size of the impurity region .
nbgs and ws are condensed matter systems where analogs of the atomic collapse of qed can be experimentally detected .
related phenomena can be also observed in graphene .
indeed , graphene possesses the linear dispersion law analogous to that of ws and microscopic parameters similar to nbgs which leads to a small value for the critical charge for promotion of electrons from the valence band to the conduction band .
such a problem has been considered elsewhere @xcite and experimental signatures of the `` atomic collapse '' in graphene were recently reported @xcite .
the graphene problem is mathematically different from what we discuss , because graphene is a two - dimensional semimetal embedded in a three - dimensional space . below we will determine the ground - state properties of nbgs and ws in the presence of a finite - size positive coulomb impurity ( a negative charge leads to the same discussion due to particle - hole symmetry ) .
the arguments given above imply that at modest @xmath3 electrons are promoted from the valence band to form a space charge around the impurity while the holes leave the physical picture ; the properties of the space charge vary with @xmath3 and @xmath10 and are determined by the interplay of attraction to the impurity ( promoting the creation of electron - hole pairs ) , and electron - electron repulsion combined with the pauli principle ( limiting the creation of the space charge ) .
the qed analysis of the physical properties of the space charge was carried out in two limits : \(i ) @xmath3 close to @xmath4 , where there are very few electrons promoted to the conduction band for which the single - particle picture is a good starting point @xcite ; and \(ii ) @xmath164 , where the number of screening electrons is large and the electron - electron interactions can not be ignored @xcite .
below we demonstrate that the physics in the @xmath164 limit exhibits a large degree of universality .
although we are mostly concerned with the nbgs setting , our findings are equally applicable in qed as both problems share the same mathematics ; a solution to the ws problem benefits the understanding of the nbgs / qed case . to help the readers orient themselves between three physically different manifestations of the problem and to provide them with a condensed matter - qed translation dictionary , in table i we summarized pertinent properties of electrons in vacua of qed , nbgs and ws .
.summary of properties of electrons in vacua of quantum electrodynamics ( qed ) , narrow band gap semiconductors ( nbgs ) and weyl semimetals ( ws ) .
[ cols= " < , < , < , < " , ] [ dictionary ] the entries not yet specified are : \(i ) _ the fermion degeneracy factor @xmath165 _ which is @xmath166 in qed while in nbgs it is twice the number of dirac valleys ( [ dispersion ] ) within the first brillouin zone ; an isotropic valley - independent limiting velocity @xmath129 is assumed for simplicity . in the ws case @xmath165 counts the number of weyl points within the first brillouin zone : @xmath167 in pyrochlore iridates @xcite and @xmath168 in a topological insulator multilayer @xcite .
\(ii ) _ the coupling constant @xmath169 _ plays a role analogous to that of the fine structure constant @xmath10 in polarization effects , as will be made clear below .
\(iii ) _ the zener field @xmath170 _ is the semiconductor analog of the schwinger field ( [ schwinger ] ) defined as @xmath171 comparing the values of the fields @xmath47 and @xmath170 explains why nbgs are so well suited to study strong field qed effects ; the situation is even more favorable in ws where due to the zero band gap , an arbitrarily weak field is `` strong '' as far as the space charge phenomena are concerned .
since for @xmath164 a large number of electrons are in the conduction band , the properties of the system consisting of the impurity and its interacting cloud of electrons can be understood semi - classically with the help of the tf theory @xcite .
the main object of the latter is a physical electrostatic potential @xmath172 felt by an electron that is due to both the electrostatic potential of the impurity @xmath173 and that of the space charge characterized by the number density @xmath174 : @xmath175 the external potential @xmath173 is a pseudopotential that represents the perturbation of the system caused by the impurity ; even though @xmath176 is not entirely of electrostatic origin , we will define @xmath177 .
we assume that the impurity charge density @xmath178 is spherically - symmetric and localized within a mesoscopic region of size @xmath86 so that for @xmath179 the potential @xmath173 reduces to a purely coulomb form @xmath180 of a net charge @xmath0 .
there are several reasons why the impurity region has to be mesoscopic in size : first of all , in practice charged atomic scale defects can not have @xmath181 .
second , a large charge localized within a small region implies a large electrostatic potential .
however all our analysis is based on approximating the exact dispersion law by its low - energy limit ( [ dispersion ] ) . for that to remain valid the order of magnitude of the potential within the impurity region
should not exceed a volt .
like in graphene , this corresponds to the electronvolt energy scale which is significantly smaller than the width of the conduction band .
finally the conditions of semi - classical description inherent within the tf theory must be met .
all these constraints along with the requirement @xmath182 can be satisfied in @xmath183 impurity clusters .
promotion of electrons to the conduction band also takes place in smaller ( down to @xmath184 ) regions but the number of these electrons may not be large enough for the predictions of the tf theory to be quantitatively reliable .
given @xmath172 , one can deduce that the electron number density @xmath174 is different from zero only in the region of space where the electron potential energy @xmath185 drops below @xmath186 , thus defining the edge of the space charge region as @xmath187 the radius of the space charge region @xmath188 is given by the equalities @xmath189 , @xmath190 ; outside the region we have @xmath191 and @xmath192 where @xmath193 is the observable charge as seen at large distances from the source center .
continuity of the potential @xmath194 across the shell boundary relates @xmath195 and @xmath2 while the nbgs electron radius @xmath145 sets the length scale as indicated in the last two steps in ( [ observable_charge_definition ] ) meaning that we can speak of the shell size or the observable charge interchangeably .
we note that in the ws case , @xmath196 , eq.([observable_charge_definition ] ) predicts @xmath197 , i.e. the electron shell extends all the way to infinity . in natural units of charge the relationship between the observable charge @xmath198 and the radius of the electron shell
is given by @xmath199 from the thermodynamical standpoint , creation of electron ( @xmath200)-hole ( @xmath201 ) pairs by the field of a coulomb impurity accompanied by escape of a hole to infinity may be viewed as a `` chemical reaction '' @xmath202 ( the ground state of the semiconductor is the `` vacuum '' ) @xcite ; the condition of equilibrium for this reaction has the form @xmath203 where @xmath204 and @xmath205 are the chemical potentials of the electrons and holes , respectively , and @xmath206 is the fermi momentum which we assume is a slowly varying function of position @xmath207 . the condition of equilibrium ( [ chem_potentials ] ) together with eq.([fermi_momentum ] ) implies a relationship between the physical potential and the number density of the space charge @xcite : @xmath208\bigg\}^{3/2}\ ] ] where @xmath209 is the coupling constant characterizing the relative strength of electron - electron interactions and zero - point motion .
the four orders of magnitude disparity between its condensed matter and qed values ( see table i ) is yet another indication that the space charge phenomenon is more relevant to semiconductors than to qed .
since the electron chemical potential in ( [ chem_potentials ] ) is set at the boundary of the lower continuum , in the nbgs/ qed cases the screening of the external charge is incomplete ; only in the ws ( @xmath210 ) case do we have complete screening .
the latter statement can be rigorously proven by setting @xmath210 in ( [ chem_potentials ] ) and combining the outcome with eqs.([definition_potential ] ) and ( [ fermi_momentum ] ) : @xmath211 taking in eq.([weyl_int_eq ] ) the @xmath212 limit gives a relationship @xmath213 whose consequences are that the electron number density @xmath174 must decay faster than @xmath214 at @xmath215 large and that @xmath216 i.e. according to tf theory , a ws succeeds in giving complete screening of the impurity charge . applying the laplacian operator to both sides of eq.([definition_potential ] ) and using ( [ n_of_phi ] )
we find the relativistic tf equation @xmath217 that was investigated in qed for the case of localized source term @xcite .
applicability of the zero - temperature tf theory to experiments to be conducted at finite temperature requires further justification .
since the @xmath218 energy gap of nbgs significantly exceeds the room temperature scale of @xmath219 , the zero temperature theory adequately describe room temperature experiments .
however ws have a zero energy gap and in equilibrium the conduction and valence bands are populated with electrons and holes , respectively .
this effect can be also neglected as we are working at potential close to @xmath220 significantly exceeding the room - temperature energy scale . in order to establish the range of applicability of the tf theory we note that the observable charge @xmath221 is the critical charge of the single - particle problem for a charged region of scale @xmath195 which is due to both the external charge and that of the space charge . then replacing @xmath222 in eq.([sc_dim_analysis ] )
we arrive at the definition @xmath223 that is consistent with the tf result ( [ qed_observable_charge_definition ] ) only in the classical limit @xmath224 ( recall that @xmath225 , section ii ) .
however ws are characterized by @xmath226 and so the semiclassical condition can never be met .
the consequence is that the prediction of complete screening in the ws case , @xmath227 , an exact consequence of the tf theory , is an artifact . in reality
the ws will screen an overcritical impurity charge by space charge only until the point at which the single - particle description is restored .
this sets a limit on the applicability of the tf theory to the ws case at large distances from the source center and implies that the space charge region has a finite radius to be estimated below . since the weyl - kepler problem with @xmath29
does not have a discrete spectrum , coulomb impurities in ws can never be fully neutralized .
their observable charge @xmath221 is either @xmath112 ( and then there is a space charge of weyl electrons ) or @xmath29 ( when there is no space charge ) .
more generally eq.([exact_definition_obs_charge ] ) implies that in the nbgs / qed setting the screening is never so great ( in the case @xmath228 ) that the observable charge @xmath221 is less than unity . in the point charge limit
@xmath229 the space charge region must also shrink to a point ( @xmath230 ) , and then eq.([exact_definition_obs_charge ] ) predicts @xmath231 i.e. disallowance for a point charge to have observable charge exceeding unity @xcite .
since the dirac - kepler problem always has bound states , coulomb impurities in nbgs can be neutral or ionized with the outer electron shells partially or fully filled with ( @xmath232 ) or without ( @xmath233 ) the space charge being present .
here we only consider the problem of an overcritical @xmath232 ion with all outer shells empty . to summarize
, the condition of applicability of the tf theory can be stated in two equivalent forms , @xmath234 or @xmath235 .
ultimately , the tf treatment of the space charge in the presence of the supercritical source @xmath236 is applicable because the fine structure constant is significantly smaller than unity . in order to see what physics is missing from the tf theory ,
we observe that the relationship between the observable charge and the radius of the electron shell ( [ qed_observable_charge_definition ] ) resembles the @xmath237 limit of the semi - classical expression ( [ localization_scale ] ) for the localization scale of an electron ( replacing @xmath238 and @xmath239 ) .
the latter is sensitive to the fall to the center occurring at @xmath240 while the tf result ( [ qed_observable_charge_definition ] ) is not .
the same can be seen more generally by solving the main equation of the tf theory ( [ chem_potentials ] ) relative to @xmath241 : @xmath242 with the identifications @xmath243 , @xmath244 , @xmath245 , this is the energy relationship eq.([r_energy_conservation ] ) , but the term involving the angular momentum is missing . we propose including this into the right hand side of eq.([diff_eq_phi ] ) : @xmath246 at the boundary of the space charge region where the potential is given by eq.([observable_charge_definition ] ) the expression in the second line vanishes leading to a relationship between observable charge @xmath221 and the radius of the space charge region @xmath195 @xmath247 generalizing the tf result ( [ qed_observable_charge_definition ] ) and correctly capturing the limiting cases of @xmath224 and @xmath248 .
this is equivalent to choosing the function @xmath111 in ( [ exact_definition_obs_charge ] ) in the @xmath249 form .
assessing the status of our modification of the tf theory ( [ modified_tf ] ) requires a separate investigation which we postpone until the future . at the very least what is proposed qualifies as an interpolation . until we learn more about the deficiencies of the tf theory , we focus on its standard version accumulated in eqs.([n_of_phi ] ) and ( [ diff_eq_phi ] ) . some other desirable features of the modified tf equation ( [ modified_tf ] )
are mentioned in section vii .
the phenomenon of screening is a manifestation of the electron - electron interactions quantified by the coupling constant @xmath169 .
its smallness ( see table i ) does not imply that the screening response is necessarily weak .
previous analysis @xcite established that the strength of screening is determined by the dimensionless combination @xmath251 which for @xmath164 can take on an arbitrary value .
below we additionally show that in the nbgs / qed cases the regime of strong screening further subdivides into that of super - strong screening @xmath252 ( that is hardly accessible in practice ) and the regime of moderately strong screening @xmath253 which is treated in detail .
since the source region is mesoscopic in size we find it useful to start with the problem of screening of a uniformly charged half - space ( `` half - infinite '' nuclear matter in the qed setting ) by space charge .
this requires solution of the one - dimensional version of eq.([diff_eq_phi ] ) : @xmath254 where @xmath255 while @xmath256 .
the solution to eq.([1d_diff_eq ] ) within the source region far away from the boundary @xmath257 corresponds to the state of local neutrality @xmath258 @xcite : @xmath259 the quantity @xmath260 is the work function , i.e. the energy needed to remove the electron from the source region . at this point
we observe that in the qed ( @xmath261 , @xmath262 , @xmath263 ) and ws ( @xmath196 ) versions of the problem , eq.([phi_-infty ] ) simplifies to @xmath264 in the condensed matter setting our theory is applicable provided the potential @xmath265 does not exceed a volt .
then for @xmath266 and @xmath267 the maximal external charge density within the impurity region can be estimated as @xmath268 , two orders of magnitude smaller than the free - electron density in normal metals .
this corresponds to @xmath269 and justifies eq.([sc_matter ] ) . with these parameter values
the approximation ( [ phi_-infty_m=0 ] ) also holds in nbgs .
given the charge concentration of @xmath147 , a @xmath270 impurity region would contain a bare charge of about @xmath271 which is significantly larger than @xmath152 .
yet larger values of @xmath164 can be obtained by choosing @xmath272 : a @xmath273 region will host an external charge of about @xmath274 .
the source boundary represents a perturbation to the constant @xmath275 ; assuming the effect is weak we substitute @xmath276 , @xmath277 , into eq.([diff_eq_phi ] ) and linearize about @xmath278 : @xmath279 where the length scale @xmath280 parallels the debye length of the tf theory of screening in a fermi gas @xcite ; @xmath281 is the scale over which the potential @xmath194 recovers to @xmath282 when disturbed by an inhomogeneity .
the last estimate in ( [ tf_length ] ) is only applicable to the nbgs or qed ( @xmath261 ) cases . in the condensed matter
setting with @xmath283 the tf screening length is of the order several nanometers . in qed @xmath281
is an order of magnitude larger than the classical electron radius .
applicability of the concept of the screening length to a finite size system is limited by the constraint @xmath284 which is a statement of strong screening @xmath250 @xcite .
the crossover in the screening response occurs at a charge @xmath285 in condensed matter applications we find @xmath286 both the regimes of weak @xmath287 and strong @xmath288 screening are experimentally accessible .
in qed we have @xmath289 which is only of academic interest . assuming the potential at the impurity boundary is not significantly smaller than @xmath265
, the @xmath290 dependence can be inferred from the linearized form ( [ 1d_linearized_tf ] ) , which gives @xmath291 : while local neutrality holds far away from the boundary , it is violated in a boundary layer whose size has the order of magnitude of the tf screening length @xmath281 .
outside the impurity region @xmath292 the potential will continue to decrease from its value at the boundary @xmath293 until it reaches the edge of the space charge region defined as @xmath294 .
the length scale @xmath295 has a meaning of the thickness of the layer of space charge outside the impurity region .
we thus see that a layer of net positive charge of thickness @xmath281 localized next to the boundary is followed by a layer of negative charge of thickness @xmath295 outside the source region @xcite
. the net charge of this double layer is positive thus implying that the electric field for @xmath296 is finite and uniform .
even though eq.([1d_diff_eq ] ) can be integrated in quadratures , an approximate solution is more illuminating . within the source region
the potential is approximated by the solution to eq.([1d_linearized_tf ] ) finite at @xmath297 : @xmath298 where it is assumed ( and later justified ) that @xmath299 . outside the impurity region
@xmath292 the full non - linear equation ( [ 1d_diff_eq ] ) becomes @xmath300 in the ws case or when @xmath301 , eq.([full_tf_x>0 ] ) simplifies to @xmath302 an analytic solution to the problem of screening of a supercharged nucleus in the strong screening limit @xmath250 that approximates the finite nucleus by half - infinite nuclear matter and relies on eqs.([1d_linearized_tf ] ) and ( [ full_tf_x>0_weyl ] ) was proposed by migdal , voskresenski , and popov ( mvp ) @xcite .
for @xmath292 the solution to eq.([full_tf_x>0_weyl ] ) satisfying the conditions of zero electric field and zero potential at @xmath303 has the form @xmath304 continuity of the potential and of the electric field at the source boundary @xmath305 determines the integration constants @xmath306 and @xmath307 in eqs.([phi_x<0 ] ) and ( [ phi_x>0 ] ) to be @xmath308 the length scale @xmath307 naturally has the order of magnitude of the tf screening length @xmath281 . the profile of the electron number density for @xmath292 is implied by eqs.([n_of_phi ] ) and ( [ phi_x>0 ] ) : @xmath309 in the @xmath310 limit the mvp solution ( [ phi_x>0 ] ) and ( [ 1d_density ] ) exhibits universality , i.e. it becomes independent of the parameters of the source region . the mvp solution eqs.([phi_x<0])-([1d_density ] ) with @xmath311 is partly relevant to the problem of screening response of nbgs or ws to the spherically symmetric charge distribution of radius @xmath86 . specifically , eq.([phi_x<0 ] ) adequately solves the problem within the impurity region in the strong - screening regime @xmath250 . for example , the net charge within the source region can be estimated as @xcite @xmath312 this is significantly smaller than the bare charge @xmath3 thus illustrating substantial screening of the source region . on the other hand , the density profile eq.([1d_density ] ) with @xmath311
integrates to an infinite charge in three dimensions . therefore outside a spherically symmetric charge distribution eqs.([phi_x>0 ] ) and ( [ 1d_density ] ) are only applicable as long as approximating the spherical surface by a plane is valid , i. e. for @xmath313 . in order to go beyond the limitation of the mvp approximation outside the source region we need to solve the full three - dimensional equation ( [ diff_eq_phi ] ) .
outside of the source in the ws ( @xmath210 ) case one has to look at the full non - linear equation ( [ diff_eq_phi ] ) whose radially - symmetric solution is sought in the form @xmath314 where , via gauss s theorem , the function @xmath315 is related to the charge @xmath316 within a sphere of radius @xmath215 as : @xmath317 substituting ( [ def_chi ] ) into ( [ diff_eq_phi ] ) for @xmath318 and @xmath210 we obtain the equation @xmath319 for @xmath320 we can neglect here the first - order derivative term @xmath321 compared to @xmath322 ; then @xmath323 .
the solution to ( [ diff_eq_chi_of_x ] ) in this limit is the mvp result ( [ phi_x>0 ] ) in disguise @xmath324 in the strong - screening limit @xmath250 the parameter @xmath325 drops out of eq.([chi_near_boundary ] ) , and the solution to the full eq.([diff_eq_chi_of_x ] ) has the form @xmath326 where @xmath327 is a parameter free universal function satisfying singular boundary condition @xmath328 .
the latter behavior is no longer an accurate representation of the true dependence @xmath327 past @xmath329 .
therefore the solution ( [ chi_near_boundary ] ) is only applicable up to a crossover scale @xmath330 , i.e. within several source radii as was already observed earlier . within this range the rescaled potential @xmath331 drops from a value of the order @xmath332 at the source boundary to @xmath333 at the crossover scale @xmath334 , and the charge within a sphere of radius @xmath335 drops from the value given by eq.([q(r < a ) ] ) at the source boundary to @xmath336 at the crossover scale @xmath334 . for @xmath337 we can neglect in eq.([diff_eq_chi_of_x ] ) the second - order derivative term @xmath322 compared to @xmath321
; then @xmath338 and for arbitrary screening strength and in natural units of charge eq.([diff_eq_chi_of_x ] ) acquires a form @xmath339 that is mathematically identical to the gell - mann - low ( renormalization - group ) equation @xcite for the physical charge in qed reflecting the effects of vacuum polarization .
eq.([gl ] ) exhibits the landau `` zero charge '' effect @xcite : for any `` initial '' value of @xmath340 the system `` flows '' to the zero charge fixed point @xmath341 as @xmath342 ( @xmath212 ) , i.e. the source charge has been completely screened . alternatively , for @xmath215 fixed complete screening is reached in the point source limit @xmath229 .
zero observable charge is an exact property of the tf theory discussed in section iv . in the strong - screening regime
the last equation is applicable at @xmath343 . as a result
the charge @xmath344 inside a sphere of radius @xmath345 will be given by @xmath346 where the integration constant @xmath347 is the charge within a sphere of radius @xmath348 .
since @xmath349 , the constant can be estimated as @xmath350 thus implying that the observable charge will be accurately given by the last representation in ( [ 0_charge_solution ] ) at distances exceeding several impurity radii .
substituting @xmath351 into eqs.([def_chi ] ) and ( [ n_of_phi ] ) we find corresponding expressions for the potential @xmath352 and the electron density @xmath353^{3/2}}=\frac{1}{16\pi r^{3}}\sqrt{\frac{3\pi}{g\alpha^{3}}}\ln^{-3/2}\left ( \frac{r}{a^{*}}\right ) \ ] ] both valid for @xmath354 .
we note that due to the logarithmic factor the density profile is integrable .
this feature , a reflection of the three - dimensional character of the problem , is missing from the mvp result ( [ 1d_density ] ) .
the hallmark of eqs.([0_charge_solution])-([w_density_strong_screening ] ) is their near universality : a weak logarithmic dependence on the source size @xmath355 with universal amplitudes .
since both @xmath86 and @xmath348 appear within arguments of the logarithm , in what follows for simplification purposes the difference between them will be neglected .
we conclude that for @xmath250 the solution to the screening problem within several impurity radii from the source boundary is universal and given by the mvp results , eqs .
( [ phi_x>0 ] ) and ( [ 1d_density ] ) , that turns nearly - universal , eqs.([0_charge_solution])-([w_density_strong_screening ] ) , at larger distances .
we argued previously ( section iva ) that the complete screening effect is an artifact - the observable charge of an overcritical source region must be always equal to unity ( @xmath23 ) . substituting this value into eq .
( [ 0_charge_solution ] ) provides us with a length scale @xmath356 which is the radius of the space charge region : for @xmath357 the electron density is negligible and the potential is that of unit ( @xmath23 ) charge .
the tf results ( [ 0_charge_solution])-([w_density_strong_screening ] ) are applicable at distances @xmath358 .
the exact magnitude of the exponential is explained in section vii . in qed the exponential factor in ( [ w_cloud_size ] ) is about @xmath359 .
then @xmath360 is the largest length scale of the problem and for all practical purposes tf theory is exact in the @xmath164 regime . in ws with @xmath167 and
@xmath361 the exponential factor in ( [ w_cloud_size ] ) is close to @xmath362 which means that the whole spatial structure of the overcritical weyl ion is experimentally accessible .
this system is particularly interesting because both the tf , @xmath358 , and the non - tf , @xmath363 , regions can be probed . on the other hand
, choosing @xmath168 gives the exponential factor of the order @xmath364 which for a nanometer scale impurity region corresponds to the weyl ion of @xmath365 radius . in the latter case
the tf theory provides practically exact description .
we already learned that at a distance of a few source radii the potential drops to a value of the order @xmath366 .
this corresponds to the energy scale @xmath367 which is much larger than the energy gap @xmath368 if the charge @xmath3 is significantly smaller than the characteristic charge @xmath369 then the analysis just given for ws will also be applicable in the nbgs / qed case with the conclusion that within several impurity radii the solution to the problem continues to be given by the mvp results , eqs.([phi_x<0])-([1d_density ] ) , with @xmath311 .
the characteristic charge ( [ xxover_charge ] ) separates the regime of moderately strong screening @xmath253 to which the results of this subsection apply , from that of super - strong screening @xmath252 . the characteristic charge @xmath370 significantly exceeds the crossover charge @xmath371 , eq.([weak_strong_xover_charge ] ) , separating the regimes of weak and strong screening response .
in nbgs with @xmath372 we find @xmath373 which even by condensed matter standards is very large .
in qed we obtain @xmath374 . in view of unrealistically large value of @xmath370 the analysis of the regime of super - strong screening @xmath375 is not pursued here . at distances exceeding several source radii
we need to look at the equation @xmath376 which generalizes eq.([gl ] ) .
now the charge decreases faster with position than its ws counterpart ; when the right - hand side vanishes , i. e. the edge of the electron shell @xmath377 is reached , the charge acquires its observable value @xmath221 ( see eq.([qed_observable_charge_definition ] ) ) and stops changing thereafter .
the form of the solution can be approximately captured by the ws result ( [ 0_charge_solution ] ) ( that remains relevant at distances @xmath378 ) ; the value of @xmath221 at @xmath195 can be estimated with logarithmic accuracy by equating @xmath379 to @xmath368 .
this is equivalent to terminating the flow equation ( [ gl ] ) at the scale @xmath380 and identifying @xmath381 . then the observable charge
@xmath221 will satisfy the equation @xmath382 whose consequence is that for @xmath86 fixed there exists a nearly universal lower limit on the observable charge @xmath221 . in the point source limit @xmath229 we find @xmath383 , i.e. there is a complete screening of the field of a point charge at an arbitrary distance from it . for a realistic extended source the approximate solution for the charge is @xmath384 where in the last step we specified to the practically important case of @xmath385 .
the size of the space charge region is then given by eq.([qed_observable_charge_definition ] ) .
the electric field at the edge of the space charge region can be estimated as @xmath386 the fact that the field at the shell edge ( [ el_field_shell _ edge ] ) is much smaller than the zener field demonstrates the sharpness of the edge .
the solution ( [ obs_charge_strong_screening ] ) is accurate provided @xmath387 . for nbgs with @xmath388 ( @xmath389 impurity region ) and @xmath390
we find @xmath391 .
this is not really in the regime where eq.([obs_charge_strong_screening ] ) applies , but suffices to estimate the charge ( ignoring the logarithmic factor ) as @xmath392 , and the size of the space charge region @xmath393 . in qed
we find @xmath394 . as the bare charge @xmath3 continues to increase within the @xmath253 range of moderately strong screening , the observable charge ( [ obs_charge_strong_screening ] ) and size of the space charge region @xmath195 remain nearly constant increasing very slowly with @xmath3 .
the size of the source region @xmath385 grows faster with @xmath3 than @xmath195 , at @xmath395 the two meet , and the result ( [ obs_charge_strong_screening ] ) ceases to be applicable . to summarize , in nbgs and qed in the regime of moderately strong screening @xmath396 the expressions for the observable charge ( [ obs_charge_strong_screening ] ) and radius of the space charge region ( [ qed_observable_charge_definition ] ) are nearly - universal .
they are manifestations of the nearly - universal `` zero charge '' behavior ( [ 0_charge_solution ] ) in the ws case .
the physical mechanism by which the zero charge situation is avoided is purely classical : when it is no longer energetically favorable , the creation of further space charge terminates . to put our analysis of the regime of moderately strong screening @xmath253 onto solid footing
we solved the full eq.([diff_eq_phi ] ) numerically .
the results are shown in the figure , where we additionally displayed the charge @xmath316 within a sphere of radius @xmath215 as an indicator of the strength of screening .
( [ definition_potential ] ) and charge @xmath316 within a sphere of radius @xmath215 ( [ charge_connection ] ) as functions of distance ( double logarithmic representation ) .
the source region is @xmath397 .
the gap value @xmath368 is indicated by the horizontal line ; the electron cloud is limited to the region where @xmath398 , which defines @xmath195 .
the curves are drawn for @xmath399 , @xmath400 . for large @xmath3 ,
@xmath195 approaches a @xmath3-independent limit , indicating that @xmath316 tends to an upper bound @xmath2 . ]
the very weak dependence of the observable charge @xmath401 and size @xmath195 of the space charge region on the bare charge @xmath3 has its origin in the @xmath3-dependence of the size of the source region ( [ sc_matter ] ) . for @xmath402 ,
our theory predicts @xmath3-independent limit on @xmath2 and @xmath195 .
therefore in order to single out this effect we chose @xmath402 .
specifically we set @xmath403 ( other values of @xmath404 are equivalent to a rescaling @xmath405 and @xmath406 ) .
several features of the numerical solution illustrating our analysis deserve mentioning : \(i ) as expected the screening effect of the space charge becomes noticeable for @xmath407 .
\(ii ) the crossing of the charge curves for @xmath408 and @xmath409 at @xmath215 small is a direct illustration of screening : the tf screening length @xmath281 ( see eq.([tf_length ] ) for @xmath402 ) is smaller for @xmath409 than @xmath408 , so that the screening at the central region is more complete in the former case .
\(iii ) the large drop of the potential within a few radii of the impurity , clearly seen in the @xmath410 and @xmath411 curves , is an illustration of our observation made in the analysis of the ws problem that within this range the potential must drop by a large factor of @xmath412 .
\(iv ) remarkably , for @xmath413 there exists a @xmath3-independent limit on @xmath195 and @xmath2 whose values agree with our estimates .
the analysis carried out so far relied on the concept of the tf screening length @xmath281 , eq.([tf_length ] ) , which in the weak - screening regime @xmath415 loses its meaning as a length scale characterizing the source region , and one has to start anew . on the other hand ,
weak screening means that electron - electron interactions can ( possibly ) be treated by a perturbation theory .
to lowest order in @xmath416 we set @xmath417 , and then eq.([n_of_phi ] ) gives @xmath418 for @xmath419 we have @xmath420 .
then the density of the space charge inside the source region can be estimated as @xmath421 , implying that the number of electrons residing at @xmath419 , is of the order @xmath422 .
the latter must be much smaller than the bare charge @xmath3 ( to justify the approximation @xmath423 ) thus specifying the condition of weak screening as @xmath415 . outside of the impurity region eq.([density_weak screening ] ) becomes @xmath424 and @xmath191 otherwise .
the total screening charge is then of order @xmath425 , so that @xmath426 ( or @xmath427 ) , consistent with the tf relationship , eq.([qed_observable_charge_definition ] )
. the electric field at the boundary of the space charge region can be estimated as @xmath428 which in view of the condition @xmath237 demonstrates the sharpness of the boundary of the space charge region . since the space charge residing at @xmath419
is small , eq.([outside_n_weak screening ] ) can be used to compute with logarithmic accuracy the net charge @xmath344 within a sphere of radius @xmath318 . in the ws case
when @xmath196 the density of the space charge is given by @xmath429 and we find @xmath430 this expression is applicable provided @xmath431 , i.e. it inevitably fails at sufficiently large distance from the source .
alternatively , the weak screening @xmath415 analysis can be carried out by treating the cubic term of ( [ diff_eq_chi_of_x ] ) perturbatively .
then the lowest order solution outside the source is @xmath432 .
the next order gives for @xmath318 @xmath433 we observe that the expression for charge ( [ charge_connection ] ) ( in natural units ) computed with the help of eq.([chi_weak screening ] ) agrees with eq.([charge_weak_screening ] ) to logarithmic accuracy which we adopt .
then the perturbative expression ( [ chi_weak screening ] ) may be regarded as a charge itself : it tells us that within the cloud , the physical potential @xmath194 and the density of space charge @xmath434 decrease with @xmath215 faster than @xmath76 and @xmath435 , respectively . on the other hand ,
no matter what the strength of screening is , at sufficiently large distances from the source center the charge @xmath344 is given by the asymptotic limit of eq.([0_charge_solution ] ) .
all these results can be summarized in a simple interpolation formula for the charge @xmath436
@xmath437 if the parameter @xmath22 is viewed more broadly as the _ net _ charge within the source region , then this equation with @xmath438 also covers the regime of strong screening @xmath439 ( see eq.([0_charge_solution ] ) ) .
this choice additionally guarantees that the results of section vb2 pertinent to the nbgs / qed case are automatically captured . substituting @xmath440 into eqs.([def_chi ] ) and ( [ n_of_phi ] )
we find corresponding interpolation formulas for the potential @xmath441 and the electron density @xmath442^{3/2}}\ ] ] the logarithmic terms in the denominators of eqs.([interpolation_charge])-([interpolation_density ] ) are relevant at the scales @xmath215 exceeding @xmath443 this quantity is the screening length within the space charge of weyl electrons .
deviations from the coulomb law become substantial for @xmath444 and the asymptotic regimes of eqs.([0_charge_solution])-([w_density_strong_screening ] ) are reached at @xmath445 . specifically , as the strength of screening increases from small to large @xmath251 the screening radius ( [ shell_screening_length ] ) decreases from a very large value to the scale comparable to the source size . since the zero charge effect is an artifact , consistency of the theory requires that the screening length ( [ shell_screening_length ] ) to be significantly shorter than the radius of the electron cloud ( [ w_cloud_size ] ) . since @xmath237 this is indeed true
. the nbgs / qed case will be handled in exactly the same manner as that of the regime of moderately strong screening - by terminating the ws solution at the scale of the space charge region ( [ termination_point ] ) .
then the observable charge follows from eq.([interpolation_charge ] ) as @xmath446 we see that the initial growth of @xmath447 as @xmath22 , in the regime of weak screening @xmath448 , slows down eventually saturating , in the strong screening regime @xmath449 , at @xmath22-independent value implied by eq.([selfcons_obs_charge_eq ] ) . in the regime of weak
screening the solution to eq.([sc_interpolation_charge ] ) one step beyond the zero order @xmath450 $ ] reproduces previous findings @xcite .
the dependence of the observable charge @xmath2 of a supercharged heavy nucleus on the bare charge @xmath3 was evaluated in ref .
@xcite by numerically solving the tf theory discussed in our paper .
the @xmath451-dependence was found to be a monotonically increasing function with growth rate decreasing with @xmath3 ; for @xmath452 the function @xmath451 was found to grow slower than @xmath3 .
the mvp theory @xcite explained the @xmath451 behavior in the regime of weak screening @xmath453 and laid out a foundation to understand the regime of strong screening @xmath250 ; its place in the problem of screening of overcritical external charge was explained earlier .
however the zero - charge type solution of the tf theory in the ws case was missed whose consequences are : \(i ) for @xmath86 fixed the observable charge @xmath2 saturates as @xmath452 at a @xmath3-independent value ; \(ii ) for realistic @xmath385 the @xmath451 dependence is a nearly universal slowly increasing function of @xmath3 , eq.([obs_charge_strong_screening ] ) , which goes beyond explanation of numerical results @xcite .
even though in the nbgs / qed case complete screening does not occur , there are other respects in which the tf theory is internally inconsistent . solving eq.([sc_interpolation_charge ] ) for the bare charge @xmath22 we find @xmath454 while correctly predicting that the latter is always larger than the observable charge , eq.([landau_pole ] ) also tells us that for fixed @xmath221 and @xmath229 the denominator vanishes for finite @xmath86 given by @xmath455 before the limit of point source is reached . at @xmath456
the bare charge is infinite while for @xmath457 it is imaginary ; the latter feature is certainly unacceptable .
these conclusions having their origin in the `` zero charge '' solution in the ws case , like the zero charge effect itself , are artifacts . even if the vanishing of the observable charge did occur , the scale ( [ pole_position ] ) is too small to be of any practical importance .
in section iva we introduced a modified tf equation ( [ modified_tf ] ) with built - in quantum - mechanical fall to the center .
it is straightforward to realize that such a modification removes the zero charge effect in the ws case ; the nbgs / qed problems are also liberated of the difficulty of the vanishing denominator .
specifically , in the ws ( @xmath210 ) case substituting ( [ def_chi ] ) into ( [ modified_tf ] ) we obtain ( instead of ( [ gl ] ) ) the following flow equation @xmath458 now any `` initial '' charge @xmath459 will be carried to the stable fixed point @xmath460 which is reached as @xmath461 .
eq.([w_flow_eq ] ) can be integrated in quadratures for arbitrary initial @xmath22 with the result @xmath462 where the scale @xmath463 is determined by the condition @xmath464 .
the corresponding spatial scale @xmath465 characterizes the size of the space charge cloud .
for @xmath363 the ( overcritical ) impurity charge appears as being poised at the critical value . as @xmath466 ( a weakly overcritical source ) ,
the cloud size ( [ better_w_cloud_size ] ) diverges because the creation of spacial charge is a critical phenomenon .
the kosterlitz - thouless - type essential singularity in ( [ better_w_cloud_size ] ) is typical of localization transitions related to the quantum - mechanical fall to the center @xcite . for the supercritical source
@xmath237 the cloud size becomes eq.([w_cloud_size ] ) .
the point source limit @xmath229 ( @xmath467 ) of the solution ( [ better_w_solution ] ) also describes the nbgs / qed problems with the conclusion that at arbitrary distance @xmath215 from overcritical point source the observable charge of the latter is unity .
this work was supported by us afosr grant no .
fa9550 - 11 - 1 - 0297 . ya .
b. zeldovich and v. s. popov , usp .
nauk * 105 * , 403 ( 1971 ) [ sov .
uspekhi * 14 * , 673 ( 1972 ) ] , and references therein ; v. s. popov , yad . fiz . *
64 * , 421 ( 2001 ) [ phys . at .
nucl . , * 64 * , 367 ( 2001 ) ] , and references therein . l. v. keldysh , zh . eksp
. fiz . * 45 * , 364 ( 1963 ) [ sov .
jetp * 18 * , 253 ( 1964 ) ] ; see also p. a. wolff , j. phys .
sol . * 25 * , 1057 ( 1964 ) ; a. g. aronov and g. e. pikus , zh .
. fiz . * 51 * , 281 ( 1966 ) [ sov .
jetp * 24 * , 188 ( 1967 ) ] , and m. h. weiler , w. zawadzki , and b. lax , phys . rev .
* 163 * , 733 ( 1967 ) . l. d. landau and e. m. lifshitz , _ quantum mechanics : non - relativistic theory _
, 3rd edition , vol.iii , ( butterworth - heinemann , 1981 ) , sections 35 ( fall to the center ) , 49 ( semi - classical langer correction @xmath468 ) , and 70 ( thomas - fermi theory ) .
a. b. migdal , d. n. voskresenski , and v. s. popov , pisma zh .
* 24 * , 186 ( 1976 ) [ sov .
jetp lett .
* 24 * , 163 ( 1976 ) ] ; a. b. migdal , v. s. popov , and d. n. voskresenski , zh .
72 * , 834 ( 1977 ) [ sov .
jetp * 45 * , 436 ( 1977 ) ] .
v b berestetskii , l. p. pitaevskii , and e.m .
lifshitz , _ quantum electrodynamics _ ,
second edition : volume 4 ( course of theoretical physics ) ( butterworth - heinemann , 1982 ) , sections 36 ( the dirac - kepler problem ) and 133 ( the landau `` zero charge '' effect ) .
v. d. mur and v. s. popov , teor . mat .
* 27 * , 204 ( 1976 ) [ theor . and math . phys .
* 27 * , 429 ( 1976 ) ] , v. s. popov , v. l. eletski , and v. d. mur , zh .
. fiz . * 71 * , 856 ( 1976)[sov .
jetp , * 44 * , 451 ( 1976 ) ] .
d. p. divincenzo and e. j. mele , phys .
b * 29 * , 1685 ( 1984 ) ; m. i. katsnelson , phys .
b * 74 * , 201401 ( r ) ( 2006 ) ; a. v. shytov , m. i. katsnelson , and l. s. levitov , phys . rev . lett . *
99 * , 236801 ( 2007 ) ; m. m. fogler , d. s. novikov , and b. i. shklovskii , phys . rev .
b * 76 * , 233402 ( 2007 ) .
y. wang , d. wong , a. v. shytov , v. w. brar , s. choi , q. wu , h .- z .
tsai , w. regan , a. zettl , r. k. kawakami , s. g. louie , l. s. levitov , and m. f. crommie , _ observing atomic collapse resonances in artificial nuclei on graphene _ , science 1234320 published online 7 march 2013 [ doi:10.1126/science.1234320 ] .
|
the field of charged impurities in narrow - band gap semiconductors and weyl semimetals can create electron - hole pairs when the total charge @xmath0 of the impurity exceeds a value @xmath1 the particles of one charge escape to infinity , leaving a screening space charge .
the result is that the observable dimensionless impurity charge @xmath2 is less than @xmath3 but greater than @xmath4 .
there is a corresponding effect for nuclei with @xmath5 , however in the condensed matter setting we find @xmath6 .
thomas - fermi theory indicates that @xmath7 for the weyl semimetal , but we argue that this is a defect of the theory . for the case of a highly - charged recombination center in a narrow band - gap semiconductor ( or of a supercharged nucleus ) , the observable charge takes on a nearly universal value . in weyl semimetals
the observable charge takes on the universal value @xmath8 set by the reciprocal of material s fine structure constant .
|
according to the so called `` no - hair '' conjecture , a stationary black hole is uniquely described in terms of a small set of asymptotically measurable quantities .
this hypothesis was disproved more than ten years ago , when several authors presented a counterexample within the framework of su(2 ) einstein - yang - mills ( eym ) theory @xcite .
although the new solution was static with vanishing yang - mills ( ym ) charges , it was different from the schwarzschild black hole and , therefore , not characterized by its total mass ( see @xcite for a comprehensive review of this topic and an extensive bibliography ) .
however , much on the literature on hairy black hole solutions is restricted to the case of an asymptotically flat spacetime . since
asymptotic flatness is not always an appropriate theoretical idealisation , and is never satisfied in reality , it may be important to consider other types of asymptotics , in particular solutions with a cosmological constant @xmath6 .
asymptotically anti - de sitter ( aads ) black hole solutions with su(2 ) nonabelian fields have been presented in @xcite .
the properties of these configurations are strikingly different from those valid in the asymptotically flat case ( for example there are stable solutions in which the gauge field has no nodes ; also solutions exist for continuous intervals of the parameter space , rather than discrete points ) . in an unexpected development
, it has been shown recently that for @xmath7 even a conformally coupled scalar field can be painted as hair @xcite . much of the discussion on ads hairy black holes has concerned the case when the matter fields fall off sufficiently fast such that the conserved charges can be written as surface integrals involving only the metric and its derivatives . however , recently it became clear that the usual ads - invariant boundary conditions do not include all aads configurations .
several solutions involving a minimally coupled self - interacting scalar field in the bulk action and preserving the asymptotic ads symmetry group despite the fact that the standard gravitational mass diverges , have appeared in the literature @xcite-@xcite .
an exact four - dimensional black hole solution of gravity with a minimally coupled self - interacting scalar field has been presented in @xcite by martinez , troncoso and zanelli ( mtz ) .
hairy black hole solutions of @xmath8 gauged supergravity in four and five dimensions are described in ref .
solutions describing a gravitating self - interacting scalar field whose mass saturates the breitenlohner - freedman bound are discussed in @xcite . in all these cases ,
the scalar fields drop off so slowly in the asymptotic region , such that they add a nonzero contribution to the conserved charges .
the mass of these solutions is computed by using an hamiltonian method , such that the divergencies from the gravity and scalar parts cancel out , yielding a finite total charge . in this paper
we consider the case of einstein gravity coupled to a dilaton field @xmath9 with a dilaton potential @xmath10 , looking for solutions satisfying a weakened set of boundary conditions at infinity , which implies a diverging adm mass , despite the fact that the spacetime is still aads . to simplify the general picture
, we restrict ourselves to the case @xmath11 .
also , since nonabelian fields usually occur together with the dilaton in the bosonic sector of many gauged supergravity theories , we include a su(2 ) nonabelian field in the action ( abelian solutions of this theory are discussed in @xcite , however for a particular set of boundary conditions which implies a finite adm mass ) . we suppose that the dilaton field approaches asymptotically a constant value @xmath12 , which corresponds to an extremum of the potential such that @xmath13 and @xmath14 .
therefore the configurations present an effective negative cosmological constant . since a negative cosmological constant allows for the existence of black holes whose horizon has nontrivial topology , we consider apart from spherically symmetric solutions , topological black holes also .
the mass of these solutions is computed by using a counterterm method .
although we discuss a general case , assuming only the existence of a power series expansion at infinity , numerical results are presented mainly for the case of @xmath15 gauged supergravity model of gates and zwiebach @xcite .
the paper is structured as follows : in the next section we explain the model and derive the basic equations and the asymptotic form of the solutions .
the boundary stress tensor and the associated conserved charge are computed in section 3 , by using a counterterm prescription adapted to our case .
the counterterm choice is tested for the mtz black hole solution . in section 4
we present the numerical results , the case of regular solutions being also briefly discussed .
we conclude with section 5 where the results are compiled .
we start with the following action principle @xmath16 where @xmath17 is the gravitational constant , @xmath18 is the ricci scalar associated with the spacetime metric @xmath19 .
@xmath20 is the gauge field strength tensor defined as @xmath21 \ , \ ] ] where the gauge field is @xmath22 @xmath23 being the pauli matrices .
the constant @xmath24 governs the coupling of @xmath9 to the nonabelian field while @xmath25 is the gauge coupling constant .
the last term in ( [ action ] ) is the hawking - gibbons surface term @xcite , where @xmath26 is the trace of the extrinsic curvature for the boundary @xmath27 and @xmath28 is the induced metric of the boundary .
the field equations are obtained by varying the action ( [ action ] ) with respect to the field variables @xmath29 and @xmath9 @xmath30&=&0,\end{aligned}\ ] ] where the energy - momentum tensor is defined by @xmath31 since for a negative cosmological constant topological black holes may appear ( whose topology of the event horizon is no longer the two - sphere @xmath32 ) , we consider a general metric ansatz @xmath33 where @xmath34 is the metric on a two - dimensional surface of constant curvature @xmath35 .
the discrete parameter @xmath36 takes the values @xmath37 and @xmath38 and implies the form of the function @xmath39 @xmath40 when @xmath41 , the metric takes on the familiar spherically symmetric form , for @xmath42 the @xmath43 sector is a space with constant negative curvature , while for @xmath44 this is a flat surface ( see @xmath45 the discussion in @xcite ) .
taking into account the symmetries of the line element ( [ metric ] ) , we find the expression of the purely magnetic ym ansatz @xcite @xmath46 which gives the ym curvature @xmath47 where a prime denotes a derivative with respect to @xmath48 .
inserting this ansatz into the action ( [ action ] ) , the field equations reduce to @xmath49 we assume that the solution of the above equations admits at large @xmath48 a power series expansion of the form @xmath50 from the lowest order term in equation of @xmath9 we find that @xmath51 ( we shall note @xmath52 ) .
we remark that , by using a suitable redefinition , we can always set @xmath53 , with no loss of generality .
the effective cosmological constant is @xmath54 the generic solution has @xmath55 , which , from the field equations , implies the following consistency conditions on the dilaton potential @xmath56 note that the scalar field mass @xmath57 is larger than the breitenlohner - freedman bound @xmath58 .
the assumption ( [ asym1 ] ) leads to the asymptotic expansion at large @xmath48 @xmath59 where @xmath60 are arbitrary constants , which implies the asymptotic form of the metric function @xmath61 for any @xmath62 , this set of asymptotics preserve the full ads symmetry group .
different from the asymptotically flat case , there are no obvious restrictions on the value of @xmath63 .
in order to compute quantities like the action and mass one usually encounters infrared divergences , associated with the infinite volume of the spacetime manifold . the traditional approach to this problem is to use the a background subtraction whose asymptotic geometry matches that of the solutions .
however , this approach breaks down when there is no appropriate or obvious background . in the ads
/ cft inspired counterterm method , this problem is solved by adding additional surface terms to the theory action .
these counterterms are built up with curvature invariants of a boundary @xmath64 ( which is sent to infinity after the integration ) and thus obviously they do not alter the bulk equations of motion .
this yields a well - defined boundary stress tensor and a finite action and mass of the system . as found in @xcite ,
the following counterterms are sufficient to cancel divergences in four dimensions , for vacuum solutions with a negative cosmological constant @xmath65 @xmath66\ , \end{aligned}\ ] ] where @xmath67 is the ricci scalar for the boundary metric @xmath28 .
using these counterterms one can construct a divergence - free boundary stress tensor @xmath68 from the total action @xmath69 by defining @xmath70 where @xmath71 is the einstein tensor of the boundary metric , @xmath72 is the extrinsic curvature , @xmath73 being an outward pointing normal vector to the boundary .
if @xmath74 is a killing vector generating an isometry of the boundary geometry , there should be an associated conserved charge .
we suppose that the boundary geometry is foliated by spacelike surfaces @xmath75 with metric @xmath76 @xmath77 thus the conserved charge associated with time translation @xmath78 is the mass of spacetime @xmath79 here @xmath80 is the proper energy density while @xmath81 is a timelike unit normal to @xmath75 .
the presence of the additional matter fields in ( [ action ] ) brings the potential danger of having divergent contributions coming from both the gravitational and matter action @xcite . for a @xmath82 ( or faster ) decay of the dilaton field in the asymptotic region
, we find that the prescription ( [ ct ] ) ( with @xmath83 removes all divergences of the total action , and implies the usual configurations mass @xmath84 @xmath85 being the area of the ( @xmath86 ) surface ( @xmath87 for @xmath41 ) . as expected , a dilaton field which behaves asymptotically as @xmath88 necessarily contributes to the action and its variations in the asymptotic region .
the counterterms ( [ ct ] ) will not yield in this case a finite action or mass .
however , similar to the case of three dimensional gravity with a minimally coupled scalar field @xcite , it is still possible to obtain a finite mass by allowing @xmath89 to depend not only on the boundary metric @xmath90 , but also on the scalar field .
this means that the quasilocal stress - energy tensor ( [ s1 ] ) also acquires a contribution coming from the matter field . since a polynomial in @xmath9
does not remove the divergencies , we are forced to consider terms containing the normal derivative of @xmath9 . following @xcite , we find that by adding a counterterm of the form @xmath91 to the expression ( [ ct ] ) , the divergence disappears . in ( [ ict ] ) with a suitable combination of @xmath92 and @xmath93 , in which case the matter conterterm reads @xmath94 , without changing the expressions for boundary stress - tensor , mass and total action .
] this yields a supplementary contribution to ( [ s1 ] ) , @xmath95 .
the nonvanishing components of the resulting boundary stress - tensor are ( here we choose @xmath96 to be a three surface of fixed @xmath48 , while @xmath97 ) @xmath98 we remark that , to leading order , this stress tensor is traceless as expected from the ads / cft correspondence , since even dimensional bulk theories are dual to odd dimensional cfts which have a vanishing trace anomaly . employing the ads / cft correspondence
, this result can be interpreted as the expectation value of the stress tensor in the boundary cft @xcite .
the mass of these solutions , as computed from ( [ mass ] ) is @xmath99 this coincides with the mass of the @xmath100 gauged supergravity solutions considered in @xcite , as computed by using an hamiltonian method ( the asymptotics of those solutions is a particular case of ( [ expansion1 ] ) ) .
also , it can be proven that the above counterterm choice yields a finite euclidean action .
recently martinez , troncoso and zanelli have found an exact black hole solution of the field equations ( [ einstein - eqs ] ) , corresponding to a @xmath101 truncation of the action ( [ action ] ) and a scalar potential @xcite @xmath102 the metric and the scalar field are given by @xmath103 , \\
\nonumber \phi & = & \sqrt{\frac{3}{4\pi g}}\;\mbox{arctanh}\frac{g\mu } { r+g\mu } \;. \label{scalar}\end{aligned}\ ] ] + the only singularities of the curvature and the scalar field occur at @xmath104 and at @xmath105 .
these singularities are surrounded by an event horizon located at @xmath106 while the hawking temperature is @xmath107 .
a straightforward computation yields the nonvanishing components of boundary stress tensor @xmath108 thus , from ( [ mass ] ) we find a total black hole mass @xmath109 while the total euclidean action is @xmath110 which , from @xmath111 gives an entropy which is one quarter of the event horizon area @xmath112 , the first law of thermodynamics being also satisfied .
these results coincide with those found in @xcite by using an hamiltonian formalism .
to solve the field equation ( [ eqs ] ) , we change to dimensionless variables by using the rescaling @xmath113 , @xmath114 and @xmath115 , together with a rescaling of the potential .
given the inherent difficulties involved in studies of these models , we will restrict ourselves to the case of a potential on the form @xmath116 ( where we suppose @xmath117 ) .
this type of potential can be obtained when a higher dimensional theory is compactified to four dimensions , including various supergravity models ( see @xcite for a recent discussion of these aspects ) . for @xmath118 ,
this is the dilaton potential appearing in the @xmath119 gauged supergravity model of gates and zwiebach . for @xmath120 ,
the action ( [ action ] ) is also a consistent truncation of the bosonic sector of this model .
spherically symmetric bps regular solutions of this theory with the general asymptotics ( [ expansion1 ] ) have been constructed recently in @xcite . although a liouville - type potential plus a cosmological constant ( obtained for @xmath121 or @xmath122 ) might be an interesting choice
, one can prove that such a model does not possess solutions with ads asymptotics .
for a nonzero @xmath0 term in the asymptotic expansion of the dilaton field , the conditions ( [ cond - pot ] ) impose the following relations between the potential parameters @xmath123 by using the scaling properties of the system @xmath124 we can always set @xmath53 , resulting in the simple potential @xmath125 where @xmath126 .
we observe that the potential of the gz model can also be written in this form ( for @xmath127 @xmath128 a negative effective cosmological constant ) , with @xmath129 and @xmath130 . in this case , following the approach in @xcite , we find the set of bogomolnyi equations @xmath131 ( with @xmath132 , @xmath133 , the equation ( [ cons ] ) being a differential consequence of the first four equations .
one can verify that these bogomolnyi equations are compatible with the eqs .
( [ eqs ] ) . also , there are no black hole solutions of the above equations .
numerical arguments for the existence of @xmath41 regular solutions are presented in @xcite .
the @xmath134 configurations preserving any supersymmetry present naked singularities .
for example , a @xmath44 solution of the above equations is ( with @xmath135 an arbitrary constant ) @xmath136 presenting a naked singularity at @xmath104 .
we are interested in black hole solutions having a regular event horizon at @xmath137 .
the field equations implies the following behaviour as @xmath138 in terms of three parameters ( @xmath139 @xmath140 the condition for a regular event horizon is @xmath141 , which places a bound on @xmath142 @xmath143 and implies the positiveness of the quantity @xmath144 . in the @xmath42 case , ( [ mhbound ] ) implies the existence of a minimal value of @xmath145 , @xmath128 for a given @xmath146 @xmath147 and @xmath148 are plotted for two @xmath41 typical black hole solutions .
+ + by going to the euclidean section ( or by computing the surface gravity ) one find the hawking temperature @xmath149 using the initial conditions on the event horizon ( [ eh ] ) , the equations were integrated for @xmath150 , several values of @xmath151 , and a large set of @xmath152 .
also , although we present here only the case @xmath153 , similar solutions seem to exist for any value of @xmath146 . since the equations ( [ eqs ] ) are invariant under the transformation @xmath154 , only values of @xmath155 are considered .
nontrivial black hole solutions with scalar hair exist also in the absence of a gauge field .
and the dilaton potential ( [ potmtz ] ) we have found a family of solutions in terms of @xmath156 , the exact mtz solution ( [ mtz - sol ] ) being a particular case . ] the basic properties of these solutions are very similar to the known eym-@xmath6 black holes @xcite . for any @xmath156 and @xmath157
, solutions appear for continuous intervals of @xmath142 ( for @xmath129 , we find always only one interval ) . for small @xmath158 ,
these intervals are separated by intervals on which there are no solutions with the asymptotics ( [ expansion1 ] ) . as @xmath142 approaches some critical value @xmath159 , the metric function @xmath160 approaches a zero value on the event horizon . the value of @xmath159 increases as @xmath158 increases . also , there are black hole solutions for which @xmath161 although @xmath162
. typical spherically symmetric solutions of the gz model are presented in figure 1 . for @xmath134 , in contrast to the spherically symmetric case
, we find only nodeless solutions , for all values of the parameters .
this can be analytically proven by integrating the equation for @xmath163 , @xmath164 between @xmath146 and @xmath48 ; thus obtaining @xmath165 for every @xmath166 . for @xmath41 and @xmath158 sufficiently large ( i.e. @xmath167 ) , there also exist solutions for which the gauge function @xmath163 has no nodes . increasing the value of @xmath158 ,
the ratio @xmath168 remains close to one for most of the @xmath142 interval , and we find nodeless solutions only .
the properties of typical solutions are presented in figure 2 for @xmath169 and a gz potential ( a similar picture is found for @xmath44 ) .
we observe that the generic solutions we find do not have the usual ads asymptotics since @xmath170 .
however , for any @xmath171 , solutions with @xmath172 exist for a discrete set of @xmath156 .
both the parameter @xmath173 and the total mass @xmath174 of some @xmath42 solutions is negative , a common situation in the topological black hole physics . for @xmath175
we find @xmath176 for all boundary conditions we have considered .
( 18,7 ) ( 2,0.0 ) ( 19,8.5 ) ( 2.6,0.0 ) + + the asymptotic parameters @xmath177 , the total mass @xmath174 , the hawking temperature @xmath178 and the value @xmath179 of the metric function function @xmath160 at the event horizon are shown as a function of @xmath180 for spherically symmetric and @xmath42 topological black hole solutions of the gz model .
+ it can be proven that the entropy of these black holes is one quarter of the event horizon area , as expected . by integrating the killing identity @xmath181 for the killing field @xmath182 , together with the einstein equation ( here we have set @xmath183 ) @xmath184 it is possible to isolate the bulk action contribution at infinity and on the event horizon .
the counterterms discussed in section 3 regularize the infrared divergencies , such that the contribution from the asymptotic region to the total action is found to be @xmath185 , the relation @xmath186 resulting straightforwardly from ( [ entropy ] ) , ( [ th ] ) .
( 18,7 ) ( 2,0.0 ) + + the asymptotic parameters @xmath177 , the total mass @xmath174 and the value @xmath187 of the metric function function @xmath160 at the origin are shown as a function of @xmath188 for typical spherically symmetric regular solutions of the gz model .
the existence of regular counterparts of the black hole solutions is possible in the spherically symmetric case only . for @xmath189 , a direct inspection of the system ( [ eqs ] )
reveals the absence of solutions with a regular origin . in this case , it is not possible to take a consistent set of boundary conditions at the origin without introducing a curvature singularity at @xmath104 .
this fact has to be attributed to the particular form of the potential term @xmath190 , which is fourth order in the ym function @xmath163 , in the reduced lagrangean of the system , .
we observe the similarity with the eym system with @xmath7 , where the absence of @xmath191 regular configurations has been noticed in @xcite . for completeness
we present here the basic properties of the @xmath41 regular configurations .
the behaviour of regular solutions near the origin is @xmath192 where @xmath193 and @xmath194 are arbitrary parameters .
we note that for bps solutions of the gz model , there are only two independent parameters since in this case @xmath195 .
the overall picture we find is similar to the one described in @xcite for the eym-@xmath6 system ( here we consider again the case of the gz model ) . by varying the parameters
@xmath196 , a continuum of monopole solutions is obtained .
as @xmath188 increases , the value at the origin of the metric function @xmath160 decreases and , for some finite values of @xmath188 , a singularity appears .
the total mass of the solutions as given by ( [ mct ] ) is an increasing function of @xmath188 .
for any value of @xmath188 it seems to be always possible to find a initial value of the scalar field such that @xmath172 . as seen in figure 3 ,
the solution with @xmath197 ( @xmath128 @xmath198 ) is not the vacuum ads spacetime .
thus , rather unexpectedly , for these asymptotics there are regular solutions even without a nonabelian field .
for the gz model , we find a continuum of ( nonsupersymmetric- ) scalar solitons , as a function of the value of the scalar field at the origin @xmath194 . a similar property has been noticed recently for a truncation of @xmath199 gauged supergravity @xcite .
the expansion as @xmath200 is valid also for these regular solutions . for the gz model ,
the coefficients of bps solutions in the asymptotics ( [ expansion1 ] ) satisfy the relations @xmath201 , @xmath202 ( we recall that for @xmath191 , these describe configurations with a naked singularity at @xmath104 ) .
we note also the intriguing expression for the total mass of bps configurations @xmath203 with @xmath204 .
if the gravitating matter fields do not fall off sufficiently fast at infinity , the asymptotic behaviour of the metric can be different from that in pure gravity . by relaxing the standard asymptotic conditions for aads solutions
, it is possible to preserve the original symmetries at infinity , while the conserved charges are modified by including matter field terms .
the aim of this letter was to consider this situation for a theory including , apart from an su(2 ) nonabelian field , a dilaton field with a nontrivial potential , playing the role of a cosmological term .
we have found that the addition of the scalar potential greatly increases the wealth of possible solutions , preserving at the same time all features familiar from the eym-@xmath6 case . for solutions admitting an asymptotic power series expansion ,
we have proposed a counterterm choice which gives finite mass and euclidean action .
the results we have found are in agreement with those obtained via the hamiltonian method .
numerical solutions have been presented mainly for the case of a consistent truncation of @xmath15 gz gauged supergravity model .
both regular and black hole solutions have been presented , the solutions with a finite adm mass constituing a discrete set . by using the relations in @xcite
, we can uplift these configurations to @xmath205 , which may suggest a holographic interpretation for them . as observed in @xcite ,
according to the general ads / cft correspondence , there should be a dual cft corresponding to each choice of boundary conditions . also , it is possible to relax the asymptotic asumptions ( [ asym1 ] ) , allowing a generic noninteger decay at infinity . in this case , apart from @xmath206 there are no other restrictions on the dilaton potential and we find the asymptotic behaviour ( with @xmath207 ) @xmath208 where @xmath209 both regular and black hole solutions with these asymptotics are likely to exist .
it can be proven that the counterterm choice ( [ ict ] ) gives a finite mass and action also in this case , yielding very similar results to those derived in this paper for @xmath210 .
it would be interesting to generalize these solutions to higher dimensions and to find the general matter counterterms expression .
+ + * acknowledgement * + this work was carried out in the framework of enterprise ireland basic science research project sc/2003/390 .
99 m. s. volkov and d. v. galtsov , jetp lett .
* 50 * ( 1989 ) 346 ; + h. p. kuenzle and a. k. masood- ul- alam , j. math . phys .
* 31 * ( 1990 ) 928 ; + p. bizon , phys .
* 64 * ( 1990 ) 2844 .
m. s. volkov and d. v. galtsov , phys .
rept . *
319 * ( 1999 ) 1 [ arxiv : hep - th/9810070 ] .
e. winstanley , class .
* 16 * ( 1999 ) 1963 [ arxiv : gr - qc/9812064 ] . j. bjoraker and y. hosotani , phys .
d * 62 * ( 2000 ) 043513 [ arxiv : hep - th/0002098 ] .
e. winstanley , found .
phys . *
33 * ( 2003 ) 111 [ arxiv : gr - qc/0205092 ] ; + d. sudarsky and j. a. gonzalez , phys .
d * 67 * ( 2003 ) 024038 [ arxiv : gr - qc/0207069 ] . c. martinez , r. troncoso and j. zanelli , arxiv : hep - th/0406111 .
m. henneaux , c. martinez , r. troncoso and j. zanelli , phys . rev .
d * 65 * ( 2002 ) 104007 [ arxiv : hep - th/0201170 ] .
t. hertog and k. maeda , arxiv : hep - th/0404261 .
t. hertog and k. maeda , arxiv : hep - th/0409314 .
t. hertog , arxiv : hep - th/0409160 .
m. henneaux , c. martinez , r. troncoso and j. zanelli , arxiv : hep - th/0404236 . j. t. liu and w. a. sabra , arxiv : hep - th/0405171 . s. j. poletti , j. twamley and d. l. wiltshire , phys . rev .
d * 51 * ( 1995 ) 5720 [ arxiv : hep - th/9412076 ] ; + s. j. poletti , j. twamley and d. l. wiltshire , class .
* 12 * ( 1995 ) 1753 [ erratum - ibid .
* 12 * ( 1995 ) 2355 ] [ arxiv : hep - th/9502054 ] .
s. j. j. gates and b. zwiebach , phys .
b * 123 * ( 1983 ) 200 ; + b. zwiebach , nucl .
b * 238 * ( 1984 ) 367 .
g. w. gibbons and s. w. hawking , phys .
d * 15 * ( 1977 ) 2752 .
r. b. mann , arxiv : gr - qc/9709039 ; + j. p. lemos , arxiv : gr - qc/0011092 .
j. j. van der bij and e. radu , phys .
b * 536 * ( 2002 ) 107 [ arxiv : gr - qc/0107065 ] .
v. balasubramanian and p. kraus , commun .
* 208 * ( 1999 ) 413 [ arxiv : hep - th/9902121 ] .
m. m. taylor - robinson , arxiv : hep - th/0002125 . j. gegenberg , c. martinez and r. troncoso , phys .
d * 67 * ( 2003 ) 084007 [ arxiv : hep - th/0301190 ] .
r. c. myers , phys .
d * 60 * ( 1999 ) 046002 .
u. nucamendi and m. salgado , phys .
d * 68 * ( 2003 ) 044026 [ arxiv : gr - qc/0301062 ] .
k. g. zloshchastiev , arxiv : hep - th/0408163 .
s. b. giddings , phys . rev .
d * 68 * , 026006 ( 2003 ) [ arxiv : hep - th/0303031 ] .
a. h. chamseddine and m. s. volkov , arxiv : hep - th/0404171 .
|
we consider black hole solutions with a dilaton field possessing a nontrivial potential approaching a constant negative value at infinity .
the asymptotic behaviour of the dilaton field is assumed to be slower than that of a localized distribution of matter .
a nonabelian su(2 ) gauge field is also included in the total action .
the mass of the solutions admitting a power series expansion in @xmath0 at infinity and preserving the asymptotic anti - de sitter geometry is computed by using a counterterm subtraction method .
numerical arguments are presented for the existence of hairy black hole solutions for a dilaton potential of the form @xmath1 , special attention being paid to the case of @xmath2 gauged supergravity model of gates and zwiebach . * new hairy black hole solutions + with a dilaton potential * + * eugen radu@xmath3 * * and d. h. tchrakian @xmath4 * _ @xmath5school of theoretical physics
dias , 10 burlington road , dublin 4 , ireland _
|
anderson localization has attracted renewed interest since a plenty of new universality classes were discoverd @xcite .
one is a system with particle - hole symmetry studied by gade @xcite .
although generic disorderd systems in two dimensions are believed to be insulators @xcite , she showed that such a system has a random critical point at the band center .
typical examples studied so far are the random flux model @xcite and the random - hopping model with @xmath0-flux @xcite .
this class is often referred to as @xmath5iii and @xmath6i in the case with broken and unbroken time reversal symmetry , respactively , which corresponds to the chiral gaussian unitary and chiral gaussian orthogonal ensamble of the random matirx theory in the zero dimensional limit @xcite .
other examples of new universality classes are found in disordered @xmath7-wave superconductors described by a mean - field bogoliubov - degennes hamiltonian @xcite .
altland and zirnbauer @xcite classified them by two basic symmetries , i.e , spin - ratation and time - reversal into four kinds of classes . with ( without )
spin - rotation invariance , disordered @xmath7-wave superconductors belong to class @xmath8 and @xmath8i ( @xmath9 and @xmath9iii ) , respectively , in the case with broken and unbroken time - reversal symmetry .
the beta function of the renormalization group for corresponding nonlinear sigma models in two dimensions tells that the classes @xmath8 and @xmath8i have localized states only , whereas the classes @xmath9 and @xmath9iii may have delocalized states .
what is particularly interesting is that some of them allow in the nonlinear sigma model description the topological term @xcite which plays a crucial role , for example , in quantum hall transitions @xcite , or the wess - zumino - witten ( wzw ) term @xcite which gives rise to a nontrivial strong - coupling fixed point .
therefore , disordered @xmath7-wave superconductors are expected to have quite rich phase diagram .
these developments can shed new light on the localization problems .
for example , the question whether the density of states of the random flux model diverges had been a long - standing problem .
recognizing , however , that it belongs to the gade s class of @xmath5iii type , furusaki concluded that the dos actually diverges .
another example is the disordered fermion model with @xmath0-flux we will revisit in this paper .
this model was already studied by fisher and fradkin @xcite in 1984 .
they showed that the effective field theory of the lattice model with random on - site potentials belongs to the universality class of o(@xmath10)/o(@xmath11o(@xmath3 ) nonlinear sigma model , concluding thereby that all states are localizaed .
a point here is that the continuum limit of the model includes random potentials specified by four independent coupling constants .
since fisher and fradkin studied only the case where all the coupling constants are equal , we can expect richer phase diagram in the whole space of the coupling constants .
actually , although the generic points in this space should flow into the localized phase mentioned - above , the model has enhanced symmetry in special cases and accordingly renormalization group flows in them are toward different fixed points . another point in this paper
is concerned with chiral symmetry .
it was shown that generic disordered @xmath7-wave superconductors in class @xmath8i , described by dirac fermions , yield no wzw term @xcite although those dirac fermions have chiral sp(@xmath1sp(@xmath2 ) symmetry .
the reason may be that chirality is not unique in the @xmath7-wave dirac fermions , since the lattice fermion gives the species doubling in the continuum limit . here
, chiral symmetry means that there exist a matrix which anticommutes with a hamiltonian . in this paper
, we investigate in detail the phase diagram of the disordered fermion model with @xmath0-flux by means of the fermionic replica method to clarify these points .
we show that this model , described by double dirac fermions , has three kinds of chiral symmetry in special cases .
interestingly , these different chiral symmetries lead to different universality classes
. to be concrete , an effective action of the green functions has sp(@xmath1sp(@xmath2 ) symmetry in one of them and u@xmath12 symmetry in the others , both of which are broken down to sp(@xmath2 ) .
it turns out that corresponding goldstone mode can be described by sp(@xmath4 wzw model and u(@xmath2)/sp(@xmath2 ) nonlinear sigma model , respectively .
this paper is organized as follows . in the next section ,
we introduce a lattice model , which is the same as fisher and fradkin studied , and derive dirac hamiltonian in the continuum limit .
although the most generic model has only time - reversal invariance , symmetry is enhanced on three surfaces in the space of the coupling constants .
we focus our attention to one of them with sp(@xmath1sp(@xmath2 ) symmetry ( others have u(@xmath3 ) symmetry ) and give calcualtions in detail . based on the continuum hamiltonian
, we derive an effective generating functional of ensemble - averaged green functions in sec . [
s : repmet ] .
we show that sp(@xmath1sp(@xmath2 ) symmetry is spontaneously broken down to sp(@xmath2 ) . in sec .
[ s : actgol ] , we derive an effective action for the goldstone mode , resulting in sp(@xmath4 wzw model ( u(@xmath3)/sp(@xmath2 ) nonlinear sigma model for the others ) . in sec . [
s : phadia ] , we present the phase diagram . in sec . [
s : sum ] , we give a summary and concluding remarks .
we define a baisc lattice model on the square lattice in two dimensions . for convenience ,
we introduce similar notations used by ludwig et al @xcite in the following way .
the square lattice in two dimensions is divided into unit cells , each composed of a plaqutte enclosing four sites .
a plaquette is labled by a set of integers @xmath13 and four sites on it are numbered 1,2,3 and 4 counterclockwise , as illustrated in fig .
[ f : fig1 ] .
with such a notational convention , a pure and disorder hamiltonian denoted respectively as @xmath14 and @xmath15 are defined by @xmath16 where @xmath17 , @xmath18 , and @xmath19 is random real number due to nonmagnetic impurities .
negative signs of the third and the sixth terms in @xmath14 are due to @xmath0 flux imposed .
now let us construct an effective hamiltonian near the band center in the case of weak disorder . to this end
, we first derive the continuum limit of the pure model , and next we take into account the effects of disorder .
fourier transformation @xmath20 for @xmath21 leads to @xmath22 where @xmath23 .
it is readily seen that the hamiltonian gives zero - energies at @xmath24 . in the vicinity of them
, the model can be effectively described by dirac fermions . to see this , set @xmath25 with the lattice constant @xmath26 , expand the hamiltonian to the first order of @xmath27 , and introduce a field operator in the continuum limit @xmath28 for @xmath21 .
then we have @xmath29 where @xmath30 . accordingly
, effective hamiltonian near the zero energy can be expressed by the following field theoretical hamiltonian @xmath31 where @xmath32 , @xmath33 , and @xmath34 matrices are defined as @xmath35 and @xmath36 . by using the relation between the lattice operator @xmath37 and the continuum field operator @xmath38 , @xmath39 the continuum limit of the disorder potentials
are readily derived ; @xmath40 , where @xmath41 is the slowly - varying part of @xmath42 defined as @xmath43 .
taking suitable linear combination of @xmath44 , we can rewrite the random potentials as @xmath45 where @xmath46 and @xmath47 .
we assume that the four disorder potentials obey gaussian probability distribution @xmath48\propto e^{-\frac{2}{g_\nu}v_\nu^2(x ) } , \quad
\mbox{for}\quad \nu=1, ... ,4 . \label{prodis}\ ] ] it may be more convenient to switch into another basis via suitable rotation for @xmath49 , @xmath50 in what follows , we refer to the space where these matrices live as @xmath51 .
it is distinguished from the replica space @xmath52 introduced momentarily .
the total hamiltonian we investigate in this paper is @xmath53 in eqs .
( [ conpurham ] ) and ( [ condisham ] ) with ( [ gamalp ] ) .
it is noted that this is the same model as fisher and fradkin @xcite studied in the case @xmath54 .
they showed that the effective field theory is o(@xmath10)/o(@xmath11o@xmath55 ) nonlinear sigma model for the bosonic replicas [ or sp(@xmath3)/sp(@xmath1sp(@xmath2 ) nonlinear sigma model for the fermionic ones ] , concluding thereby that all states are localized .
according to zirnbauer , this class is referred to as @xmath5ii @xcite .
although a slight relaxation of the condition @xmath54 still leads to the same universality class , several subspaces in @xmath56 have not only time - reversal symmetry but also additional enhanced symmetry specified below , and therefore belong to different universality classes . in this subsection
, we examine symmetries of the present model , which plays a crucial role in the classification of the universality classes for disordered systems .
first of all , the most general hamiltonian has unbroken time - reversal symmetry . in the original basis , it is expressed by @xmath57 , whereas in the rotated basis where the hamiltonian is described by the matrices in eq .
( [ gamalp ] ) , it is expressed by @xmath58 if the model does not have any other symmetries , the universality class is @xmath5ii and hence all states are localized .
there are several subspaces in @xmath56 , however , on which symmetry is enhanced .
to be concrete , the hamiltonian ( [ conpurham ] ) has following symmetry @xmath59 recently , hatsugai et.al @xcite proposed the same model as eq .
( [ latham ] ) but with random hoppings instead of random on - site potential , which we will refer to as hwkm model .
they have shown that such hamiltonian has not only time - reversal symmetry but also another one often called chiral symmetry .
it has been shown @xcite that this symmetry is responsible for the existence of delocalized critical point at the zero energy .
chiral symmetry in this case is expressed as @xmath60 with @xmath61 , similarly to eq .
( [ chisym ] ) , and the universality class is specified as @xmath6i or in other words u@xmath62sp(@xmath2 ) ( gl(2@xmath2,r)/o(2@xmath2 ) ) nonlinear sigma model for fermionic ( bosonic ) replica method . considering these
, it is easily guessed that the symmetry ( [ chisym ] ) also plays a crucial role in anderson localization for the present model . in what follows
, we refer to the symmetry in eq .
( [ chisym ] ) also as chiral symmetry .
it is so tedious to repeat similar calculations that we concentrate on the case @xmath63 , and hence we often denote @xmath63 just as @xmath64 for short . but with slight modification , any @xmath65 can apply to the formulation . at the moment
, the difference between @xmath63 and the others is whether @xmath65 commutes or anticommutes with @xmath66 and whether it is real symmetric or antisymmetric .
surprisingly , this difference , which seems trivial at first glance , leads to a different universality class . in order to study the trasport properties of the model near the zero energy
, we need to compute the green functions @xmath67 with the @xmath64 symmetry , however , we do not need to distinguish the retarded and advanced green functions , since they are related with each other via the relation @xcite , @xmath68 the green function @xmath69 can be computed by a replicated functional @xmath70 with @xmath71 where the subscript @xmath72 denotes the @xmath2 species in the replica space @xmath52 , @xmath73 , and the usual convention of the matrix product is applied to the fermion matrix field @xmath74 and @xmath75 .
although such a notational convention is not always necessary , it is quite convenient for us to save many complicated subscripts .
it should be noted that the fields @xmath76 and @xmath77 are completely independent grassmann variables .
therefore , they obey in general different transformation laws . based on the lagrangian ( [ lagbefave ] ) , we will construct an effective theory of order parameter fields for the replicas by averaging - over disorder . in this process
, it is quite important to fully incorporate the symmetries ( [ timrev ] ) and ( [ chisym ] ) in the space @xmath51 . to this end
, we introduce an auxiliary space to reflect these into the replicas , according to the method developed by zirnbauer @xcite .
namely , we extend the replica space as @xmath78 , where @xmath79 @xmath80 stands for the space associtaed with time - reversal ( chiral ) symmetry . we then introduce a @xmath81 matrix field @xmath82 , @xmath83 where @xmath84 @xmath85 , and @xmath86 stands for the transpose of @xmath49 . since the components of these fields are not independent of each other , the fields are subject to the constraints @xmath87 note that the matrices @xmath34 and @xmath0 are defined in @xmath88 , whereas @xmath66 and @xmath64 are defined in @xmath51 .
we immediately find @xmath89 these relations tell that the matrix @xmath90 in @xmath51 has the same symmetries as @xmath91 has .
it is also easy to find alternative relations for the matrix @xmath92 in @xmath88 , @xmath93 these relations actually specify the symmetry of the model , as we shall see lator . although the potentials @xmath94 and @xmath95 have time - reversal and chiral symmetries , the converse is not true : the symmetries ( [ timrev ] ) and ( [ chisym ] ) admit other potentials @xmath96 and @xmath97 . the reason why there are no beta matrices in eq .
( [ condisham ] ) is that the basic lattice model ( [ latham ] ) includes only on - site random potential .
namely , if we take all kinds of possible disorder for the lattice hamiltonian including random ( nearest neighbor and diagonal ) hoppings as well as random on - site potentials , we have a continuum hamiltonian with all kinds of possible disorder potentials specified by the relations ( [ timrev ] ) and ( [ chisym ] ) .
such a model may be referred to as a model with maximum entropy .
since we are interested in the model with on - site random potentials , we have to remove these @xmath98 matrices . for this purpose , we take account of additional conditions @xmath99 it is readily seen that the potentials including beta matrices do not satisfy these relations .
let us then introduce the fields @xmath100 we refer to this extra space as @xmath101 .
it is easy to verfy that the matrix @xmath102 obeys the same relations that @xmath103 does in ( [ adicon ] ) and in ( [ psitpsi ] ) with extension @xmath104 and @xmath105 , which means that the undesirable potentials are in fact excluded . by using these fields
we can reexpress the lagrangian ( [ lagbefave ] ) as @xmath106 where @xmath107 and @xmath108 .
now we are ready to ensemble - average the lagrangian ( [ lagreexp ] ) over disorder . in this subsection , we assume @xmath109 for simplicity .
more general case @xmath110 is briefly discussed in sec .
[ s : anieff ] . the probability distribution ( [ prodis ] )
can be expressed as @xmath111\propto e^{-\frac{1}{2g}{{\rm tr}}_v{{\cal h}}_{{\rm d}}^2 } .
\label{prodis2}\ ] ] therefore , the lagrangian is converted into @xmath112 where we have used a relation @xmath113 .
it should be noted that integration over @xmath103 is readily performed , since the matrix @xmath102 has completely the same symmetries as @xmath103 has .
this is the reason why we have introduced several auxiliary spaces in the last subsection . by using @xmath114 again and the definition of ( [ cirpsi ] ) , the gaussian integration over @xmath103 leads to @xmath115 where @xmath116 , @xmath117 , and @xmath118 .
we mention in passing that if the model have maximun entropy , there appears only one interaction term with @xmath119 and @xmath120 .
( [ lagaftdis ] ) is an effective lagrangian which generates the emsanble - averaged green functions for the present model .
it is easy to confirm that new matrices are subject to @xmath121 so far we have obtained an effective action including four fermi interactions by averaging over disorder .
the next task is to introduce auxiliary ( order - parameter ) fields to decouple them . to this end , we add the following terms of auxiliary fields @xmath122 @xmath123 to the lagrangian ( [ lagaftdis ] ) : @xmath124 then , the lagrangian becomes @xmath125 eq .
( [ intauxfie ] ) requires that @xmath122 should obey the same symmetrirs that @xmath126 does in eq .
( [ wpsiopsi ] ) .
therefore , @xmath127 the antihermiticity ensures that the integration over @xmath122 in eq .
( [ lagordpar ] ) converges .
this lgrangian ( [ lagordpar ] ) has a global sp(@xmath1sp(@xmath2 ) symmetry . to see this ,
let us onsider a transformation @xmath128 , @xmath129 , and @xmath130 . since this transformation should keep the relations ( [ conq ] ) ,
@xmath131 is subject to @xmath132 and @xmath133 .
therefore , @xmath131 has the following form @xmath134 where @xmath135 matrices @xmath136 belongs to @xmath137 and the explicit matrix above represent the space @xmath138 .
the conditions for @xmath136 require @xmath139 sp(@xmath2 ) .
next let us solve eq .
( [ conq ] ) , @xmath140 where @xmath141 matrices @xmath142 and @xmath143 are subject to @xmath144 , @xmath145 , and @xmath146 . using the definition in eq .
( [ defpsi ] ) , the lagrangian is now written as @xmath147 -\frac{i\epsilon}{2g'}{{\rm tr}}_{{w_{{\rm rt}}}}(m_1-m_1^\dagger ) \nonumber\\ & & -\sum_{\tau=1}^2{{\rm tr}}_{{w_{{\rm rt}}}}\left ( m_\tau{{\widetilde\chi}}_-{{\cal o}}_\tau\chi_+ -m_\tau^\dagger{{\widetilde\chi}}_+{{\cal o}}_\tau\chi_- + a_\tau^+{{\widetilde\chi}}_+{{\cal o}}_{\tau+2}\chi_+ + a_\tau^-{{\widetilde\chi}}_-{{\cal o}}_{\tau+2}\chi_- \right ) .
\label{chilag}\ ] ] the tranformation laws under @xmath131 are @xmath148 so far we have examined the case @xmath63 . it has been shown that the lagrangian has sp(@xmath1sp(@xmath2 ) symmetry and hence the model belongs to @xmath8i , which is the same class as @xmath7-wave superconductors with unbroken time - reversal and spin - rotation symmetries .
contrary to this , in the other cases associated with @xmath149 and @xmath150 , symmetry of the lagrangian is quite different .
similar calculations above show that these lagrangians have u(@xmath3 ) symmetry insted of sp(@xmath1sp(@xmath2 ) dut to the fact that @xmath34 defined in ( [ relpsi ] ) is @xmath151 if we take the same @xmath0 .
this is the same class as hkwm model , which is often referred to as @xmath6i or chiral orthogonal class .
based on the lagrangian in the last subsection , we derive the effective potential to study the spontaneous symmetry breaking .
we assume that the order - paramters are diagonal in the space @xmath52 .
namely , set @xmath152 where @xmath153 and @xmath154 are real fields .
then the lagrangian ( [ chilag ] ) becomes @xmath155 where @xmath156 is @xmath157 matrix defined by @xmath158 integrating out the fermi fields , we have an effective potential @xmath159 of one - loop order , @xmath160 since the potential @xmath161 is a function of six variables , it is hard to find out the true minimum .
however , it is readily expected from the symmetry breaking term that @xmath162 has a finite value .
assuming it , we have tried to find out nontrivial sulution @xmath163 , but in vain .
although possibility of some other solutions still remains , the solution @xmath164 and others=0 is of course a natural solution . in this case , the saddle point equaiton for @xmath162 becomes @xmath165 which gives nontrivial solution @xmath166 .
this result suggests that the operator @xmath167 is relevant whereas others are irrelaevant around the present saddle - point .
therefore , in the following analysis , we neglect the irrelavant interactions and take only the relevant interaction into account . in the next section , we will discuss slow fluctuation aroung the saddle point .
but before it , we will see how the anisotropy @xmath110 modifies the theory so far obtained . in the case
where @xmath168 , we could treat them simultaneously in eq .
( [ prodis2 ] ) .
contrary to this simple case , we have to treat the two potentials @xmath94 and @xmath95 separately in anisotropic case .
to this end , first introduce a new matrix @xmath169 under which they transform @xmath170 we next introduce @xmath171 let us denote @xmath172 for @xmath173 and the extra auxiliary space as @xmath174 . since the matrix @xmath175 has completely the same symmetry property as @xmath176 , we can average over disorder in the following way : @xmath177 where arrow menas that the integration over @xmath176 is carried out .
this equaiton shows that anisotropy @xmath110 yields an additional term including @xmath178 .
however , it turns out that this term is also irrelevant and therefore can be neglected , following similar discussions to the previous subsection .
the saddle - point @xmath179 is invariant under @xmath180 type transformation in eq .
( [ tralaw ] ) but not invarinat under @xmath181 .
this means that sp(@xmath1sp(@xmath2 ) symmetry is broken to sp(@xmath2 ) .
therefore , fluctuation around the saddle - point can be incorporated by parametrizing the order - parameter field @xmath182 as @xmath183 where @xmath184 and @xmath185 sp(@xmath2 ) , i.e. , @xmath186 , @xmath187 .
the field @xmath188 descirbes the goldstone mode around the saddle point .
then , the lagrangian becomes @xmath189 + \frac{\epsilon m}{2g'}{{\rm tr}}_{{w_{{\rm rt}}}}(u+u^\dagger)+\frac{2nm^2}{g ' } \nonumber\end{aligned}\ ] ] the last constant term will be neglected , since it vanishes in the replica limit @xmath190 . in the following subsection , we construct an effective theory for small fluctuation @xmath188 around the saddle point .
so far we have treated @xmath191 and @xmath192 as matrix fields to save complicated subscripts .
it has actually worked well especially for the discussions of the symmetry . however , in order to extract an effective lagrangian for @xmath188 via chiral gauge transformation introduced below , we can not treat the space @xmath51 and @xmath88 separately any longer . therefore , we switch into the standard notation for fermi fields as follows : note the following identity , for example , @xmath193 , where the last arrow means that we have made a @xmath194 rotation to @xmath191 and @xmath192 and we have regarded @xmath195 as a vector with @xmath196 components in @xmath197 .
then the fermion part of the lagrangian can be denoted as @xmath198 \chi .
\label{chistalag}\ ] ] although this lagrangian seems to be standard , the integration over @xmath192 is not straightforward .
it is because the elementary fields @xmath199 and @xmath49 are doubly included in @xmath191 and @xmath192 defined in eq .
( [ defchi ] ) .
it is possible , however , to rewrite this lagrangian as @xmath200 where @xmath201 and @xmath202 have @xmath203 components whose explicit subscripts in the above definition are those for @xmath51 .
it should be noted that this lagrangian is described by independent field variables @xmath201 and @xmath202 only . as a result
, we can integrate out fermi fields , which results in @xmath204 .
\label{ferdet}\ ] ] based on this result , we derive an effective action for the fluctuation @xmath188 in the next subsection .
suppose that the fluctuation @xmath205 around the saddle - point varies slowly with @xmath206 .
then , we can expand , with respect to derivatives , the determinant derived in the last subsection . to this end , make the folowing transformation @xmath207 here and hereafter , we often denote @xmath208 just as @xmath188 for simplicity . without confusion ,
we also use e.g. , @xmath64 instead of @xmath209 .
the lagrangian is now converted into @xmath210 where @xmath211 with @xmath212 integration over fermi fields @xmath213 and @xmath214 gives also a determinant , and the action for @xmath188 is @xmath215 where @xmath216 is a jacobian @xcite due to the transformation ( [ chitra ] ) .
it turns out that expansion of the determinant yields no second order terms with respect to the derivative , and hence we neglect the contribution from it in the leading order approximation . in what follows ,
we then concnetrate on the calculation of the jacobian .
we mention in passing that we will use the lagrangian ( [ chistalag ] ) to compute the jacobian , although ( [ simlag ] ) is less confusing .
let us consider the following eigenvalue problem for the dirac opeartor ( [ dirope ] ) to define and regularize the jacobian of the path integration , @xmath217 where right- and left - eigenvalue equations are introduced , since the dirac operator is non - hermitian .
the eigenfunctions are normalized via the inner product @xmath218 . assuming the completeness of the eigenfunctions , the fields @xmath191 and @xmath192
are expanded as @xmath219 the measure is then defined as @xmath220 it should be noted that if one uses the lagrangian ( [ simlag ] ) described by independent field variables @xmath201 and @xmath202 only , the measure is defined by @xmath221 as usual . in the present case , however , either @xmath222 or @xmath223 is enough , since each already includes all independent variables .
it seems to be unusual , to be sure , but we follow the calculation with the latter notation . using the definitions for the path integration measure , we can compute the jacobian for infinitesimal small change for the gauge transformation ( [ chitra ] ) , @xmath224 and @xmath225 with @xmath226 and @xmath227 as follows @xcite : @xmath228 where @xmath229 therefore , the measure changes as @xmath230 and @xmath231 , where @xmath232 stands for the space for the eigenmodes of the dirac operator . noting that @xmath233 valid for small @xmath234 , we reach @xmath235 where @xmath236 and @xmath237 . here
@xmath238 stands for the trace for @xmath206 and @xmath232 . in principle , we can choose either @xmath239 or @xmath240 as the jacobian in eq .
( [ actchitra ] ) if they can be regularized . in
what follows , we will apply standard regularization scheme in refs .
@xcite . nevertheless , it turns out that they still give rise to divergent terms for each @xmath241 .
fortunately , as we shall see below , they cancel each other in @xmath242 . therefore , it is suitable to define the jacobian in eq .
( [ chitra ] ) as a square root of the product of the naive jacobians as @xmath243 .
namely , @xmath244 it is possible to show that the jacobian based on @xmath221 for the lagrangian ( [ simlag ] ) indeed yields the same jacobian .
we now give the outline of how to calculate the jacobian following refs .
@xcite . using the definition of the @xmath238
, we have @xmath245 where ultraviolet cut - off @xmath246 is introduced , which will be set @xmath247 at the end of the calculation . using the relation @xmath248 ,
we have @xmath249 \nonumber\\ & = & \frac{1}{2}\int\frac{d^2k}{(2\pi)^2 } { { \rm tr}}_{v\otimes{{w_{{\rm rt}}}}}e^{-ikx}u^\dagger\delta u \left [ { { \cal p}}(e^{\not\partial\not d/\lambda^2 } + e^{\not d\not\partial/\lambda^2 } ) + ( e^{\not\partial\not d/\lambda^2 } -e^{\not d\not\partial/\lambda^2 } ) \right]e^{ikx } , \nonumber\end{aligned}\ ] ] where in the last line we have switched the basis for @xmath232 from the eigenmodes of the dirac operator into plane waves .
expanding @xmath246 and taking the limit @xmath247 , we reach @xmath250 .
\label{delgam}\ ] ] where @xmath251 and @xmath252 with @xmath253 = \epsilon_{\mu\nu}\sigma_3\otimes\sigma_3 , \quad g_{\mu\nu}=\frac{1}{2}\{\gamma_\mu,\gamma_\nu\ } = \delta_{\mu\nu}1\otimes1 . \label{sigg}\ ] ] therefore , taking trace in @xmath206- and @xmath51-space , we have @xmath254 it is nowadays well - known that the action function satisfying above includes not only the usual kinetic term but also the wzw term , @xmath255 with @xmath256 .
adding the symmetry - breaking term in ( [ actchitra ] ) , we end up with an action @xmath257 therefore , it turns out that the goldstone mode of the model with chiral symmetry of type @xmath63 can be described by sp(@xmath4 wzw model ( [ wzwact ] ) with a symmetry breaking term as in ( [ finact ] ) .
so far we have considered the case where hamiltonian has the chiral symmetry specified by @xmath63 in eq .
( [ chisym ] ) . as to the other cases , similar calculations lead to u(@xmath3)/sp(@xmath2 ) nonlinear sigma model , which is so - called chiral orthogonal class or @xmath6i , the same universality class as hwkm model .
we present a schematic phase diagram at the zero energy in fig . [
f : fig2 ] , which is valid for weak disorder , i.e , for small @xmath258 . in the case
@xmath259 , which is denoted by a white surface in the figure , the model belongs to class @xmath8i and is described by sp(@xmath4 wzw model with some irrelevant perturbations discussed in sec .
[ s : effpot ] and [ s : anieff ] .
the reason why the present wzw model is level-2 is that the continuum limit ( [ conpurham ] ) leads to double dirac fermions .
one of the main origin of the irrelevant interactions is that the lattice model has only on - site potentials without any hopping randomness .
namely , the model does not have maximum entropy .
another origin is the anisotropy @xmath110 , as discussed in sec .
[ s : anieff ] . besides , the continuum model should have basically many higher order irrelevant interactions . therefore , at the zero energy any points on this surface flow to the sp(@xmath4 wzw fixed point , denoted by a black circle in the figure .
it turns out that around the fixed point the density of states obeys the scaling law @xmath260 , following from the exact results obtained by knizhnik and zamolodchikov @xcite . on the other hand
, the universality class of the two surfaces @xmath261 and @xmath262 , shaded in the figure , is different .
they belong to the class of u(@xmath3)/sp(@xmath2 ) nonlinear sigma model or in other words , chiral orthogonal class or class @xmath6i , which is the same universality class as that of hwkm model .
it is well - known that the beta function of the renormalization group vanishes and the coupling constant of the nonlinear sigma model is not renormalized .
hence any points on these surfaces are critical .
the density of states is conjectured to be divergent at the zero energy , according to @xmath263 $ ] with a positive constant @xmath72 @xcite .
these surfaces , one @xmath8i and two @xmath6i s , are unstable : out the surfaces renormalization group flows are toward the strong - coupling limit of sp(@xmath3)/sp(@xmath1sp(@xmath2 ) nonlinear sigma model , as shown by fisher and fradkin for @xmath54 case .
therefore , all states away from the surfaces are expected to be localized .
we have studied the phase diagram of the disordered fermion model with random on - site potential defined on a two - dimensional square lattice with @xmath0-flux .
the continuum limit near the zero energy yields dirac fermions with random potentials specified by four independent coupling constants .
it has been argued that in addition to time - reversal symmetry , this model has so - called chiral symmetry in three subspaces in the space of the coupling constants . here
, chiral symmetry means that there exist a matrix anticommuting with the hamiltonian .
what is interesting is that this model alows three kinds of chiral symmetry , which is due to the species doubling of the lattice fermion .
it has been shown that one of them leads to sp(@xmath1sp(@xmath2 ) symmety of the lagrangian for generating functional of emsemble - averaged green functions whereas the others lead to u(@xmath3 ) symmetry , both of which is broken to sp(@xmath2 ) .
it has been found that the fluctuation around the saddle point is described , respectively , by sp(@xmath4 wzw model for the former and u(@xmath3)/sp(@xmath2 ) nonlinear sigma model for the latter .
based on these results , we have proposed a phase diagram .
the dirac fermion model with @xmath63 symmetry calculated in detail in the text is quite similar to that for the disordered @xmath7-wave superconductors with spin - ratation and time - reversal symmetries .
actually , effective lagrangians for these systems have sp(@xmath1sp(@xmath2 ) symmetry and hence both the models belong to the same universality class @xmath8i . however , fixed - points are quite different due to the wzw term
. it was shown @xcite that in the @xmath7-wave superconductors the wzw terms for several species of dirac fermions cancel each other and the resultant theory predicts localization of quasiparticles at the band center . contrary to this , the model treated in this paper is described by level-2 wzw model , which means that it has delocalized states .
the difference lies in eq .
( [ sigg ] ) : in the present model the wzw term survives owing to the relation @xmath264/2i=\epsilon_{\mu\nu}{{\cal p}}_1 $ ] , which is due to @xmath265 .
this is just the same definition of so - called @xmath266 in the particle physics literature , although the present @xmath34 matrices are not @xmath267 but @xmath157 .
conversely , we can say that chiral symmetry only in this sense yields the chiral anomaly and hence the wzw term , although we often use the same words for other similar symmetry in the anderson localization literature . considering the fact that the lattice model having @xmath63 chiral symmetry is realized in the case @xmath268 and @xmath269 in eq .
( [ latham ] ) , which is indeed a fine - tuned model , we conclude that the delocalization due to the existence of the wzw term seldom occures in nature .
email : [email protected] r. gade , nucl .
* b398 * , 499 ( 1993 ) ; r. gade and f. wegner , _ ibid .
_ , * b360 * , 213 ( 1991 ) . m. r. zirnbauer , j. math . phys .
* 37 * , 4986 ( 1996 ) .
a. altland and m. r. zirnbauer , phys
* b55 * , 1142 ( 1997 ) .
e. abrahams , p. w. anderson , d. c. licciardello , and t. v. ramarkrishnan , phys .
42 * , 673 ( 1979 ) .
a. furusaki , phys .
* 82 * , 604 ( 1999 ) , and references therein .
a. altland and b. d. simons , nucl .
phys . * b562 * , 445 ( 1999 ) .
s. guruswamy , a. leclair , and a. w. w. ludwig , cond - mat/9909143 .
m. fabrizio and c. castellani , nucl . phys . *
b583 * , 542 ( 2000 ) .
y. hatsugai , x .- g . wen , and m. kohmoto , phys . rev .
* b56 * , 1061 ( 1997 ) ; y. morita and y. hatsugai , phys . rev . lett . * 79 * , 3728 ( 1997 ) ; y. morita and y. hatsugai , phys . rev . * b58 * , 6680 ( 1998 ) . t. fukui , nucl .
b562 * , 477 ( 1999 ) .
a. a. nersesyan , a. m. tsvelik , and f. wenger , phys .
lett . * 72 * , 2628 ( 1994 ) .
k. ziegler , m. h. hettler , and p. j. hirschfeld , phys .
* 77 * , 3013 ( 1996 ) .
t. senthil , m. p. a. fisher , l. balents , and c. nayak , phys .
lett . * 81 * , 4704 ( 1998 ) : t. senthil and m. p. a. fisher , phys . rev . * b60 * , 6893 ( 1999 ) .
r. bundschuh , c. cassanello , d. serban , and m. r. zirnbauer , phys .
* b59 * , 4382 ( 1999 ) . t. senthil and m. p. a. fisher , phys . rev .
* b61 * , 9690 ( 2000 ) .
m. bocquet , d. serban , and m. r. zirnbauer , cond - mat/9910480 .
t. fukui , cond - mat/0002002 .
s. vishveshwara and m. p. a. fisher , cond - mat/0003018 .
p. fendley and r. m. konic , cond - mat/0003436 .
w. a. atkinson , p. j. hirschfeld , and a. h. macdonald , cond - mat/0002333 .
zhu , d. n. sheng , and c. s. ting , cond - mat/0005266 .
w. a. atkinson , p. j. hirschfeld , a. h. macdonald , and k. ziegler , cond - mat/0005487 .
a. altland , b. d. simons , and m. r. zirnbauer , cond - mat/0006362 .
a. m. m. pruisken , nucl . phys . *
b235 * , 277 ( 1984 ) ; h. levine , s. b. libby , and a. m. m. pruisken , _ ibid _ * b235 * , 30 , 49 , 71 ( 1984 ) .
m. p. a. fisher and e. fradkin , nucl .
b251 * , 457 ( 1985 ) .
a. w. w. ludwig , m. p. a. fisher , r. shankar , and g. grinstein , phys . rev . * b50 * , 7526 ( 1994 ) .
r. bundschuh , c. cassanello , d. serban , and m. r. zirnbauer , nucl .
b532 * , 689 ( 1998 ) .
k. fujikawa , phys . rev . * d21 * , 2848 ( 1980 ) ; _ ibid _ * d22 * , 1499 ( 1980 ) ( e ) l. alvarez - gaum and p. ginsparg , nucl . phys . * b243 * , 449 ( 1984 )
. v. g. knizhnik and a. b. zamolodchikov , nucl
* b247 * , 83 ( 1984 ) .
|
a fermion model with random on - site potential defined on a two - dimensional square lattice with @xmath0-flux is studied . the continuum limit of the model near the zero energy yields dirac fermions with random potentials specified by four independent coupling constants .
the basic symmetry of the model is time - reversal invariance .
moreover , it turns out that the model has enhanced ( chiral ) symmetry on several surfaces in the four - dimensional space of the coupling constants .
it is shown that one of the surfaces with chiral symmetry has sp(@xmath1sp(@xmath2 ) symmety whereas others have u(@xmath3 ) symmetry , both of which are broken to sp(@xmath2 ) , and the fluctuation around a saddle point is described , respectively , by sp(@xmath4 wzw model and u(@xmath3)/sp(@xmath2 ) nonlinear sigma model .
based on these results , we propose a phase diagram of the model .
|
the possibility of information loss in black hole evaporation has been with us for many years . there have been many attempts over the years to resolve the issue one way or another , but it is probably fair to say that there is still no consensus on what the correct picture of a black hole and its evolution should be . indeed , the possibility that the evolution of a black hole may be a non - unitary process has many disturbing consequences .
we will not attempt here to provide a definitive answer to the question of whether black hole evaporation is unitary or not .
we believe that this is a question that can only be answered in the context of a theory that we do not yet possess .
instead what we want to do here is to highlight some of the semi - classical issues involved in the problem and the implicit assumptions that are often under - appreciated in the analysis .
specifically we wish to examine the role that the event horizon plays and its connection to the global causal structure of spacetime .
it is precisely the non - trivial causal structure related to event horizons that leads to problems with unitarity . because we can `` throw away '' entangled particles in an absolute sense , we can have pure states evolving into mixed states . to this end
we wish to argue that there are relatively few options for doing away with information loss other than doing away with event horizons .
however , even if the spacetime does not contain event horizons it is possible to have a physically meaningful notion of a black hole . to replace event horizons
, we propose a formalism for black holes that is already well established in the literature , based on marginally trapped surfaces . to free us from this event horizon constraint we will use a quasi - local definition of the black hole , in terms of trapping horizons . at least at the semi - classical level
, trapping horizons will be useful whatever the ultimate resolution of the issue of black hole evolution is . to clarify what one might expect from black hole evaporation and give as clear a picture as possible , without solving the full back - reacting semi - classical einstein equations with renormalised stress - energy tensor , we present here a simple , spherically symmetric picture making use of the misner - sharp mass function .
in spherical symmetry the misner - sharp mass function can be interpreted as measuring the amount of energy within a sphere of areal radius @xmath0 at a time @xmath1 .
therefore the misner - sharp mass function gives a reasonable , quasi - local definition to the concept of the curvature - producing energy contained within a black hole .
we can use the misner - sharp mass to locate black hole horizons .
it can give us insight into black hole thermodynamics and clarify our picture of what is happening when the black hole evaporates .
this paper is organised as follows . in section [ sec : infoloss ] we review the standard arguments for how black holes can be used to turn pure states into mixed states .
we emphasize the role played by the event horizon and by extension the non - trivial causal structure related to it , usually in terms of a spacelike singularity . in section
[ trapping ] we give a brief review of trapping horizons in spherical symmetry and how they give rise to black hole thermodynamics and hawking radiation without any event horizon necessary . in section [ misnersharp ]
we describe how the misner - sharp mass can be used in spherical symmetry to follow the evolution of the black hole spacetime and how the various possible outcomes can be explained in terms of it .
information loss in black holes is related to the possibility of pure states evolving into fundamentally mixed states ( for example see @xcite ) .
a fundamentally mixed state is one that can not be turned into a pure state by finer - graining or further measurements . in a black hole context
, the mutation of pure states into mixed states comes about most strikingly in the context of hawking radiation .
essentially virtual pair - production can lead to hawking radiation if one of the virtual particles falls into the black hole , allowing the other particle to escape to infinity with real positive energy .
the two particles are initially entangled with one another and are described by an entangled state [ hawkingstate ] |= c e^b_1^c_1^ | 0 , where @xmath2 is a normalisation factor and @xmath3 is a creation operator for the particles that escape to infinity and @xmath4 is a creation operator for particles that fall into the black hole . for simplicity we can write a similar state as ( [ hawkingstate ] ) if we just ignore all two and more particle states in the expansion of @xmath5 .
this state is entangled because we can not write it as a factored state of the form @xmath6 .
while we still have both particles we can write the density operator for the system as a pure state _ pure = || & = & | 0_b_1| 0_c_1 0 |_b_1 0 |_c_1 + & & + | 0_b_1| 0_c_1 1 |_b_1 1 |_c_1 + & & + | 1_b_1| 1_c_1 0 |_b_1 0 |_c_1 + & & + | 1_b_1| 1_c_1 1 |_b_1 1 |_c_1 .
however , if we ` ignore ' the particle that falls into the black hole , then we must sum or partial trace over the unknown state and we obtain a mixed state _ mixed = | 0 _ b_10 |_b_1 + | 1 _
b_11 |_b_1 .
this state is mixed because it can not be written as single state @xmath7 in the form @xmath8 . in ordinary quantum systems
this procedure of taking the partial trace over unmeasured subsystems is common and leads to ` statistically ' mixed states .
the subsystem we focus on is described by a mixed state but this is only because we choose to focus on a part of the entire system .
this subsystem is entangled with some other subsystem and the entire state , including the subsystem we choose not to measure , is still a pure state . if we chose to measure the full system we would recover the full pure quantum state .
however , when one of the subsystems has fallen over an event horizon or perhaps more importantly , has reached a boundary of spacetime and terminated , then there is no way , even in principle , that we could measure the state of the entire system , because one of the subsystems will never return to our past light cone .
for all intents and purposes it is gone and we can not choose to measure it sometime in the future . therefore we must take a trace over the lost subsystem and the remaining state that is accessible to us will be a fundamentally mixed state .
it is this tracing over the unknowable degrees of freedom that leads to pure states turning into fundamentally mixed states .
while both particles still exist one can claim that the full quantum state is pure .
one could even claim that a particle that had fallen over the event horizon was still in principle measurable . however , if there is a spacetime boundary inside the black hole , such as a singularity , then the particle will reach the end of its worldline in finite proper time . after this
it will cease to exist .
if the black hole evaporates completely then the outgoing hawking radiation particles will be entangled with nothing and they must represent a mixed state . whether one wants to associate
the tracing over the unknowable states with them falling over the event horizon or with them falling into a singular spacetime boundary and ending is largely a matter of taste .
particles that fall over the event horizon can not , by definition , return to the exterior asymptotic region , unless they move acausally . in the classical picture of a schwarzschild black hole , particles that fall into black holes
reach the singularity in a finite proper time and at the singularity they meet a spacetime boundary where their worldlines can not be continued . in a more realistic , perhaps quantum description , whether their worldines actually come to an end at some finite parameter time depends on the full causal structure and the details of what one believes happens in the singular regions of general relativity
. a similar thing can happen if we allow naked timelike singularities .
one of the entangled particles can fall into the singularity and disappear while the other particle escapes to infinity and becomes a fundamentally mixed state .
this would appear to provide a mechanism for turning pure states into mixed states without event horizons , although the existence of a naked singularity would cause other problems as well . in some sense black holes
are most commonly associated with information loss , only because the horizon readily creates the entangled particles and the spacetime singularity / boundary allows one of the particles to be erased .
the possibility of pure states turning into mixed states is therefore related to the possibility of the worldlines of particles coming to an end .
this suggests that in order to understand whether information can be lost in black holes one must look at the possible endings of particle worldlines .
there are in fact several possibilities for worldlines .
worldlines will either end at finite parameter time , in which case we would say they have reached a spacetime boundary , or they can be continued to infinite parameter time , in which case we would say that they end at infinity . if continued to infinite parameter time they can either end up in the same infinity of external observers or some other infinity , as occurs for example , for certain worldlines in the reissner - nordstrm spacetime .
the are other possibilities for worldlines that are slightly less clearcut .
one can consider the possibility that worldlines become undefined and it is meaningless to talk about their continuation .
this may occur if spacetime because somehow ` fuzzy ' because of quantum effects .
one can also consider the possibility that some new physics conspires to induce a breakdown of locality , or allows particles and observers to move along spacelike curves , so that causal penrose diagrams are largely meaningless .
this might occur for example if the interior boundary is defended by some membrane - like structure that prevents all freely - falling particles from passing through itself , along their geometric geodesic paths @xcite . in most situations
a spacelike boundary will give rise to an event horizon .
this is the case with the spacelike singularity in the schwarzschild solution .
the spacelike boundary casts a shadow on light rays continued back from future infinity .
of course , spacelike singularities are not a necessary condition for event horizons .
the reissner - nordstrm spacetime has no spacelike boundaries but does have event horizons . the singularities in the reissner - nordstrm spacetime are timelike and occur because of the disjoint null infinity structure .
if we want to avoid pure states evolving into mixed states we must not throw away the ingoing hawking particles .
if the ingoing particles reach a boundary in finite parameter time there will be hypersurfaces in the spacetime that their worldlines do not intersect and it is hard to argue that they should not be ` thrown away ' .
there are several possibilities to avoid this outcome .
the first , perhaps most severe approach , is to deny the physical existence of black holes .
this can be done , for example , by arguing that some non - trivial physics comes into play when matter attempts to collapse to form a black hole , as in for example @xcite .
the second approach is to deny the existence of a spacetime boundary within the black hole that entangled particles can fall over .
this is for example the purely classical proposal of dymnikova @xcite and hayward @xcite and the quantum geometrical proposal of @xcite .
the possibility that one could have horizons without singularities seems to run foul of the singularity theorems but these can either be circumvented with non - global hyperbolicity @xcite or energy condition violation @xcite or a breakdown of the differentiable manifold picture @xcite .
violation of the energy conditions is perhaps not as drastic as it may sound , because hawking radiation is known to violate nearly all of the energy conditions @xcite .
if one follows the approach of denying the existence of spacetime boundaries within black holes , but one does not want to deny the existence of a black hole , one is left with the question of what one means by the black hole region .
traditionally , one defines the black hole using an event horizon .
if there is no internal boundary within the black hole and no wormholes that connect the interior region to a different disjoint asymptotic region , since the worldlines of particles have to end somewhere , the only place left for them to end is at asymptotic future infinity .
this means that technically there will be no event horizon .
if there is no event horizon , then how should one define the black hole ? here
we suggest that one simply adopt the proposal of using some quasi - local definition of the black hole in terms of for example , a trapping horizon .
trapping horizons are defined quasi - locally in terms of the geometrical structure and are independent of the global causal structure . to see how trapping horizons provide an acceptable definition of black holes and even give rise to black hole thermodynamics and hawking radiation , we give here a brief summary of trapping horizons in spherical symmetry .
further details can be found in the literature , see @xcite and references therein . while spherical symmetry is not a reasonable assumption for astrophysical black holes it simplifies the discussion sufficiently to justify its adoption for ` conceptual clarity ' .
spherical symmetry is a reasonable assumption to study the formation and evaporation of black holes at a toy model level and it makes the analysis much simpler and clearer .
furthermore , in spherical symmetry , we can use the misner - sharp mass as a definition of quasi - local mass , which as we will see , also provides conceptual clarity to the issue of black hole evaporation .
a trapping horizon , more properly a future outer trapping horizon , is defined by hayward @xcite as the closure of a three - surface which is foliated by marginal surfaces , for which @xmath9 , and which , in addition , satisfies 1 .
@xmath10 ( to distinguish between white holes and black holes ) .
2 . @xmath11 ( to distinguish between inner and outer horizons of , for example , the non - extremal reissner - nordstrm solution ) .
any spherically symmetric metric in four dimensions can be put in the form s^2 & = & - e^-2(t , r)(1-)t^2 + + & & + r^2 ^ 2 , in so - called schwarzschild coordinates , where @xmath12 is immediately recognizable as the misner - sharp mass function .
the misner - sharp mass function can be interpreted as the quasi - local mass contained within a sphere of radius @xmath0 at a time @xmath13 .
unfortunately , this coordinate system is undefined at @xmath14 and can not be continued through this hypersurface and can not be relied upon for calculating properties at @xmath15 .
a more appropriate coordinate system is perhaps the -gullstrand coordinate system s^2 & = & -e^-2(,r)(1-)^2 + + & & 2e^-(,r)r + r^2 + r^2 ^ 2 .
this coordinate system is regular at future horizons .
the radial null geodesics for this metric can be easily found by setting @xmath16 .
for this we find [ dtdtau ] = -e^-(,r ) ( 1 ) , where the plus sign denotes the ingoing geodesics .
thus we can find outgoing geodesics @xmath17 and ingoing geodesics @xmath18 with components [ l ] l^a = ( e^(,r),1-,0,0 ) , [ n ] n^a = ( e^(,r),-1-,0,0 ) .
the factor of two ensures that the cross normalisation is the conventional @xmath19 .
then we can compute _
l = ( 1- ) , _
n = -(1 + ) .
we see that the expansion of @xmath18 is always negative and that at @xmath20 the expansion of @xmath17 is zero .
we can also compute the value of @xmath21 at @xmath15 ( n^a_a_l)_h = - ( 1 + ) , where a subscript @xmath22 denotes functions to be evaluated at the horizon and we have used a dash to denote partial derivative with respect to @xmath0 and a dot to denote the partial derivative with respect to the time @xmath1 ( here , since @xmath23 is only a function of @xmath1 it is actually an ordinary derivative ) . for the horizon to be an outer horizon in a spacetime with a regular asymptotic region ,
we require @xmath24 , since @xmath25 must be less than @xmath0 for large @xmath0 . in addition
, we can see from ( [ dtdtau ] ) for the ingoing null geodesic @xmath18 that @xmath26 .
thus we see that we have a trapping horizon at @xmath15 if the horizon is outer and not moving inwards faster than ingoing null geodesics .
the normal @xmath27 to the surface @xmath15 has norm n^an_a = -4_h^2e^2_h-4_he^_h(1 - 2m_h ) .
if @xmath28 the trapping horizon will be a null hypersurface , and , assuming @xmath29 , it will be a spacelike hypersurface if @xmath30 .
for @xmath31 the trapping horizon will be a timelike hypersurface .
this opens the possibility that one can move along a causal curve from inside an evaporating horizon to the outside . for @xmath32
the horizon is spacelike , but evaporating ` faster than the speed of light ' and so all timelike curves from a region just inside the horizon must move to the outside @xcite . the surface @xmath33 does not however , define the location of the event horizon in a dynamical spacetime .
the event horizon is always a null surface and so the spherically symmetric trapping horizon at @xmath15 can only be an event horizon if @xmath34 ( note however that this is necessary but not sufficient ) . to find the event horizon ,
firstly one would need an explicit form for @xmath35 and then one would look for radial null vectors that are not able to reach infinity by propagating them outwards from the centre of the spacetime .
it has been well - known for a long time that locally defined horizons can give rise to equations of black hole thermodynamics @xcite . at the horizon
we have [ firstlaw ] = , where @xmath36 and @xmath1 is a parameter labeling ` time - slicings ' of the horizon .
this equation has the same form as the first law of black hole thermodynamics @xmath37 with a surface gravity that agrees with other definitions of surface gravity @xcite . in order to obtain a version of the second law we can just compute @xmath38 , where @xmath39 is the einstein tensor .
this gives g_abl^al^b = - ( 1-)^2 . rearranging , and imposing ( [ firstlaw ] ) at @xmath20 we find [ secondlaw ] = g_abl^al^b .
thus we see that the area of the horizon @xmath40 is increasing if @xmath41 . by the einstein equations we can write this condition as @xmath42 , which is just the null energy condition ( nec ) .
the area of the horizon is increasing if the nec is satisfied and can decrease only if the nec is violated .
furthermore , using the parikh - wilczek tunneling approach to hawking radiation we may expect to have hawking radiation produced in the vicinity of the trapping horizon @xcite .
this is perhaps not surprising , since the effect is just a result of quantum field theory on a curved background and quantum field theory is a local theory that is unlikely to depend on global properties like an event horizon @xcite .
consider the equation for a massless scalar field on a curved background _ a ( g^ab_b)= 0 .
we look for solutions of the form @xmath43 . taking this limit as @xmath44 , to lowest order this equation gives the hamilton - jacobi equation g^ab_as_bs
= 0 . invoking the geometrical optics approximation s(,r ) =
t - k(r)r , this equation gives ^2 + 2e^-k - e^-2(1-)k^2 = 0 .
rearranging gives k = .
the plus sign denotes the outgoing modes and the negative sign denotes the ingoing modes .
the outgoing modes contain a pole at the horizon @xmath15 .
the integral can be performed by deforming the contour into the lower half of the complex plane , which gives a complex part to @xmath45 s = . at this level of approximation
this corresponds to thermal radiation with a temperature t = ( 1 - 2m_h ) , which agrees with the calculations in @xcite .
and mass @xmath46 . ] ] the misner - sharp mass @xcite can be very useful for investigating the behaviour of black holes @xcite . for a spherically symmetric metric ,
using the einstein equations , we have [ t00 ] t_0 ^ 0 = , where @xmath47 denotes partial differentiation with respect to the coordinate @xmath0 .
this will correspond to the local energy density @xmath48 as measured by an observer with constant coordinate @xmath0 .
hypersurfaces of constant -gullstrand time @xmath1 are spacelike since the normal to the surface @xmath49 satisfies ^a_a= -e^2(,r ) . on a @xmath50 hypersurface
@xmath51 we can write a quasi local misner - sharp mass as m = _ v , where @xmath52 is the volume element on the hypersurface . in the words of misner and sharp @xcite , the function @xmath53 includes contributions from the kinetic energy and gravitational potential energy .
since the volume element is just @xmath54 we obtain m = m r = m. the misner - sharp mass reduces to the arnowitt - deser - misner ( adm ) mass at spacelike infinity and reduces to the bondi - sachs mass at null infinity . in the newtonian limit
is also gives the newtonian mass to first order @xcite .
we can also see from ( [ t00 ] ) that if the local energy density is negative then @xmath47 will also be negative .
a negative energy density corresponds to a violation of the weak energy condition ( wec ) .
using the page approximation @xcite for the renormalised energy - momentum tensor in the hartle - hawking vacuum of a schwarzschild black hole we know that hawking radiation is expected to violate the wec and nec @xcite .
thus we might expect , in a fully backreacting spacetime , that in the region near the black hole horizon @xmath47 should be negative .
in fact , we know that the null energy condition must be violated in the vicinity of an evaporating , area decreasing horizon ( [ secondlaw ] ) .
this also means that the surface gravity and therefore temperature of the black hole can be greater than the corresponding temperature of a schwarzschild black hole with the same mass @xcite .
the misner - sharp mass allows us to construct the following heuristic picture of black hole formation and evaporation . by taking equal time
snapshots " of the misner - sharp mass function on various constant @xmath1 coordinate slices we can imagine the evolution of a star as it undergoes gravitational collapse to form a black hole and how hawking radiation evaporates away the mass - energy of the star .
an important feature of this picture is the @xmath15 condition that allows us to see the location of the trapping horizon boundary of the black hole .
this is indicated in the plots by a red line .
initially , fig.[mstar ] , the star has some stellar mass profile , without necessarily any prejudice as to whether the density goes like @xmath55 as depicted here , and vacuum outside . at this time , nowhere is the condition @xmath15 reached . as the star collapses , a horizon forms , initially with a small radius , fig.[mstarii ] .
as the collapse proceeds the area of the horizon grows , fig.[mstariii ] .
hawking radiation is expected to occur in the vicinity of the horizon , fig.[mstariv ] .
this means that positive energy flows outwards and a corresponding negative energy flows inwards .
the size of this effect for a macroscopically sized black hole is greatly exaggerated here . as the hawking evaporation proceeds , the horizon area shrinks , fig.[mstarv ] .
the flux of positive energy moves outwards to greater and greater radius , and the flux of negative energy moves inwards to the centre .
eventually , sufficient energy has evaporated away from the black hole to leave the intermediate spacetime effectively flat , fig.[mstarvi ] .
an observer in this region would feel almost no gravitational field from the collapsed star . how the ingoing flux of negative energy interacts with the interior mass of the star depends on the details of the calculation at small radii .
this is the region where the matter of the originally collapsed star is compressed to great densities and the region where one might expect quantum gravitational effects to be important .
if the negative energy and original star matter annihilate completely , one will end up with a spacetime that is effectively flat minkowskian space , fig.[mstarvii ] .
if we want to view the collapse purely classically then the boundary term at @xmath56 is crucial . in the above plots we have imposed @xmath57 as was done , for example , in @xcite . to investigate the behaviour at @xmath56 we can compute the kretschmann scalar for the simple static case with @xmath58 .
if we take @xmath59 to leading order we get r_abcdr^abcd ~(n^4 - 6n^3 + 17n^2 - 20n + 12 ) . this quartic equation in @xmath60
has no real roots and therefore if @xmath59 with @xmath61 we will get a singularity at @xmath56 in the kretschmann scalar . of course , a finite kretschmann scalar is necessary but not sufficient for there to be no singularity at @xmath56 . for the schwarzschild solution we have the condition @xmath62 for all @xmath1 and @xmath0 , fig.[mstarschwarz ] .
this suggests that the collapse may proceed in the way depicted in fig.[mstarsing ] .
if @xmath63 at @xmath56 then we obtain a spacelike singularity at @xmath56 .
indeed , it is even possible that we could have @xmath64 at @xmath56 in which case we would have an untrapped ( weakly naked ) timelike singularity @xcite .
, for all values of @xmath1 and @xmath0 . ] . ]
the final thing that it is easy to show using the misner - sharp mass is the creation of an inner and outer horizon , as for example considered in @xcite .
the simple condition for this to occur is that m > , at the inner horizon where @xmath15 . in hayward s scenario
the trapping horizon forms initially as an extremal horizon at a finite radius .
then the inner and outer horizons separate to form a trapped region ( with the region close to @xmath56 untrapped ) and then the inner and outer horizons join up again forming an instantaneously extremal horizon before disappearing altogether .
this is shown in fig.[mstardouble ] .
the region between the inner and outer horizons contains trapped surfaces , while the region inside the inner horizon does not .
we have seen how the causal structure of spacetime plays an important role in the semi - classical information loss argument . in this way ,
information loss is very closely tied up with defining black holes via event horizons .
one often reasons that in order to have a black hole one must have an event horizon . in order to have an event horizon one must have a non - trivial causal structure .
if one has a non - trivial causal structure then one always has the possibility of information loss .
current notions about what constitutes a black hole are , of course , a result of the history of the development of the subject .
the early days of finkelstein , wheeler and others started out with a purely classical picture of a black hole based largely on exact solutions , which by their very nature , required a large amount of symmetry in order to be discovered as solutions of the einstein equations .
penrose , hawking and others then laid down the fundamental concepts of what it means to be a black hole as a region that one can never escape from and developed techniques for dealing with situations beyond exact solutions .
then the laws of black hole thermodynamics were discovered , still in the context of black holes that eventually settle down to just being black before finally , hawking discovered black hole radiation and the possibility that black holes could disappear arose . had hawking radiation been discovered at an earlier stage , the community might not be so insistent on defining black holes in terms of event horizons .
we have presented a picture that hopefully clarifies the issue of how event horizons are related to information loss .
it is not based on an exact solution to the semi - classical einstein equations but gives the properties one might expect from a full solution .
it is also important to recognise what assumptions have gone into the above picture .
firstly , we have assumed spherical symmetry .
this greatly simplifies the formalism and allows us to use the misner - sharp mass as a quasi - local measure of energy .
it is of course , not a valid model for describing astrophysical black holes , but it is a valid assumption to study black hole evaporation in its purest form . it is not expected that deviations from spherical symmetry will qualitatively change the physics of black hole evaporation . secondly , we have chosen to define the black hole in terms of a trapping horizon .
this is not the usual definition of a black hole , but arises from the desire to define black holes without appealing to event horizons .
there are many issues that remain to be clarified about the use of trapping horizons and they may ultimately turn out not to be the optimal choice .
questions about the uniqueness of trapping horizons have not been studied much in the literature and one can certainly ask questions about how a spacelike trapping horizon should intersect a given constant time " spatial hypersurface .
thirdly , we have of course assumed that it is meaningful to talk about the matter that falls into the black hole , and this irrespective of whether it is the original collapsing star or matter that is subsequently accreted .
this aspect is often ignored in other studies , partly because whatever falls through the event horizon , can , by definition , be ignored in terms of its effect of the outside region and secondly because there is in some quarters an implicit belief that whatever the state of the interior black hole is , it can somehow be read off from physics at the horizon . here
we argue instead that the collapse of the central matter is essential to determining the causal structure of the spacetime and this is ultimately responsible for determining whether one can turn pure states into mixed states .
black holes can not be truly understood until we understand what happens to all the neutrons ( and protons and other particles ) of the original star as they collapse under gravity to ever higher densities .
it is perhaps tempting to think that since quantum gravity effects are only likely to come into play in regions of high curvature , that the existence of an event horizon , since it occurs in a region of relatively low curvature , is unlikely to be affected .
however , it is important to remember that the event horizon is not defined in terms of the local geometry , but causally in terms of the global causal structure . if the collapse is somehow halted or a singularity avoided then small changes in the microscopic central region can have large effects on the global causal structure , in the same way that adding one electron s worth of charge to an otherwise supermassive black hole can change its causal structure from schwarzschild to reissner - nordstrm .
this change , while insignificant in terms of local physics , corresponds to a radical change of the causal structure , turning spacelike singularities into timelike singularities amongst other things .
the picture presented above also downplays several features commonly associated with information loss in black holes .
firstly , the uniqueness theorems ( often called no - hair theorems ) that hold for stationary , asymptotically flat , electro - vac , einstein - maxwell black hole spacetimes @xcite play little role . nor does the issue of whether the hawking radiation is exactly thermal or not .
all that is essential is that previously entangled particles become entangled with nothing when their entangled
partners " are thrown away .
one could imagine a black hole spacetime with infinitely many independent measurable charges at infinity @xcite , if the spacetime still has a spacetime boundary inside the black hole , over which entangled particles can be lost , then there is still the possibility of information loss .
one could also imagine a situation with a single pair of entangled particles .
if one of them is lost to a spacetime boundary the other particle will be in a fundamentally mixed state , and a single particle can hardly be said to describe a thermal spectrum . another related feature to uniqueness is the possibility of retrodicting the original state that collapsed to form the black hole from measurements of the final state .
this should also be conceptually separated from the loss of unitarity in black holes .
such loss of retrodictablility occurs in ordinary particle physics where one knows that the s - matrix is perfectly unitary .
for example , since both electron - positron pairs and muon - anti muon pairs can co - annihilate to produce high energy photons , if one just observes the high energy photons at late time , one can not retrodict whether they arose due to electron - positron co - annihilation or a different particle - antiparticle pair .
another implicit assumption that one often meets is that the evaporating hawking radiation somehow is linked to the originally collapsed matter .
this is because energy is conserved and if the hawking radiation is carrying away energy , then that energy must somehow be the energy of the original star .
what we have shown using the misner - sharp mass function is that this need not be the case .
one can think of the hawking radiation as a pure quantum field effect on a curved background .
it is the spacetime that it evaporating and not the collapsed star .
the original mass of the star can remain clumped at the centre ( in some as yet unknown state ) while the hawking effect proceeds to produce an outgoing flux of energy from the vicinity of the horizon .
to conclude we highlight some of the outstanding questions that remain to be answered .
what happens to the collapsing matter as it gets compressed to ever higher densities ?
what is the physics that is truly responsible for generating the hawking radiation and does it effect the evolution of the black hole , and how does the negative energy produced in the hawking radiation interact with the collapsed , highly compact remnant of the star ?
we believe that the answers to these questions will be needed before we can truly say we understand black hole evolution .
s. d. mathur , arxiv:0803.2030 [ hep - th ] .
d. h. yeom and h. zoe , arxiv:0802.1625 [ gr - qc ] .
g. chapline , e. hohlfeld , r. b. laughlin and d. i. santiago , int .
j. mod .
phys . a * 18 * ( 2003 ) 3587 [ arxiv : gr - qc/0012094 ] .
p. o. mazur and e. mottola , proc .
* 101 * ( 2004 ) 9545 [ arxiv : gr - qc/0407075 ] . t. vachaspati , d. stojkovic and l. m. krauss , phys . rev .
d * 76 * ( 2007 ) 024005 [ arxiv : gr - qc/0609024 ] .
s. d. mathur , fortsch .
* 53 * ( 2005 ) 793 [ arxiv : hep - th/0502050 ] .
i. dymnikova , _
8th marcel grossmann meeting on recent developments in theoretical and experimental general relativity , gravitation and relativistic field theories ( mg 8) , jerusalem , israel , 22 - 27 jun 1997 _
s. a. hayward , phys .
* 96 * ( 2006 ) 031103 [ arxiv : gr - qc/0506126 ] .
a. ashtekar and m. bojowald , class .
* 22 * ( 2005 ) 3349 [ arxiv : gr - qc/0504029 ] .
s. ansoldi , arxiv:0802.0330 [ gr - qc ] .
m. visser , phys .
d * 54 * ( 1996 ) 5103 [ arxiv : gr - qc/9604007 ] .
a. b. nielsen and m. visser , class .
* 23 * ( 2006 ) 4637 [ arxiv : gr - qc/0510083 ] .
s. a. hayward , phys .
d * 49 * ( 1994 ) 6467 .
w. collins , phys .
d * 45 * ( 1992 ) 495 a. b. nielsen and j. h. yoon , class .
* 25 * ( 2008 ) 085010 [ arxiv:0711.1445 [ gr - qc ] ] . m. visser , int . j. mod .
d * 12 * ( 2003 ) 649 [ arxiv : hep - th/0106111 ] .
r. di criscienzo , m. nadalini , l. vanzo , s. zerbini and g. zoccatelli , phys .
b * 657 * ( 2007 ) 107 [ arxiv:0707.4425 [ hep - th ] ] .
t. clifton , arxiv:0804.2635 [ gr - qc ] .
a. b. nielsen , arxiv:0802.3422 [ gr - qc ] .
p. szekeres , j. math .
* 6 * ( 1965 ) 1387 . c. w. misner and d. h. sharp , phys .
rev . * 136 * ( 1964 ) b571 .
s. a. hayward , phys .
d * 53 * ( 1996 ) 1938 [ arxiv : gr - qc/9408002 ] .
d. n. page , phys .
d * 25 * ( 1982 ) 1499 .
m. heusler , living rev .
* 1 * ( 1998 ) 6 .
b. m. leith and a. b. nielsen , phys .
d * 76 * ( 2007 ) 124024 [ arxiv:0709.2541 [ gr - qc ] ] .
|
understood in terms of pure states evolving into mixed states , the possibility of information loss in black holes is closely related to the global causal structure of spacetime , as is the existence of event horizons .
however , black holes need not be defined by event horizons , and in fact we argue that in order to have a fully unitary evolution for black holes , they should be defined in terms of something else , such as a trapping horizon .
the misner - sharp mass in spherical symmetry shows very simply how trapping horizons can give rise to black hole thermodynamics , hawking radiation and singularities .
we show how the misner - sharp mass can also be used to give insights into the process of collapse and evaporation of locally defined black holes .
|
bragg gratings ( bgs ) are broadly used as a basis for spectral filters are other elements of optical telecommunication networks @xcite .
simultaneously , bgs offer a unique medium for the theoretical @xcite @xcite and experimental @xcite study of _ gap solitons _ ( gss ) , supported by the balance between the bg - induced dispersion and kerr ( cubic ) nonlinearity dispersion .
the latter topic was surveyed in an earlier review @xcite and recent one @xcite .
matter - wave pulses similar to the optical bg solitons were very recently observed in bose - einstein condensates @xcite .
the bg solitons may exist not only as temporal pulses in fiber gratings , but also as spatial solitons in 2d ( two - dimensional ) waveguides @xcite . in the latter case ,
the bg is represented by a set of parallel grooves on the surface of the waveguide .
a generalization , that makes it possible to increase the number of waves coupled by the bg , is based on a superposition of more than one bgs in the spatial domain . in particular , a system of three gratings which form a triangular lattice was considered in ref .
it couples three waves , giving rise to new types of stable gss for instance ,
peakons " . in recent papers @xcite and @xcite , a model of a _
cross - grating _ was introduced , as a superposition of two mutually transverse bgs written on a 2d waveguide with the kerr nonlinearity . in this case , one is dealing with a system of four waves .
the evolution of their amplitudes @xmath3 , @xmath4 , obeys the system of normalized equations @xcite , @xmath5 v_{1}+2v_{2}v_{3}v_{4}^{\ast } & = & { \small 0 , } \nonumber \\ i\frac{\partial v_{2}}{\partial t}-\frac{i}{\sqrt{2}}\frac{\partial v_{2}}{\partial z}+\frac{i}{\sqrt{2}}\frac{\partial v_{2}}{\partial x}+v_{1}+\mu v_{4}+\left [ \left\vert v_{2}\right\vert ^{2}+2(\left\vert v_{1}\right\vert ^{2}+\left\vert v_{3}\right\vert ^{2}+\left\vert v_{4}\right\vert ^{2})\right ] v_{2}+2v_{1}v_{4}v_{3}^{\ast } & = & 0 , \nonumber \\ i\frac{\partial v_{3}}{\partial t}+\frac{i}{\sqrt{2}}\frac{\partial v_{3}}{\partial z}-\frac{i}{\sqrt{2}}\frac{\partial v_{3}}{\partial x}+v_{4}+\mu v_{1}+\left [ \left\vert v_{3}\right\vert ^{2}+2(\left\vert v_{1}\right\vert ^{2}+\left\vert v_{2}\right\vert ^{2}+\left\vert v_{4}\right\vert ^{2})\right ] v_{3}+2v_{1}v_{4}v_{2}^{\ast } & = & 0 , \nonumber \\ i\frac{\partial v_{4}}{\partial t}-\frac{i}{\sqrt{2}}\frac{\partial v_{4}}{\partial z}-\frac{i}{\sqrt{2}}\frac{\partial v_{4}}{\partial x}+v_{3}+\mu v_{2}+\left [ \left\vert v_{4}\right\vert ^{2}+2(\left\vert v_{1}\right\vert ^{2}+\left\vert v_{3}\right\vert ^{2}+\left\vert v_{2}\right\vert ^{2})\right ] v_{4}+2v_{2}v_{3}v_{1}^{\ast } & = & 0 .
\label{tzx}\end{aligned}\]]here , @xmath6 is time , @xmath7 and @xmath8 are the coordinates in the waveguide , the bragg reflectivity ( strength ) of one grating is normalized to be @xmath9 , and @xmath10 is a relative strength of the second grating
. coefficients in front of the nonlinear terms in eqs .
( [ tzx ] ) , which represent the self - phase - modulation , cross - phase - modulation and four - wave - mixing , correspond to the normal kerr effect in isotropic materials .
the same model finds a different interpretation in terms of a 3d photonic crystal with a transverse chess - board structure @xcite . in that case ,
( [ tzx ] ) , with @xmath6 replaced by the third coordinate @xmath11 , along which the medium is homogeneous , govern the spatial evolution of the time - independent fields . in the latter context , a model which is tantamount to the particular symmetric case of the system ( [ tzx ] ) , with @xmath12 ,
was earlier introduced in ref .
@xcite ( in fact , this case is a degenerate one , as it corresponds to the gap in the model s linear spectrum collapsing to a point @xcite ) .
@xcite was dealing with 2d solitons , treating their stability by means of a method which is tantamount to the known vakhitov - kolokolov ( vk ) criterion @xcite .
in fact , this method is not sufficient for the verification of the stability , and our direct simulations ( to be reported elsewhere ) demonstrate that _ all _ the 2d solitons , both in the general system ( [ tzx ] ) and in its special case with @xmath12 that was considered in ref .
@xcite , are _
unstable_. a simpler possibility is to consider 1d solitons , which are generated by the substitution @xmath13 in eqs .
( [ tzx ] ) , @xmath14 u_{1}+2u_{2}u_{3}u_{4}^{\ast } & = & 0 , \nonumber \\
i\frac{\partial u_{2}}{\partial t}+\frac{k}{\sqrt{2}}u_{2}+\frac{i}{\sqrt{2}}\frac{\partial u_{2}}{\partial x}+u_{1}+\mu u_{4}+\left [ \left\vert u_{2}^{2}\right\vert + 2(\left\vert u_{1}^{2}\right\vert + \left\vert u_{3}^{2}\right\vert + \left\vert u_{4}^{2}\right\vert ) \right ] u_{2}+2u_{1}u_{4}u_{3}^{\ast } & = & 0 , \nonumber \\
i\frac{\partial u_{3}}{\partial t}-\frac{k}{\sqrt{2}}u_{3}-\frac{i}{\sqrt{2}}\frac{\partial u_{3}}{\partial x}+u_{4}+\mu u_{1}+\left [ \left\vert u_{3}^{2}\right\vert + 2(\left\vert u_{1}^{2}\right\vert + \left\vert u_{2}^{2}\right\vert + \left\vert u_{4}^{2}\right\vert ) \right ] u_{3}+2u_{1}u_{4}u_{2}^{\ast } & = & 0 , \nonumber \\
i\frac{\partial u_{4}}{\partial t}+\frac{k}{\sqrt{2}}u_{4}-\frac{i}{\sqrt{2}}\frac{\partial u_{4}}{\partial x}+u_{3}+\mu u_{2}+\left [ \left\vert u_{4}^{2}\right\vert + 2(\left\vert u_{1}^{2}\right\vert + \left\vert u_{3}^{2}\right\vert + \left\vert u_{2}^{2}\right\vert ) \right ]
u_{4}+2u_{2}u_{3}u_{1}^{\ast } & = & 0 .
\label{tx}\end{aligned}\]]note that eqs .
( [ tx ] ) conserve the hamiltonian , momentum , and the norm ( which is frequently called energy , in the applications to optics ) , @xmath15 the model based on eqs .
( [ tx ] ) is of special interest , as previously considered multi - wave gs models , such as the three - wave one in ref .
@xcite , did not contain a parameter similar to @xmath16 , which accounts for _ frustration _
, i.e. , symmetry breaking in the wave pairs @xmath17 and @xmath18 in eqs .
( [ tx ] ) .
in fact , the frustration makes the present four - wave distinct from the usual two - wave bg model .
the same system ( [ tx ] ) also finds an interpretation in terms of the bg in optical fibers . in that case , the fields @xmath19 and @xmath20 in eqs .
( [ tx ] ) represent pairs of right- and left - travelling waves with two orthogonal circular polarizations , @xmath10 is the bragg - reflectivity strength , while the linear - coupling terms with the coefficient @xmath9 account for linear mixing between the circular polarizations due to an elliptic deformation of the fiber s core , and the terms proportional to @xmath16 take into regard birefringence due to a twist of the fiber . looking for solutions to the linearized version of eqs .
( [ tx ] ) as @xmath21 , one arrives at a dispersion relation,@xmath22which generates a bandgap in the spectrum , @xmath23 , provided that @xmath24 . besides the gap , the two branches of the dispersion relation ( [ omega^2 ] ) give rise to _ semi - gaps _
, @xmath25 , which are also found in the case of @xmath26when the full gap is absent .
it was recently shown , in a general form , that families of bg solitons may exist in semi - gaps even if the full gap is absent @xcite .
quiescent ( zero - velocity ) solitons in eqs .
( [ tx ] ) , of the form @xmath27 , were investigated in detail in ref . @xcite
[ solitons generated by eqs .
( [ tzx ] ) in the spatial domain , corresponding to @xmath28 with fixed @xmath29 , were studied in ref .
they obey a reduction @xmath30 , which is not valid for moving solitons , see below . if @xmath31 , the solitons fall into three categories : symmetric and anti - symmetric ones , with @xmath32 , and more general asymmetric solitons .
the latter ones , and the solitons obtained by their continuation to @xmath33 , are always subject to a weak instability .
the symmetric and anti - symmetric solitons , as well as those stemming from them at @xmath33 , were found , respectively , to be stable chiefly ( but not exclusively ) in the full gap ( if any ) , and in the semi - gaps outside the full gap .
moving solitons are to be looked for solitons in the form @xmath34straightforward consideration of the underlying system ( [ tx ] ) shows that moving solitons may exist for @xmath35 @xcite . on the other hand
, quiescent bg solitons have never been observed in the experiment , the minimum velocity at which they could be created being , roughly , @xmath36 @xcite ( a possibility to diminish the velocity through collisions between the moving solitons was proposed in ref .
@xcite ) . in ref .
@xcite , moving solitons in the present model were considered , in a brief form , inside the full gap .
it was found that they loose their stability at quite small values of the velocity , @xmath37 .
our aim is to study the stability of the moving solitons in a systematic way , as well as collisions between them .
we will consider moving solitons in the most interesting case , when the full bandgap does not exist , but , nevertheless , solitons may be found in the semi - gaps ( see above ) . to the best of our knowledge , the moving solitons in this case has never been considered before .
to look for solitons in the form ( [ c ] ) , one can rewrite the underlying equations ( [ tx ] ) in terms of the variables @xmath38 .
the linearization of the equations gives rise to a dispersion relation , which differs from ( [ omega^2 ] ) by the replacement @xmath39 . in the case when the spectrum contains no full gap , straightforward analysis yields an equation to for the values of @xmath29 at edges of the semi - gaps , where we aim to find the solitons:@xmath40 ^{2}=\mu ^{2}\left ( 1 - 2c^{2}\right ) ~. \label{edge}\]]it immediately follows from eq .
( [ edge ] ) that no semi - gaps , hence no solitons either , are possible for @xmath41 .
typical examples of the semi - gap in the plane @xmath42 are shown in fig .
[ semigaps ] .
stationary equations , which are obtained by the substitution of eq .
( [ c ] ) into eqs .
( [ tx ] ) , were solved by means of the shooting method .
the numerical solution shows that the semi - gaps are , indeed , completely filled by solitons .
note that the moving solitons can not be generated automatically from quiescent ones , as eqs .
( [ tx ] ) bear no galilean or lorentzian invariance .
stability of the solitons was identified in direct simulations of eqs .
( [ tx ] ) for perturbed solutions , by means of the split - step fourier - transform and finite - difference techniques , both methods yielding identical results .
typical examples of the stable moving solitons , found inside the semi - gaps , are displayed in fig .
[ solitons ] . as well as in the case of zero - velocity solitons @xcite
, the frustration parameter @xmath16 strongly affects the solitons shape and stability : if @xmath31 , the stability area is largest , and the stable solitons obey the reduction @xmath43 . as is seen in figs .
[ solitons ] and [ semigaps ] , nonzero frustration breaks the latter relations , and leads to strong shrinkage of the stability region .
a notable feature is that , in the case of small but finite @xmath16 , the stability interval for a fixed value of @xmath29 may be limited not only by a maximum value of the soliton s velocity , but also by a _ minimum _ one .
for example , three cross sections shown by arrows in fig .
[ semigaps](b ) ( for @xmath44 and @xmath45 ) correspond to the stability intervals@xmath46the two latter ones being limited by a minimum value of @xmath47 .
the fact that the stability of the solitons may require a finite velocity is a completely new one .
nothing similar occurs in the model supporting the usual two - wave gss @xcite .
the stability of solitons is not sensitive to the value of the relative bragg reflectivity @xmath10 , unlike the frustration @xmath16 . in particular , the variation of @xmath10
does not essentially change the shape of the stability region displayed in fig .
[ semigaps ] , nor the shape of the solitons shown in fig .
[ solitons ] .
note that stability of the solitons against non - oscillatory perturbations ( those corresponding to real stability eigenvalues ) is provided , at fixed @xmath47 , by the above - mentioned vk criterion , @xmath48 @xcite [ recall the soliton s energy @xmath49 is defined as per eq .
( [ e ] ) ] .
the applicability of the vk criterion to the present model was not rigorously proven , but in ref .
@xcite it was actually derived for the particular case of @xmath12 .
inspection of numerical data shows that all the solitons satisfy this criterion ; however , it does not preclude instability against oscillatory perturbations ( those corresponding to pairs or quartets of complex instability eigenvalues ) , that is why a part of the solitons are unstable .
further simulations demonstrate that the development of the instability always leads to complete destruction of the soliton , see a typical example in fig .
[ destruction ] . ) . here
( as well as in figs .
4 and 5 below ) shown is the evolution of the field @xmath50 ; in other components , the evolution is quite similar .
the parameters are @xmath51 .
the figure uses the original coordinate @xmath7 , rather than @xmath52 from eq .
( [ c]).,width=384 ]
the next step in the study of the moving solitons is to consider collisions between them , which is interesting in its own right , and for various applications too @xcite . as eqs .
( [ tx ] ) are not integrable , an issue is whether collisions between stable solitons will be elastic or not .
we stress that neither the stability of the solitons of the _ semi - gap _ type , considered in this paper , nor collisions between them were considered in earlier works . exploring the stability region of the moving solitons ( see fig .
[ semigaps ] ) , we have found that collisions between them may be both completely elastic , leading to no other effect but finite shifts of the colliding solitons , and strongly inelastic , resulting in a partial destruction of the solitons , and generation of large amounts of radiation .
typical examples of these outcomes are shown in figs .
[ elastic ] and [ inelastic ] .
the interaction between the elastically colliding solitons has the character of the bounce from each other . , the other parameters being @xmath53.,width=384 ] and @xmath54 .
the other parameters are @xmath55.,width=384 ] the outcome of the collision strongly depends on the solitons velocities @xmath56 .
therefore , the results can be naturally summarized in the form of a region in the plane @xmath57 where the collisions are elastic ; at the border of the region , the transition between the two types of the collision is found to be quite sharp . in turn , the size of the region of the nondestructive ( elastic ) collisions depends on proximity of the individual solitons to the instability border in their parameter planes , see fig .
[ semigaps ] . first , in fig .
[ cc ] we present these results for the zero - frustration case ( @xmath31 ) , when the stability region in fig .
[ semigaps](a ) is quite large . as it follows from the inspection of fig .
[ semigaps](a ) , the difference between the three cases displayed in fig .
[ cc ] is that in the case of @xmath58 ( a ) the colliding solitons are close to the instability border , while in the other cases , @xmath59 ( b ) and @xmath60 ( c ) , they are located deeper in the stability region ( in the latter case , the soliton is actually close to the edge of the semi - gap , i.e. , the existence border of the solitons ) .
accordingly , there is a notable difference in the shape of the elasticity region in the @xmath61 plane : in the first case , the collisions may be elastic only if the velocities are small , while in the other cases , the solitons with large but nearly opposite velocities , @xmath62 , collide elastically even if the velocities are large . as it was stressed above , in the case of finite frustration ,
the velocity interval corresponding to the stable solitons may be limited not only from above , but also from below .
the collisions between unstable solitons being definitely destructive , the latter fact strongly affects the region of elastic ( nondestructive ) collisions in the @xmath61 plane for @xmath33 , as is seen in fig .
( [ cc2 ] ) .
as it was mentioned above , in the currently feasible experiments with bg solitons , their velocity can not be made smaller than , approximately , half the maximum group velocity , which is @xmath63 in the present notation .
for this reason , it is quite important that the elasticity regions shown in figs .
[ cc](b , c ) and [ cc2](a , b ) extend to large values of @xmath47 .
lastly , it is necessary to mention that , as well as the stability of individual solitons , the outcome of the collisions is not sensitive to the relative bg reflectivity @xmath10 , on the contrary to the strong dependence on the frustration @xmath16 .
in fact , the elasticity regions show almost no dependence on @xmath10 , therefore the examples shown in figs . [ cc ] and [ cc2 ]
are quite generic ones .
in this paper , we have presented results of a systematic study of the stability and collisions of moving four - wave solitons in the recently introduced model of the bragg cross - grating in planar optical waveguides with the kerr nonlinearity .
we have focused on the case when the system s linear spectrum does not contain any true bandgap , but rather features semi - gaps , in which soliton families may be found . in recent works ,
only zero - velocity solitons in this four - wave system have been studied .
that case is far from the current experiments , which generate solitons whose velocity exceed , roughly , half the maximum group velocity .
we have found the semi - gaps for the moving solitons in the analytical form , and demonstrated that numerically found solitons completely fill the semi - gaps .
stability of the moving solitons was determined by direct simulations , revealing a stability border which strongly depends on the frustration parameter .
it is interesting that a previously unknown situation is possible in the present model , when the velocity interval for stable solitons is limited not only from above , but also from below . collisions between stable solitons may be both elastic , so that they bounce from each other , and strongly inelastic , leading to a partial destruction of the solitons , with a sharp border between the two outcomes .
a region of elastic collisions was identified , with a well - pronounced peculiarity : if the individual solitons are close to the stability border , they collide elastically only if their velocities @xmath0 and @xmath1 are small ; however , if the solitons are more stable , the collision remains elastic in a strip around the line @xmath2 . as well as
the stability of individual solitons , the elasticity region strongly depends on the frustration parameter , but is insensitive to the relative reflectivity of the two bragg gratings which constitute the cross configuration .
this work of b.a.m . was supported , in a part , by the israel science foundation through the grant no .
t.m . and b.a.m .
appreciate hospitality of the department of electronics engineering at the city university of hong kong .
b. j. eggleton , r. e. slusher , c. m. de sterke , p. a. krug , and j. e. sipe , phys .
lett . * 76 * , 1627 ( 1996 ) ; c. m. de sterke , b. j. eggleton , and p. a. krug , j. lightwave technol . *
15 * , 1494 ( 1997 ) ; b. j. eggleton , c. m. de sterke , and r. e. slusher , j. opt .
b * 16 * , 587 ( 1999 ) .
r. g. scott , a. m. martin , t. m. fromhold , s. bujkiewicz , f. w. sheard , and m. leadbeater , phys .
lett . * 90 * , 110404 ( 2003 ) ; b. eiermann , t. anker , m. albiez , m. taglieber , p. treutlein , k. p. marzlin , and m. k. oberthaler , phys .
92 * , 230401 ( 2004 ) .
|
we report results of a systematic study of one - dimensional four - wave moving solitons in a recently proposed model of the bragg cross - grating in planar optical waveguides with the kerr nonlinearity ; the same model applies to a fiber bragg grating ( bg ) carrying two polarizations of light .
we concentrate on the case when the system s spectrum contains no true bandgap , but only semi - gaps ( which are gaps only with respect to one branch of the dispersion relation ) , that nevertheless support soliton families .
solely zero - velocity solitons were previously studied in this system , while current experiments can not generate solitons with the velocity smaller than half the maximum group velocity .
we find the semi - gaps for the moving solitons in an analytical form , and demonstrated that they are completely filled with ( numerically found ) solitons .
stability of the moving solitons is identified in direct simulations .
the stability region strongly depends on the frustration parameter , which controls the difference of the present system from the usual model for the single bg .
a completely new situation is possible , when the velocity interval for stable solitons is limited not only from above , but also from below .
collisions between stable solitons may be both elastic and strongly inelastic . close to their instability border ,
the solitons collide elastically only if their velocities @xmath0 and @xmath1 are small ; however , collisions between more robust solitons are elastic in a strip around @xmath2 .
|
the discovery of the first inorganic spin - peierls ( sp ) compound cugeo@xmath3 ( ref . ) has triggered subsequent studies of the impurity - substitution effect on the spin - singlet states , @xcite and the existence of the disorder - induced transition into three - dimensional antiferromagnetic ( 3d - af ) state has been established .
@xcite since the relevant exchange energies , i.e. , intra- and interchain coupling @xmath6 and @xmath7 , do not change with doping except at the impurity sites , it is difficult to understand the impurity - induced transition from the sp to 3d - af state in the framework of the conventional " competition between dimensionalities ( where @xmath8 is an essential parameter @xcite ) and therefore such impurity - substitution effect is a matter of current interest . in cugeo@xmath3 , a small amount of impurity leads to an exotic low - temperature phase where the lattice dimerization and antiferromagnetic staggered moments simultaneously appear [ dimerized af ( d - af ) phase ] .
@xcite moreover , when the impurity concentration @xmath9 exceeds a critical concentration @xmath10 , the sp transition measured by dc susceptibility disappears @xcite and a uniform af ( u - af ) phase appears below the nel temperature @xmath11 @xmath12 4 k. @xcite the d - af ground state can be understood as a state of spatially modulated staggered moments accompanied with the lattice distortion .
@xcite however , the mechanism of the depression of the sp phase and the establishment of the disorder - induced antiferromagnetism are still to be elucitated .
a transport measurement could be a desirable tool in dealing with such a problem , because the mobility of spin excitations or spin diffusivity is often sensitive to the impurities .
nevertheless , even a crude estimation of the mobility has not been carried out for the spin excitations so far .
recently , we reported intriguing behaviors of the thermal conductivity @xmath0 of pure cugeo@xmath3 .
@xcite in ref . , the existence of the spin heat channel ( @xmath13 ) was suggested .
however , we can not separate @xmath13 and the phonon thermal conductivity @xmath14 with the measurement of the pure sample only . in this work
, we have resolved this problem with mg - doped crystals .
it turned out that the spin diffusion length , which can be calculated from @xmath13 , is much larger than the distance between adjacent spins for the pure cugeo@xmath3 , indicating coherent heat transport due to the spin excitations .
since the spin heat transport almost disappears for the heavily doped samples in which the long - range sp ordering is absent , it is suggested that the large mobility of the spin excitations plays an important role in the sp ordering .
the spin heat transport is also a probe of the spin gap .
it is found that local spin - gap opening is robust against the impurity doping , though the temperature of the long - range ordering is strongly suppressed with @xmath9 .
phonon heat channel ( @xmath14 ) provides an unusual thermal conductivity peak in the sp state , which was one of the topics of our previous report.@xcite the peak is drastically suppressed with both magnetic field and mg - doping , indicating that the spin - gap opening has produced the peak in @xmath14 via strong phonon - spin interaction .
the single crystals of cu@xmath1mg@xmath2geo@xmath3 were grown with a floating - zone method .
the mg concentration is determined by inductively coupled plasma - atomic emission spectroscopy ( icp - aes ) .
@xcite the critical concentration @xmath10 was carefully determined by masuda _
et al . _ for the same series of crystals that we have used . @xcite
they found that the nel temperature @xmath11 jumps at the impurity - driven transition from d - af to u - af phase and that phase is separated in the transition range of @xmath15 , indicating that the transition is of first order.@xcite we used six samples with @xmath16 , @xmath17 , and @xmath18 for the thermal conductivity measurement .
they were originally prepared for the neutron and the synchrotron x - ray experiments , and thus the sample homogeneity has been confirmed in detail which is described elsewhere.@xcite typical sample size is 0.2 @xmath19 2 @xmath19 7 mm@xmath20 ( @xmath21 ) . since our interest is thermal conductivity along the 1d spin chain , we need the sample with a longer dimension along the @xmath22 axis .
thus we payed particular attention to the mg homogeneity along the @xmath22 axis for this study , and the best way is to use the samples prepared for the neutron experiments .
the fundamental magnetic properties of the samples used already appeared in another paper.@xcite the thermal conductivity is measured using a one heater , two thermometers " method as the previous measurement.@xcite the base of the sample is anchored to a copper block held at desired temperatures .
a strain - gauge heater is used to heat the sample . a matched pair of microchip cernox thermometers are carefully calibrated and then mounted on the sample .
the maximum temperature difference between the sample and the block is 0.2 k. typical temperature difference between the two thermometers is @xmath120.5 k above 20 k and 0.2 k below 20 k. the temperature - controlled block is covered with a metal shield so that heat loss through radiation becomes negligible .
magnetic field is applied parallel to the @xmath22-axis direction .
the measured temperature range is below 30 k , which is well below the temperature scale of the intra - chain coupling @xmath23 ( @xmath12 120 k according to neutron scattering measurement @xcite ) , and therefore , the one - dimensional quantum spin liquid ( 1d - qsl ) is realized above the transition temperatures .
figure 1 shows the temperature dependence of @xmath0 for samples with and without mg substitution : 1(a)-1(f ) are for @xmath24 , 0.0216 , 0.0244 , 0.032 and 0.040 , respectively .
@xmath25 is well below @xmath10 , @xmath26 is close to @xmath10 , @xmath27 is on the border and @xmath28 and @xmath29 are in the u - af region .
the transition temperatures @xmath30 for the @xmath31 samples and @xmath11 for the @xmath18 samples , which were determined by susceptibility measurement , @xcite are indicated by arrows in the figures .
a sharp drop is observed just below @xmath30 for the pure cugeo@xmath3 as presented in the inset of fig .
the @xmath25 sample shows the same behavior at @xmath4 , while @xmath0 is _ enhanced _ just below @xmath4 for the @xmath26 and 0.0244 samples .
mg@xmath2geo@xmath3 single crystals in magnetic fields up to 15 or 16 t applied along the chain direction , for ( a ) @xmath16 , ( b ) @xmath25 , ( c ) @xmath32 , ( d ) @xmath27 , ( e ) @xmath28 , and ( f ) @xmath29 .
inset : a magnified view of @xmath0 near the spin - peierls transition in zero field .
the dashed line is an extrapolation of @xmath0 from above the transition temperature , width=302 ] in figs .
1(e ) and 1(f ) , a slight change in the curvature is observed at @xmath11 and the field dependence in the nel state is opposite to that above @xmath11 ; @xmath0 decreases with field below @xmath11 while @xmath0 increases with field above @xmath11 .
the three - dimensional magnon excitations , which appear in the nel state , are responsible for this behavior .
the thermal conductivity in u - af phase will be further discussed elsewhere.@xcite although the nel transition ( to d - af phase ) should be present also for the @xmath31 samples , we can not identify the corresponding feature at the transition , because @xmath11 ( @xmath33 k ) @xcite is too close to the lower temperature limit of our measurement , @xmath12 2 k. taking a look at the full scale of each figure , one can notice that @xmath0 is strongly suppressed with mg substitution in the whole temperature range .
two thermal conductivity peaks are present in the zero - field curve of fig .
1(a ) ; one is in the sp phase and the other is in the 1d - qsl state .
the broad shape of the high - temperature peak is maintained with doping . on the other hand ,
the low - temperature peak ( sp peak ) diminishes with @xmath34 , and disappears for the @xmath18 samples .
the sp peak is suppressed also in magnetic fields [ figs .
1(a)-1(d ) ] .
in order to see the field dependence in more detail , the thermal conductivity at the sp peak is plotted as a function of magnetic field in fig . 2 for the @xmath16 sample .
@xmath0 is suppressed with magnetic field and suddenly drops at the threshold field , @xmath35 , at which the system makes a transition to the incommensurate phase .
the same features were observed in our previous measurement with the crystal grown by the laser - heated pedestal growth technique.@xcite also , the typical value of @xmath0 , for example at the high - temperature peak , is almost the same as that of the previous result .
we thus believe that the above features are intrinsic in this material regardless of the growth method of the crystal , as long as high quality single crystal is used.@xcite , demonstrating the suppression of the low - temperature peak with field application . ] in a temperature range close to @xmath30 , @xmath0 _ increases _ with field for the mg - doped samples , while little field dependence is seen in the pure sample [ figs .
1(a)-1(d ) ] .
note that the field dependence extends even above @xmath30 for the mg - doped samples .
the variation in the field dependence implies qualitative difference of the sp transition between pure cugeo@xmath3 and doped cugeo@xmath3 . this field dependence is apparent also in the @xmath28 sample , which is above @xmath10 [ fig .
1(e ) ] , but rapidly diminishes in the heavily doped sample [ fig .
cugeo@xmath3 is an insulator and heat conduction by electrons or holes is absent unlike metals . instead
, low - energy spin excitations@xcite can carry heat . in the 1d - qsl state , total @xmath0
is given by a sum of @xmath13 and @xmath14 as , @xmath36 in contrast , @xmath13 is suppressed at the lowest temperature region in sp phase because few spin excitations are present due to the spin gap , and the total @xmath0 represents a phonon contribution only , @xmath37 the field and impurity dependence of @xmath14 below @xmath4 will be discussed in subsection a. the value of @xmath13 in 1d - qsl will be crudely estimated in subsection b. finally , the problem of local spin - gap formation will be dealt with in subsection c , using this @xmath13 as a probe .
since the low - temperature peak is drastically suppressed with the application of field as shown in figs . 1 and 2 , the feature should be related to the spin - excitation spectrum .
we have proposed the following explanation for this peak.@xcite at temperatures well below @xmath4 , where low - energy spin excitations are negligible , heat is carried mostly by phonons and @xmath14 increases with temperature because of the growing population of phonons like usual insulating crystals . with increasing temperature , the increasing number of thermally excited spin excitations scatter phonons , and @xmath14 diminishes . as a result
, the thermal conductivity peak shows a peak .
although the thermal spin excitations begin to carry heat , the increase in @xmath13 appears only in the temperature region slightly below @xmath4 [ see inset in fig .
1(a ) ] . in order to confirm the above understanding , the impurity substitution is helpful because the spin gap is absent in heavily substituted samples .
we normalized the size of the peak for each sample as @xmath0(5 k)/@xmath0(15 k ) , and plotted it against @xmath9 in fig .
one can see the suppression of the peak with @xmath9 similarly to the case of field application ( fig . 2 ) .
moreover , the peak disappears at the concentration @xmath18 where the long - range sp order no longer exists , as shown in figs .
1(e ) and 1(f ) .
we have thus confirmed the correlation between spin gap and the peak also by mg substitution , giving the additional evidence for the scenario described in the previous paragraph .
( 5 k)/@xmath0(15 k ) , is plotted against mg - concentration . a suppression of the peak with
mg doping is evident . ] the @xmath27 sample , which is on the border from sp ( d - af ) to u - af demonstrates enhancement of @xmath0 below @xmath38 , as observed in fig .
since the enhancement is suppressed in magnetic fields , the existence of a well - defined spin gap is suggested even at @xmath39 .
we can notice the rapid change in @xmath13 in the ( low - temperature ) vicinity of @xmath4 , which is another sign of the spin - gap opening , for the @xmath16 and 0.016 samples [ figs .
1(a ) and 1(b ) ] .
however , this feature is absent for the @xmath26 and 0.0244 samples [ figs .
1(c ) and 1(d ) ] , since @xmath13 is significantly reduced with @xmath9 because of impurity scattering . figure
4(a ) shows the zero - field thermal conductivity of all the samples .
the thermal conductivity is strongly suppressed with mg - substitution in the whole measured temperature range .
the @xmath9 dependence of @xmath0 below @xmath12 4 k can be understood as the difference in the scattering rate of phonons . since we showed in ref .
that the scattering by planer defects is dominant for the @xmath16 sample , difference in the number of planer defects may cause the variation of @xmath14 in our series of samples .
the number of planer defects does not necessarily increase with increasing @xmath9 ; note that @xmath0 in the @xmath28 sample is smaller than that in the @xmath29 sample below @xmath12 5 k. mg@xmath2geo@xmath3 single crystals .
( b ) the thermal conductivity at 17.5 k as a function of mg concentration .
the spin - peierls temperatures measured by dc susceptibility are plotted together . ] remembering the origin of the low-@xmath40 thermal conductivity peak , we can notice that scattering by the spin excitations , in turn , becomes dominant in @xmath14 above the peak temperature for the @xmath31 samples .
noting that all the @xmath41 curves looks parallel to one another above @xmath12 15 k , @xmath14 in the @xmath18 samples should probably have a temperature dependence similar to that in the @xmath25 , 0.0216 and 0.0244 samples above @xmath12 15 k. therefore , it is expected that phonon heat transport is governed by the spin - scattering also for the @xmath18 samples , at least above @xmath12 15 k. in contrast to rather complicated temperature dependence below @xmath4 , @xmath42 in the 1d - qsl region is simpler , which is an advantage of discussing @xmath9 dependence in the region . as a crude approximation , it is assumed that the phonon part is not so much @xmath9 dependent and that most of the @xmath9-dependence comes from @xmath13 , because the detailed characterization @xcite guarantees good homogeneity in both pure and mg - doped crystals and the direct modification in the phonon modes due to the mg substitution can be estimated to be negligible .
the scattering rate of phonons due to the point defects is more than two orders of magnitude smaller than the total scattering rate.@xcite one may notice that the robustness of @xmath14 to the mg - doping is natural , considering that phonons are mainly scattered by spin excitations above @xmath12 15 k , which is the conclusion of the previous paragraph . since the population of both phonons and spin excitations does not change so much with @xmath9 in the temperature range above @xmath4 , as has been reported in specific heat results , @xcite @xmath9 dependence of @xmath14 is expected to be small in the 1d - qsl state.@xcite figure 4(b ) shows the @xmath9 dependence of @xmath0 at 17.5 k. @xmath0 rapidly decreases with @xmath9 up to @xmath10 , and it saturates when @xmath9 exceeds @xmath10 .
therefore , it is naturally assumed that @xmath13 is close to zero in the heavily doped samples with @xmath9 @xmath43 @xmath10 .
this result suggests some correlation between the mobility of the spin excitations and the sp ordering .
following the above interpretation , it is possible to estimate @xmath13 by subtracting @xmath0 of the @xmath29 sample from the measured @xmath0 for each sample .
the result indicates that most part of the @xmath0 of pure cugeo@xmath3 is due to @xmath13 , which is approximately 0.4 w / cm k at 17.5 k. this gives the diffusivity of the spin excitations @xmath44 of @xmath121.6@xmath1910@xmath45 m@xmath46/s , if we assume @xmath47 @xmath121.8 j / mol k.@xcite @xmath44 in 1d systems can be written as @xmath48 , where @xmath49 is the velocity of the spin excitations and @xmath50 is the mean free path , in other words , the spin diffusion length . assuming the spin - wave like dispersion given by des cloizeaux and pearson @xcite @xmath51=(@xmath52)@xmath53 and @xmath23 @xmath12 120 k , @xcite one can evaluate @xmath49 @xmath12 ( @xmath54)@xmath55/@xmath56= 1.3@xmath1910@xmath57 m / s and @xmath50 @xmath12 130 nm ( @xmath51 is the energy of the spin excitations
, @xmath58 is the wave vector and @xmath22 ( @xmath12 0.3 nm ) is the distance between adjacent spins ) .
note that the spin diffusion length of the same order of magnitude was obtained from the nmr relaxation measurement for another 1d spin system , agvp@xmath59s@xmath60.@xcite since @xmath61 430 is much larger than unity , we can conclude that the spin heat transport for the pure cugeo@xmath3 is coherent .
if the same estimation is applied also to the mg - doped samples , @xmath61 100 for the @xmath25 ( @xmath62 ) and @xmath61 50 even for @xmath39 at this temperature .
since @xmath63 is approximately calculated as 40 , the above estimation indicates that the spin diffusion length in the 1d - qsl state exceeds the mean impurity distance @xmath64 , as long as @xmath9 is less than @xmath10 . whereas , the spin excitations are not so mobile when @xmath18 .
figure 5 shows @xmath65 ( @xmath66 @xmath67 ) of the three mg - doped samples which have sp transition .
one can find that @xmath67 is almost temperature independent above @xmath5 @xmath12 15 k and that @xmath67 of all the samples deviates from this temperature - independent value below @xmath5 .
since @xmath5 is close to @xmath4 of the pure cugeo@xmath3 , the deviation can be attributed to a precursor of the sp transition .
note that similar behavior is observed in the raman scattering measurement on zn- and si - dopd cugeo@xmath3 .
@xcite [ @xmath68 , for the @xmath25 , 0.0216 and 0.0244 samples .
the local spin gap opens below @xmath12 15 k in all the samples . ]
recently , it is shown by synchrotron x - ray diffraction measurement that fwhm of the bragg peak from the lattice dimerization reaches the resolution limit at a temperature @xmath69 which is far below @xmath38 .
@xcite the authors claimed that the long - range order takes place only below @xmath69 and that lattice dimerization is short - range just below @xmath38.@xcite the gap - like feature in @xmath0 is also to be explained along the idea of the short - range order ( sro ) .
however , the data in fig .
5 suggests that sro grows from @xmath5 @xmath12 15 k , which is even above @xmath70 .
the field dependence of @xmath0 is a strong indication of the presence of the sro below @xmath5 .
figure 6 shows the temperature dependence of @xmath67 for mg - doped samples in various magnetic fields . in all cases , we observed the recovery of @xmath67 with increasing fields below @xmath5 .
the results are consistent with the notion that the reduction of @xmath67 below @xmath5 is due to the development of the local spin gap . since the magnetic field
is thought to reduce the magnitude of spin gap even above @xmath38 , the number of spin excitations responsible for the heat transport will increase with the field .
moreover , the field dependence only appears below @xmath5 in our observation .
we have no field dependence above @xmath5 , which is consistent with the notion that even the local spin gap is absent . for the ( a ) @xmath71 , ( b ) @xmath26 , ( c ) @xmath27 and ( d ) @xmath28 samples .
magnetic fields are applied along the chain direction . ]
it should be emphasized that the heat transport by spin excitations is very sensitive to the spin - peierls sro .
we observed a magnetic field dependence in @xmath67 below @xmath5 even for the sample with @xmath72 , where the spin - peierls lro no longer exist in the whole temperature range . according to the synchrotron x - ray measurement
, sp - sro develops below 7.5 k in the @xmath28 sample .
@xcite for the @xmath29 sample , however , there was no evidence that sp - sro actually develops .
in contrast , one can see that the magnetic field dependence of @xmath0 still exists for the @xmath29 sample ( fig .
1(f ) ) even though the change is not so obvious as in the @xmath28 sample .
we think that the difference of a length scale of probed area between the two measurement techniques causes such discrepancy .
the above results tell us that the impurity - substitution modifies the ground state and suppresses the spin - peierls ordering in cugeo@xmath3 system through different mechanisms . the importance of the inter - chain coupling and the tendency to the 3d - antiferromagnetism increases below @xmath73 . as a result
, @xmath11 increases with @xmath9 and the ground state changes from d - af to u - af at @xmath10 .
@xcite on the other hand , the local spin - gap formation , which occurs at @xmath5 , is goverened only by one - dimensional nature of the spin system coupled with 3d phonons and has nothing to do with 3d - af fluctuation .
therefore , the temperature of sp - sro is not modified with @xmath9 , as long as the spin - singlet state is energetically favored .
( the spin - singlet state will be no more favored in a heavily disordered system where long - wave - length spin excitations , whose energy is less than the spin - gap energy , are absent.@xcite ) the spin - diffusion length @xmath50 , in turn , rapidly diminishes with @xmath9 owing to the growing impurity - scattering of the spin excitations .
since the spin correlation length is directly related to @xmath50,@xcite the length scale of the sp domains is reduced with @xmath9 . as a result ,
signals of the spin gap , detected with any probe , diminishes in size with @xmath9 .
the thermal conductivity of the mg - doped cugeo@xmath3 helps to understand both the spin - gapped state below the sp transiton and the 1d - qsl state above the transition .
large spin heat transport is observed in the 1d - qsl state for pure cugeo@xmath3 , and rapidly diminishes with mg - doping , accompanied by the suppression of the long - range sp ordering .
the spin gap opening suppresses @xmath13 and enhances @xmath14 .
examining the @xmath9 dependence of the spin - gap features , it turned out that the local spin gap opens at a temperature ( @xmath5 ) independent of @xmath9 , suggesting that the suppression of sp ordering is attributed to the reduction of the spin diffusion due to the impurity scattering .
it is expected that the above analysis of the impurity - substitution effect on the thermal conductivity is applicable to other one - dimensional spin systems and that thus obtained transport properties may reveal new aspects on such materials .
we thank a. kapitulnik for helpful advices both on the experimental techniques and in understanding the results in the early stage of this project .
we can show that the scattering rate @xmath74 by point defects , which is introduced by the substitution with the ion with different mass , is very small even for the most heavily doped sample ( @xmath29 ) .
the dominant - phonon approximation gives,@xcite [ @xmath76 ( @xmath77 ) is density ratio of point defects , @xmath78 ( @xmath79 cm ) is lattice constant , @xmath80 ( @xmath81 ) is number of phonon modes ( 3 times the number of atoms per unit cell ) , @xmath82 ( @xmath83 cm / s ) is phonon velocity and @xmath84 ( @xmath85 ) is the ratio ( mass difference between mg and cu atoms)/(mass of cu atom ) ] .
@xmath74 can be estimated to be about @xmath86 ( s@xmath87 ) at 30 k. assuming phonon specific heat @xmath90 to be @xmath91 and using the value of @xmath92 ( j / cm@xmath20 k@xmath94 ) , shown in ref . .
taking @xmath95 ( w / cm k ) from our result of the @xmath0 measurement , @xmath96 ( s@xmath87 ) at 30 k. since this @xmath88 is more than 600 times larger than @xmath74 ( the ratio is even larger below 30 k ) , the scattering by the point defects is not important for the three - dimensional phonon heat transport in the measured temperature range .
m. hase , i. terasaki , and k. uchinokura , phys .
* 70 * , 3561 ( 1993 ) .
m. hase , i. terasaki , y. sasago , k. uchinokura , and h. obara , phys .
lett . * 71 * , 4059 ( 1993 ) .
m. hase , n. koide , k. manabe , y. sasago , k. uchinokura , and a. sawa , physica ( amsterdam ) * 215b * , 164 ( 1995 ) . s. b. oseroff , s .- w .
cheong , b. aktas , m. f. hundley , z. fisk , and l. w. rupp , jr .
74 * , 1450 ( 1995 ) .
s. inagaki and h. fukuyama , j. phys .
. jpn . * 52 * , 3620 ( 1983 ) .
t. masuda , a. fujioka , y. uchiyama , i. tsukada , and k. uchinokura , phys .
lett . * 80 * , 4566 ( 1998 ) .
y. sasago , n. koide , k. uchinokura , m. c. martin , m. hase , k. hirota , and g. shirane , phys .
b * 54 * , r6835 ( 1996 ) . m. c. martin , m. hase , k. hirota , g. shirane , y. sasago , n. koide , and k. uchinokura , phys .
b * 56 * , 3173 ( 1997 ) .
h. fukuyama , t. tanimoto , and m. saito , j. phys .
jpn . * 65 * , 1182 ( 1996 )
. k. m. kojima , y. fudamoto , m. larkin , g. m. luke , j. merrin , b. nachumi , y. j. uemura , m. hase , y. sasago , k. uchinokura , y. ajiro , a. revcolevschi , and j. p. renard , phys .
lett . * 79 * , 503 ( 1997 ) .
y. ando , j. takeya , d. l. sisson , s. g. dttinger , i. tanaka , r. s. feigelson , and a. kapitulnik , phys .
b * 58 * , r2913 ( 1998 )
. t. masuda , i. tsukada , k. uchinokura , y. j. wang , and v. kiryukhin , r. j. birgeneau , phys .
b , * 61 * , 4103 ( 2000 ) .
m. nishi , o. fujita , and j. akimitsu , phys .
b * 50 * , r6508 ( 1994 ) .
recently , vasilev _ et al .
_ reported @xmath0 of cugeo@xmath3 single crystal , which does not show the low - temperature peak [ a. m. vasilev , m. i. kaganov , v. v. pryadun , g. dhalenne , and a. revcolevschi , jetp lett . *
66 * , 868 ( 1997 ) ; a. m. vasilev , v. v. pryadun , d. i. khomskii , g. dhalenne , a. revcolevschi , m. isobe , and y. ueda , phys .
lett . * 81 * , 1949 ( 1998 ) ] .
considering the fact that the absolute value of @xmath0 is smaller compared to our result , the absence of the peak is probably because of larger defect - scattering of phonons in their crystals
. j. takeya _
et al . _ , in preparation .
j. des cloizeaux and j. j. pearson , phys . rev . *
128 * , 2131 ( 1962 ) .
r. berman , _ thermal conduction in solids _
( oxford university press , oxford , 1976 ) , p. 73 . t. lorenz , h. kierspel , s. kleefisch , b. bcher , e. gamper , a. revcolevschi , and g. dhalenne , phys . rev .
b * 56 * , r501 ( 1997 ) .
t. masuda _
et al_. , in preparation .
it is hardly possible to assume that defect scattering is mainly responsible in @xmath14 in the 1d - qsl state and that it causes the parallel shift in the @xmath42 curves for the @xmath41 samples , because such scattering provides @xmath97 proportional to @xmath98 .
m. takigawa , t. asano , y. ajiro , m. mekata , and y. j. uemura , phys .
* 76 * , 2173 ( 1996 ) .
h. kuroe , h. seto , j. sasaki , t. sekine , n. koide , y. sasago , k. uchinokura , m. hase , m. isobe , and y. ueda , j. magn .
magn . mater . *
177 - 181 * , 679 ( 1998 ) .
y. j. wang , v. kiryukhin , r. j. birgeneau , t. masuda , i. tsukada , and k. uchinokura , phys .
* 83 * , 1676 ( 1999 ) .
m. saito and h. fukuyama , j. phys .
jpn , * 66 * , 3259 ( 1997 ) .
g. s. uhrig , f. schnfeld , j - p .
boucher , and m. horvati , phys .
b * 60 * , 9468 ( 1999 ) .
b. i. halperin and p. c. hohenberg , phys . rev . * 177 * , 952 ( 1969 ) .
|
the thermal conductivity @xmath0 is measured in a series of cu@xmath1mg@xmath2geo@xmath3 single crystals in magnetic fields up to 16 t. it has turned out that heat transport by spin excitations is coherent for lightly doped samples , in which the spin - peierls ( sp ) transition exists , at temperatures well above the sp transition temperature @xmath4 .
depression of this spin heat transport appears below @xmath5 , at which the spin gap _ locally _ opens .
@xmath5 is not modified with mg - doping and that @xmath5 for each mg - doped sample remains as high as @xmath4 of pure cugeo@xmath3 , in contrast to @xmath4 which is strongly suppressed with mg - doping .
the spin - gap opening enhances phonon part of the heat transport because of reduced scattering by the spin excitations , producing an unusual peak .
this peak diminishes when the spin gap is suppressed both in magnetic fields and with the mg - doping . 2
|
the fundamental split between integrable and non - integrable systems in classical mechanics has not been comprehensively mirrored in quantum mechanics @xcite .
the issue seems to hinge on finding a suitable definition for _ quantum _ chaos . the sensitive dependence to initial conditions that characterises classical chaos
is wholly understood in terms of trajectories of classical phase space points which have no direct quantum analog @xcite .
this has led to two distinct ways of identifying variables to measure quantum chaos .
one involves investigating various quantum variables that can act as signatures of chaos by clearly distinguishing between quantum systems whose classical counterparts are integrable and those which are non - integrable .
the implementation of an expanding array of energy spectra properties have made this approach highly successful ( e.g. @xcite ) .
a second approach seeks an intrinsically quantum definition of quantum chaos by investigating the quantum parallels for variables such as lyapunov exponents and various entropy measures which define and quantify chaos in classical mechanics @xcite . in this paper
we adopt the second approach and develop an original technique involving the analysis of entropy production measures to reveal a clear correspondence between the quantum and classical formulations of a seminal system which has a rich structure of delicately interwoven regular and chaotic dynamics - the standard map .
section ii a briefly outlines the main characteristics of the standard map and then section ii b describes how classical entropy contours are calculated and used to give a comprehensive account of these characteristics . in a similar fashion , section iii a briefly outlines the quantisation of the standard map and then section iii b describes how quantum entropy contours are generated through interaction with an environment .
the similarities and differences between the classical and quantum entropy contours are explained in section iv a and section iv b concludes the paper with a discussion on the use of entropy measures to describe quantum chaos .
the standard map describes the local behaviour of nonintegrable dynamical systems in the separatrix region of non - linear resonances .
the name results from its extensive use in the investigation of chaos , especially the mechanisms involved in the transition to global chaos in conservative systems .
it is derived from the kicked rotor model of a one dimensional pendulum and is a hamiltonian ( area preserving ) dynamical system .
though the map has been extensively analysed @xcite , here we describe some relevant details .
the standard map can be represented by the equations of motion : @xmath0 where @xmath1 is a real variable which acts as the chaos parameter . with unit mass and discrete time , @xmath2 and
@xmath3 have the same dimensions .
the phase space of the map is periodic in @xmath3 , by definition , and as a special feature of the standard map it is also periodic in @xmath2 , with the same period 1 .
@xmath4 is the case for a free rotor , a trivial and completely regular map . in fig.1(a)-(c ) , we show the well known sequence of standard map plots for increasing values of @xmath1 .
the map has several axes of symmetry and the unit periodicity in both @xmath2 and @xmath3 means that only a unit square in phase space need be viewed .
the interval @xmath5 is used for all results in this paper .
fig.1(a ) shows the mapping when @xmath6 .
much of the phase space is composed of kam ( kolmogorov - arnold - moser ) tori @xcite stretching horizontally from @xmath7 to @xmath8 and serve to isolate one region of phase space from another .
periodic orbits in the central resonance can be easily seen , as well as a few other dominant non - linear resonances ( small ellipses ) between kam tori . as @xmath1 is increased , kam tori that horizontally span the phase space are destroyed by resonances and are replaced with smaller kam islands . beyond a critical value at @xmath9 , the last phase space spanning kam torus is broken and the map becomes globally chaotic ( fig.1(b ) ) . now
the remaining resonances are clearly visible as _ stable islands _ in a _
chaotic sea _ of trajectories . on reaching @xmath10 ,
most structure is wiped out ( fig.1(c ) ) . a positive lyapunov exponent , which quantifies the exponential divergence in time of two closely neighbouring phase space trajectories , is a primary definition of classical chaos @xcite . for one - dimensional maps such as
the standard map the positive lyapunov exponent is equal to the kolmogorov - sinai ( ks ) entropy , @xmath11 , which is a measure of the rate of information production in the system @xcite .
thus @xmath12 only for completely regular dynamics .
the ks entropy ( also called dynamical entropy or metric entropy ) of a chaotic mapping can be calculated using the formula @xmath13 where @xmath14 is the changing distance between two initially close neighbouring points , @xmath15 and @xmath16 , in phase space . @xmath17 and @xmath18 are evolved by iterating a linearised form of the chaotic map .
this _ tangent map _ is rescaled after every iteration as follows : the @xmath19 iteration of the map produces the values @xmath20 and @xmath21 from which @xmath22 is calculated .
these values are then rescaled to @xmath23 and @xmath24 which are fed back into the tangent map for the next iteration @xcite .
use of the base-2 logarithm in ( [ ks - ent ] ) allows the entropy to be measured in _
bits _ of information .
the standard map is linearised to give its associate tangent map @xmath25 which can be employed in ( [ ks - ent ] ) to calculate @xmath11 .
the tangent map ( [ tangent ] ) clearly shows that the value of @xmath11 depends on the initial position in phase space @xmath15 for the standard map .
this is not always the case .
@xmath11 is generally used as a _ global _ measure of the level of chaos in a given system , but this is only valuable if all chaotic trajectories in the system can reach into all regions of its phase space .
well known examples of such systems include the cat and baker s maps @xcite .
however for @xmath26 the standard map has a predominantly _ mixed _ phase space in which different chaotic regions are not connected .
kam tori act as boundaries so that trajectories originating in one chaotic region can not escape to another .
this isolation inhibits the exponential divergence of chaotic trajectories so that the positive lyapunov exponent , and consequently @xmath11 , will vary from region to region .
the kicked top is another example of a well known mixed phase space system @xcite . to reveal a _
complete _ description of the standard map at a specific @xmath1 in terms of ks entropy , many values of @xmath11 corresponding to many initial positions in phase space can be plotted as a contour map on phase space .
this has been done in fig.1(d)-(f ) . using ( [ ks - ent ] ) and
( [ tangent ] ) , and setting @xmath27 iterations , values for @xmath11 were calculated for each point on a 64 x 64 grid spanning the same unit of phase space and the same @xmath1 values as in fig.1(a)-(c ) .
shading intensity reflects the relative sizes of the ks entropy .
@xmath28 is shown as white on the maps while darker and darker shades of grey reflect an increasing @xmath11 .
the black areas show the largest @xmath11 values corresponding to the most chaotic regions of the standard map .
the resemblance between fig.1(a)-(c ) and fig1.(d)-(f ) is striking .
stable islands in the classical maps translate to stark white patches in the contour maps .
this is because @xmath28 for _ all _ periods in non - chaotic dynamics .
the chaotic sea of trajectories in the classical maps are also faithfully reproduced as very dark patches of similar shape and size in the contour maps .
all these correlations indicate that @xmath11 presented in this way can comprehensively display all the essential features of the standard map as it becomes globally chaotic .
the quantised model of the standard map is governed by the hamiltonian @xmath29 this `` kicked rotor '' describes a free particle of unit mass which experiences impulses ( kicks ) at intervals @xmath30 .
following @xcite and @xcite , the kinematics are that of finite dimensional quantum mechanics with periodic boundary conditions .
position and momentum space are thus discretized , placing the lattice sites at integer values @xmath31 for @xmath32 .
the dimension @xmath33 of hilbert space is taken as even and , for consistency of units , the quantum scale on phase space is taken to be @xmath34 . position and momentum basis kets are denoted by @xmath35 and @xmath36 .
initial states @xmath37 are assumed to be _ coherent states _
( minimum - uncertainty states ) .
the fiducial initial coherent state @xmath38 is defined as the ground state of a special harper operator @xcite , which can be displaced with the appropriate operators to produce all the other possible initial coherent states @xmath39 , i.e. , @xmath40 , the system can be described by the density operator @xmath41 which changes according to the evolution equation @xmath42 where the kicked rotor unitary evolution operator , @xmath43 parallelling the classical case , we look for an entropy measure to reveal the dynamics of the quantum system .
entropy in quantum statistical mechanics is referred to as von neumann ( vn ) entropy , @xmath44 , ( the equivalent measure in classical mechanics is the gibbs entropy ) and can be defined in terms of the density matrix @xmath45 of a system as @xmath46 @xmath44 is a quantative measure of disorder and can be measured in bits .
however , the unitarity of hamiltonian dynamical evolution dictates that @xmath44 remain constant at all times .
the situation can be altered by perturbing the system through the interaction with an environment .
averaging over the various possible effects of this environment will then lead to an entropy increase @xmath47 which can then be employed to measure the system s chaotic nature .
this is more than a convienient mathematical construction . to produce a quantum kicked rotor in an experimental situation
, the free particle motion must be periodically opened up to an environment to allow the `` kick '' to be introduced @xcite .
( though ( [ ev - eq ] ) defines the evolution operator for free motion experiencing an _ instantaneous _ periodic kick , it is equally valid , as long as the free motion is not concurrent @xcite , for a _ finite _ time periodic kick which is what is required to realise this experimentally . ) in doing this the environment itself effects the system which naturally causes the entropy increase required .
thus we choose the environmental coupling to mirror the form of the kick in ( [ hstan ] ) viz . the @xmath3 dependence and interaction time .
we also choose the environment model to be a collection of degenerate two - state atoms with a range of interaction strengths governed by a normal distribution ( this is a generalisation of the class of environments considered by schack and caves @xcite ) , so that @xmath48 thus during the _ n_th kick the rotor interacts with a single two state system with pauli operator @xmath49 and interaction strength @xmath50 .
each of the two - state environment systems is equally likely to be in the `` up '' state @xmath51 , where @xmath52 , where @xmath53 .
also , @xmath50 is drawn from a collection of @xmath54 independent interaction strengths such that @xmath55 for @xmath56 .
the distribution @xmath57 for @xmath58 is the normal distribution @xmath59 .
the combined hamiltonian for the coupled system and environment is thus @xmath60 and the corresponding density operator evolution equation is @xmath61 where the combined evolution operator , @xmath62 with @xmath63 is the result of measuring the two state environment after each interval to determine whether it is an up or down state .
( as before , this same operator would result if ( [ hint ] ) was turned on for the _ finite _ time required for an experimental realisation of this system . )
the effect of this environmental coupling is to produce a multiple stochastic perturbation at the end of each time interval .
after each interval , there are @xmath64 different measurement results leading to @xmath64 possible pure states for the system .
averaging over all these possible outcomes in the position basis produces the density operator evolution equation @xmath65 where @xmath66 now contains all the perturbation effects due to the environment . this causes a vn entropy increase which can be determined by tracing over the system so that , @xmath67 where @xmath68 is the average density matrix of the system after @xmath69 time intervals . one final step will allow us to see the quantum chaotic dynamics
zurek and paz @xcite have conjectured that for an _ open _ quantum system with minimal dissipation which displays classical chaos , the _ rate _ of vn entropy production , @xmath70 , of its quantum analogue , after an initial decoherence time , @xmath71 , will rise to a maximum value which is solely dependent on the sum of its positive lyapunov exponents .
this will continue to be the case until the system begins to approach equilibrium when @xmath70 will slowly decrease reaching zero at time @xmath72 .
in contrast , the entropy production rate of the quantum analogue of a regular systems will asymptotically tend to zero well before @xmath72 . applying this to the standard map , @xmath70 for
the quantised system interacting with an environment should be comparable to the ks entropy of its classical ( unperturbed ) counterpart .
thus for @xmath73 , @xmath74 a uniformly spaced 64 x 64 grid of initial coherent states ( corresponding to an even spread over unit phase space ) were numerically evolved in time . the maximum value of @xmath70 for each evolution was plotted on a contour map in a similar fashion to the classical case .
fig.1(g)-(i ) displays the results for the same three values of @xmath1 with @xmath75 , @xmath76 , @xmath77 , @xmath78 and @xmath79 .
there are remarkable similarities between fig.1(d)-(f ) and fig.1(g)-(i ) . for the same @xmath1 values ,
the size and location of the various stable islands is analogous , dark patches are prevelant in the heavily chaotic regions , the axes of symmetry are consistent and the overall complexity of the dynamics is clearly visible in both .
there are also differences .
the quantum contour maps are generally much smoother than their classical counterparts .
this is because each initial coherent state in the quantum system has a support area causing their evolution to imitate that of a _ density _ of points on phase space . thus neighbouring coherent states
will fail to achieve dramatically different rates of vn entropy production .
increasing @xmath33 reduces the supports of the initial coherent states as well as reducing the overlap between neighbouring states .
it was found that this led to a reduction in the smoothness of the quantum contour maps making them more greatly resemble the classical contour maps .
the process of calculating these entropy contour maps was repeated with variations to @xmath80 . for @xmath81 , when any entropy increase is due primarily to the chaotic dynamics of the system and not the interaction , similar results were achieved . we have demonstrated that an entropy based approach allows classical chaotic dynamics to be accurately measured in a trajectory independent way which in turn makes it eminently suitable to measure @xcite and analyse the corresponding quantum chaotic dynamics .
we have also given numerical support to the zurek and paz conjecture in a chaotic system which folds phase space , a characteristic that their conjecture did not directly take into account .
entropy measures for diagnosing chaotic dynamics can also be employed in other maps .
the sawtooth map @xcite ( which becomes arnold s cat map @xcite for a specific value of the chaos parameter @xmath1 ) does not have a mixed phase space so entropy contours would be of little interest .
however , correspondence _
can _ be investigated by comparing quantum and classical entropy measures for a _ range _ of @xmath1 values .
even more interesting is the kicked top @xcite which is described by a map on the unit sphere . like the standard map
, it has a predominantly mixed phase space for lower values of its chaos parameter making it ideal for comparing classical and quantum entropy contours .
the kicked top also has a special `` order - within - chaos '' @xcite feature .
in general @xmath11 increases monotonically with the chaos parameter @xmath1 of a given map .
however , the kicked top has islands of stability reappearing for specific higher @xmath1 values when the map is already globally chaotic .
this leads to an intricate relationship between @xmath11 and @xmath1 which can be compared to numerical results for the corresponding @xmath70 .
we will discuss the results for these maps elsewhere .
finally , ks entropy is information - theoretically defined as the rate of production of shannon entropy ( also called the shannon information measure ) @xcite .
thus , within well defined parameters , we have shown that the quantum - classical correspondence of chaotic dynamical systems may be realised by viewing the shannon entropy production rate as the classical measure corresponding to the quantum measure of the von neumann entropy production rate .
these results provide a new diagnostic for investigating the chaotic nature of quantum systems .
we are grateful to the esprc for their financial support .
rz would like to thank rdiger schack for useful discussions .
|
von neumann entropy production rates of the quantised kicked rotor interacting with an environment are calculated .
a significant correspondence is found between the entropy contours of the classical and quantised systems .
this is a quantitative tool for describing quantum - classical correspondence in the transition to chaos . 2
|
proteins are fascinating molecules due to their ability to play many roles in biological systems .
their functions often involve complex configurational changes .
therefore the familiar aphorism that `` form is function '' should rather be replaced by a view of the `` dynamic personalities of proteins''@xcite .
this is why proteins are also intriguing for theoreticians because they provide a variety of yet unsolved questions .
besides the dynamics of protein folding , the rise in the time averaged mean square fluctuation @xmath0 occurring at temperatures around @xmath1 , sometimes called the `` protein dynamic transition '' @xcite is arguably the most considerable candidate in the search of unifying principles in protein dynamics .
protein studies lead to the concept of _ energy landscape _ @xcite . according to this viewpoint
a protein is a system which explores a complex landscape in a highly multidimensional space and some of its properties can be related to an incomplete exploration of the phase space .
the protein glass transition , in which the protein appears to `` freeze '' when it is cooled down to about @xmath2k is among them .
protein folding too can be related to this energy landscape .
the famous kinetic limitation known as the levinthal paradox , associated to the difficulty to find the native state among a huge number of possible configurations , is partly solved by the concept of a funneled landscape which provides a bias towards the native state .
these considerations suggest that the dynamics of the exploration of protein phase space deserves investigation , particularly at low temperature where the dynamic transition occurs .
but , in spite of remarkable experimental progress which allows to `` watch protein in action in real time at atomic resolution '' @xcite , experimental studies at this level of detail are nevertheless extremely difficult .
further understanding from models can help in analyzing the observations and developing new concepts .
however , studies involving computer modeling to study the dynamics of protein fluctuations are not trivial either because the range of time scales involved is very large .
this is why many meso - scale models , which describe the protein at scales that are larger than the atom , have been proposed . yet , their validity to adequately describe the qualitative features of a real protein glass remains to be tested . in this paper we examine a model with an intermediate level of complexity .
this frustrated g model @xcite is an off - lattice model showing fluctuations at a large range of time scales .
it is though simple enough to allow the investigation of time scales which can be up to @xmath3 times larger than the time scales of small amplitude vibrations at the atomic level .
the model , which includes a slight frustration in the dihedral angle potential which does not assume a minimum for the positions of the experimentally determined structure , exhibits a much richer behavior than a standard g model .
besides folding one observes a rise of fluctuations above a specific temperature , analogous to a dynamical transition @xcite , and the coexistence of two folded states .
this model has been widely used and it is therefore important to assess to what extent it can describe the qualitative features of protein dynamics beyond the analysis of folding for which it was originally designed .
this is why we focus our attention on its low - temperature properties in an attempt to determine if a fairly simple model can provide some insight on the protein dynamical transition .
the purpose of the present article is to clarify the origin of the transition in the computer model , and to determine similarities and differences with respect to experimental observations .
although the calculations are performed with a specific model , the methods are more general and even raise some questions for experiments , especially concerning the non - equilibrium properties .
this article is organized as follows .
the numerical findings relating to the dynamical transitionfrom previous studies @xcite are presented in sec .
[ sec : model ] . as
a very large body of experimental studies of protein dynamics emanates from neutron scattering experiments , it is rational to seek a connection between theory and experiment by studying the most relevant experimental observable for dynamics , the incoherent structure factor ( isf ) .
we calculate the isf from molecular dynamics simulations of the model in sec .
[ sec : strfact ] .
we show that its main features can be well reproduced by a theoretical analysis based on the one - phonon approximation , which indicates that , at low temperature , the dynamics of the protein within this model takes place in a single minimum of the energy landscape .
[ sec : is ] proceeds to an inherent structure analysis to examine how the transitions among energy states start to play a role when temperature increases . as the freezing of the protein dynamics at low temperature is often called a `` glass transition '' , this raises the question of the properties of the model protein in non - equilibrium situations . in sec .
[ sec : fdt ] we examine the violation of the fluctuation
dissipation theorem after a sudden cooling of the protein .
we find that the effective temperature of the quenched protein , deduced from the fluctuation - dissipation theorem ( fdt ) deviates from the temperature of the thermostat , however it relaxes towards the actual temperature with an an arrhenius behavior as the waiting time increases .
this would imply that the dynamics of the protein model is very slow but not actually glassy .
this method could be useful to distinguish very slow dynamics from glassy dynamics , in experimental cases as well as in molecular dynamics simulations .
finally sec .
[ sec : discussion ] summarizes and discusses our results .
following earlier studies @xcite we chose to study a small protein containing the most common types of secondary structure elements ( @xmath4helix , @xmath5sheets and loops ) , protein @xmath6 , the @xmath7 domain of immunoglobulin binding protein @xcite ( protein data bank code 2gb1 ) .
it contains 56 residues , with one @xmath8helix and four @xmath5strands forming a @xmath9sheet .
we describe it by an off - lattice g model with a slight frustration which represents its geometry in terms of a single particle per residue , centered at the location of each @xmath10 carbon in the experimentally determined tertiary structure .
the interactions between these residues do not distinguish between the type of amino acids .
details on the simulation process and the parametrization of the model are presented in the appendix . in spite of its simplicity
, this model appears to exhibit properties which are reminiscent of the protein dynamical transition .
this shows up when one examines the temperature dependence of its mean - squared fluctuations @xcite by calculating the variance @xmath11 of the residue distances to the center of mass as a function of temperature , defined by @xmath12 here , @xmath13 denotes the number of residues , and @xmath14 is distance of residue @xmath15 with respect to the instantaneous center of of mass .
the average @xmath16 of the observable @xmath17 is the time average @xmath18 . the variances of 20 trajectories ( langevin dynamics simulations , each @xmath19 time units long ) were averaged for each temperature point ( @xmath20 denotes the average over independent initial conditions ) . as a function of temperature for protein g. data adapted from @xcite .
the temperatures marked @xmath21 , @xmath22 , @xmath23 are the temperatures studied in sec .
[ sec : is].,width=287 ] figure [ fig : deltar2 ] shows the evolution of @xmath24 as a function of temperature .
it exhibits a crossover in the fluctuations in the temperature range @xmath25 $ ] resembling the transition observed for hydrated proteins in neutron scattering and mssbauer spectroscopy experiments @xcite , the so - called _
dynamical transition_. above @xmath26 , the fluctuations increase quickly with temperature whereas a smaller , linear , growth is observed below .
one may wonder whether the complexity of the protein structure , reflected by the g model , is sufficient to lead to a dynamical transition or whether , notwithstanding the resemblance of the onset of fluctuations in the present model and the experimentally determined transition , different physical and not necessarily related events may contribute to the curves which by coincidence look similar .
one can already note that , for a folding temperature in the range @xmath27 , the range @xmath28 corresponds to @xmath29 , lower than the experimentally observed transition occurring around @xmath30 .
if it had been confirmed the observation of a dynamical transition in a fairly simple protein model would have been very useful to shine a new light on this transition which is still not fully understood .
it is generally agreed that it is hydration - dependent @xcite , but still different directions for a microscopic interpretation are being pursued , suggesting the existence @xcite or non - existence @xcite of a transition in the solvent coinciding at the dynamical transition temperature .
recently , a completely different mechanism based on percolation theory for the hydration layer has been proposed @xcite .
the precise nature of the interaction between the solvent and proteins , and the driving factor behind the transition , hence still remain to be understood .
the dynamical transition has often been called the _ protein glass transition _ due its similarity with some physical properties of structural glasses at low temperatures . in particular ,
it was pointed out that , for both glasses and protein solutions , the transition goes along with a crossover towards non - exponential relaxation rates at low temperatures .
the comparison is however vague since the glass transition itself and notably its mechanism are ongoing subjects of research and debate .
our goal in this paper is to clarify the origin of the numerically observed transition , which moreover gives hints on the possibilities and limits of protein computer models .
if computer models of proteins are to be useful they must go beyond a simple determination of the dynamics of the atoms , and make the link with experimental observations .
this is particularly important for the `` dynamical transition '' because its nature in a real protein is not known at the level of the atomic trajectories .
it is only observed indirectly through the signals provided by experiments .
therefore a valid analysis of the transition observed in the computer model must examine it in the same context , i.e. determine its consequences on the experimental observations . along with nmr and mssbauer spectroscopy ,
neutron scattering methods have been among the most versatile and valuable tools to provide insight on the internal motion of proteins @xcite .
indeed , the thermal neutron wavelength being of the order of ngstrms and the kinetic energy of the order of @xmath31s , neutrons provide an adequate probe matching the length- and frequency scales of atomic motion in proteins .
an aspect brought forward in the discussion of the dynamical transition in view of the properties of glassy materials is the existence of a _ boson peak _ at low frequencies in neutron scattering spectra @xcite .
such a broad peak appears to be a characteristic feature of unstructured materials as compared to the spectra of crystals . in
neutron scattering the vibrational and conformational changes in proteins appear as a quasielastic contribution to the dynamic structure factor @xmath32 which contains crucial information about the dynamics on different time- and length scales of the system . in scattering experiments
one measures the double - differential scattering cross section @xmath33 which gives the probability of finding a neutron in the solid angle element @xmath34 with an energy exchange @xmath35 after scattering .
the total cross - section of the experiment is obtained by integration over all angles and energies . neglecting magnetic interaction and only considering the short - range nuclear forces ,
the isotropic scattering is characterized by a single parameter @xmath36 , the scattering length of the atomic species @xmath15 @xcite , which can be a complex number with a non - vanishing imaginary part accounting for absorption of the neutron .
if one defines the average over different spin states @xmath37 as the coherent scattering length , and the root mean square deviation @xmath38 as the incoherent scattering length , the double - differential cross section arising from the scattering of a monochromatic beam of neutrons with incident wave vector @xmath39 and final wave vector @xmath40 by @xmath13 nuclei of the sample can be expressed as @xcite @xmath41 where @xmath42 is the wave vector transfer in the scattering process and @xmath43 denote the time - dependent positions of the sample nuclei and the coherent and incoherent dynamical structure factors are @xmath44 the coherent structure factor contains contributions from the position of all nuclei .
the interference pattern of @xmath45 contains the average ( static ) structural information on the sample , whereas the incoherent structure factor @xmath46 monitors the average of atomic motions as it is mathematically equivalent to the fourier transform in space and time of the particle density autocorrelation function . in experiments on biological samples ,
incoherent scattering from hydrogens dominates the experimental spectra @xcite unless deuteration of the molecule and/or solvent are used .
since the g - model represents a reduced description of the protein and the locations of the individual atoms in the residues are not resolved , we use effective incoherent weights of equal value for the effective particles of the model located in the position of the @xmath10-atoms .
such a coarse grained view assumes that the average number of hydrogens atoms and their location in the residues is homogeneous , which is of course a crude approximation in particular in view of the extension and the motion of the side chains .
these approximations are nevertheless acceptable here as we do not intend to provide a quantitative comparison with experimental results considering the simplifications and the resulting limitations of the model .
we generated langevin and nos - hoover dynamic trajectories of length @xmath47 time units , i.e. about 1000 periods of the slowest vibrational mode of the protein , after an equilibration of equal length for protein g at temperatures in the interval @xmath48 $ ] . to compute the incoherent structure factor for the g - model of protein g , we use
_ nmoldyn _ @xcite to analyze the molecular dynamics trajectories generated at different temperatures .
the data are spatially averaged over @xmath49 wave vectors sampling spheres of fixed modules @xmath50 @xmath51 , and the fourier transformation is smoothed by a gaussian window of width @xmath52 of the full length of the trajectory .
prior to the analysis , a root - mean square displacement alignment of the trajectory onto the reference structure at time @xmath53 is performed using virtual molecular dynamics ( vmd ) @xcite .
such a procedure is necessary in order to remove the effects of global rotation and translation of the molecule .
figure [ fig425 ] shows the frequency dependence of the incoherent structure factor @xmath54 for a fixed wave vector @xmath55 @xmath51 for a simulation with the nos - hoover thermostat . on fig .
[ fig425]-a , the evolution of the low frequency range of the structure is shown for a range of temperatures including the supposed dynamic transition region @xmath56 $ ] . at low temperatures up to @xmath57 ,
individual modes are clearly distinguishable and become broadened as temperatures increases .
the slowest mode , located around @xmath58 is also the highest in amplitude .
it has a time constant of about @xmath59 in reduced units ( @xmath60ps ) .
these well - defined lines are observed to be shifting towards lower frequencies with increasing temperature , similar to the phonon frequency shifts that are frequently observed in crystalline solids .
as we show in the following section , the location of these lines can be calculated from a harmonic approximation associated to a single potential energy minimum .
therefore , the shift in frequency and the appearance of additional modes can be seen as a signature of increasingly anharmonic dynamics involving several minima associated to different conformational substates .
to control temperature in the molecular dynamics simulations , several types of numerical thermostats were used .
most of the calculations use underdamped langevin simulations @xcite with a time step @xmath61t.u . and friction constants in the range @xmath62 .
the mass of all the residues is assumed to be equal to @xmath63 .
some calculations were also performed with the multi - thermostat nose - hoover algorithm using the specifications defined in @xcite . for simulation purposes ,
the variables in the g - model are chosen dimensionless ( reduced units ) .
lengths are expressed in units of @xmath64 representing ngstrms , and the average mass of an amino - acid , @xmath65da , has been expressed as @xmath66 units of mass of the model . as the empirical potentials are defined at the mesoscopic scale of the amino - acids , values for the interaction constants in the effective potentials
can not be easily estimated in absolute units .
it is possible to estimate the energy scale of the model by comparing the reduced folding temperature of the g - model with a realistic order of magnitude of the folding temperature @xmath67 in units of @xmath68 and setting @xmath69 where the variables on the left - hand side are given in si units ( @xmath67 is the estimate of the folding temperature ) , and unprimed variables are written in reduced units ; @xmath70 is the required energy scale in units of @xmath71 to match between both . in our calculations
@xmath72 ( meaning that reduced temperatures are expressed in reduced energy units ) and @xmath73 is deduced from equilibrium simulations .
then a simple dimensional analysis gives the time unit of the model as @xmath74 one arrives at an estimated time unit of @xmath75ps . in the paper
we refer to @xmath76ps as the unit of time ( t.u . ) for the simulations of the g - model , keeping in mind that this can merely be seen as an order of magnitude in view of the approximations that lead to this number .
the dissimilarity factor is a weighted distance map between two conformations .
let @xmath77 and @xmath78 be two conformations of the protein ( for our purpose @xmath78 will be the native structure ) and @xmath79 , @xmath80 the distance between residues @xmath15 and @xmath81 in these conformations .
the dissimilarity @xmath82 between the two conformations is defined by @xmath83 where @xmath79 and @xmath80 are the distances between residues @xmath15 and @xmath81 in the @xmath77 and @xmath78 conformations and @xmath84 an integer which determines how much residues which are far apart contribute .
we use here @xmath85 . in this article
the reference conformation a is chosen as the nmr structure , see @xcite for details .
|
prompted by results that showed that a simple protein model , the frustrated g model , appears to exhibit a transition reminiscent of the protein dynamical transition , we examine the validity of this model to describe the low - temperature properties of proteins .
first , we examine equilibrium fluctuations .
we calculate its incoherent neutron - scattering structure factor and show that it can be well described by a theory using the one - phonon approximation . by performing an inherent structure analysis , we assess the transitions among energy states at low temperatures .
then , we examine non - equilibrium fluctuations after a sudden cooling of the protein . we investigate the violation of the fluctuation dissipation theorem in order to analyze the protein glass transition .
we find that the effective temperature of the quenched protein deviates from the temperature of the thermostat , however it relaxes towards the actual temperature with an arrhenius behavior as the waiting time increases .
these results of the equilibrium and non - equilibrium studies converge to the conclusion that the apparent dynamical transition of this coarse - grained model can not be attributed to a glassy behavior .
|
physicists conduct experiments through iterative cycles of design , testing , and refinement , yet undergraduate physics students rarely have opportunities to do the same @xcite.while laboratory ( lab ) courses can provide opportunities for authentic practice , they often employ uninspiring recipe - style activities with little emphasis on designing , building , or troubleshooting experimental approaches @xcite .
this has prompted numerous calls to reform physics lab experiences @xcite , with some arguing authentic inquiry should begin as early as students freshman year @xcite to improve student persistence in physical science majors @xcite .
moreover , there is growing recognition that physics education must focus on self - learning , critical thinking , collaboration and other such skills to help students succeed @xcite .
this paper describes a transformed lab course called _ intro to measurement _ , the third and final course in the berkeley compass project s three - part sequence for first - year undergraduate students @xcite .
the course borrows elements from other transformed lab courses known to be effective @xcite and enhances them with attention to the sociocultural aspects of learning physics @xcite .
_ intro to measurement _ is grounded in the compass value of supporting students holistically @xcite , which means working with students on inter- and intra - personal challenges that may affect their academic success in addition to teaching them science concepts and practices .
this is accomplished by developing a strong student learning community through a summer bridge program ( e.g. , ref .
@xcite ) and strengthening that community through collaborative , project - based coursework during the academic year . by fostering close communities within cohorts of students ,
compass seeks to provide a network of support that empowers undergraduate students , particularly those from underrepresented backgrounds @xcite , to complete physical sciences degrees .
as the final freshman - level compass course , _ intro to measurement _ is intended to help first - year undergraduates develop the skills necessary to be successful as lifelong science learners through guided self - reflection , personalized instructor feedback , and authentic research practices .
the cycle of student self - reflection and instructor feedback described herein is a strategy for providing students with sociocultural support , which is particularly important for students from underrepresented groups in the sciences .
this type of support can take many forms , including : buffering critical feedback by highlighting high expectations and assuring students of their capacity to succeed @xcite ; responding to struggling students by presenting them with strategies for improvement @xcite ; increasing students confidence and enhancing their self - efficacy @xcite ; and generally promoting in students a _ growth mindset _ , _ i.e. _ , the belief that intelligence is malleable and can be changed with effort @xcite . in addition , fostering supportive student / instructor relationships addresses some of the major factors contributing to student attrition in the sciences , namely , the perception of faculty as unapproachable and the rejection of faculty and teaching assistants as role models @xcite . the major connection between the laboratory and reflection activities lies not in anticipated outcomes for students , but in the framing of the activities themselves : in the former context , students are working to improve an experimental apparatus or procedure ; in the latter , they are working to improve themselves as learners .
our goal is not to make quantitative connections between the outcomes of these two activities , but instead to document the compass approach to blending lab instruction with holistic student support .
we report on a study of 80% of the 10 students who completed a recent iteration of _ intro to measurement_. the present work has three goals : to document the design of a laboratory course based on compass holistic learning principles , to understand the course s impact on students measurement concepts , and to describe students reflections and corresponding instructor feedback as related to development of lifelong learning skills .
.[tab:0]demographic breakdown of participants ( @xmath0 ) .
[ cols="<,^",options="header " , ] due to the small size of the cohort in the present work , it is difficult to draw definitive conclusions about the relative impact of our intervention on students reasoning about scientific measurement and uncertainty .
we can , however , provide context for these results by considering previous studies of point and set paradigms in student reasoning . specifically , we consider studies in which the pmq was used to measure shifts in reasoning in both traditional @xcite and reformed @xcite introductory physics lab courses at the university of cape town and in which a survey based on the pmq was used to evaluate a reformed lab course at the university of maryland @xcite . each of the studies involved 50 - 120 students , a much larger population than is considered in the present work .
consistent with the reform courses at maryland @xcite and cape town @xcite , our students had relatively large shifts towards set - like reasoning in comparing data sets ( see table [ tab:3 ] ) .
in contrast , students in the traditional lab courses at cape town @xcite did not show such gains .
we analyzed 73 student reflections and the related instructor feedback from the weekly written reflections over the course of the semester .
in 54 of 73 reflections , students explicitly stated the skills they were focused on ( e.g. , saying `` this week i will focus on self - compassion '' ) . in 12 of the remaining 19 reflections , students used language directly from the rubrics without explicitly stating the skill ( e.g. , writing about `` frustration '' reflects language in the persistence skill prompt ) .
the remaining 7 reflections did not follow the prompt for any particular skill .
students were not limited to one skill per reflection ; in fact , 30% of the reflections contained two or more learning skills . the most popular skills for reflection were organization ( 35% ) , connections ( 18% ) , persistence ( 15% ) , self - compassion ( 9% ) , and courage ( 8% ) .
the other skills were only reflected on a few times each .
thus , more than half of the reflections focused on just three skills , namely , organization , connections , and persistence .
reflections related to organization focused primarily on time management ( 86% ) .
most of these reflections referred to time management specifically ; others used phrases like , `` i want to be efficient with the little time i ve got , '' or `` i did not have the time to keep up with my work . ''
most students were not used to the work load of college , and struggled to keep up .
students made statements such as : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` due to the volume of work to be done ( and the limited length of time ) , i sometimes find myself going through the motions '' _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ and , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` i stayed up all night and managed to get 2 and a half hours of sleep last night . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ reflections about connections focused on creating collaborations with other students , such as a study group ( 63% ) , or in using the office hours of their gsi or instructor ( 37% ) .
we note that the rubric treats building connections as a distinct skill from collaboration which includes , among other things , working productively with others ( appendix a ) .
in contrast , reflections about persistence did not have any clear themes .
students reflected on many different areas , such as : staying focused and not giving up on long homework assignments , personal difficulties staying focused on school , and continuing towards a college degree despite feeling unsuccessful .
instructor responses to student reflections focused on a variety of themes . to support students holistically ,
instructor responses typically included both an affective and instructive component .
the affective responses focused on supporting students through personal struggles they faced ; here are examples of the types of affective responses instructors gave : * affirmation : `` i , personally , am glad you are a part of compass and physics so that i ve had a chance to get to know you . ''
* empathizing : `` figuring out that it is okay and even beneficial to drop a class is something that is hard to deal with .
i know this is something i struggled with as a student . ''
* normalizing : `` impostor syndrome basically means that you think you are the only one struggling and eventually everyone else will figure out that you do nt understand things as well .
everyone ( including myself ) thinks this from time to time . '' on the instructive side of feedback , the instructors suggested resources or strategies that students could use to improve their physics learning .
they also suggested resources that could help students be more successful in their college transition . * suggesting strategies : `` [ m]ake sure you get the most out of every problem you do .
what i mean is to take a few moments after you read a problem and think generally about what concepts or techniques might be useful before you start working .
after you finish the problem , reflect on what you did . ''
* pointing to resources : `` prof .
[ smith ] teaches a class that uses root for some of the homeworks .
his website is here : '' * non - academic support : `` have you looked at padmapper.com for apartments ?
it is useful for off campus housing and it is how i found my apartment . ''
most instructor responses included some aspect of affective support ( 82% ) and concrete instructions for working on a challenge ( 85% ) .
the quality of student and instructor interactions suggested collegial , supportive , and non - traditional student - instructor relationships , which may be connected to students persistence in the sciences @xcite and warrants future study .
the compass project s _ intro to measurement _ course provides an example of a physics laboratory course that emphasizes the iterative nature of research practice through authentic inquiry projects . by using uncertainty analysis to inform improvements on their experimental designs and theoretical models ,
students developed greater understanding of the nature of measurement .
this course further supported students in development of lifelong learning skills through guided self - reflection and personalized instructor feedback . the reflection and
feedback process provided students with affective support , a unique feature of the course .
student responses to the pmq showed that nearly all 8 of the students in the study achieved set - like reasoning by the end of the course .
however , gains in student understanding of the nature of measurement were limited by the ceiling effect . on the pre - instruction survey ,
half of students were already using set - like reasoning most of the time , and none of the students used point - like reasoning most of the time .
most of the students had already completed a compass class or attended the compass summer program , which also emphasize authentic science experiences and metacognitive skills .
this suggests that we should administer future questionnaires to compass students before the summer program to better assess gains from compass interventions .
further research is required to develop a large enough data set to confirm any trends about gains seen in set like reasoning .
student reflections and instructor responses supported meaningful exchanges about topics ranging from coping with the workload of college to dealing with impostor syndrome .
an analysis of students reflections showed that students focused primarily on three areas : organization , connections , and persistence .
moreover , analysis of responses within these categories showed that the majority of students focused on the same issues , such as time management and networking with others .
meanwhile , instructor responses included both affective and instructive components , simultaneously empathizing with students challenges and pointing them towards strategies and resources to help them overcome those challenges .
we believe that targeted support for students to focus lifelong learning skills could be a promising direction for future research . given the importance of authentic introductory lab coursework @xcite and supportive student - instructor relationships @xcite to students persistence in the sciences , _ intro to measurement _ plays an important role in ongoing efforts by compass to promote retention in the sciences at uc berkeley , particularly among students from underrepresented groups .
moreover , due to the introductory nature of the physics concepts covered in the course , elements of _ intro to measurement _ may be adapted for use in high school settings .
the thermal expansion experiment , for instance , is an ideal activity for engaging high school students in developing and using models , analyzing and interpreting data , and other scientific and engineering practices outlined in the _ next generation science standards _ @xcite .
finally , the case of the _ intro to measurement _ course represents an intersection of educational transformation and student empowerment .
in contrast to both traditional , faculty - led transformation efforts ( cf . , ref .
@xcite ) as well as cogenerative reform processes where teachers and their students work together to identify pedagogical improvements ( cf . ,
@xcite ) , _ intro to measurement _ was designed , implemented , and refined exclusively by graduate and undergraduate students @xcite .
such authentic engagement of students as leaders is considered a fundamental aspect of promoting student voice @xcite .
moreover , research on institutional transformation highlights that change efforts which do not afford the actors involved in the change efforts such agency are much less likely to be effective @xcite .
while the general compass approach to educational and organizational reform is outlined elsewhere @xcite , future studies are needed to fully understand the impact of student - led course transformation efforts .
the authors gratefully acknowledge andy disessa and saalih allie for their mentorship and guidance throughout the project , andy buffler and the university of cape town physics education research group for sharing their pmq coding schemes , and ben spike , ben zwickl , gina quan , hilary jacks , joel corbo , and rebbekah semel for fruitful discussions and invaluable feedback .
this work was supported by the berkeley compass project , which in turn receives support from the university of california , berkeley as well as from private donations .
44ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty http://www.nap.edu/catalog.php?record_id=13362[__ ] , edited by , , and ( , ) link:\doibase 10.1038/484440a [ * * , ( ) ] link:\doibase 10.1119/1.19042 [ * * , ( ) ] @noop _ _ , ( , ) @noop _ _ , ( , ) @noop _ _ , ( , ) http://www.nap.edu/openbook.php?record_id=18312[__ ] ( , ) link:\doibase 10.1063/1.4789638 [ * * , ( ) ] link:\doibase 10.1119/1.4768890 [ * * , ( ) ] link:\doibase 10.1119/1.3381077 [ * * , ( ) ] \doibase http://dx.doi.org/10.1119/1.2983148 [ * * , ( ) ] in http://www.per-central.org/document/servefile.cfm?id=4988[__ ] , vol . , ( , ) @noop _ _ , vol .
( , ) @noop _ _ , vol .
( , ) http://www.ncbi.nlm.nih.gov/books/nbk83369/[__ ] ( , ) chap .
link:\doibase 10.1119/1.4772035 [ * * , ( ) ] link:\doibase 10.1177/0146167299258011 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.jesp.2011.12.012 [ * * , ( ) ]
@noop _ _ ( , ) @noop _ _ ( , ) and , eds . ,
@noop _ _ ( , ) _ _ , @noop ph.d .
thesis , ( ) @noop _ _ ( , ) in @noop _ _ , ( , , ) pp .
link:\doibase 10.3102/003465430298487 [ * * , ( ) ] link:\doibase 10.1207/s15430421tip4102_2 [ * * , ( ) ] link:\doibase 10.1119/1.4789875 [ * * , ( ) ] @noop `` , '' ( ) @noop `` , '' ( ) @noop _ _ ( , ) link:\doibase 10.1103/physrevstper.2.020103 [ * * , ( ) ] http://stacks.iop.org/0143-0807/29/i=3/a=024 [ * * , ( ) ] link:\doibase 10.1103/physrevstper.4.010108 [ * * , ( ) ] link:\doibase 10.1002/sce.1012 [ * * , ( ) ] link:\doibase 10.1080/09500690802189807 [ * * , ( ) ] link:\doibase 10.1177/001316446002000104 [ * * , ( ) ] \doibase 10.1002/(sici)1097 - 0258(20000315)19:5<723::aid - sim379>3.0.co;2-a [ * * , ( ) ] link:\doibase 10.1119/1.1881253 [ * * , ( ) ] @noop `` , '' ( ) @noop * * , ( ) in @noop _ _ , ( , ) pp .
link:\doibase 10.1080/10476210600680358 [ * * , ( ) ] http://www.tandfonline.com/doi/abs/10.1080/0141192042000195236 [ * * , ( ) ] http://onlinelibrary.wiley.com/doi/10.1002/tea.20439/full [ * * , ( ) ]
|
aligning science instruction with authentic scientific practices is a national priority in education . in particular , undergraduate laboratory courses have been criticized as employing recipe - style activities with little emphasis on inquiry and design .
this paper offers an alternative laboratory - style course , offered via the compass project at uc berkeley . using a model - based approach , undergraduate physics students engaged in authentic research practices by iteratively refining their experimental and theoretical designs .
the course also promoted lifelong learning skills , such as persistence and organization , through a cycle of student self - reflection and personalized instructor feedback .
this cycle is a strategy for providing students with sociocultural support , which is particularly important for students from underrepresented groups in the sciences .
we document growth in students understanding of scientific measurement and , drawing on student reflections , we suggest areas for future research focused on improving students lifelong learning skills .
|
coronal mass ejections ( cmes ) are large - scale transient eruptions of magnetized plasma from the solar corona that propagate outward into interplanetary space .
their statistical , physical , and morphological properties have been well studied using satellite- and ground - based observational data , especially by using white - light coronagraph data from the large angle and spectrometric coronagraph ( lasco , * ? ? ?
* ) aboard the @xmath0 spacecraft ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
on the other hand , in an effort to understand the initiation mechanism of cmes , a number of numerical simulations have been carried out with different initial magnetic configurations and prescribed motions at the photosphere , i.e. , ( 1 ) a reconnection - favored emerging magnetic flux model ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , ( 2 ) a flux cancellation model ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , ( 3 ) a breakout model ( e.g. , * ? ? ?
* ) , ( 4 ) ideal mhd instabilities such as kink instability ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and torus instability ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , ( 5 ) shear motion of footpoints of magnetic field lines ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. meanwhile , observational studies have demonstrated the common features of magnetic fields and plasma flows in the cme - producing active regions : ( 1 ) highly sheared magnetic fields ( e.g. , * ? ? ?
* ; * ? ? ?
* ) and strong shear flows ( e.g. , * ? ? ?
* ; * ? ? ?
* ) were often observed in the vicinity of the photospheric magnetic polarity inversion lines of cme - producing active regions and ( 2 ) newly emerging magnetic flux ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and sunspot rotations ( e.g. , * ? ? ?
? * ; * ? ? ?
* ) were reported in some of cme - productive active regions as possible mechanisms for cme initiation . however , the magnetic structure and the plasma flow in the cme - prodcuing active regions as well as the triggering mechanisms are still yet to be determined .
it is believed that cmes occur in solar active regions with complex magnetic structures , whose complexity and non - potentiality can be measured by magnetic helicity quantitatively , in terms of twists , kinks and inter - linkages of the magnetic field lines @xcite .
magnetic helicity studies therefore have been carried out to understand magnetic field configurations and pre - eruption conditions in eruptive solar active regions .
@xcite studied a filament in the active region noaa 8375 , and found a steady injection of magnetic helicity through the photosphere over a period of @xmath128 hours involving four eruptive events in the filament .
it was also found that major flares were preceded by a large amount of photospheric helicity injection in their source active regions over a period of a few days ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in particular , there were some observational reports that the injection of oppositely signed magnetic helicity into a pre - existing coronal field might be a crucial ingredient for triggering eruptions of active - region filaments , flares , and cmes : e.g. , @xcite investigated coronal magnetic helicity in the flare - cme - productive active region noaa 10930 , and found that there was not only a large increase of negative magnetic helicity of @xmath23.2@xmath310@xmath4mx@xmath5 in the active region corona over @xmath11.5 days prior to the x3.4 flare and its associated halo cme on 2006 december 13 but also a significant injection of positive magnetic helicity through the photosphere around the flaring magnetic polarity inversion line of the active region ( for other observational reports , refer also to * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . recently , by investigating photospheric magnetic helicity injection in 28 active regions producing 47 cmes
, @xcite found that the 47 cmes under investigation can be categorized into two different groups by the evolution of helicity injection in their source active regions : ( 1 ) a monotonically increasing pattern with one sign of helicity ( group a ; 30 cmes ) and ( 2 ) a pattern of significant helicity injection followed by its sign reversal ( group b ; 17 cmes ) .
in addition , it was found that there was a difference in cme speed , acceleration , and occurrence frequency between group a and group b. the main difference between these two cme groups in the temporal variation of helicity injection is that the helicity sign reversal phase appears in group b , but not in group a. in the study of @xcite , however , were not examined ( 1 ) where oppositely signed helicity is significantly injected during the sign reversal phase and ( 2 ) how its magnitude and distribution evolve with time . in this study , we therefore investigate carefully the spatial distribution and temporal variation of magnetic helicity flux density ( i.e. , magnetic helicity injection per unit area per unit time ) on the photosphere of the very eruptive active region noaa 9236 .
this active region clearly showed the sign reversal phase and produced eight large cmes during the sign reversal phase .
we expect that this study will develop a better understanding of a pre - cme magnetic structure and a trigger mechanism for cme initiation .
the rest of this paper is organized as follows : data selection and analysis for the calculation of magnetic helicity flux density in noaa 9236 are explained in section [ sec2 ] . in section [ sec3 ] ,
we present the observational findings that : ( 1 ) all of eight cme - associated flares occurred in newly emerging magnetic flux regions in the vicinity of the main positive - polarity sunspot of noaa 9236 and ( 2 ) magnetic fields in the cme - associated flaring regions have favorable conditions for making reconnection with large - scale and overlying magnetic fields in the main sunspot regions of positive and negative magnetic polarities .
finally , the results are summarized and discussed in section [ sec4 ] .
the active region noaa 9236 appeared on the east limb of the solar disk on 2000 november 18 , and it was continuously observed during the time interval of its entire disk passage by the @xmath0 spacecraft . at an early stage of the development of noaa 9236 ( i.e. , 2000 november 1822 ) , a group of sunspots in the active region showed a general @xmath6 magnetic configuration of two main sunspots of positive ( preceding ) and negative ( following ) magnetic polarities with a simple and distinct division between the two polarities . from 2000
november 23 , it then changed to a @xmath7 configuration : i.e. , a number of newly emerging small sunspot groups with complex polarities appeared around the main positive - polarity sunspot .
there have been some studies on the properties of ( 1 ) flare - cme events that occurred in the active region noaa 9236 and ( 2 ) magnetic fields on the photosphere .
@xcite examined 6 recurrent halo cmes and their associated flares in noaa 9236 , and reported that the flares were not long - decay events ( lde , * ? ? ?
* ) in terms of soft x - ray light curves and morphologies even though they produced soft x - ray plasma ejections during their impulsive phases .
they also found that soft and hard x - ray emissions from the flares were seen in and around the preceding positive - polarity sunspot .
@xcite studied temporal variation of magnetic flux of small moving magnetic features ( mmfs ) in the vicinity of the main positive sunspot in noaa 9236 by tracking 452 mmfs from birth to death , and found that magnetic flux emergence and disappearance rates in the form of mmfs are 2.3@xmath310@xmath8mx@xmath9hr@xmath10 and 1.9@xmath310@xmath8mx@xmath9hr@xmath10 , respectively , during the period of 2000 november 2324 .
@xcite analyzed multi - wavelength observational data of three x - class homologous flares in noaa 9236 that occurred on 2000 november 24 .
as a result , they found a good correlation between the separation velocity of the two ribbons and the time derivative of the soft x - ray flux for each of the homologous flares , which can be explained by the magnetic energy release rate estimated from magnetic reconnection models for solar flares ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . in this study
, we only took into account the time span of @xmath0/michelson doppler imager ( mdi , * ? ? ?
* ) observations from 2000
november 23 , 11:12 ut to 2000 november 26 , 20:47 ut when the active region was located near the center of the apparent solar disk .
this was to reduce the uncertainty of determined magnetic helicity flux density in noaa 9236 produced in the derivation of magnetic fields perpendicular to the photosphere from mdi measurements of line - of - sight magnetic fields . during the time interval under consideration , eight large cmes , including six halo cmes , occurred in the active region accompanied with flares ( see table [ sec2:tab1 ] for the detailed information of the multiple flare - associated cme events ) .
note that the cmes were identified in the originating active region noaa 9236 by carefully inspecting the locations of soft x - ray and extreme - ultraviolet ( euv ) brightenings of their associated flares , observed by the @xmath11/soft x - ray telescope ( sxt , * ? ? ?
* ) and @xmath0/extreme - ultraviolet imaging telescope ( eit , * ? ? ?
* ) , respectively . by helicity
, we refer to the _ relative _ magnetic helicity @xmath12 in the rest of this paper , i.e. , the gauge - invariant amount of magnetic helicity derived by @xcite : @xmath13 where @xmath14 is the potential field having the same normal component as the magnetic field @xmath15 on the boundary surface of the volume @xmath16 of a given magnetic field system . @xmath17 and @xmath18 are the vector potentials for @xmath15 and @xmath14 , respectively .
helicity injected through the photospheric surface of a solar active region can be determined by taking into account helicity flux density @xmath19 ( i.e. , helicity injection per unit area per unit time ) .
we calculated @xmath20 at a position @xmath21 on the entire surface @xmath22 of noaa 9236 and at a specific time @xmath23 using the well - recognized formula of @xmath19 proposed by @xcite with the help of the numerical calculation method developed by @xcite : @xmath24 \right)_{n } b_{n } ' \,ds',\ ] ] where the subscript @xmath25 indicates the component perpendicular to @xmath22 .
@xmath26 is the magnetic field component perpendicular to @xmath22 , and it is derived from the line - of - sight component @xmath27 of @xmath0/mdi magnetogram data with a spatial resolution of 2@xmath28 per pixel by assuming that the horizontal magnetic field component to @xmath22 is negligible compared to @xmath27 ( i.e. , @xmath26 = @xmath27/@xmath29 where @xmath30 is the heliocentric angle ) .
@xmath31 is the velocity of the apparent horizontal motion of magnetic field line footpoints at the surface @xmath22 determined by the differential affine velocity estimator ( dave ) method @xcite . in applying the dave
, we set the full width at half maximum of a localizing window and the time interval between two successive aligned images as 10@xmath28 and 96 minutes , respectively .
refer to the study of @xcite for the details of the @xmath19 calculation .
note also that many recent studies have used @xmath19 for calculating helicity injection rate in solar active regions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
figure [ sec2:sub2:fig1 ] presents some of @xmath26 , @xmath31 , and @xmath19 maps of the active region noaa 9236 during the time span of 2000
november 2326 : i.e. , the velocity @xmath31 is superposed on a gray - scale image of @xmath26 as red and yellow arrows on positive and negative @xmath26 , respectively , in the left panels , and @xmath19 is shown by a gray - scale image in the right panels . the velocity maps clearly show that there are steady outward motions of mmfs from the main positive sunspot with an average speed of 0.15@xmath9km@xmath9s@xmath10 during the observation time .
it is also shown in @xmath19 maps that there is a continuous injection of positive helicity inside the main positive sunspot , while there is a remarkable injection of negative helicity in mmf regions around the positive sunspot .
the details of temporal variations and spatial distributions of @xmath19 in noaa 9236 are described in section [ sec3 ] .
we calculated the total helicity injection rate @xmath32 through a local region @xmath33 of noaa 9236 by integrating @xmath20 in the area of @xmath33 : @xmath34 then , the accumulated amount of the total helicity injection @xmath35 through @xmath33 was determined by integrating @xmath32 with respect to time : @xmath36 where @xmath37 is the start time of the mdi magnetogram data set of the active region .
positive and negative values of @xmath32 indicate that a dominant sign of helicity injected through @xmath33 per unit time is right - handed and left - handed senses , respectively . as a reference parameter
, we also derived the total unsigned magnetic flux @xmath38 at @xmath33 : @xmath39 it is important to check whether the calculation of @xmath32 is sensitively affected by the per - pixel noise ( @xmath130@xmath9 g ) in mdi magnetograms .
we therefore estimated the uncertainty of @xmath32 that is originated from measurement uncertainty of mdi magnetograms as follows .
first , pseudo - random noises which have normal distribution with the standard deviation of 30 g were added to each magnetogram .
then @xmath32 through the entire region @xmath22 of noaa 9236 was calculated .
the calculation of @xmath32 was repeated 10 times with different sets of errors to calculate the standard deviations of @xmath32 . finally , we calculated the average of the standard deviations during the measurement period of @xmath32 , and considered it as the uncertainty of @xmath32 .
it was found that the uncertainty is 6.8@xmath310@xmath40mx@xmath41hr@xmath10 , i.e. , 3@xmath42 of the average of @xmath32 .
this indicates that the mdi noise does not significantly affect the helicity calculation .
we first checked how helicity injection and magnetic flux , respectively , through and in the entire photospheric surface of the active region evolve in time .
figure [ sec3:fig1 ] presents the time profiles of @xmath32 ( top ) , @xmath35 ( middle ) , and @xmath38 ( bottom ) calculated for @xmath22 of noaa 9236 .
note that the vertical lines in figures [ sec3:fig1 ] , [ sec3:fig4 ] , and [ sec3:fig5 ] represent the first appearance of each of the eight cmes in the field of view ( fov ) of lasco / c2 .
the temporal variations of @xmath32 and @xmath35 show that a relatively large amount of negative helicity started to be steadily injected through @xmath22 for the first 28 hours ( november 23 , 12:00 ut
november 24 , 16:00 ut ) of the helicity measurement with an average helicity injection rate of @xmath211@xmath310@xmath43mx@xmath41hr@xmath10 .
and then the injection of a significant amount of oppositely signed ( positive ) helicity through @xmath22 took place for the next 24 hours ( november 24 , 16:00 ut
november 25 , 16:00 ut ) with an average rate of 15@xmath310@xmath43mx@xmath41hr@xmath10 .
afterwards , @xmath32 through @xmath22 changed its sign every 6 hours until the end of the measurement ( november 25 , 16:00 ut
november 26 , 20:00 ut ) .
this result indicates that both positive and negative helicities were concurrently , significantly , and continuously injected through @xmath22 during 80 hours of the entire measurement period . on the other hand , @xmath38 in @xmath22 kept increasing from 5.6@xmath310@xmath44mx to 7.5@xmath310@xmath44mx with an average rate of 6@xmath310@xmath45mx@xmath9hr@xmath10 .
the increase of @xmath38 is mainly due to newly emerging magnetic flux around the main positive sunspot .
note that the characteristic evolution pattern of significant helicity injection at the photosphere followed by the sign reversal of injected helicity has been well reported by previous studies of photospheric helicity injection in flare - cme - productive active regions ( e.g. , * ? ? ?
* ; * ? ? ?
however , the understanding of the occurrence of flare - assocaited cmes requires further study on the spatial distribution and temporal evolution of the positive and negative components of the injected helicity .
we therefore aligned mdi magnetograms of noaa 9236 with eit 195@xmath9 images to identify local regions where intense euv emission occurred from the cme - associated flares .
figure [ sec3:fig2 ] shows the active region mdi maps ( gray - scale images ) with the aligned eit contours at the times of the eight cme - associated flares .
as shown in figure [ sec3:fig2 ] , all the cme - associated flares occurred in the vicinity of the main positive sunspot : i.e. , the m1.0 flare on 2000 november 23 and the following seven flares occurred in the northern and western neighboring regions of the main positive sunspot , respectively . for a more detailed understanding of morphological characteristics of magnetic fields in the flaring regions , we selected a local area enclosing emerging flux regions around the main positive sunspot ( see the dotted - line box in figure [ sec3:fig3]a ) and examined a time series of @xmath26 maps of the local area for the period of 2000 november 2326 with a time cadence of 5 hours , as shown in figure [ sec3:fig3]b . in figure [ sec3:fig3]a ,
r@xmath46 and r@xmath47 represent the main positive and negative - polarity sunspot regions .
e@xmath48e@xmath49 indicate the emerging flux regions which are located in the eastern , northern , western , and southern part of the positive sunspot , respectively : e@xmath50 and e@xmath51 regions correspond to the flaring regions , while e@xmath52 and e@xmath49 represent non - flaring regions .
figure [ sec3:fig3]b shows that the emerging flux regions are mainly composed of small - scale bipolar magnetic fragments .
we especially marked some bipoles of emerging magnetic fragments which appear around the times and in the locations of the eight flares with rectangular boxes in the panels of figure [ sec3:fig3]b .
the bipolar magnetic fragment in e@xmath50 ( as indicated by the box at the panel b4 of figure [ sec3:fig3]b ) first appeared at 16:00 ut on 2000 november 23 , developed over a few hours , and gradually disappeared after the m1.0 flare at 23:18 ut on 2000 november 23 . in the multiple - flare - productive region e@xmath51 , we also found relatively larger and more elongated bipolar fragments compared to that in e@xmath50 during the flare occurrence times of 2000 november 24 - 26 .
it is noteworthy that @xcite reported elongated magnetic structures in e@xmath51 , which is associated with the x1.8 flare on 2000 november 24 , by examining a high - resolution line - of - sight magnetogram obtained with the big bear solar observatory s ( bbso ) digital vector magnetograph system ( dvmg , * ? ? ?
in addition , two bipolar pairs of emerging magnetic fragments located in e@xmath52 and e@xmath49 , respectively , are marked with the circles in the panel b12 of figure [ sec3:fig3]b .
next , we determined the temporal evolution of @xmath32 , @xmath35 , and @xmath38 in the local region of the main positive and negative sunspots ( see figure [ sec3:fig4 ] ) , taking account of equations [ sec2:sub2:equ3 ] , [ sec2:sub2:equ4 ] , and [ sec2:sub2:equ5 ] with the integrated area of r@xmath46 and r@xmath47 combined . in figure [ sec3:fig4 ] , the time variations of @xmath32 and @xmath35 show the following evolution patterns : ( 1 ) a phase of nearly constant @xmath35 of 80@xmath310@xmath43mx@xmath53 lasted for the first 28 hours ( november 23 , 12:00 ut
november 24 , 16:00 ut ) after an abrupt and large injection of positive helicity at the beginning of the helicity measurement , ( 2 ) @xmath35 kept on significantly and monotonically increasing up to 500@xmath310@xmath43mx@xmath53 at an average rate of 12.5@xmath310@xmath43mx@xmath41hr@xmath10 over the next 40 hours ( november 24 , 16:00 ut
november 26 , 8:00 ut ) , and ( 3 ) it decreased by 100@xmath310@xmath43mx@xmath53 for the next 4.5 hours and remained in a phase of its constant value of 400@xmath310@xmath43mx@xmath53 until the end of the measurement . on the other hand , in the case of @xmath38 , it changed little for the entire measurement period with only a slight decrease by @xmath110@xmath42 from 160@xmath310@xmath8mx to 140@xmath310@xmath8mx . finally , as shown in figure [ sec3:fig5 ] , we also calculated and plotted the time variations of @xmath32 , @xmath35 , and @xmath38 in each of the emerging flux regions e@xmath48e@xmath49 .
the following are the main findings from figure [ sec3:fig5 ] : ( 1 ) the flaring region e@xmath50 shows a considerable injection of negative helicity ( i.e. , a decrease of @xmath35 of @xmath220@xmath310@xmath43mx@xmath53 ) for 4.5 hours before the m1.0 flare and the associated cme on 2000-nov-23 , accompanied by an increase of @xmath38 from 6@xmath310@xmath8mx to 13@xmath310@xmath8mx with an average rate of 15.6@xmath310@xmath54mx@xmath9hr@xmath10 , ( 2 ) the multiple flaring region e@xmath51 also presents a similar pattern in the temporal variation of @xmath35 during the three days of seven cme events : i.e. , a significant amount of negative helicity was injected through the region e@xmath51 over 72 hours so that @xmath35 reached @xmath2180@xmath310@xmath43mx@xmath53 with an average rate of @xmath22.5@xmath310@xmath43mx@xmath41hr@xmath10 .
in addition , there was a large increase of @xmath38 from 10@xmath310@xmath8mx to 100@xmath310@xmath8mx with an average rate of 10.9@xmath310@xmath54mx@xmath9hr@xmath10 during the entire time period , ( 3 ) the non - flaring region e@xmath52 shows a continuous increase in the magnitude of the negative helicity , however , @xmath38 only increased from 30@xmath310@xmath8mx to 40@xmath310@xmath8mx with a much smaller average rate of 1.8@xmath310@xmath54mx@xmath9hr@xmath10 during the entire period , and ( 4 ) the non - flaring region e@xmath49 has a significant amount of injection of positive helicity , which is of the same sign as in the main sunspot , accompanied by an increase of @xmath38 .
we now have the question of why cme - related flares occurred in e@xmath50 and e@xmath51 , but not in e@xmath52 and e@xmath49 . in the long - term ( a few days )
variation of helicity injection , there seems to be no significant difference between the flaring and non - flaring regions because they in general show a similar trend in the time variation of @xmath35 , i.e. , a monotonically increasing trend with one sign of helicity during the entire investigation .
furthermore , the magnitudes of @xmath32 , @xmath35 , and @xmath38 in the flaring regions are comparable to those in the non - flaring regions .
however , comparison of the helicity in the bipolar magnetic fragments of e@xmath50 and e@xmath51 and the main sunspots suggests that the magnetic fields in e@xmath50 and e@xmath51 ( especially in the emerging bipoles outlined by the rectangular boxes in figure [ sec3:fig3]b ) have substantially favorable conditions for making reconnection with those in r@xmath46 and r@xmath47 .
the simplified cartoon of figure [ sec3:fig6 ] describes a top view of the magnetic field lines connecting r@xmath46 and r@xmath47 as well as bipoles of the magnetic fragments in each of e@xmath48e@xmath49 .
the large - scale kinked structures of a coronal magnetic field line can be inferred from the sign of injected helicity through the photosphere , as shown by previous studies ( e.g. , * ? ? ?
* ; * ? ? ?
* ) : i.e. , coronal fields in general display a forward - s ( inverse - s ) shape with positive ( negative ) helicity . in figure [ sec3:fig6 ] ,
the small - scale , under - lying , and inverse - s - shaped field lines ( red lines ) in the emerging bipoles of magnetic fragments in e@xmath50 and e@xmath51 may interact and reconnect with the large - scale , overlying , s - shaped , and oppositely directed field lines ( green lines ) in r@xmath46 and r@xmath47 .
on the other hand , we speculate that the inferred field lines ( blue lines ) of an inverse - s and s shape in e@xmath52 and e@xmath49 , respectively , may pass underneath the field lines ( green lines ) in r@xmath46 and r@xmath47 so that it would be difficult to make a reconnection between the emerging and overlying fields .
another noticeable difference is that the average increasing rate of the total unsigned magnetic flux @xmath55 in e@xmath51 during the entire time period is relatively larger than in the other regions : i.e. , 10.9@xmath310@xmath54mx@xmath9hr@xmath10 in the most eruptive region e@xmath51 where seven out of the eight flares occurred , but ( 1.8 , 1.1 , and 7.0)@xmath310@xmath54mx@xmath9hr@xmath10 in e@xmath52 , e@xmath50 , and e@xmath49 , respectively .
note that the flaring region e@xmath50 also shows a large value of @xmath55 of 15.6@xmath310@xmath54mx@xmath9hr@xmath10 during 4.5 hours right before the m1.0 flare .
in addition , we found that e@xmath51 shows not only a continuous increase of @xmath38 during the seven flares but also an impulsive injection of negative helicity a few hours before each of the flares , as shown in figure [ sec3:fig5 ] .
we therefore conjecture that the continuous emergence of magnetic bipoles with an opposite helicity to the overlying fields may be responsible for the occurrence of the multiple flare events in e@xmath51 .
the detailed investigation of the spatial distribution and temporal variation of magnetic helicity injection in a cme - productive active region has been carried out to better understand a preferred magnetic field configuration and its evolution during cme events . in this study , we determined and examined the time variations of the total helicity injection rate @xmath32 , the accumulated amount of the total helicity injection @xmath35 , and the total unsigned magnetic flux @xmath38 not only in the entire active region noaa 9236 but also in some local regions ( i.e. , the main positive and negative sunspot regions , r@xmath46 and r@xmath47 , and the emerging flux regions , e@xmath48e@xmath49 , around r@xmath46 ) .
note that eight cmes and their associated flares occurred in e@xmath50 and e@xmath51 during the investigation period ( 2000 november 2326 ) . as a result
, we found that : ( 1 ) a large amount of positive helicity was injected through r@xmath46 and r@xmath47 during most of the measurement time , ( 2 ) a continuous injection with negative sign of helicity took place in the cme - associated flaring regions e@xmath50 and e@xmath51 during the entire measurement period , accompanied by an increase of magnetic flux ( especially , with a large increasing rate of @xmath38 in the most eruptive region e@xmath51 ) , and ( 3 ) e@xmath50 and e@xmath51 are mainly composed of emerging bipoles of magnetic fragments in which magnetic field lines have substantially favorable conditions for making reconnection with large - scale , overlying , and oppositely directed magnetic field lines in r@xmath46 and r@xmath47 .
we also found that the observational results during the eight flare - associated cmes events can be well explained by reconnection - favored emerging magnetic flux models ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
for example , in the two - dimensional cme model of @xcite , they set up a newly emerging magnetic flux system , of which magnetic fields oppositely directed to and may reconnect with those of a nearby large - scale magnetic flux system , and showed that the large - scale system could be totally destabilized and erupted due to a localized reconnection between the two systems . although the coronal magnetic field illustrated in figure [ sec3:fig6 ] is much more complicated than that in the simple two - dimensional model , it is likely that some internal reconnections between large - scale overlying field lines ( r@xmath46 to r@xmath47 ) and small - scale emerging bipolar fields ( e@xmath50 and e@xmath51 ) may destabilize the magnetic system and give rise to cmes .
in fact , @xcite recently found that the emergence of small - scale flux could trigger large - scale solar eruption even in three - dimensional ( 3d ) systems , if the emerging flux has the magnetic helicity with the sign opposite to the overlying field or the magnetic polarity of the emerging flux is oppositely directed to the large - scale field .
the small - scale magnetic fluxes in regions e@xmath50 and e@xmath51 are well consistent with the former case .
furthermore , @xcite showed in their 3d numerical simulations of an emerging magnetic flux rope that the successful eruption of the flux rope results from the effective reconnection between the ambient field overlaying the rope and pre - existing coronal field : i.e. , the downward tension force of the ambient field above the flux rope is reduced via the reconnection so that the rope can erupt .
we therefore conclude that reconnection - favored magnetic fields in the flaring emerging flux regions @xmath56 and @xmath57 play a crucial role in producing the multiple flare - associated cme events in the active region noaa 9236 . as a future work , more
flare - associated cmes will be examined by using high spatial and temporal resolution data observed by the solar dynamics observatory ( @xmath58 , * ? ? ?
* ) to better understand the long - term precondition and onset mechanism of reconnection - driven cmes .
the authors thank the anonymous referee for highly detailed and valuable comments to improve the level of this paper .
we thank the teams of soho / eit , soho / lasco , and soho / mdi for providing the valuable data .
soho is a project of international cooperation between esa and nasa .
we acknowledge use of the cme catalog generated and maintained at the cdaw data center by nasa and the catholic university of america in cooperation with the naval research laboratory . this research has made use of nasa s astrophysics data system abstract service .
this work has been support by the development of korea space weather prediction center " project of kasi and the kasi basic research fund .
this work was partially supported by a grants - in - aid for scientific research ( b ) understanding and prediction of triggering solar flares " ( 23340045 , head investigator : k. kusano ) from the ministry of education , science , sports , technology , and culture of japan .
antiochos , s. k. , devore , c. r. , & klimchuk , j. a. 1999 , , 510 , 485 archontis , v. , & trk , t. 2008 , , 492 , l35 bateman , g. 1978 , mhd instability ( cambridge , ma : mit press ) berger , m. a. , & field , g. b. 1984 , journal of fluid mechanics , 147 , 133 brueckner , g. e. , howard , r. a. , koomen , m. j. , et al .
1995 , , 162 , 357 chae , j. 2007 , advances in space research , 39 , 1700 chen , p. f. , & shibata , k. 2000 , , 545 , 524 delaboudinire , j .- p . , artzner , g. e. , brunaud , j. , et al .
1995 , , 162 , 291 dmoulin , p. , & pariat , e. 2009 , advances in space research , 43 , 1013 falconer , d. a. , moore , r. l. , & gary , g. a. 2002 , , 569 , 1016 falconer , d. a. , moore , r. l. , & gary , g. a. 2006 , , 644 , 1258 fan , y. , & gibson , s. e. 2004 , , 609 , 1123 feynman , j. , & martin , s. f. 1995 , , 100 , 3355 finn , j. m. & antonsen , t. m. 1985 , comments plasma phys . controlled fusion , 9 , 111 gopalswamy , n. 2006 , journal of astrophysics and astronomy , 27 , 243 gopalswamy , n. , yashiro , s. , michalek , g. , et al . 2009 ,
earth moon and planets , 104 , 295 hood , a. w. , & priest , e. r. 1981 , geophysical and astrophysical fluid dynamics , 17 , 29 isobe , h. , yokoyama , t. , shimojo , m. , et al .
2002 , , 566 , 528 jing , j. , yurchyshyn , v. b. , yang , g. , xu , y. , & wang , h. 2004 , , 614 , 1054 jing , j. , park , s .- h . , liu , c. , et al .
2012 , , 752 , l9 kahler , s. 1977 , , 214 , 891 kliem , b. , & trk , t. 2006 , physical review letters , 96 , 255002 kumar , p. , manoharan , p. k. , & uddin , w. 2010 , , 710 , 1195 kumar , p. , park , s .- h . , cho , k .- s . , & bong , s .- c . 2013 , , 282 , 503 kusano , k. , maeshiro , t. , yokoyama , t. , & sakurai , t. 2003a , advances in space research , 32 , 1917 kusano , k. , yokoyama , t. , maeshiro , t. , & sakurai , t. 2003b , advances in space research , 32 , 1931 kusano , k. , maeshiro , t. , yokoyama , t. , & sakurai , t. 2004 , , 610 , 537 kusano , k. , bamba , y. , yamamoto , t. t. , et al .
2012 , , 760 , 31 labonte , b. j. , georgoulis , m. k. , & rust , d. m. 2007 , , 671 , 955 linker , j. a. , lionello , r. , miki , z. , & amari , t. 2001 , , 106 , 25165 liu , y. , & schuck , p. w. 2012 , , 761 , 105 luoni , m. l. , dmoulin , p. , mandrini , c. h. , & van driel - gesztelyi , l. 2011 , , 270 , 45 manchester , w. 2008 , subsurface and atmospheric influences on solar activity , 383 , 91 mikic , z. , barnes , d. c. , & schnack , d. d. 1988 , , 328 , 830 nitta , n. v. , & hudson , h. s. 2001 , , 28 , 3801 pariat , e. , dmoulin , p. , & berger , m. a. 2005 , , 439 , 1191 park , s .- h .
, lee , j. , choe , g. s. , chae , j. , jeong , h. , yang , g. , jing , j. , & wang , h. 2008 , , 686 , 1397 park , s .- h . ,
chae , j. , & wang , h. 2010a , , 718 , 43 park , s .- h . ,
chae , j. , jing , j. , tan , c. , & wang , h. 2010b , , 720 , 1102 park , s .- h . , cho , k .- s . ,
bong , s .- c . , et al .
2012 , , 750 , 48 pesnell , w. d. , thompson , b. j. , & chamberlin , p. c. 2012 , , 275 , 3 pevtsov , a. a. 2008 , journal of astrophysics and astronomy , 29 , 49 romano , p. , contarino , l. , & zuccarello , f. 2003 , , 218 , 137 romano , p. , pariat , e. , sicari , m. , & zuccarello , f. 2011 , , 525 , a13 romano , p. , & zuccarello , f. 2011 , , 535 , a1 scherrer , p. h. , bogart , r. s. , bush , r. i. , et al .
1995 , , 162 , 129 schuck , p. w. 2006 , , 646 , 1358 shibata , k. 1999 , , 264 , 129 smyrli , a. , zuccarello , f. , romano , p. , et al .
2010 , , 521 , a56 takasaki , h. , asai , a. , kiyohara , j. , et al .
2004 , , 613 , 592 tan , c. , chen , p. f. , abramenko , v. , & wang , h. 2009 , , 690 , 1820 titov , v. s. , & dmoulin , p. 1999 , , 351 , 707 trk , t. , & kliem , b. 2005 , , 630 , l97 trk , t. , temmer , m. , valori , g. , et al .
2013 , , 70 tsuneta , s. , acton , l. , bruner , m. , et al .
1991 , , 136 , 37 van ballegooijen , a. a. , & martens , p. c. h. 1989 , , 343 , 971 wang , h. , denker , c. , spirock , t. , et al .
1998 , , 183 , 1 wang , h. , gallagher , p. , yurchyshyn , v. , yang , g. , & goode , p. r. 2002 , , 569 , 1026 wang , j. , zhou , g. , & zhang , j. 2004 , , 615 , 1021 yamamoto , t. t. , kusano , k. , maeshiro , t. , yokoyama , t. , & sakurai , t. 2005 , , 624 , 1072 yan , x. l. , qu , z. q. , kong , d. f. , & xu , c. l. 2012 , , 754 , 16 yang , g. , xu , y. , cao , w. , et al .
2004 , , 617 , l151 yashiro , s. , gopalswamy , n. , michalek , g. , et al .
2004 , journal of geophysical research ( space physics ) , 109 , 7105 yokoyama , t. , kusano , k. , maeshiro , t. , & sakurai , t. 2003 , advances in space research , 32 , 1949 zhang , j. , & wang , j. 2002 , , 566 , l117 zuccarello , f. p. , romano , p. , zuccarello , f. , & poedts , s. 2011 , , 530 , a36 ccccrccccc 1 & 2000 - 11 - 23 23:54 & 336/157 & 690 & 1.8 & 23:18 & 23:28 & 23:37 & m1.0 + 2 & 2000 - 11 - 24 05:30 & halo/360 & 1289 & 2.1 & 04:55 & 05:02 & 05:08 & x2.0 + 3 & 2000 - 11 - 24 15:30 & halo/360 & 1245 & @xmath23.3 & 14:51 & 15:13 & 15:21 & x2.3 + 4 & 2000 - 11 - 24 22:06 & halo/360 & 1005 & @xmath20.8 & 21:43 & 21:59 & 22:12 & x1.8 + 5 & 2000 - 11 - 25 09:30 & halo/360 & 675 & @xmath24.7 & 09:06 & 09:20 & 09:40 & m3.5 + 6 & 2000 - 11 - 25 19:31 & halo/360 & 671 & @xmath210.8 & 18:33 & 18:44 & 18:55 & x1.9 + 7 & 2000 - 11 - 26 03:30 & 259/118 & 495 & @xmath222.9 * & 02:47 & 03:08 & 03:20 & m2.2 + 8 & 2000 - 11 - 26 17:06 & halo/360 & 980 & 5.8 & 16:34 & 16:48 & 16:56 & x4.0 + crrrr entire & @xmath21.0 & @xmath270 & 660 & 23.1 + r@xmath59r@xmath47 & 4.7 & 390 & 160 & @xmath21.6 + e@xmath52 & @xmath23.0 & @xmath2250 & 40 & 1.8 + e@xmath50 & @xmath20.8 & @xmath270 & 10 & 1.1 + e@xmath51 & @xmath22.1 & @xmath2170 & 50 & 10.9 + e@xmath49 & 4.0 & 330 & 40 & 7.0 +
|
to better understand a preferred magnetic field configuration and its evolution during coronal mass ejection events , we investigated the spatial and temporal evolution of photospheric magnetic fields in the active region noaa 9236 that produced eight flare - associated cmes during the time period of 2000 november 2326 .
the time variations of the total magnetic helicity injection rate and the total unsigned magnetic flux are determined and examined not only in the entire active region but also in some local regions such as the main sunspots and the cme - associated flaring regions using @xmath0/mdi magnetogram data . as a result
, we found that : ( 1 ) in the sunspots , a large amount of postive ( right - handed ) magnetic helicity was injected during most of the examined time period , ( 2 ) in the flare region , there was a continuous injection of negative ( left - handed ) magnetic helicity during the entire period , accompanied by a large increase of the unsigned magnetic flux , and ( 3 ) the flaring regions were mainly composed of emerging bipoles of magnetic fragments in which magnetic field lines have substantially favorable conditions for making reconnection with large - scale , overlying , and oppositely directed magnetic field lines connecting the main sunspots .
these observational findings can also be well explained by some mhd numerical simulations for cme initiation ( e.g. , reconnection - favored emerging flux models ) .
we therefore conclude that reconnection - favored magnetic fields in the flaring emerging flux regions play a crucial role in producing the multiple flare - associated cmes in noaa 9236 .
|
what is _ causality _ ?
| the notion has been defined in different ways and turned out to be highly problematic , both in physics and philosophy .
this observation is not new , as is nicely shown by _
bertrand russell _ s quote @xcite from more than a century ago : + _ `` the law of causality [ ] is a relic of a bygone age , surviving , like the monarchy , only because it is erroneously supposed to do no harm . '' _ + indeed , a number of attempts have been made to abandon causality and replace global by only local assumptions ( see , _
e.g. _ , @xcite ) .
a particular motivation is given by the difficulty of explaining quantum non - local correlation according to _
reichenbach s principle _ @xcite .
the latter states that in a given ( space - time ) causal structure , correlations stem from _ a common cause _ ( in the common past ) or a _ direct influence _ from one of the events to the other . in the case of violations of bell s inequalities ,
a number of results indicate that explanations through some mechanism as suggested by reichenbach s principle either fail to explain the correlations @xcite or are unsatisfactory since they require infinite speed @xcite , @xcite , @xcite , @xcite or precision @xcite .
all of this may serve as a motivation for dropping the assumption of a global causal structure in the first place .
closely related to causality is the notion of _ randomness _ : in @xcite , a piece of information is called _ freely random _ if it is statistically independent from all other pieces of information except the ones in its future light cone .
clearly , when the assumption of an initially given causal structure is dropped , such a definition is not possible any longer .
one may choose to consider freely random pieces of _ information _ as being more fundamental than a space - time structure | in fact , the latter can then be seen as emerging from the former : if a piece of information is free , then any piece correlated to it is in its causal future . but how can we _ define _ the randomness of an object _ purely intrinsically _ and independently of any context ? for further motivation , note that colbeck and renner s definition of randomness @xcite is consistent with full determinism : a random variable with trivial distribution is independent of every other ( even itself ) .
how can we exclude this and additionally ask for the possibility in principle of a counterfactual outcome , _
i.e. _ , that the random variable @xmath0 _ could have taken a value different from the one it actually took _ ? intuitively , this is a necessary condition for _ freeness_. the question whether the universe ( or a _ closed _
system ) starting from a given state @xmath1 always ends up in _ the same _ state @xmath2 seems to be meaningless : even if rewinding were possible , and two runs could be performed , the outcomes @xmath3 and @xmath4 that must be compared never exist in the same reality since rewinding erases the result of the rewound run @xcite : `` @xmath5 ? '' is _ not a question which can not be answered in principle , but that can not even be formulated precisely_. in summary , defining freeness of a choice or a random event , understood as the _ actual possibility of two ( or more ) well - distinguishable options _
, seems hard even when a causal structure _ is _ in place .
we look for an _ intrinsic _ definition of randomness that takes into account only the `` factuality , '' _
i.e. _ , the state of the closed system in question .
clearly , such a definition is hard to imagine for a single bit , but it _ can _ be defined in a natural way for ( long ) strings of bits , namely its length minus the work value ( normalized through dividing by @xmath6 ) of a physical representation of the string with respect to some extraction device ; we relate this quantity to the string s `` best compression . ''
we test the alternative view of randomness for physical meaning .
more specifically , we find it to be functional in the context of _ non - local correlations _ : a reasoning yielding a similar mechanism as in the probabilistic regime is realized which has the conceptual advantage not to require relating the outcomes of measurements that can not all actually be carried out .
that mechanism is : random inputs to a non - local system plus no - signaling guarantee random outputs . in the second half of this text
, we consider consequences of abandoning ( space - time ) causality as being fundamental . in a nutshell
, we put _ logical reversibility _ to the center of our attention here .
we argue that if a computation on a turing machine is logically reversible , then a `` second law '' emerges : the complexity of the tape s content can not decrease in time .
this law holds without failure probability , in contrast to the `` usual '' second law , and implies the latter . in the same spirit
, we propose to define causal relations between physical points , modeled by bit strings , as given by the fact that `` the past is entirely contained in the future , '' _
i.e. _ , nothing is forgotten . in this view
, we also study the relationship between full causality ( which we aim at dropping ) and mere logical consistency ( that we never wish to abandon ) in the complexity view : they are different from each other as soon as more than two parties are involved .
let @xmath7 be a fixed universal turing machine ( tm ) . for a finite or infinite string @xmath8 , the _ kolmogorov complexity _
@xcite , @xcite is the length of the shortest program for @xmath7 such that the machine outputs @xmath8 . note that @xmath9 can be infinite if @xmath8 is .
let @xmath10 be an infinite string .
then @xmath11}:=(a_1,\ldots , a_n,0,\ldots)\ .\ ] ] we study the asymptotic behavior of @xmath12})\ , : \ , { \bf n}\rightarrow { \bf n}$ ] . for this function ,
we simply write @xmath13 , similarly @xmath14 for @xmath12}\ , |\ , b_{[n]})$ ] , the latter being the length of the shortest program outputting @xmath15}$ ] upon input @xmath16}$ ] . we write @xmath17})}{n}\right ) = 1\ .\ ] ] we call a string @xmath18 with this property _ incompressible_. we also use @xmath12})=\theta(n)$ ] , as well as @xmath19})}{n}\right ) = 0 \longleftrightarrow k(a_{[n]})=o(n).\ ] ] note that _ computable _ strings @xmath18 satisfy @xmath20 , and that incompressibility is , in this sense , the extreme case of uncomputability . generally , for functions
@xmath21 and @xmath22 , we write @xmath23 if @xmath24 .
_ independence of @xmath18 and @xmath25 _ is then @xmath26 or , equivalently , @xmath27 if we introduce @xmath28 independence of @xmath18 and @xmath25 is @xmath29 . in the same spirit
, we can define _ conditional independence _ : we say that _ @xmath18 and @xmath25 are independent given @xmath30 _ if @xmath31 or , equivalently , @xmath32 or @xmath33
in our search for an _ intrinsic _ notion of randomness | independent of probabilities or the existence of alternatives | expressed through the properties of the object in question , we must realize , first of all , that such a notion is impossible for single bits , since neither of the two possible values , @xmath34 nor @xmath35 , is in any way more an argument for the `` randomness '' of that bit than not . the situation , however , changes for _ long _ strings of bits : no one would call the one - million - bit string @xmath36 _ random _ ( even though , of course , it is not impossible that this string originates from a random process such as a million consecutive tosses of a fair coin ) . in the spirit of rolf landauer s
@xcite famous slogan `` information is physical , '' we may want to test our intuition physically : if the @xmath37 bits in a string encode the position , being in the left ( @xmath34 ) as opposed the right ( @xmath35 ) half of some container , of the molecules of a gas , then the @xmath34-string means that the gas is all concentrated in one half and , hence , allows for extracting work from the environmental heat ; the amount is @xmath38 if @xmath39 is boltzmann s constant and @xmath40 is the temperature of the environment .
this fact has also been called the _ converse of landauer s principle_. note that any other system which can be transformed by a reversible process into that maximally asymmetric gas has the same work value ; an example is a physical representation of the first @xmath37 bits of the binary expansion of @xmath41 of the same length | although this string may look much more `` random '' at first sight .
this reversible process is , according to the _ church - turing thesis _ , imagined to be carried out by a turing machine in such a way that every step is _ logically reversible _
( such as , _
e.g. _ , a toffoli gate ) and can be _
un_computed by the same device ; the process is then also possible in principle in a _ thermodynamically _ reversible way : no heat is dissipated @xcite .
it is clear that most @xmath37-bit strings can not have any work value provided there is no _ perpetuum mobile of the second kind_. for a given string @xmath42 , its _ length minus the work value of a physical representation _ ( divided by @xmath6 ) may be regarded as an intrinsic measure for the _ randomness _ of @xmath42 .
we address the question what in general the fuel value is of ( a physical representation of ) @xmath42 .
since ( the _ reversible _ extraction of ) the string @xmath43 from @xmath42 is equivalent to ( the gain of ) free energy of @xmath38 , we have a first answer : _ work extraction is data compression_. _ state of the art . _
bennett @xcite claimed the fuel value of a string @xmath42 to be _ its length minus @xmath44 _ : @xmath45 bennett s argument is that ( the physical representation of ) @xmath42 can be | logically , hence , thermodynamically @xcite | reversibly transformed into the string @xmath46 , where @xmath47 is the shortest program for @xmath7 generating @xmath42 and the length of the generated @xmath48 is @xmath49 ( see figure [ fig:1 ] ) .
it was already pointed out by zurek @xcite that whereas it is true that the _ reverse direction _ exists and is _ computable _ by a universal turing machine , its _ forward direction _ , _
i.e. _ , @xmath47 from @xmath42 , is _
not_. this means that the demon that can carry out the work - extraction computation on @xmath42 from scratch does not physically exist if the church - turing hypothesis is true .
we will see , however , that bennett s value is an _ upper bound _ on the fuel value of @xmath42 .
dahlsten _ et al . _
@xcite follow szilrd @xcite in putting the _ knowledge _ of the demon extracting the work to the center of their attention .
more precisely , they claim @xmath50 where the `` defect '' @xmath51 is bounded from above and below by a smooth rnyi entropy of the distribution of @xmath42 from the demon s viewpoint , modeling her ignorance .
they do not consider the algorithmic aspects of the demon s actions extracting the free energy , but the effect of the demon s _ a priori knowledge on @xmath42_. if we model the demon as an algorithmic apparatus , then we should specify the _
form _ of that knowledge explicitly : vanishing conditional entropy means that @xmath42 is _ uniquely determined _ from the demon s viewpoint .
does this mean that the demon possesses a _ copy _ of @xmath42 , or the _ ability _ to produce such a copy , or pieces of _ information _ that uniquely determine @xmath42 ?
this question sits at the origin of the gap between the two described groups of results ; it is maximal when the demon fully `` knows '' @xmath42 which , however , still has maximal complexity even given her internal state ( an example see below ) . in this case , the first result claims @xmath52 to be @xmath34 , whereas @xmath53 according to the second .
the gap vanishes if `` knowing @xmath42 '' is understood in a _ constructive _ | as opposed to entropic | sense , meaning that `` the demon possesses or can produce a copy of @xmath42 represented in her internal state : '' if that copy is included in bennett s reasoning , then his result reads @xmath54 in this case , knowledge has immediate work value . _ the model .
_ we assume the _ demon _ to be a _
universal turing machine @xmath7 _ the memory tape of which is sufficiently long for the tasks and inputs in question , but _
finite_. the tape initially contains @xmath42 , the string the fuel value of which is to be determined , @xmath0 , a finite string modeling the demon s _ knowledge about @xmath42 _ , and @xmath34 s for the rest of the tape . after the extraction computation ,
the tape contains , at the bit positions initially holding @xmath42 , a ( shorter ) string @xmath47 plus @xmath55 , whereas the rest of the tape is ( again ) the same as before work extraction .
the demon s operations are _ logically _ reversible and can , hence , be carried out _ thermodynamically _ reversibly @xcite .
logical reversibility in our model is the ability of the same demon to carry out the backward computation step by step , _
i.e. _ , from @xmath56 to @xmath57 .
we denote by @xmath58 the _ maximal amount of @xmath34-bits extractable logically reversibly from @xmath42 given the knowledge @xmath0 _ , _ i.e. _ , @xmath59 if @xmath47 s length is minimal ( see figure [ fig:3 ] ) . according to the above , the work value of any physical representation of @xmath42 for a demon knowing @xmath0 is @xmath60 _ lower bound on the fuel value . _
let @xmath61 be a computable function @xmath62 such that @xmath63 is injective .
we call @xmath61 a _ data - compression algorithm with helper_. then we have @xmath64 this can be seen as follows . first , note that the function @xmath65 is computable and bijective .
given the two ( possibly irreversible ) circuits computing the compression and its inverse , one can obtain a _
reversible _ circuit realizing the function and where no further input or output bits are involved .
this can be achieved by first implementing all logical operations with toffoli gates and uncomputing all junk @xcite in both of the circuits .
the resulting two circuits have now both still the property that the input is part of the output . as a second step
, we can simply combine the two , where the first circuit s first output becomes the second s second input , and _
vice versa_. roughly speaking , the first circuit computes the compression and the second reversibly uncomputes the raw data .
the combined circuit has only the compressed data ( plus the 0 s ) as output , on the bit positions carrying the input previously .
( the depth of this circuit is roughly the sum of the depths of the two irreversible circuits for the compression and for the decompression , respectively . )
we assume that circuit to be hard - wired in the demon s head .
a typical example for a compression algorithm that can be used is ziv - lempel @xcite . _
upper bound on the fuel value .
_ we have the following upper bound on @xmath58 : @xmath66 the reason is that the demon is only able to carry out the computation in question ( logically , hence , thermodynamically ) reversibly _ if she is able to carry out the reverse computation as well_. therefore , the string @xmath47 must be at least as long as the shortest program for @xmath7 generating @xmath42 if @xmath0 is given .
although the same is not true in general , this upper bound is _ tight _ if @xmath67 .
the latter means that @xmath0 itself is a program for generating an additional copy of @xmath42 .
the demon can then bit - wisely xor this new copy to the original @xmath42 on the tape , hereby producing @xmath68 _ reversibly _ to replace the original @xmath42 ( at the same time preserving the new one , as reversibility demands ) .
when bennett s `` uncomputing trick '' is used | allowing for making any computation by a turing machine logically reversible @xcite | , then a history string @xmath69 is written to the tape during the computation of @xmath42 from @xmath0 such that after the xoring , the demon can , going back step by step , _ uncompute _ the generated copy of @xmath42 and end up in the tape s original state | except that the original @xmath42 is now replaced by @xmath68 : this results in a maximal fuel value matching the ( in this case trivial ) upper bound .
note that this harmonizes with @xcite if vanishing conditional entropy is so established .
_ discussion .
_ we contrast our bounds with the entropy - based results of @xcite : according to the latter , a demon _ having complete knowledge of @xmath42 _ is able to extract maximal work : @xmath70 . _
what means `` knowing @xmath42 ? '' _ ( see figure [ fig:2 ] ) .
+ + we have seen that the results are in accordance with ours if the demon s _ knowledge _ consists of ( a ) a _ copy _ of @xmath42 , or at least of ( b ) the _ ability to algorithmically reconstruct @xmath42 _ , based on a known program @xmath47 , as discussed above .
it is , however , possible ( c ) that the demon s knowledge is of different nature , merely _ determining @xmath42 uniquely without providing the ability to build @xmath42_. for instance , let the demon s knowledge about @xmath42 be : `` @xmath42 equals the first @xmath37 bits @xmath71 of the binary expansion of @xmath72 . '' here
, @xmath72 is the so - called _ halting probability _ @xcite of a fixed universal turing machine ( _ e.g. _ , the demon @xmath7 itself ) .
although there is a _ short description _ of @xmath42 in this case , and @xmath42 is thus uniquely determined in an entropic sense , there is no _ set of instructions shorter than @xmath42 enabling the demon to generate @xmath42 _
| which would be required for work extraction from @xmath42 according to our upper bound . in short ,
this gap reflects the one between the _ `` unique - description complexity '' _ and the _ kolmogorov complexity_.
non - local correlations @xcite are a fascinating feature of quantum theory . the conceptually challenging aspect is the difficulty of explaining the correlations origin _ causally _ , _ i.e. _ , according to _ reichenbach s principle _ , stating that a correlation between two space - time events can stem from a _ common cause _
( in the common past ) or a _ direct influence _ from one event to the other @xcite . more specifically , the difficulty manifests itself when _ alternatives _ hence , counterfactuals are taken into account : the argument leading up to a bell inequality relates outcomes of _ alternative _ measurements only one of which can actually be realized .
does this mean that if we drop the assumption of _ counterfactual definiteness _
@xcite , _ i.e. _ , the requirement to consistently understand counterfactual events , the paradox or strangeness disappears ?
the answer is _ no _ : even in the `` _ factual - only view _ , '' the joint properties in terms of mutual compressibility of the involved ( now : fixed ) pieces of information are such that consequences of non - local correlations , as understood in a common probability - calculus , persist : an example is the significant complexity forced upon the output given the input s maximal complexity plus some natural translation of no - signaling to the static scenario ( see figure [ ganz ] ) . in the traditional , probabilistic view ,
a _ popescu - rohrlich ( pr ) box _
@xcite gives rise to a mechanism of the following kind : let @xmath1 and @xmath2 the respective input bits to the box and @xmath0 and @xmath73 the output bits ; the ( classical ) bits satisfy @xmath74 this system is _ no - signaling _ , _ i.e. _ , the joint input - output behavior is useless for message transmission .
( interestingly , on the other hand , the _ non - locality _ of the correlation means that classically speaking , signaling _ would _ be required to _ explain _ the behavior since shared classical information is insufficient . )
according to a result by fine @xcite , the non - locality of the system ( _ i.e. _ , conditional distribution ) @xmath75 , which means that it can not be written as a convex combination of products @xmath76 , is equivalent to the fact that there exists no `` roof distribution '' @xmath77 such that @xmath78 for all @xmath79 .
in this view , non - locality means that the outputs to _ alternative inputs _ can not consistently coxist .
the _ counterfactual _ nature of this reasoning has already been pointed out by specker @xcite : `` in einem gewissen sinne gehren aber auch die scholastischen spekulationen ber die _ infuturabilien _
hieher , das heisst die frage , ob sich die gttliche allwissenheit auch auf ereignisse erstrecke , die eingetreten wren , falls etwas geschehen wre , was nicht geschehen ist . ''
`` in some sense , this is also related to the scholastic speculations on the _ infuturabili _ , _ i.e. _ , the question whether divine omniscience even extends to what would have happened if something had happened that did not happen . ''
zukowski and brukner @xcite suggest that non - locality is to be understood in terms of such _ infuturabili _ , called there `` counterfactual definiteness . ''
we intend to challenge this view .
let us first restate in more precise terms the counterfactual reasoning .
such reasoning is intrinsically assuming or concluding statements of the kind that that some piece of classical information , such as a bit @xmath80 , _ exists _ or _ does not exist_. what does this mean ? _ classicality _ of information is an idealized notion implying that it can be measured without disturbance and that the outcome of a measurement is always the same ( which makes it clear this is an idealized notion requiring the classical bit to be represented in a redundantly extended way over an _
infinite _ number of degrees of freedom ) .
it makes thus sense to say that a _
classical bit @xmath80 exists _ , _
i.e. _ , has taken a definite value . in this way of speaking
, fine s theorem @xcite reads : `` the outputs can not _ exist _ before the inputs do . ''
let us make this qualitative statement more precise .
we assume a perfect pr box , _ i.e. _ , a system always satisfying @xmath81 .
note that this equation alone does not uniquely determine @xmath75 since the marginal of @xmath0 , for instance , is not determined .
if , however , we additionally require _ no - signaling _ , then the marginals , such as @xmath82 or @xmath83 , must be perfectly unbiased under the assumption that all four @xmath84-combinations , _ i.e. _ , @xmath85 , and @xmath86 , are possible . to see this ,
assume on the contrary that @xmath87 . by the pr condition
, we can conclude the same for @xmath73 : @xmath88 . by no - signaling
, we also have @xmath89 . using symmetry , and no - signaling again
, we obtain both @xmath90 and @xmath91 .
this contradicts the pr condition since _ two bits which are both biased towards @xmath34 can not differ with certainty_. therefore , our original assumption was wrong : the outputs _ must _ be perfectly unbiased .
altogether , this means that @xmath0 as well as @xmath73 can not exist ( _ i.e. _ , take a definite value actually , there can not even exist a classical value arbitrarily weakly correlated with one of them ) _ before _ for some nontrivial deterministic function @xmath92 , the classical bit @xmath93 exists . the paradoxical aspect of non - locality at least if a causal structure is in place now consists of the fact that _ fresh _ pieces of information _ come to existence _ in a _ spacelike - separated _ way but that are nonetheless _ perfectly correlated_. we propose an understanding of non - locality that refrains from using counterfactual definiteness but invokes solely the data at hand , _
i.e. _ , existing in a single reality @xcite .
_ uncomputability of the outputs of a pr box .
_ let first @xmath94 be infinite binary strings with @xmath95 obviously , the intuition is that the strings stand for the inputs and outputs of a pr box . yet
, no dynamic meaning is attached to the strings anymore ( or to the `` box , '' for that matter ) since there is _ no free choice of an input i.e. , a choice that `` could also have been different '' ( a notion we discussed and suspect to be hard to define precisely in the first place ) and no generation of an output in function of an input _ ; all we have are four fixed strings satisfying the pr condition .
however , nothing prevents us from defining this ( static ) situation to be _ no - signaling _ : @xmath96 recall the mechanism which the maximal non - locality displayed by the pr box enables : _ if the inputs are not entirely fixed , then the outputs must be completely unbiased as soon as the system is no - signaling .
_ we can now draw a similar conclusion , yet entirely within actual and without having to refer to counterfactual data : _ if the inputs are incompressible and independent , and no - signaling holds , then the outputs must be uncomputable_. for a proof of this , let @xmath97 with @xmath98 ( bit - wisely ) , no - signaling , and @xmath99 _ i.e. _ , the `` input '' pair is incompressible .
we conclude @xmath100 note first that @xmath101 implies @xmath102 , and second that any further compression of @xmath103 , given @xmath25 , would lead to `` structure in @xmath104 , '' _
i.e. _ , a possibility of describing ( programming ) @xmath18 given @xmath25 in shorter than @xmath105 and , hence , @xmath104 in shorter than @xmath106 . observe now @xmath107 which implies @xmath108 on the other hand , @xmath109 now , no - signaling together with and implies @xmath110 and @xmath111 the string @xmath112 must be uncomputable .
we have seen that if the pair of inputs @xmath104 is maximally incompressible , then the outputs @xmath112 and @xmath113 must at least be uncomputable .
this observation raises a number of natural questions : does a similar result hold with respect to the _ conditional _ complexities @xmath114 and @xmath115 ?
with respect to _ quantum _ non - local correlations ? can we give a suitable _ general definition _ of non - locality and does a similar result as the above hold with respect to _ any _ non - local correlation ?
can we strengthen and tighten our arguments to show , for instance , that _ uncomputable _ inputs plus no - signaling and maximal non - locality leads to _ incompressibility _ of the outputs ? what results might turn out to be _ incompressibility - amplification _ methods .
let us address these questions .
_ conditional uncomputability of the outputs of a pr box . _ with respect to the same assumptions as in the previous section , we now consider the quantities @xmath114 and @xmath115 , respectively .
note first @xmath116 _
i.e. _ , the two expressions vanish simultaneously .
we show that , in fact , they both _ fail to be of order _ @xmath117 . in order to see this ,
assume @xmath118 and @xmath119 .
hence , there exist programs @xmath120 and @xmath121 ( both of length @xmath117 ) for functions @xmath122 and @xmath123 with @xmath124 for fixed ( families of ) functions @xmath122 and @xmath123 , asymptotically how many @xmath125 can at most exist that satisfy ? the question boils down to a _ parallel - repetition _ analysis of the _ pr game _ : a result by raz @xcite implies that the number is of order @xmath126 .
therefore , the two programs @xmath120 and @xmath121 together with the index , of length @xmath127 of the correct pair @xmath104 within the list of length @xmath126 lead to a program , generating @xmath104 , that has length @xmath128 in contradiction to the assumption of incompressibility of @xmath104 .
_ conditional uncomputability from quantum correlations .
_ in the `` traditional view '' on non - locality , the pr box is an idealization unachievable by the behavior of any quantum state .
if it _ did _ exist , on the other hand , it would be a most precious resource , _ e.g. _ , for cryptography or randomness amplification .
the reason is that as we have discussed above under the minimal assumption that the inputs are _ not completely determined _ , the outputs are _ perfectly random _ , even given the inputs .
perfect pr boxes are not predicted by quantum theory , but sometimes , the best approximations to pr boxes that are quantum physically achievable ( @xmath129 ) can be used for information - processing tasks , such as key agreement @xcite . for our application here , however , we found this not to be the case . on the other hand ,
it has been shown @xcite , @xcite , @xcite that correlations which _ are _ achievable in the laboratory @xcite allow for similar applications ; they are based on the _ chained bell inequality _ instead of perfect pr - type non - locality .
we show the same to hold here . to
the chained bell inequality belongs the following idealized system : let @xmath130 be the inputs .
we assume the `` promise '' that @xmath2 is congruent to @xmath1 or to @xmath131 modulo @xmath132 . given this promise , the outputs @xmath133 must satisfy @xmath134 where @xmath135 is the characteristic function of the event @xmath136 .
barrett , hardy , and kent @xcite showed that if @xmath1 and @xmath2 are random , then @xmath0 and @xmath73 must be perfectly unbiased if the system is no - signaling .
more precisely , they were even able to show such a statement from the gap between the error probabilities of the best classical @xmath137 and quantum @xmath138 strategies for winning this game . in our framework , we show the following statement . _ let @xmath139 be such that the promise holds , and such that @xmath140 _ i.e. _ , the string @xmath141 is maximally incompressible given the promise ; the system is no - signaling ; the fraction of quadruples @xmath142 , @xmath143 , satisfying is of order @xmath144
. then @xmath145 . _
let us prove this statement .
first , @xmath146 being maximal implies @xmath147 the fractions of @xmath35 s in @xmath25 must , asymptotically , be @xmath148 due to the string s incompressibility
. if we condition on these positions , the string @xmath149 is incompressible , since otherwise there would be the possibility of compressing @xmath104 .
now , we have @xmath150 since one possibility for `` generating '' the string @xmath149 , from position @xmath35 to @xmath105 , is to generate @xmath151}$ ] and @xmath152}$ ] as well as the string indicating the positions where is violated , the complexity of the latter being at most is the binary entropy @xmath153 .
usually , @xmath154 is a probability , but @xmath155 is invoked here merely as an approximation for binomial coefficients.]@xmath156 let us compare this with @xmath148 : although the binary entropy function has slope @xmath157 in 0 , we have @xmath158 if @xmath132 is sufficiently large . to see this , observe first that the dominant term of @xmath159 for small @xmath112 is @xmath160 , and second that @xmath161 for @xmath132 sufficiently large . together with
, we now get @xmath162 if @xmath132 is chosen sufficiently large . on the other hand , @xmath163 now , , , and together imply @xmath164 in particular , @xmath112 must be uncomputable . for any non - local behavior characterizable by a condition that is always satisfiable with entanglement , but not _ without _
this resource | so called `` pseudo - telepathy '' games @xcite | , the application of raz _ parallel - repetition theorem _ shows that incompressibility of the inputs leads to uncomputability of at least one of the two outputs _ even given the respective input _ , _
i.e. _ , @xmath165 we illustrate the argument with the example of the _ magic - square game _
@xcite : let @xmath166 be the quadruple of the inputs and outputs , respectively , and assume that the pair @xmath104 is incompressible as well as @xmath167 .
then there exist @xmath117-length programs @xmath120 , @xmath121 such that @xmath151}=p_n(a_{[n]})$ ] and @xmath152}=q_n(b_{[n]})$ ] .
the parallel - repetition theorem @xcite implies that the length of a program generating @xmath168},b_{[n]})$ ] is , including the employed sub - routines @xmath120 and @xmath121 , of order @xmath169},b_{[n]})$ ] in contradiction to the incompressibility of @xmath104 . _
an all - or - nothing flavor to the church - turing hypothesis . _
our lower bound on @xmath114 or on @xmath115 means that if the experimenters are given access to an incompressible number ( such as @xmath72 ) for choosing their measurement bases , then the measured photon ( in a least one of the two labs ) is forced to generate an uncomputable number as well , even given the string determining its basis choices .
roughly speaking , there is either no incompressibility at all in the world , or it is full of it .
we can interpret that as an all - or - nothing flavor attached to the church - turing hypothesis : either _ no _ physical system at all can carry out `` beyond - turing '' computations , or _ even a single photon can_. _ general definition of ( non-)locality without counterfactuality .
_ we propose the following definition of when a no - signaling quadruple @xmath97 ( where @xmath170 are the `` inputs '' and @xmath171 the outputs ) is _ local _ : there must exist @xmath172 such that @xmath173 sufficient conditions for locality are then @xmath174 since we can set @xmath175 . at the other end of the scale ,
we expect that for any non - local `` system , '' the fact that @xmath146 is maximal implies that @xmath112 or @xmath113 is conditionally uncomputable , given @xmath18 and @xmath25 , respectively .
it is a natural question whether the given definition harmonizes with the probabilistic understanding .
indeed , the latter can be seen as a special case of the former : if the ( fixed ) strings are _ typical sequences _ of a stochastic process , our non - locality definition implies non - locality of the corresponding conditional distribution .
the reason is that a hidden variable of the distribution immediately gives rise , through sampling , to a @xmath176 in the sense of .
note , however , that our formalism is _ strictly more general _ since asymptotically , almost all strings fail to be typical sequences of such a process .
it has already been observed that the notion of kolmogorov complexity can allow , in principle , for _ thermodynamics independent of probabilities or ensembles _ : zurek @xcite defines physical entropy @xmath177 to be @xmath178 where @xmath179 stands for the collected data at hand while @xmath180 is the remaining conditional shannon entropy of the microstate @xmath42 given @xmath179 .
that definition of a macrostate is _ subjective _ since it depends on the _ available _ data .
how instead can the macrostate | and _ entropy _
, for that matter | be defined _ objectively _ ?
we propose to use the _ kolmogorov sufficient statistics _
@xcite of the microstate : for any @xmath181 , let @xmath182 be the smallest set such that @xmath183 and @xmath184 hold .
let further @xmath185 be the value of @xmath39 at which the function @xmath186 becomes linear with slope @xmath187
. intuitively speaking , @xmath185 is the point beyond which there is no more `` structure '' to exploit for describing @xmath42 within @xmath188 : @xmath42 is a `` typical element '' of the set @xmath188 .
we define @xmath189 to be _ @xmath42 s macrostate_. it yields a program generating @xmath42 of minimal length @xmath190 the fuel value ( as discussed in section [ fuel ] ) of a string @xmath191 is now related to the macrostate @xmath192 by @xmath193 ( see figure [ fig:6 ] ) : decisive is neither the complexity of the macrostate nor its log - size _ alone _ , but their _
sum_. a notion defined in a related way is the _ sophistication _ or _ interestingness _ as discussed by aaronson @xcite investigating the process where milk is poured into coffee ( see figure [ fig:4 ] ) .
whereas the initial and final states are `` simple '' and `` uninteresting , '' the intermediate ( non - equilibrium ) states display a rich structure ; here , the sophistication | and also @xmath194 for our macrostate @xmath179 | becomes maximal . during the process under consideration , neither the macrostate s complexity nor its size is monotonic in time : whereas @xmath194 has a _ maximum _ in the non - equilibrium phase of the process , @xmath195 has a _ minimum _ there ( see figure [ fig:5 ] ) . on the other hand , the complexity of the _ microstate _ , @xmath196 is a candidate for a ( essentially ) monotonically nondecreasing quantity : is this the _ second law of thermodynamics _ in that view ?
this law , which claims a certain quantity to be ( essentially ) monotonic in time , is by many believed to be the origin of our ability to distinguish the future from the past . _ the second law , traditional view .
_ let a closed system be in a thermodynamical equilibrium state of entropy @xmath197 at time @xmath198 .
assume that the system evolves to another equilibrium state , of entropy @xmath199 , at some fixed later time @xmath200 .
then , for @xmath201 , @xmath202=2^{-s}\ .\ ] ] it is a rare example | outside quantum theory | of a physical `` law '' holding only with some probability .
_ is there an underlying fact in the form of a property of an evolution holding with certainty and also for all intermediate states ? _ clearly , that fact would not talk about the _ coarse - grained _ behavior of the system , which we have seen in the discussed example to be _
non_-monotonic in time .
if , however , we consider the _ micro_state , then _ logical reversibility _
| meaning that the past can be computed step by step from the future ( not necessarily _ vice versa _ ) | is a good candidate : indeed , also landauer s principle links the second law to logical irreversibility .
a logically reversible evolution is potentially asymmetric in time if the backward direction is _ not _ logically reversible . in the spirit of the _ church - turing hypothesis _ , we see the state of a closed system in question as a finite binary string and its evolution ( through discretized time ) as being computed by a universal turing machine . _ the second law , revisited . _
the evolution of a closed system is _ logically reversible _ and the past at time @xmath198 can be computed from the future at time @xmath203 by a constant - length program on a turing machine .
it is somewhat ironic that this view of the second law puts forward the _
reversibility _ of the computation , whereas the law is usually linked to the opposite : _ irreversibility_. a consequence of the law is that the decrease of the kolmogorov complexity of the string encoding the system s state is limited . _
consequence of the second law , revisited .
_ let @xmath204 and @xmath205 be the contents of a reversible turing machine s tape at times @xmath206 . then @xmath207 if the turing machine is _
deterministic _ , the complexity increases at most logarithmically in time . on the other hand
, this growth can of course be arbitrarily faster for _ probabilistic _ machines .
turned around , kolmogorov complexity can yield an _
criterion for the distinction between determinism and indeterminism ( see figure [ figueres ] ) . in the case of randomness , a strong asymmetry and
an objective arrow of time can arise .
a context - free definition of randomness ( or free will for that matter ) has the advantage not to depend on the `` possibility that something could have been different from how it was , '' a metaphysical condition we came to prefer to avoid . _ the traditional second law from complexity increase .
_ it is natural to ask what the connection between logical reversibility and complexity on one side and the traditional second law on the other is .
we show that the latter emerges from increasing complexity | including the exponential error probabilities .
let @xmath204 and @xmath205 be the microstates of a closed system at times @xmath206 with @xmath208 . if the macrostates @xmath209 and @xmath210 of @xmath204 and @xmath205 , respectively , have _ small kolmogorov complexity _
( such as traditional thermodynamical equilibrium states characterized by global parameters like volume , temperature , pressure , etc . )
, then @xmath211 if the macrostates are simple , then their size is non - decreasing .
note that this law is still compatible with the exponentially small error probability ( @xmath212 ) in the traditional view of the second law for a spontaneous immediate drop of entropy by @xmath213 : the gap opens when the simple thermodynamical equilibrium macrostate of a given microstate differs from our macrostate defined through the kolmogorov statistics .
this can occur if , say , the positions and momenta of the molecules of some ( innocent- , _ i.e. _ , general - looking ) gas encode , _ e.g. _ , @xmath41 and have essentially zero complexity .
we can now finish up by closing a logical circle .
we have started from the converse of landauer s principle , went through work extraction and ended up with a complexity - theoretic view of the second law : we have returned back to our starting point . _
landauer s principle , revisited . _
the ( immediate ) transformation of a string @xmath42 to the @xmath34-string of the same length requires free energy at least @xmath214 which is then dissipated as heat to the environment . for every concrete lossless compression algorithm @xmath61 , @xmath215 is , on the other hand , an upper bound on the required free energy . finally , landauer s principle can be combined with its converse and generalized as follows . _
generalized landauer s principle .
_ let @xmath1 and @xmath2 two bit strings of the same length .
the ( immediate ) transformation from @xmath1 to @xmath2 costs at least @xmath216 free energy , or it releases at most the absolute value of if this is negative .
if the turing machine is a closed physical system , then this principle reduces to the complexity - non - decrease stated above .
this suggests that the physical system possibly _ simulated _ by the machine | in the spirit of the church - turing hypothesis | also follows the second law ( _ e.g. _ , since it is a closed system as well ) .
the fading boundaries between what the machine _ is _ and what is _ simulated _ by it are in accordance with wheeler s @xcite `` it from bit : '' _ every `` it '' | every particle , every field of force , even the spacetime continuum itself | derives its function , its meaning , its very existence entirely [ ] from the apparatus - elicited answers to yes or no questions , binary choices , `` bits . '' _ if we try to follow the lines of such a view further , we may model the environment as a binary string @xmath217 as well .
the goal is a unified discourse avoiding to speak about complexity with respect to one system and about free energy , heat , and temperature to the other .
the transformation addressed by landauer s principle and its converse then looks as in figure [ wheel ] : the low - complexity zero - string can be swapped with `` complexity '' in the environment which in consequence becomes more redundant , _
i.e. _ , cools down but receives free energy , for instance in the form of a weight having been lifted .
if , motivated by the above , we choose to regard _ information _ as being more fundamental than space and time , how can the latter be imagined to emerge from the former ? can such a causal structure be understood to be of _ logical _ rather than _ physical _ nature ?
in other words , is it more accurate to imagine causal relations to be a property of logical rather than physical spaces @xcite ?
we address these questions here , continuing to avoid speaking about `` what could have been different , '' _
i.e. _ , the counterfactual viewpoint . in section
[ td ] , an arrow of time has emerged under the assumption of ( uni - directional ) logical reversibility . here , we refine the same idea in an attempt to derive a causal structure based on the principle that any point carries complete information about its space - time past .
let us start with a finite set @xmath218 of strings on which we would like to find a causal structure arising _ from inside _ , _
i.e. _ , from the properties of , and relations between , these strings .
the intuition is that an @xmath219 encodes the totality of momentary local physical reality in a `` point , '' _
i.e. _ , parameters such as mass , charge , electric and magnetic field density .
let @xmath220 be finite .
we define the following order relation on @xmath218 : , for instance , the condition @xmath113 is assumed to be the full ( infinite ) string , whereas the asymptotic process runs over @xmath151}$ ] .
the reason is that very insignificant bits of @xmath113 ( intuitively : the present ) can be in relation to bits of @xmath112 ( the past ) of much higher significance .
the past does not disappear , but it fades.]@xmath221 we say that @xmath112 is a _ cause _ of @xmath113 , and that @xmath113 is an _ effect _ of @xmath112 .
so , @xmath113 is in @xmath112 s future exactly if @xmath113 contains the entire information about @xmath112 ; no information is ever lost .
the intuition is that _ any _ `` change '' in the cause affects _
each one _ of its effects if sufficient precision is taken into account .
we write @xmath222 if @xmath223 as well as @xmath224 hold . if @xmath225 and @xmath226 , we write @xmath227 and call @xmath112 and @xmath113 _ spacelike separated_. we call the pair @xmath228 a _
causal structure_. for a set @xmath229 and @xmath230 , we say that @xmath113 is the _ first common effect _ of the @xmath231 if it is the least upper bound : @xmath232 holds for all @xmath231 , and for any @xmath233 with @xmath234 for all @xmath231 , also @xmath235 holds .
the notion of _ last common cause _ is defined analogously .
a minimum ( maximum ) of @xmath228 is called _ without cause _ ( _ without effect _ ) .
if @xmath218 has a smallest ( greatest ) element , this is called _ big bang _ ( _ big crunch _ ) .
we call a causal structure _ deterministic _ if , intuitively , every @xmath113 which is not without cause is completely determined by all its causes . formally , for some @xmath230 , let @xmath236 be the set of all @xmath237 such that @xmath232 holds
. then we must have @xmath238 otherwise , @xmath218 is called _
probabilistic_. + observe first that _ every deterministic causal structure which has a big bang is trivial _ : we have @xmath239 this can be seen as follows .
let @xmath25 be the big bang , _
i.e. _ , @xmath240 for all @xmath219 . on the other hand , @xmath241 if @xmath242 is the set of predecessors of @xmath112 .
since the same is true for each of the @xmath243 , we can continue this process and , ultimately , end up with only @xmath25 : @xmath244 , _ i.e. _ , @xmath245 , and thus @xmath246 for all @xmath219 . in this case
, we obviously can not expect to be able to explain space - time .
( note , however , that there can still exist deterministic @xmath218 s without big bang with non - trivial structure . )
however , the world as it presents itself to us with _ both _ big bang _ and _ arrow of time seems to direct us away from determinism ( in support of @xcite ) .
the situation is very different in _ probabilistic _ causal structures : here , the partial order relation @xmath247 gives rise to a non - trivial picture of causal relations and , ideally , causal space - time including the arrow of time .
obviously , the resulting structure depends crucially on the set @xmath218 .
challenging open problems are to understand the relationship between sets of strings and causal structures : can every partially ordered set be implemented by a suitable set of strings ? what is the property of a set of strings that gives rise to the `` usual '' space - time of relativistic light - cones ?
is it helpful to introduce a _
metric _ instead of just an order relation ? as a first step , it appears natural to define @xmath248 as the _ distance of @xmath112 from the set of effects of @xmath113_. in case @xmath113 is an effect of @xmath112 , this quantity intuitively measures the _ time _ by which @xmath112 happens _ before _
@xmath113 .
generally in such a model , what is a `` second law , '' and under what condition does it hold ? can it and the arrow of time be compatible even with determinism ( as long as there is no big bang ) ? what singles out the sets displaying quantum non - local correlations as observed in the lab ?
( what is the significance of tsirelson s bound in the picture ? )
a recent framework for quantum @xcite and classical @xcite correlations without causal order is based on _ local _ assumptions only .
these are the local validity of quantum or classical probability theory , that laboratories are closed ( parties can only interact through the environment ) , and that the probabilities of the outcomes are linear in the choice of local operation .
global _ assumption of a _ fixed global causal order _ is replaced by the assumption of _ logical consistency _ : all probabilities must be non - negative and sum up to @xmath35 . some correlations termed _ non - causal _ that can be obtained in this picture can not arise from global quantum or classical probability theory .
similarly to the discovery of non - local correlations that showed the existence of a world between the local and the signaling ; in a similar sense , we discuss here _ a territory that lies between what is causal and what is logically inconsistent : it is not empty_. in the spirit of section [ nl ] , where we studied the consequences of non - locality , we show that the results from non - causal correlations carry over to the picture of ( conditional ) compressibility of bit strings , where we do not employ probabilities , but consider _ actual _ data only . in that sense , these are the _ non - counterfactual _ versions of results on non - causal correlations .
we define causal relations operationally , where we use the notion of parties .
a _ party _ can be thought of as a system , laboratory , or an experimenter , performing an operation . in the traditional view , the choice of operation is represented by _ randomness _ , and thus by a probability distribution . here
, in contrast , we refrain from this counterfactual approach ( probabilities ) , and consider _ actual _ as opposed to _ potential _
choices only . the traditional view with probabilities
has a _ dynamic _ character : systems undergo ( randomized ) evolutions . like in section [ nl ] ,
we obtain a _
situation if we consider actual data only .
all statements are formulated with bit strings and _ relations _ between these strings modeling the `` operations . ''
a party @xmath1 is modeled by two bit strings @xmath249 and @xmath250 .
we restrict ourselves to pairs of bit strings that satisfy some relation @xmath251 . within a party , we assume a fixed causal structure ( @xmath249 precedes @xmath250 ) ( see figure [ fig : party ] ) .
\(i ) @xmath249 ; ( o ) @xmath250 ; ( i )
( o ) node[midway , left ] ( l ) relation @xmath251 ; the relation @xmath251 is called _ local operation of @xmath1 _ , the string @xmath249 is called _ input to @xmath1 _ , and @xmath250 is _
@xmath1 s output_. if we have more than one party , we consider only those input and output bit strings that satisfy some _
global relation_. these relations are , as in section [ nl ] , to be understood to act _ locally _ on the involved strings : a relation involves only a finite number of instances ( bit positions ) , and it is repeated @xmath252 times for obtaining the global relation . for two parties @xmath1 and @xmath2 , we say that @xmath1 is in the _ causal past _ of @xmath2 , @xmath253 , if and only if @xmath254 intuitively , @xmath1 is in the causal past of @xmath2 if and only if @xmath2 s input is _ uncomputable _ otherwise @xmath2 could simply obtain it herself and better _ compressible _ with @xmath1 s output than without , _
i.e. _ , the two strings depend on each other .
expressed according to intuitive dynamic thinking , the definition means that @xmath1 is in the causal past of @xmath2 if and only if @xmath2 _ learns _ parts of an incompressible string from @xmath1 . the causal relation among parties defined here
is extended straight - forwardly to the scenario where _ one or more _ parties are in the causal past of _ one or more _ parties .
this definition is different from the one proposed in section [ spacom ] .
there , a string @xmath112 is said to be the _ cause _ of another string @xmath113 ( the _ effect _ ) if and only if @xmath255 .
the intuition there is _ logical reversibility _ : future events contain all information about past events , no information is ever lost , and @xmath112 and @xmath113 are understood to encode complete physical reality in some space - time point .
in contrast , the definition here only relates pieces of information chosen and processed by the parties : if one party s input depends on another s output , then she is in the causal future of the latter .
( reversibility , the central notion in section [ spacom ] , does not play a role here . ) since the strings now just correspond to the pieces of information manipulated by the parties , we can not simply define freeness as an attribute of complexity .
instead , we _ postulate _ the output strings to be free .
the rationale of definition above is similar to the one we propose in @xcite for the probability picture .
there , @xmath253 holds if and only if both random variables @xmath1 and @xmath2 are _ correlated _ and @xmath1 is postulated _
free_. the motivation is to define causal relations _ based _ on freeness , and not the other way around ( see figure [ fig : free ] ) .
\(s ) ; ( st ) @xmath42 ; ( m ) ; ( m.east ) arc ( 70:110:1.18 ) ; ( m.south)++(0cm,0.05cm ) + + ( 0.3cm,0.2 cm ) ; ( t ) ; ( tt ) @xmath40 ; ( k ) ; ( k2 ) at ( k ) ; in -13 + 0 * 30 , -13 + 1 * 30 , -13 + 2 * 30 , -13 + 3 * 30 , -13 + 4 * 30 , -13 + 5 * 30 , -13 + 6 * 30 , -13 + 7 * 30 , -13 + 8 * 30 , -13 + 9 * 30 , -13 + 10 * 30 , -13 + 11 * 30 , -13 + 12 * 30 ( k.center)++(:0.05cm )
+ ( : 0.15 cm ) ; ( t.west )
( s.east ) ; intuitively , if you flip a switch that is correlated to a light bulb , then flipping the switch is in the causal past of the light turning on or off the definition of a causal relation relies on what we call _ free _ ( the switch in this case ) .
such a definition based on postulated freeness is similar to the interventionist s approach to causality , see _
e.g. _ , @xcite . in the approach
studied here , the analog to _ correlation _ is _ dependence_. the distinction between _ free _ and _ not free _
variables is done in the same way by distinguishing between _ input _ and _ output _ bit strings .
_ causal scenarios _ describe input and output strings of the parties where the resulting causal relations reflect a _ partial ordering _ of the parties ( see figure [ fig : causalscenario ] ) . in the most general case ,
the partial ordering among the parties of a set @xmath42 , who are all in the causal future of some other party @xmath256 , _ i.e. _ , for all @xmath257 , can depend ( _ i.e. _ , satisfy some relation with ) the bit strings of @xmath1 @xcite , @xcite .
a causal scenario , in particular , implies that _ at least one party is not in the causal future of some other parties_. if no partial ordering of the parties arises , then the scenario is called _ causal _ ( see figure [ fig : noncausalscenario ] ) .
a trivial example of a causal scenario is a communication channel over which a bit is perfectly transmitted from a party to another .
this channel , formulated as a global relation , is @xmath258 , with @xmath259 , and where the first bit belongs to @xmath1 ( sender ) and the second to @xmath2 ( receiver ) ( see figure [ fig : channel ] ) .
consider the @xmath252-fold sequential repetition of this global relation , and assume that both output bit strings are incompressible and independent : @xmath260 .
the bit string @xmath249 is @xmath261 according to the global relation .
in contrast , @xmath262 is equal to @xmath250 . since @xmath263 and @xmath264 ,
the causal relation @xmath253 holds , restating that @xmath1 is in the causal past of @xmath2 .
conversely , @xmath265 and , therefore , @xmath266 : the receiver is _ not _ in the causal future of the sender .
consider the global relation @xmath267 which describes a _ two - way channel _ : @xmath1 s output is equal to @xmath2 s input and @xmath2 s output is equal to @xmath1 s input ( see figure [ fig:2wayprocess ] ) .
this global relation can describe a non - causal scenario .
if @xmath260 , then indeed , the causal relations that we obtain are @xmath253 and @xmath268 .
what we want to underline here is that for _ this particular choice _ of local operations of the parties , input bit strings that are consistent with the relation exist . in stark contrast , if we fix the local operations of the parties to be @xmath269 ( the output equals the bit - wise flipped input ) for party @xmath1 and @xmath270 for party @xmath2 , then _ no choice of inputs _
@xmath249 and @xmath262 satisfies the desired global relation .
this inconsistency is also known as the _ grandfather antinomy_. if no satisfying input and output strings exist , then we say that the global relation is _ inconsistent with respect to the local operations_. otherwise , the global relation is _ consistent with respect to the local operations_. for studying bit - wise global relations , _
i.e. _ , global relations that relate single output bits with single input bits , that are consistent _ regardless _ the local operations , we set the local operation to incorporate all possible operations on bits .
these are the constants @xmath34 and @xmath35 as well as the identity and bit - flip operations .
the parties additionally hold incompressible and independent strings that define which of these four relations is in place at a given bit position . for party @xmath47 ,
let this additional bit string be @xmath271 .
formally , if we have @xmath39 parties @xmath272 , then @xmath273 the local operation of a party @xmath47 is @xmath274 where a superscript @xmath275 selects the @xmath276-th bit of a string . depending on pairs of bits on @xmath271 ,
the relation states that a given output bit is either equal to @xmath34 or @xmath35 or equal to or different from the corresponding input bit .
an example is presented in figure [ fig : c ] . of a party @xmath47 that satisfy the relation defined by eq .
( 15 ) . ] since all pairs of bits appear equally often in @xmath271 ( asymptotically speaking ) , in half the cases bits of the output string are identical to bits of the ( incompressible ) string @xmath271 in the respective positions .
thus , the output satisfies @xmath277 .
we call the local operation of eq
. of a party _ universal local operation_. if a global relation is consistent with respect to universal local operations , then we call it
_ logically consistent_. if we consider _ all _ ( bit - wise ) operations , the global relation becomes _ inconsistent _ : no input and output strings exist that satisfy the desired global relation . to see this
, note that since we are in the asymptotic case , there exist positions @xmath276 where the relation of @xmath1 states that the @xmath276th output bit is equal to the @xmath276th input bit , and the relation of @xmath2 states that the @xmath276th output bit is equal to the negated @xmath276th input bit , which results in a contradiction the global relation _ can not _ be satisfied . in more detail , there exists an @xmath276 such that the bit string @xmath278 contains the pair @xmath279 at position @xmath280 , and such that the bit string @xmath281 contains the pair @xmath282 at the same position : @xmath283 on the one hand , the input to @xmath1 has a value @xmath18 on the @xmath276th position , and , because of @xmath278 , the same value is on the @xmath276th position of @xmath250 : @xmath284 the input and output bit strings of @xmath2 , on the other hand , must , due to @xmath281 , have opposite bits on the @xmath276th position : @xmath285 a contradiction arises : no choice of bits @xmath18 and @xmath25 exist that satisfy the global relation : the global relation , which is depicted in figure [ fig:2wayprocess ] , is _ logically inconsistent_. we show that there exist _ logically consistent global relations that are non - causal _ @xcite .
suppose we are given three parties @xmath1 , @xmath2 , and @xmath61 with universal local operations .
there exist global relations where the input to any party is a function of the outputs from the remaining two parties .
an example @xcite of such a global relation is @xmath286 where all variables represent bits , and where @xmath287 is the input to @xmath288 and @xmath289 is the output from @xmath288 , respectively .
this global relation can be understood as follows : depending on the _ majority _ of the output bits , the relation _ either _ describes the identity channel from @xmath1 to @xmath2 to @xmath61 , and back to @xmath1 , _ or _
it describes the bit - flip channel from @xmath1 to @xmath61 to @xmath2 , and back to @xmath1 ( see figure [ fig : maj ] ) .
( -,0)++(90- : ) arc ( 90-:-30 + : ) ; ( -,0)++(-30- : ) arc ( -30-:-150 + : ) ; ( -,0)++(-150- : ) arc ( -150-:-270 + : ) ; ( -,0)++(90 : ) node ( a ) @xmath1 ; ( -,0)++(-30 : ) node ( b ) @xmath2 ; ( -,0)++(-150 : ) node ( c ) @xmath61 ; ( , 0)++(90- : ) arc ( 90-:-30 + : ) ; ( , 0)++(-30- : ) arc ( -30-:-150 + : ) ; ( , 0)++(-150- : ) arc ( -150-:-270 + : ) ; ( , 0)++(90 : ) node ( a2 ) @xmath1 ; ( , 0)++(-30 : ) node ( b2 ) @xmath2 ; ( , 0)++(-150 : ) node ( c2 ) @xmath61 ; ( , 0)++(90 - 60:+.3 ) node ( n1 ) @xmath290 ; ( , 0)++(-30 - 60:+.3 ) node ( n2 ) @xmath290 ; ( , 0)++(-150 - 60:+.3 ) node ( n3 ) @xmath290 ; ( ---,+0.3 ) node ( half1 ) @xmath291 ; ( --,+0.3 ) node ( half2 ) @xmath292 ; we study the causal relations that emerge from @xmath252 sequential repetitions of this global relation , _
i.e. _ , infinite strings that satisfy the global relation .
the input to party @xmath1 is uncomputable , even @xmath293 , because some bit positions of the outputs from @xmath2 and @xmath61 are uncomputable .
yet , the outputs from @xmath2 and @xmath61 _ completely _ determine the input to @xmath1 , _ i.e. _ , @xmath294 .
therefore , the causal relation @xmath295 holds . due to symmetry , the causal relations @xmath296 and @xmath297 hold as well . all together
imply that _ every _ party is in the causal future of some other parties the scenario is _ non - causal_. on the other hand , it is _ logically consistent _ : there exist input and output bit strings that satisfy the global relation at every bit - position . in the probability view , there exists an example of a _ randomized _ process that results in non - causal correlations @xcite , shown in figure [ fig : prob ] .
( -,0)++(90- : ) arc ( 90-:-30 + : ) ; ( -,0)++(-30- : ) arc ( -30-:-150 + : ) ; ( -,0)++(-150- : ) arc ( -150-:-270 + : ) ; ( -,0)++(90 : ) node ( a ) @xmath1 ; ( -,0)++(-30 : ) node ( b ) @xmath2 ; ( -,0)++(-150 : ) node ( c ) @xmath61 ; ( , 0)++(90- : ) arc ( 90-:-30 + : ) ; ( , 0)++(-30- : ) arc ( -30-:-150 + : ) ; ( , 0)++(-150- : ) arc ( -150-:-270 + : ) ; ( , 0)++(90 : ) node ( a2 ) @xmath1 ; ( , 0)++(-30 : ) node ( b2 ) @xmath2 ; ( , 0)++(-150 : ) node ( c2 ) @xmath61 ; ( , 0)++(90 - 60:+.3 ) node ( n1 ) @xmath290 ; ( , 0)++(-30 - 60:+.3 ) node ( n2 ) @xmath290 ; ( , 0)++(-150 - 60:+.3 ) node ( n3 ) @xmath290 ; ( -/2,0 ) node ( plus ) @xmath298 ; ( ---,0 ) node ( half1 ) @xmath299 ; ( --,0 ) node ( half2 ) @xmath299 ; in every run , the process models with probability @xmath300 the clockwise identity channel or the clockwise bit - flip channel . in the probabilistic view
channels appear with equal probability .
this leads to every party s inability to influence its past .
for instance , if parties @xmath2 and @xmath61 copy the input to the output and party @xmath1 has @xmath18 on the output , then @xmath1 has a random bit on the input party @xmath1 can not influence its past , and the grandfather antinomy does not arise . if , however , the probabilities of the mixture are altered slightly , a contradiction arises @xcite .
furthermore , the process from figure [ fig : prob ] can not be embedded into a process with more inputs and outputs such that the larger process becomes _ deterministic _ and remains _ logically consistent _ @xcite . since in the view studied here ,
we look at single runs _ without _ probabilities either of the global relations from the left or from the right channel must hold . thus ,
if all parties use the universal local operation , a contradiction always arises , showing the inconsistency of the process .
_ discussion . _ in section [ eins ] we saw that one consequence of dropping the notion of an _ a priori _ causal structure is that randomness becomes hard to define .
thus , we are forced to take the `` _ factual - only view _
: '' no probabilities are involved . here ,
we show two facts that we formulate without considering _ counterfactuals_.
the first is that causal relations among parties can be _ derived _ by considering fixed bit strings only , without the use of the probability language .
these causal relations are an _ inherent _ property of the bit strings of the parties . in other words ,
these strings are understood to be _ logically prior _ to the causal relations ( just as in section [ spacom ] ) .
the second consequence is that the causal relations that stem from certain strings can describe _ non - causal _ scenarios .
this means that _ logical consistency _ does not imply a causal scenario : _ causality is strictly stronger than logical consistency . _
whereas for _ parmenides of elea _ , time was a mere illusion | `` no was nor will , all past and future null '' | , _ heraclitus _ saw space - time as the pre - set stage on which his play of permanent change starts and ends .
the follow - up debate | two millennia later and three centuries ago | between _ newton _ and _ leibniz _ about as how fundamental space and time , hence , _ causality _ , are to be seen was decided by the course of science in favor of newton : in this view , space and time can be imagined as fundamental and given _ a priori_. ( this applies also to relativity theory , where space and time get intertwined and dynamic but remain fundamental instead of becoming purely relational in the sense of _ mach s principle_. ) today , we have more reason to question a fundamental causal structure | such as the difficulty of explaining quantum non - local correlations according to reichenbach s principle .
so motivated , we care to test refraining from assuming space - time as initially given ; this has a number of consequences and implications , some of which we address in this text . when causality is dropped , the usual definitions of randomness stop making sense . motivated by this ,
we test the use of intrinsic , context - independent `` randomness '' measures such as a string s length minus its ( normalized ) fuel value .
we show that under the church - turing hypothesis , kolmogorov complexity relates to this value .
we argue that with respect to quantum non - locality , complexity allows for a reasoning that avoids comparing results of different measurements that can not all be actually carried out , _
i.e. _ , that is _ not counterfactual_. some may see this as a conceptual simplification .
it also leads to an all - or - nothing flavor of the church - turing hypothesis : _ either no physical system can generate uncomputable sequences , or even a single photon can_. finally , it is asked whether _
logical reversibility _ is connected to the second law of thermodynamics | interpreted here in complexities and independent of any context expressed through probabilities or ensembles | and potentially to the arrow of time , past and future .
finally , we have speculated that if a causal structure is not fundamental , how it may emerge from data - compressibility relations . when causality is dropped , one risks antinomies .
we show , in the complexity - based view , that sticking to logical consistency does not restore causality but is strictly weaker .
this observation has recently been extended to _ computational complexity _
@xcite : circuits solely avoiding antinomies are strictly stronger than causal circuits .
the authors thank mateus arajo , veronika baumann , charles bdard , gilles brassard , harvey brown , caslav brukner , harry buhrman , matthias christandl , sandro coretti , fabio costa , bora dakic , frdric dupuis , paul erker , adrien feix , jrg frhlich , nicolas gisin , esther hnggi , arne hansen , marcus huber , lorenzo maccone , alberto montina , samuel ranellucci , paul raymond - robichaud , louis salvail , l. benno salwey , andreas winter , and magdalena zych for inspiring discussions , and the einstein kaffee as well as the reitschule bern for their inspiring atmosphere . _
| grazie mille ! _
our work was supported by the swiss national science foundation ( snf ) , the national centre of competence in research `` quantum science and technology '' ( qsit ) , the cost action on fundamental problems in quantum physics , and the hasler foundation .
bancal , s. pironio , a. acn , y .- c .
liang , v. scarani , n. gisin , quantum non - locality based on finite - speed causal influences leads to superluminal signalling , _ nature physics _
, vol . 8 , pp .
867870 , 2012 .
. baumeler , s. wolf , perfect signaling among three parties violating predefined causal order , _ proceedings of ieee international symposium on information theory 2014 _ , pp
. 526530 , ieee , piscataway , 2014 .
. baumeler , s. wolf , non - causal computation avoiding the grandfather and information antinomies , _ arxiv preprint _ , arxiv:1601.06522 [ quant - ph ] , 2016 ; accepted for publication in _ new journal of physics _ , 2016 .
a. stefanov , h. zbinden , n. gisin , a. suarez , quantum correlations with spacelike separated beam splitters in motion : experimental test of multisimultaneity , _ physical review letters _ ,
88 , pp . 120404 , 2002 .
l. szilrd , ber die entropieverminderung in einem thermodynamischen system bei eingriffen intelligenter wesen ( on the reduction of entropy in a thermodynamic system by the intervention of intelligent beings ) , _ zeitschrift fr physik _ ,
53 , pp . 840856 , 1929 .
c. wood and r. spekkens , the lesson of causal discovery algorithms for quantum correlations : causal explanations of bell - inequality violations require fine - tuning , _ new journal of physics _ , vol .
17 , pp .
033002 , 2015 .
|
the difficulty of explaining non - local correlations in a fixed causal structure sheds new light on the old debate on whether space and time are to be seen as fundamental . refraining from assuming space - time as given _ a priori _ has a number of consequences .
first , the usual definitions of _ randomness _ depend on a causal structure and turn meaningless .
so motivated , we propose an _ intrinsic _ , physically motivated measure for the randomness of a string of bits : its length minus its normalized work value , a quantity we closely relate to its _ kolmogorov complexity _ ( the length of the shortest program making a universal turing machine output this string ) .
we test this alternative concept of randomness for the example of non - local correlations , and we end up with a reasoning that leads to similar conclusions as , but is more direct than , in the probabilistic view since only the outcomes of measurements that can _ actually all be carried out together _ are put into relation to each other . in the same context - free spirit , we connect the logical reversibility of an evolution to the second law of thermodynamics and the arrow of time .
refining this , we end up with a speculation on the emergence of a space - time structure on bit strings in terms of data - compressibility relations .
finally , we show that logical consistency , by which we replace the abandoned causality , it strictly weaker a constraint than the latter in the multi - party case .
|
it is a big problem to better understand the vacuum structure of string / m theory . even in the case of bosonic string theory which contains the tachyon near its perturbative vacuum
, it turned out not to be a simple deal to find a stable vacuum . note
that hopes for the existence of such a vacuum have their origins already in the seventies @xcite .
however , for open strings a real breakthrough has been achieved in recent years when the open string tachyons were related with annihilation or decay of d - branes via the process of tachyon condensation @xcite ( see @xcite for a review and a list of references ) . in studying the phenomenon of tachyon condensation ,
string field theory methods are the most appropriate ones ( see @xcite for the latest developments ) .
however , in practice it turns out to be very hard to deal with them .
this was a motivation for finding simpler methods , for example toy models @xcite , that can allow one to get some intuition on the physics of the phenomenon .
recently , it was realized that the background independent open string field theory ( biosft ) of witten @xcite is a powerful tool to attack the problem .
in particular , it allows one to compute the tree level effective tachyon potentials @xcite which amusingly coincide with the potentials found from the toy models based on the exactly solvable schr " odinger problem @xcite .
the background independent open string field theory is based on the batalin - vilkovisky formalism @xcite whose master equation provides the effective action of the theory . in the bosonic case , the classical master equation gives a solution @xcite @xmath0 where @xmath1 is the renormalized partition function of the underlying two - dimensional theory defined on the unit disk in the complex plane in such a way that it is a free theory in the bulk but an interacting theory on the boundary .
so , the @xmath2 stand for coupling constants of boundary interactions and @xmath3 for their rg @xmath4-functions . in the supersymmetric case ,
the solution of the classical master equation is simply @xcite @xmath5 as we have already mentioned , it is very simple to compute the tree level effective tachyon potentials from formulae and .
however , to investigate the vacuum structure of the theory , it is necessary to have information about quantum corrections to the potentials .
unfortunately , the quantum master equation which is a proper machinery is still missing for the biosft .
the main problem here is a poor understanding of the space of boundary interactions of the underlying two - dimensional theory .
however , we think that there are enough tools at our disposal to shed some light onto the problem of interest . indeed , having the tree level effective actions , we can evaluate the loop corrections to the tachyon potentials by using ordinary field theory methods . on the other hand , following the biosft , we can also compute partition functions of open strings in the presence of a non - zero tachyon background on riemann surfaces which correspond to string loops . in general , the appearance of these surfaces is not surprising as the effective action has to reduce to the partition function on - shell . in doing so , the comparison with the field theory results is crucial
. then we need to suggest how the loop effective actions are expressed in terms of the partition functions . at the present time is not known whether the situation may be taken under control .
so we are bound to learn something if we succeed . in what follows
, we will restrict ourselves to the one - loop approximation that significantly simplifies the problem .
thus we will consider the partition functions on riemann surfaces of zero genus in the presence of a constant tachyon background .
we should also mention that there have been attempts to generalize the tree level machinery of the biosft to next order @xcite .- adic string theory see , @xcite .
] however , they did not discuss the consistency with the field theory results . as a consequence ,
it has not become completely clear whether tachyon condensation described by the biosft leads to the strong coupling regime as the field theory results assume or not .
the outline of the paper is as follows .
we start in section 2 by directly evaluating the one - loop corrections to the tachyon potentials based on the toy field theory models as well as the tree level actions provided by the biosft .
we do so for both cases , the bosonic and supersymmetric ones . we continue to pursue our program in section 3 by evaluating the partition functions of open strings with a constant tachyon background on riemann surfaces of zero genus . in doing so ,
the comparison with the field theory results is crucial . in section 4
, we define the effective action via the partition function and find the one - loop corrections to the tree level tachyon potentials .
we will also discuss the phenomenon of tachyon condensation as it is seen by closed strings .
the picture that emerges seems quite curious as in terms of closed strings , tachyon condensation looks like the splitting of an original d - brane into two copies which are running apart while the tachyon is rolling down to the local minimum .
section 5 will present our conclusions and directions for future work . in the appendix we give some technical details which are relevant for our discussion in sections 3 and 4 .
the goal of the present section is to evaluate the one - loop corrections to the tree level potentials based on some field theory models of unstable branes .
our discussion in this section is elementary .
we begin with the bosonic string . the field theory model of @xcite as well as the @xmath6 limit of the tree level effective action of the tachyon field in p - adic string theory @xcite lead to the following action on the unstable bosonic brane @xmath7 \quad.\ ] ] from now , we use the minkowski metric on the d - brane .
thus , the metric for the transverse directions is always euclidean . a formal evaluation of the one - loop correction to the potential results in @xmath8 with @xmath9 the second term is a normalization , so @xmath10 equals @xmath11 or @xmath12 .
@xmath13 stands for possible counterterms which are fixed by the renormalization conditions .
so far we have just given a model which could be used in principle as a toy model for tachyon condensation .
however , it suffers from some drawbacks : for example , it does not reproduces the descent relations for brane tensions . because of this , let us consider the effective action that comes out from the background independent open string field theory @xcite . restricting to @xmath14 order
, the action of the tachyon field on the bosonic dp - brane takes the form @xcite @xmath15 \quad,\ ] ] where @xmath16 is the tension of the @xmath17-brane and @xmath18 .
we also include a constant factor of the dilaton .
so , @xmath19 means its vacuum expectation value .
note also that modulo a numerical coefficient in front of the kinetic term , one can get one model from the other by a relation @xmath20 .
like in the previous case , the one - loop correction to the potential can be written in the form but now with @xmath21 as we see , the tachyon mass near the open string perturbative vacuum ( @xmath22 ) differs from its correct value .
this may sound like a problem , but in fact it is not , because the expansion in @xmath23 as it is used in eq .
fails near this vacuum .
so one has to use the expansion in @xmath24 or , alternatively , to take into account higher derivative terms @xcite .
it is clear that the one - loop correction to the potential receives contributions from such terms , so our result will receive corrections too .
according to the results of @xcite , the tachyon mass near the open string perturbative vacuum is finally getting corrected in a proper way . since in both cases
we have the same dependence on the background field ( modulo a field redefinition ) , it is natural to assume that the right hand side of eq .
is only corrected by @xmath25 .
now we will make some remarks about these models : ( i ) in the problem of interest the perturbation theory coupling constant is given by @xmath26 that allows one to say that the theory is getting strongly coupled near the vacuum @xmath27 ( see , e.g. , @xcite and refs .
therein ) . by mapping the field @xmath24 into @xmath28 as @xmath29 , there seems to be no way to `` escape '' the strong coupling regime in the vicinity of this vacuum without coming back to p - adic string theory with a general parameter p. ( ii ) it immediately follows from that we can trust our one - loop estimation at least for a range . ] @xmath30 ( iii ) since the field theory analysis makes no sense for @xmath31 , we do not consider d - instantons but only comment on them in section 5 .
now we want to briefly discuss what the field theory actions of unstable d - brane in superstring theory look like .
we consider a generic case , with the worldvolume of a brane parameterized by @xmath32 .
thus , we assume that the even and odd @xmath33 s are referred to type iia and iib , respectively . the field theory model of @xcite
is given by @xmath34 \quad.\ ] ] where erf stands for the error function : @xmath35 .
the one - loop correction to the potential is given by with the effective mass @xmath36 ^ 2 - 1\bigr ) \quad.\ ] ] on the other hand , the background independent open string field theory provides the following action of the tachyon field on the unstable dp - brane @xcite @xmath37 \quad.\ ] ] the correction to the potential again is given by the general form but with the effective mass @xmath38 just as in the bosonic case , this biosft model does not give the correct mass for the tachyon field near the perturbative vacuum .
the reason are again higher derivative corrections in .
note that modulo a numerical coefficient in front of the kinetic term , one can get the first model by a relation @xmath39 .
finally , let us note that the effective coupling constant of perturbation theory is @xmath40 that allows us to stay at the weak coupling regime as far as @xmath41
in this section we consider open strings in the presence of a constant tachyon background .
our primary interest is to compute the partition functions on a world - sheet with euler number zero .
since we are interested in the oriented theories , we restrict ourselves to the annulus and cylinder . before going to explicit computations ,
let us stress a couple of important points : ( i ) we will consider boundary interactions due to non - primary conformal fields , so the use of conformal maps to transform one worldsheet into another one is pointless because the transformation laws of the fields are unknown .
note that constant background fields are defined via the taylor expansion in @xmath42 of @xmath42-dependent backgrounds , as for example in the biosft , so they can also transform in a non - trivial way under the maps .
( ii ) it is clear that the physics in spacetime must be universal in a sense that it does not depend on whether one uses the cylinder or annulus .
as we have noted in the problem of interest the use of conformal machinery is problematic , so we have to involve other principles .
a related problem is that there are infinitely many non - equivalent backgrounds that makes the computation of the partition function ambiguous . to fix these problems
, we propose to use a comparison with field theory results as basic principle . indeed , given the partition function , one can in principle extract the contribution due to the tachyon from it and fix the background by comparison with the corresponding field theory result computed from the tree level effective action . as we will see shortly , this also leads to equivalent expressions for the partition functions computed for the annulus and cylinder .
first we would like to see what happens on a world - sheet whose geometry is an annulus .
a motivation for doing so comes from the fact that the biosft is defined on the disk , so it seems natural to expect that one - loop corrections come from the theory on the annulus .
moreover , there is another interesting problem which remains to be fixed : the field theory analysis results in the effective coupling constant which assumes that the theory becomes strongly coupled due to tachyon condensation while the world - sheet analysis of @xcite leads to an opposite conclusion .
to proceed , we briefly recall the main points of the theory on a disk @xcite .
we add the boundary interaction to the standard action .
@xmath43 where @xmath44 is the radius . in
what follows , we restrict ourselves to a constant value for the tachyon background , i.e. @xmath45 . the computation of the tree level effective action along the lines of the biosft is elementary .
first , we define a renormalized coupling @xmath46 that results in the following @xmath4-function @xmath47 .
next , using the fact that the renormalized partition function on the disk is @xmath48 , we get from the formula the same potential as in after the identification @xmath49 .
it is also straightforward to compute the partition function on the annulus whose inner and outer radii are @xmath50 and @xmath44 , respectively . in this case
we add two tachyon backgrounds @xmath51 \quad\ ] ] and restrict ourselves to constant backgrounds : @xmath52 and @xmath53 .
we define renormalized couplings : @xmath54 and @xmath55 .
since the boundary interaction does not depend on @xmath42 , the computation of the corresponding path integral is elementary .
we get @xmath56 where @xmath57 is the d - brane volume .
it is natural to assume that @xmath58 coincides with the corresponding background on the disk , i.e. @xmath59 . in other words
, we consider the annulus as being produced from the disk by cutting a hole .
although this procedure has no influence on the background at the outer boundary , it does not say how to define the background at the new inner boundary .
the later makes the partition function ambiguous .
as we have already mentioned , a conformal map which maps one boundary into one another does not help as the boundary interactions are not due to the primary conformal operators and their transformation laws are unknown . to fix the background , we use the comparison with the field theory result of section 2 .
up to an overall factor , the contribution due to the open string tachyon is given by @xmath60 using the fact that @xmath61 , we find @xmath62 by comparing the exponents in and @xmath63 finally , we get @xmath64 at first glance , the relation between the backgrounds on the different boundaries might seem strange .
however , it is the price to pay for the unknown transformation law of the background fields under the conformal map of one boundary into another . from this point of view
it seems reasonable to consider the cylindric geometry when both boundaries are apparently treated on an equal footing , so the backgrounds on them are the same .
moreover , the cylinder is a finite strip obtained by cutting from an infinite one . in doing so , one does not change the background field living on the boundaries
. thus the background is fixed at the tree level by a comparison with field theory .
so , the computation of the partition function may be considered as a consistency check .
to proceed , we consider the open bosonic string on the worldsheet parameterized by coordinates @xmath65 with a range for the spacial coordinate @xmath66 to be @xmath67 .
we add the boundary interaction @xmath68 \quad\ ] ] to the standard action and restrict ourselves to a constant tachyon background ( coupling constant ) @xmath69 in what follows . the tree level effective action is easily computed via treated now as an ansatz .
first , we introduce a scale in the problem at hand that restricts @xmath70 to @xmath71 and simply rescales @xmath66 .
next , we define a renormalized coupling as @xmath72 .
this implies that the renormalized partition function on the strip is simply @xmath73 and the rg beta function is @xmath74 .
finally , a simple algebra results in the same potential as in after the identification @xmath75 . having fixed the background , let us now compute the partition function on the cylinder . again , since the boundary interaction does not depend on the @xmath42 s , the path integral is easily worked out .
the result is given by @xmath76 where @xmath77 is the dedekind eta function .
as a consistency check , we extract from @xmath78 the contribution due to the open string tachyon @xmath79 a comparison with the field theory result of section 2 gives @xmath75 . here
we have found the same relation between the renormalized value of the tachyon background and the classical field as at the tree level that is of course a desired result .
moreover , a simple algebra shows that the expression for the partition function we have found on the cylinder coincides with the one on the annulus .
this is again a desired result .
note that we have not used the conformal map for mapping the annulus into the cylinder from the beginning .
the reason for doing so is that we do not know the transformation law for the background fields . at this point
a comment is in order .
we should note that in the case of a constant tachyon background the representation of the partition function as a loop of open string states is much like that without any background .
indeed , one can rewrite it as , see @xcite . ]
@xmath80 where @xmath81 and @xmath82 is defined as the coefficient of @xmath83 in @xmath84 .
this representation looks like a sum of the point - particle free energies ( in the coleman - weinberg representation ) over the open string spectrum but with the masses of the particles shifted according to eq .. it is natural to treat the particle with @xmath85 as the tachyon and compare its contribution with the field theory result .
in fact , the last assumes that the partition function coincides with the one - loop correction to the effective action .
we will return to this point in section 4 .
now , let us analyze the expression for the partition function .
be reminded what happens for @xmath86 : in this case the integrand has divergencies as @xmath87 or @xmath88 which come from the lightest string modes . in order to see what is going on in the case of a nonzero tachyon background ,
we begin with @xmath87 . in studying this limit
, we really need the representation of the partition function in terms of a loop of open string modes , namely eq . ( or equivalently )
expanding the products in @xmath89 , we get @xmath90 \quad.\ ] ] in the range ( @xmath75 ) , a possible divergence can only come from the leading term if the value of the background is less than its critical value @xmath91 .
it can be fixed by using the analytic continuation in @xmath92 .
indeed , evaluating the integral for @xmath93 as @xmath94^{\frac{d}{2}}\ , \gamma(-\frac{d}{2})\ ] ] and then defining it for @xmath95 via the analytic continuation @xmath96 , we get the desired result . at this time we do not care about the last factor and the sign in @xmath97 .
to complete the picture , let us also evaluate the contribution of massive open string modes in eq .. integrating over @xmath98 , we end up with the following sum @xmath99 it is well - known that @xmath82 rapidly grows with @xmath100 that leads to a divergency . to be more precise , for @xmath101 ,
@xmath82 is asymptotically @xmath102 thus , the contribution of massive open string modes is divergent and this is the case for any value of the background field .
we will return to this point shortly after working out the other limit .
we can readily rewrite the product in in terms of @xmath103 .
this yields another representation of the partition function which is useful in studying the limit @xmath88 .
we then get the following expansion of the partition function @xmath104 \quad.\ ] ] before continuing our discussion , we will make a detour and recall some basic results on the feynman propagator of a scalar particle of mass @xmath105 in @xmath33 dimensions that is given by the familiar formula @xmath106 and can also be rewritten as @xmath107 with @xmath108 . given the integral representation for the propagator , we can readily rewrite the expansion as @xmath109 \quad,\ ] ] where @xmath110 . at this point a couple of remarks
are in order .
( i ) such a representation makes sense only in the case of @xmath111 .
so , we will always assume this if not stated otherwise .
( ii ) in contrast with the representation in terms of open string modes ( see ) , the representation includes closed string modes whose masses are not shifted . as in the case of the representation via open string modes , a natural question to pose
is whether the sum over all massive closed string modes is convergent or not . to proceed , we will use the familiar asymptotics for the propagator of a massive scalar particle @xmath112 \quad,\quad mr\rightarrow\infty\ ] ] together with the asymptotics for @xmath82 that allows us to evaluate the contribution of the massive closed string modes in eq .
@xmath113 it is obvious that the sum is convergent if @xmath93 , i.e. the background field must be larger than its critical value .
thus , the critical value @xmath114 now means convergence of not only a single contribution to the partition function by the open string tachyon but all massive closed string states .
it is instructive to see if eq .
provides the same scaling behavior near the critical point as eq .. using the fact that in the vicinity of the critical value @xmath115 , a simple estimation results in the desired scaling @xmath116 having related the divergence due to the massive closed string modes to the open string tachyon , it seems natural to relate the divergence due to the closed string tachyon to the massive open string modes .
as we have seen , the divergence due to the massive open string modes does not crucially depend on the value of the background field .
that is also clear in terms of the closed string tachyon where it is due to the poles of the propagator .
we will now use similar methods to estimate the asymptotic behavior @xmath117 for large @xmath92 .. ] in this limit , the propagator of a massive mode is given by which allows us to estimate the contribution of all massive modes except the tachyon .
it is given by the lightest massive mode and therefore exhibits an exponential falling .
on the other hand , the contribution of massless modes which is due to the propagator @xmath118 shows a power like behavior . and
@xmath119 deserve a special consideration .
we will discuss this issue in section 4 .
] this means that in this limit we can omit all massive modes which is the expected result . however , there is also the contribution of the closed string tachyon which makes the estimation more involved . in this case
the propagator becomes divergent as it is clear from eq . or
eq .. so , one must somehow treat this divergency but then the final result is ambiguous , i.e. , it depends on the way of treatment .
we refer to the appendix for more details .
due to this , finding the asymptotics of the partition function for large @xmath92 is problematic .
moreover , it also becomes difficult to extract the real part of @xmath117 .
indeed , there are two contributions to @xmath120 : the first one comes from the open string tachyon or , equivalently , from massive closed string modes if @xmath121
. it can be readily found from eq .. explicitly , @xmath122 where @xmath123 is the heaviside step function .
the second comes from massive open string modes , or equivalently , from the closed string tachyon propagator .
so , from we have @xmath124 we will return to the ambiguities by a discussion of physics in the next section
. we will now use similar methods to compute the partition functions for the open nsr string in the presence of a constant tachyon background . since
the generalization of the results of section 3.1 is rather straightforward , we omit some technical details .
let us first recall the computation within the open nsr string on the disk .
so , we add the boundary interaction @xmath125 to the standard worldsheet action . as in the bosonic case , we consider a constant background field @xmath126 . in computing the partition function on the disk ,
the only novelty compared to the bosonic computation is that the renormalized coupling is now defined as @xmath127 .
this leads to a partition function @xmath128 that together with the definition provides the same potential as in after the identification @xmath61 .
note that the r sector gives no net contribution to @xmath129 .
having computed the partition function on the disk , we now want to consider the partition function on the annulus .
we proceed in the same way as in the bosonic case namely , we introduce two backgrounds @xmath130 \quad\ ] ] and restrict ourselves to constant backgrounds : @xmath52 and @xmath53 .
we define renormalized couplings : @xmath131 and @xmath132 . in the case of constant backgrounds
the computation of the partition function is a simple task .
we find @xmath133 \quad .
\end{split}\ ] ] note that in terms of the closed string modes , the partition function is expressed only via the ns - ns states .
this sounds good because we are interested in unstable branes .
moreover , it also assumes that there is no closed string tachyon because of the corresponding cancellation between the ns and r sectors . as in the bosonic case
, it is natural to assume that @xmath58 coincides with the background on the disk , so @xmath59 . to fix the background on the inner boundary
, we extract the contribution due to the open string tachyon @xmath134 and use the fact that @xmath61 .
thus we get for the background @xmath135 note that in order to extract the contribution due to the open string tachyon one has to consider the @xmath136 limit that makes the right hand side of eq . positive .
finally , the partition function is given by @xmath137 \quad .
\end{split}\ ] ] having computed the partition function on the annulus , let us see what happens on the cylinder . to this end , we start with the open nsr string on the strip and add the boundary interaction @xmath138 \quad\ ] ] to the standard worldsheet action . as in the bosonic case , we consider a constant background field @xmath69 . in computing the partition function on the strip ,
the only novelty compared to the bosonic computation is again that the renormalized coupling is now defined as @xmath139 .
this leads to a partition function @xmath140 that together with the definition provides the same potential as in after the identification @xmath75 .
let us now consider the partition function on a finite strip , the cylinder .
evaluating the path integral , we find @xmath141 \quad,\ ] ] where @xmath142 and @xmath143 come from the ns and r sectors , respectively .
these have the product representations : @xmath144 as a consistency check , we again extract from @xmath78 the contribution due to the open string tachyon @xmath145 a comparison with the field theory result of section 2 gives @xmath75 .
we have found the same relation between the renormalized value of the tachyon background and the classical field as at the tree level that is of course the desired result .
moreover , a simple algebra shows that the expressions for the partition function on the cylinder and annulus coincide .
this is again the desired result .
note that we have not used the conformal map for mapping the annulus into the cylinder from the beginning .
the reason for doing so is that we do not know the transformation law for the background fields .
as in the bosonic case , the partition function can be rewritten as a sum of the point - particle free energies ( in the coleman - weinberg representation ) .
it is natural to treat the leading term as the tachyon contribution and compare it with the field theory result .
having computed the partition function in terms of a loop of open string modes , one can formally analyze it by expanding the products in @xmath146 .
this yields @xmath147 \quad.\ ] ] in the limit @xmath87 a possible divergence can come only from the leading term if the absolute value of the background is less than its critical value @xmath148 . as in the bosonic case
, one can use the analytic continuation to fix the problem . to be more precise , @xmath149 where @xmath150 if @xmath151 and @xmath152 if @xmath153 .
now , we come to the question of whether the series over all open string massive modes is convergent or not . unlike the bosonic case , here the sum is alternating , so there is a chance to get a finite result
. we will not directly examine this problem but use our experience with the bosonic case where the divergency has been attributed to the closed string tachyon mode . since there is no tachyon among the ns - ns states , we claim that the sum is convergent .
we can readily rewrite the products in in terms of @xmath154 .
this yields another representation of the partition function .
expanding now in @xmath154 , we have @xmath155 \quad.\ ] ] as in section 3.1 , the fact that the feynman propagator of a scalar particle of mass @xmath105 in @xmath33 dimensions has the integral representation allows us to rewrite this expansion as @xmath156 \quad,\ ] ] where @xmath157 .
note that in the closed string modes appear with the same masses as in the case @xmath86 .
now let us check whether the sum over all massive closed string modes is convergent or not .
to do so , we proceed in precisely the same way as in section 3.1 . using the asymptotics of the propagator , we get @xmath158 where @xmath159 is defined as the coefficient of @xmath83 in @xmath160 .
a similar technique as it is used to estimate @xmath82 for large @xmath100 can be used to get the asymptotics for @xmath159 .
the result is simply @xcite @xmath161 comparing the exponent to that of , we see that the sum over massive modes is convergent if @xmath162 . in other words ,
the absolute value of the background field must be larger than its critical value @xmath114 we have found by studying the contribution to the partition function of the open string tachyon .
it should be also noted that a simple check shows that eq .
leads to the same scaling behavior near the critical point as eq . does .
it is interesting to consider the large @xmath92 limit again .
since there is no closed string tachyon , the asymptotic of the partition function is due to the massless ns - ns modes .
explicitly , @xmath163 note that the right hand side of is singular in the two cases @xmath164 and @xmath165 .
we postpone discussion of this complication until section 4 .
it is of some interest to determine the real part of @xmath117 which appears if the absolute value of the background field is less than the critical one .
for doing so , we use the representation of the partition function as a loop of open string modes . from
we get @xmath166 to summarize , we have computed the partition functions by evaluating the corresponding path integrals .
it turns out that the results admit the representation as a sum of the point - particle free energies ( in the coleman - weinberg representation ) over the open string spectrum but with the shifted masses of the particles .
this allows us to compare the tachyon contribution with the corresponding field theory result .
moreover , the partition functions are easily rewritten via closed string modes with the standard masses .
this reveals the transverse directions to the brane and looks like closed string modes appearing from one brane , propagating a distance @xmath167 , and then disappearing on the other brane .
in other words , the description in terms of closed strings assumes that the original brane splits into two parallel branes separated by a distance depending on the vev of the open string tachyon .
note that the expressions for the partition functions are similar to those for the partition functions of open strings stretched between two parallel branes ( see , e.g. , @xcite ) .
however , our situation is absolutely different as we are interested in the partition functions of open string modes on a single brane but in the presence of a constant tachyon background .
thus , it is a priori unclear why in this case closed strings `` see '' the background in such a way .
by now we have considered the one - loop corrections to the tree level potentials as they follow from some field theory models and the partition functions of open strings in the presence of a constant tachyon background .
the remaining question is how to properly combine them in order to find the one - loop corrections to the tachyon potentials .
as we have already mentioned , in the case of a constant tachyon background the partition functions can be rewritten as the sums of the point - particle contributions over the open string spectrum .
it follows from this fact that we already have the desired correspondence with the field theory results on the level of the partition functions .
thus it seems natural , modulo terms due to a normalization and renormalization conditions ( see ) , to identify the one - loop correction to the effective tachyon potential with the open string partition function in both cases , the bosonic and supersymmetric ones .
explicitly , @xmath168 this also means the identification of the renormalized background field @xmath92 in @xmath117 with the classical field @xmath169 .
a natural question to pose at this point is whether there are other indications in favor of this identification .
this time we have one which is as follows .
an interesting point of view on the formulae and proposed in @xcite is to relate them with a possible way of subtraction of the m " obius infinities from the disk partition functions .
indeed , it is well known that the m " obius group volume is linearly divergent in the bosonic case while it is finite in the supersymmetric case @xcite .
it is easy to see that the definition leads exactly to the cancellation of the divergency .
as there is no need to deal with the divergency in the supersymmetric case , it now seems natural to identify the partition function with the effective action as in . having the scenario for the disk
, it is a simple matter to generalize it for all higher genera . since in this case
there is no m " obius invariance , there is no need to deal with the m " obius infinities and , as a consequence , one can simply identify the corresponding correction to the action with the partition function , namely @xmath170 from this point of view , eq . is a special case of this relation .
note that for @xmath171 this relation has already been suggested in the literature as a possible ansatz for finding the one - loop effective actions via the corresponding open string partition functions @xcite .
moreover , in the context of the effective action for massless modes it was proposed in @xcite .
having combined the worldsheet constructions with the field theory results , we have the expressions for the one - loop corrections to the effective potentials .
so , it is time to consider them in more detail . to proceed , we will specialize first to the case of the nsr string as it turns out to be simpler than the case of the bosonic string .
as we have seen , the partition function has a real part if the absolute value of the background field is less than its critical value . as a result
, this will cause the one - loop correction to the potential to be complex for this range .
this difficulty is due to the open string tachyon , or equivalently closed massive modes .
it is advantageous to analyze it in terms of the open string tachyon as it allows us to use the well developed machinery for analyzing perturbative instabilities in field theories with nonconvex potentials ( see , e.g. , @xcite ) .
thus , a physically meaningful quantity to compute in this range is the decay rate per unit volume of an initial unstable state .
formally , in terms of the imaginary part of @xmath172 it is given by @xmath173 since this quantity must be positive , this requirement fixes the sign in our formula for the analytic continuation .
then from together with it follows that @xmath174 on the other hand , for @xmath175 there is no problem with the imaginary part .
this allows us to write down the one - loop correction to the tachyon potential @xmath176 \quad.\ ] ] at this point , let us make a remark . from our discussion in section 3
it follows that @xmath172 is finite , so @xmath177 .
thus , the use of @xmath178 in eq . or @xmath179 as a normalization of @xmath172 does not lead to a modification of our results . to be precise ,
this is the case for @xmath180 .
we will return to this point shortly .
it is of some interest for what follows to look at the one - loop correction in the limit of large @xmath169 . using the results for the asymptotical behavior of the corresponding partition functions
, we immediately get from relation @xmath181 a special consideration is needed for @xmath182 where the argument in the @xmath183-function becomes nonpositive .
it might lead to divergencies as the @xmath183-function has poles .
in fact , there is no problem for @xmath184 , as in this case we can use the familiar analytic continuation for the @xmath183-function . as a result , we end up with @xmath185 however , this procedure does not work for @xmath186 .
the problem is that the propagators of massless particles are ill - defined in the corresponding codimensions ( @xmath187 and @xmath188 ) .
first , let us consider the case @xmath164 .
as is known , one way to define the propagator of a massless scalar particle in two dimensions is to introduce a long - distance cutoff ( scale ) . then the propagator takes the form @xmath189 .
note that a constant part can be fixed by a proper normalization at @xmath190 . keeping this in mind , we proceed in the following way .
we use dimensional regularization for expression @xmath191 \quad.\ ] ] however , to emphasize the physical role of the cutoff , we exchange @xmath192 for @xmath193 and fix a constant part by a normalization at @xmath194 .
then this expression takes the form @xmath195 although for @xmath165 there is no direct analogy with field theory ( this is the case of a spacefilling brane , so the corresponding codimension is zero ) , the preceding computation formally generalizes without difficulty to this case .
the result is given by @xmath196 at this point , there is a subtle issue to discuss .
we can evaluate the propagators by cutting off the upper limit of integration in expression for @xmath197 .
in particular , a simple algebra shows that in the case of a spacefilling brane the corresponding integral is @xmath198 .
here we restrict ourselves to the leading asymptotics in @xmath199 .
in fact , the power like divergency is well - known in string theory without tachyon background ( @xmath200 ) .
it is removable by the fischler - susskind mechanism @xcite which assumes that @xmath199 is identified with the worldsheet cutoff @xmath201 .
this just provides a correction to the worldsheet beta functions . for @xmath202 ,
the situation is quite different .
as we have seen , it is possible to avoid power - like divergencies by using dimensional regularization .
however , the novelty now is the appearance of logarithms .
note that a naive attempt to apply the fischler - susskind mechanism to the logarithms runs into trouble because it would lead to worldsheet divergencies like @xmath203 .
to interpret these logarithms , let us look at the representation of the partition function via open string states .
it looks like a sum of the point - particle free energies written in the coleman - weinberg form ( see , e.g. , ) . for large values of @xmath169 , the integrals are generically divergent at the lower limit that corresponds to short - distance divergencies .
the use of dimensional regularization ( see , e.g. , ) shows a pole @xmath192 for even @xmath33 which can be exchanged for a logarithm .
note that this is the ultraviolet divergency of the theory on a brane . to get the representation via closed string states , we changed variables as @xmath204 .
it is clear that it results in the logarithmic divergencies at the upper limit of integration .
but this is exactly what we found in the closed string channel .
moreover , we found that closed string states propagate in the transverse directions to the brane i.e. , in the bulk .
thus the moral of this story in fact is known : the uv / ir correspondence .
note that the real situation is more subtle than what we sketched above .
indeed , the logarithms appear only for codimensions @xmath188 and @xmath187 while the coleman - weinberg formula ( see ) assumes their appearance for any even dimension .
an answer to this is the cancellation between contributions of different open string states .
it is clear that further work is needed to make this more rigorous .
as we have already mentioned , the relation is defined modulo a normalization and renormalization conditions ( see ) . from our analysis of the partition functions it follows that the refinement is needed only for @xmath182 .
we propose to normalize the corrections for these cases at some large @xmath194 . as a result ,
the one - loop corrections at large @xmath169 are given by @xmath205 the main conclusion we can draw from the large @xmath169 behavior of the one - loop corrections is that except for a special case at @xmath165 they are not falling at infinity .
this will have a strong impact on our discussion of tachyon condensation .
now we have all at our disposal to see how the one - loop approximation to the tachyon potential looks like .
a typical picture is that of figure 1 .. we also assume that the vev of the dilaton is large enough to discard the one - loop corrections near @xmath206 .
] there are two different scenarios that crucially dependent on the string coupling constant near the perturbative vacuum ( @xmath207 ) : i. the strong coupling regime ( @xmath208 ) appears before @xmath169 becomes large .
this means that we gain nothing essentially new by the one - loop corrections as the system rolls down into the strong coupling regime as it also does in the tree level approximation .
the strong coupling regime appears at a very large value of @xmath169 , where our approximation for the propagator is valid . in this case
, a new minimum @xmath209 appears at a finite value of the tachyon field .
is a special case .
we will return to this point in section 5 .
] so the one - loop corrections become relevant and , as a result , the system does not roll down into the strong coupling regime as the tachyon condenses .
this means that the phenomenon of tachyon condensation can be described in this case by weakly coupled theory .
the situation in the bosonic case is much more involved .
indeed , as we have seen in section 3.1 , there are two contributions to @xmath117 ( see and ) that become relevant in studying the imaginary part of the one - loop potential .
following our experience with the supersymmetric case , we can easily elaborate the contribution due to the open string tachyon .
thus , we interpret it again as the perturbative field theoretical instability and its contribution to the decay rate per unit volume of an initial unstable state is simply @xmath210 that is in agreement with the results that are already avaliable in the literature for the case of @xmath207 ( see @xcite ) .
however , the other contribution to the imaginary part of the one - loop correction makes the situation more involved .
remind that it can not be interpreted in terms of the open string tachyon , as it comes from open string massive modes or , equivalently , the closed string tachyon . in practice ,
the last turns out to be more useful .
formally , we have @xmath211 that as it follows from in the limit of large @xmath169 reduces to @xmath212 \quad.\ ] ] a naive attempt to make use of the above formula for analyzing a possible instability at @xmath179 fails . indeed , as in the supersymmetric case
, we can consider two scenarios .
the point is that for both of them @xmath179 belongs to the strong coupling regime , so this is outside the reach of our approach .
anyway , we have found that in the problem of tachyon condensation , there is the instability due to the closed string tachyon .
the physical meaning of this effect should be further clarified .
note also that there was a speculation with weak evidence for it in the literature @xcite that interprets the instability of the vacuum @xmath179 as the closed string tachyon instability which leads to the true vacuum at @xmath213 .
we will return to this issue in section 5 .
given what was said in the previous sections , the interpretation of the dynamics of tachyon condensation in terms of closed strings is clear . indeed ,
initially a d - brane is at the open string perturbative vacuum @xmath207 , i.e. at a local maximum of the potential .
then as the tachyon is rolling down , the brane is getting split into two branes with a distance in between depending on @xmath169 , as shown on figure 2 .
the further behavior depends on details : in the bosonic case , there is the instability due to the closed string tachyon and the development of this instability is not clear yet ; in the supersymmetric case , the further behavior crucially depends on the chosen scenario . for scenario i , the branes enter the strong coupling regime where our tools become inappropriate to explore the details of the process of condensation . for scenario
ii , the branes are getting stabilized at some distance as the potential develops the new minimum at @xmath209 .
it is of some interest to notice that there is an analogy with field theory models that undergo spontaneous symmetry breaking @xcite that helps to realize the brane dynamics of figure 2 at the beginning of the decay .
indeed , given an effective potential , for example @xmath214 , one can study the dynamics of spontaneous symmetry breaking by setting an initial narrow distribution of the scalar field at the maximum of the potential ( @xmath22 ) .
then , by quantum fluctuations it spreads out and shows two maxima . to reveal such a picture in the problem of interest ,
let us recall two key facts we have already seen .
the first is that in terms of closed strings @xmath169 transforms into a transverse direction to a brane .
it is the reason why we have drawn a part of the profile of the potential in figure 2 .
the second is that a brane itself can be considered as a lump solution of the higher dimensional worldvolume theory that allows one to say that this solution plays a role of the initial distribution in field theory
. finally , let us note that such a description of the decay shows that these lumps interact by the closed string exchanges .
what we have learned is the one - loop approximation to the tachyon potentials .
we did so by the worldsheet methods .
the key fact to stress at this point is that the tachyon backgrounds we have considered are not due to conformal primary fields .
this might seem like a problem as it would lead to an ambiguity . to fix this problem ,
we have used the correspondence with field theory . in the bosonic case , our analysis has also revealed the instability of d - branes due to the closed string tachyon .
moreover , it allowed us to see the dynamics of tachyon condensation in terms of closed strings that is similar to the dynamics of spontaneous symmetry breaking in field theory .
the first is the physical meaning of the instability of bosonic d - branes due to the closed string tachyon . a proposal of @xcite that relates it to the tunneling through the potential barrier from @xmath179 to @xmath213 seems inappropriate in our case .
indeed , the tunneling results in the nonperturbative instability that behaves like @xmath215 .
clearly , we do not have such a behavior as we have found the closed string tachyon instability which is perturbative .
rather , this instability can be attributed to a peculiarity of euclidean field theory with the tachyonic particle .
the second issue is how to incorporate d - instantons .
it is obvious that the field theory analysis of section 2 does not help in doing so because @xmath31 for this case .
in fact , the problem is deeper : even at the tree level the worldsheet methods , e.g. biosft , have turned out to be useful to check some of the sen conjectures about tachyon condensation .
but they have stumbled on d - instantons as it is unclear how to proceed further namely , how to get the closed string perturbative vacuum without any brane excitations .
it seems that at this point the worldsheet methods should be revised as the appearance of d - instantons usually requires a modified construction of string perturbation theory ( see , e.g. , @xcite ) .
note also that a formal substitution @xmath31 into formulae of section 4 , e.g. eq . , is not legitimate as our analysis is based on the use of the euclidean metric for the transverse directions to branes rather than the minkowski metric .
this is another subtlety we face with d - instantons .
as we have seen , the definition of the one - loop correction @xmath172 to the tachyon potential through the open string partition function @xmath117 ( see ) results in a potential which is complex for some range of @xmath169 .
it is known that in this case the physically meaningful quantity to compute is the decay rate per unit volume which is defined in terms of the imaginary part of @xmath172 .
it was suggested in @xcite that in field theory models the real part of the potential can , nevertheless , be interpreted as the energy density .
this assumes the use of an analytic continuation of @xmath172 to dangerous regions .
it is clear that an attempt to do so in the case of the bosonic string fails as there is no region where @xmath172 is real . in the supersymmetric case
the situation is better since the one - loop correction is real for @xmath216 .
so , it might be possible to define the energy density for @xmath217 as well . as an example , let us briefly illustrate this for odd @xmath33 . the results of sections 3 and 4 yield @xmath218 where @xmath219 is defined as the coefficient of @xmath83 in @xmath220 . as we have noticed in section 3 ,
the sum is convergent , so the only problem is that it becomes complex valued for @xmath221 due to the contribution from the open string tachyon .
we use the analytic continuation ( see ) to obtain the one - loop correction for this region @xmath222 it is interesting to note that our result simply assumes dropping the contribution of the tachyon .
moreover , the tree level vacuum @xmath207 remains the extremum at the one - loop level too .
the latter is also clear by applying a general argument : since @xmath172 is an even function of @xmath169 , its derivative is odd .
the case of even @xmath33 is more delicate as it requires the treatment of the divergencies .
however , if the symmetry @xmath223 is preserved then the above argument also holds .
the case of a spacefilling brane is quite intricate .
indeed , in this case our interpretation of a brane decay as the splitting of the original brane into two parallel branes becomes problematic .
moreover , the logarithmic divergency obviously deserves the better understanding .
finally , the analogy of field theory models with spontaneous symmetry breaking deserves to be studied further .
let us note at this point that the idea to use a higgs like mechanism within the process of tachyon condensation has been already proposed in the literature @xcite . however , our discussion of it is quite different .
we would like to thank j. polchinski , a. sen , s. shenker , w. taylor for useful discussions and also h. dorn and a.a .
tseytlin for comments and reading the manuscript .
part of this work was done while o.a .
was participating in m - theory program at the institute of theoretical physics at ucsb supported by the nsf under grant no .
phy99 - 07949 .
this work is also supported in part by dfg under grant no .
do 447/3 - 1 and the european commission rtn programme hprn - ct-2000 - 00131 .
t.o . would like to thank the graduiertenkolleg `` the standard model of particle physics - structure , precision tests and extensions '' , maintained by the dfg .
in this appendix we will briefly discuss the propagator of the tachyon of mass @xmath224 in @xmath33-dimensional euclidean space based on the integral representation .
for what follows , it is convenient to take such a frame with @xmath225 and @xmath226 . to gain intuition on what is going on ,
let us begin with @xmath227 . in this case
, the integral suffers from divergencies at @xmath228 .
a possible way to treat them is to slightly deform the integration contour in their vicinity .
obviously , there are four possibilities in doing so : after closing the contours at infinity , . ]
the integrals are given by the residue of the integrands .
so , we get @xmath229 let us also note at this point that there is no ambiguity for a real mass @xmath105 because in this case the poles lie on the imaginary axis
. moreover , the closing of the contour at infinity picks up a contribution only from one pole . to see ambiguity for generic @xmath33
, we note that in our special frame the propagator admits the following representation @xmath230 \quad,\ ] ] where @xmath231 .
together with , this implies that the propagator is ambiguous .
in fact , only @xmath232 are relevant for our discussion of the bosonic partition function .
indeed , the closed string tachyon contribution to the partition function can be obtained modulo a numerical coefficient from the contribution of the lightest massive states by @xmath233 .
this is equivalent to the use of @xmath232 for the propagator because only in this case one pole contributes . in particular ,
the imaginary part of the propagator behaves at large @xmath167 as @xmath234\ ] ] as it follows from eq . after the continuation to an imaginary mass .
99 k. bardakci , 331 ; + k. bardakci and m.b .
halpern , 4230 ; 285 ; + k. bardakci , 297 ; + v.a .
kostelecky and s. samuel , 263 .
a. sen , 027 ; 4061
. k. ohmori , `` a review on tachyon condensation in open string field theories '' , hep - th/0102085 .
d. gross , `` split string field theory '' , presented at avatars of m - theory , santa barbara , http://online.itp.ucsb.edu/online/mtheory_c01/gross/ ; + a. sen , `` vacuum string field theory and d - brane solutions '' , presented at avatars of m - theory , santa barbara , http://online.itp.ucsb.edu/online/mtheory_c01/sen/ ; + w. taylor , gauge invariance and tachyon condensation in open string field theory , presented at string s 2001 , mumbay , http://theory.theory.tifr.res.in/strings/proceedings/taylor/ ; + b. zwiebach , `` vacuum string field theory and the sliver state '' , presented at avatars of m - theory , santa barbara , http://online.itp.ucsb.edu / online / mtheory_c01/zwiebach/. j.a .
minahan and b. zwiebach , 029 ; 038 .
e. witten , 5467 ; 3405 ; + k. li and e. witten , 853 ; + s.l .
shatashvili , 83 ; `` on the problems with background independence in string theory '' , report no . iassns - hep-93/66 , hep - th/9311177 .
gerasimov and s. shatashvili , 034 .
d. kutasov , m. marino , and g. moore , 045 .
d. kutasov , m. marino , and g. moore , `` remarks on tachyon condensation in superstring theory '' , report no . efi-2000 - 40 , hep - th/0010108 .
i.a . batalin and g.a .
vilkovisky , 27 .
m. marino , 059 ; + v. niarchos and n. prezas , `` boundary superstring field theory '' , report no .
efi-2001 - 08 , hep - th/0103102 .
t. suyama , `` tachyon condensation and spectrum of strings on d - branes '' , report no .
ou - het 380 , hep - th/0102192 ; + r. rashkov , k. s. viswanathan , and y. yang , `` background independent open string field theory with constant @xmath235 field on the annulus '' , hep - th/0101207 ; + k. bardakci and a. konechny , `` tachyon condensation in boundary string field theory at one loop '' , hep - th/0105098 ; + b. craps , p. kraus , and f. larsen , `` loop corrected tachyon condensation '' , report no .
efi-01 - 15 , hep - th/0105227 .
k.s . viswanathan and y. yang , `` tachyon condensation and background independent superstring field theory '' , hep - th/0104099 ; + m. alishahiha , `` one - loop correction of the tachyon action in boundary superstring field theory '' , report no .
itfa-2001 - 13 , hep - th/0104164 ; + g. arutyunov , a. pankiewicz , and b. stefa ' nski , jr , boundary superstring field theory annulus partition function in the presence of tachyons
, report no .
aei-2001 - 054 , hep - th/0105238 ; + t. lee , k.s .
viswanathan , and y. yang , `` boundary string field theory at one - loop '' , hep - th/0109032 .
minahan,quantum corrections in @xmath236-adic string theory
, report no .
uuitp-02/01 , hep - th/0105312 .
m. kleban , a. lawrence , and s. shenker , `` closed strings from nothing '' , report no .
slac - pub-8735 , hep - th/0012081 .
j. iliopoulos , c. itzykson , and a. martin , rev . of mod .
47 ( 1975 ) 165 . j. polchinski , `` string theory '' , volume i ( cambridge university press , 1998 ) . j.a .
harvey , d. kutasov , and e. martinec , `` on the relevance of tachyons '' , report no .
efi-2000 - 9 , hep - th/0003101 .
green , j.h .
schwarz , and e. witten , `` superstring theory '' , volume i ( cambridge university press , 1987 ) .
danielsson , a. g " uijosa , and m. kruczenski , 011 .
tseytlin , 81 ; + o.d .
andreev and a.a .
tseytlin , 205 .
tseytlin , 165 ; 1257 .
e.j . weinberg and a. wu , 2474 .
w. fischler and l. susskind , 262 .
g.felder , j.garcia-bellido , p.greene , l.kofman , a.d.linde , and i.tkachev , phys.rev.lett.87 ( 2001 ) 011601 ; + g. felder , l. kofman , and a.d .
linde , `` tachyonic instability and dynamics of spontaneous symmetry breaking '' , report no . su - itp-01 - 26 , hep - th/0106179 .
j. polchinski , 6041 .
a.a . gerasimov and s. shatashvili , 019 .
|
we compute the one - loop corrections to the tachyon potentials in both , bosonic and supersymmetric , cases by worldsheet methods . in the process
, we also get some insight on tachyon condensation from the viewpoint of closed string modes and show that there is an instability due to the closed string tachyon in the bosonic case . +
pacs : 11.25.-w + keywords : string theory , unstable d - branes hep - th/0109187 + hu berlin - ep-01/32
|
in his original work on path integrals , feynman @xcite noted that recognizing known facts from different perspectives can lead to new and interesting insights .
quantum mechanics in particular has been an important example of this observation , having schrdinger s differential equation and heisenberg s matrix algebra .
while the two theories mathematical descriptions are seemingly distinct , dirac s transformation theory proved their equivalence . in 1948 , feynman @xcite added a third important mathematical formulation of quantum mechanics based on some of dirac s observations about the role of the classical action in quantum mechanics .
this third description is also known as feynman s path integral formalism and , in combination with feynman diagrams , proved to be fundamental for the development and study of quantum field theories ( qfts ) .
unfortunately , the path integral is a very elusive object .
in fact , only for quantum mechanics an analytically well - defined path integral construction is known . in most other cases ,
the path integral can only be evaluated `` formally '' , e.g. , by means of a formal power series in the physical variables @xcite .
thus , giving rise to perturbation theoretical approaches to qft . in quantum mechanics ,
the path integral can be defined as a continuum limit of the discretized system @xcite .
wilson @xcite further developed this idea for qfts since the path integral of a quantum mechanical system in discretized space - time is always well - defined .
thus , wilson defined the path integral fully non - perturbatively on a space - time grid , going beyond perturbation theory . using a transformation to euclidean space - time ( wick rotation ) , this discretized path integral has been successfully applied to study physical systems computationally @xcite and phase space path integrals mathematically @xcite .
non - discretized path integrals in euclidean space - time can be studied within the framework of classical pseudo - differential operators and their traces and determinants @xcite .
these traces and determinants are defined using @xmath2-regularization which also gives rise to the kontsevich - vishik trace @xcite . incidentally ,
hawing @xcite had proposed studying the path integral with a curved space - time background in a @xmath2-regularized setting long before the kontsevich - vishik trace was developed . in his approach , hawking used a power series expansion of the action and regularized the quadratic term using the spectral @xmath2-function .
furthermore , gibbons , hawking , and perry @xcite studied convergence properties of the @xmath2-regularized one - loop approximation of the path integral .
thus , the paper aims to shed light on the following questions .
* is it possible to @xmath2-regularize the partition function and expectation values of observables in minkowski space - time ? * are the regularized partition functions and expectation values of observables independent of the choices made in the construction of the @xmath2-function ? * does the regularization contain the known special cases of well - defined path integrals ( e.g. , wick rotated or space - time discretized ) ?
* is the construction physically `` meaningful '' ? given the recent developments on @xmath2-functions of fourier integral operators @xcite , we aim to consider a non - perturbative approach to @xmath2-regularization of path integrals .
in particular , we want this new approach to contain all the special cases above , i.e. , discretizations , wick rotations , and spectral @xmath2-functions . in order to achieve this goal
, the generalized kontsevich - vishik trace is the ideal candidate .
in fact , it can be shown that the kontsevich - vishik trace is the only trace on classical pseudo - differential operators ( which we obtain from wick rotating ) that restricts to the canonical trace ( which we obtain after discretization ) .
hence , we will alter feynman s definition of the path integral to incorporate the generalized kontsevich - vishik trace .
this ensures that the new definition of the path integral coincides with feynman s definition whenever feynman s path integral is well - defined .
this paper is organized as follows .
appendix [ sec : kv ] contains a non - technical overview of fourier integral operator @xmath2-functions and the generalized kontsevich - vishik trace . in section [ sec : feynman ] , we will use the results of appendix [ sec : kv ] to show that path integrals are regularizable in this sense and obtain an altered definition of the path integral , partition function , and expectation values of observables .
finally , we will consider a number of physical models in sections [ sec : harmonic - oscillator]-[sec : spontaneous ] .
first , we will give examples applying the proposed regularization to very simple models such as the harmonic oscillator , the topological oscillator , and free fermions in order to show how the regularization works in practice .
in a second step , we will apply the method to non - trivial cases such as the massive schwinger model and a @xmath1 theory .
in particular , we will show analytic computations as well as python implementations using symbolic arithmetic .
the author would like to express his gratitude to dr .
karl jansen and dr .
erhard seiler for inspiring comments and conversations which helped to develop the work presented in this article .
considering the schrdinger equation @xmath3 we obtain @xmath4 following feynman s approach @xcite ( cf . `` some remarks on mathematical rigor '' in chapter 4 - 3 @xcite and @xcite , as well ) , we will change the physics slightly and introduce a ( flat ) time torus of length @xmath5 , i.e. , @xmath6 then , we can formally introduce the partition function @xmath7 and the expectation of an observable @xmath8 @xmath9 the actual expectation value @xmath10 in the quantum theory can be recovered using the thermal limit @xmath11 unfortunately , the @xmath12 are not of trace - class , in general .
hence , they need to be regularized . based on ray and singer s work on spectral @xmath2-functions @xcite , hawking @xcite proposed @xmath2-function regularization . since most algebras of fourier integral operators do not have the holomorphic functional calculus
, we can not expect to be able to define a spectral @xmath2 function for @xmath12 , but we may consider ( generalized ) @xmath2-functions @xcite .
thus , the regularized traces are given by the generalized kontsevich - vishik trace on pseudo - differential trace - class operators @xcite .
the generalized kontsevich - vishik trace is , thus , a natural choice of regularization . ]
@xcite .
let @xmath13 be a pseudo - differential operator with symbol the symbol of an operator @xmath14 if and only if @xmath14 is an integral operator with kernel @xmath15 which ( locally ) satisfies @xmath16}e^{i\langle x - y,\xi\rangle_{\ell_2(n)}}\sigma(x , y,\xi)d\xi$ ] for some @xmath17 . ]
@xmath18 where @xmath19 , the @xmath20 are continuous , and @xmath21 has an asymptotic expansion @xmath22 .
then , @xmath23 has the symbol @xmath24 where @xmath25 in particular , @xmath26 in combination with the power series identity ( for @xmath17 ) @xmath27 where @xmath28 and @xmath29 , shows that @xmath30 has an asymptotic expansion @xmath31 .
regarding @xmath12 , we note @xmath32}\phi(x)\l(\int_{\rn[n]}e^{i\langle x,\xi\rangle}\sigma_{\omega^*}(x,\xi)\fp u(\xi)d\xi\r)^*dx\\ = & \int_{\rn[n]}\int_{\rn[n]}\phi(x)e^{-i\langle x,\xi\rangle}\sigma_{\omega^*}(x,\xi)^*\fp u(\xi)^*d\xi dx\end{aligned}\ ] ] which implies @xmath33}e^{-i\langle x,\xi\rangle}\sigma_{\omega^*}(x,\xi)^*\phi(x)dx\end{aligned}\ ] ] and , thus , @xmath34}e^{ih_2(x,\xi)}e^{ih_1(x,\xi)}b(x,\xi)e^{i\langle x,\xi\rangle}\fp\omega\phi(\xi)d\xi\\ = & \int_{\rn[n]}\int_{\rn[n]}e^{ih_2(x,\xi)}e^{ih_1(x,\xi)}b(x,\xi)e^{i\langle x,\xi\rangle}e^{-i\langle y,\xi\rangle}\sigma_{\omega^*}(y,\xi)^*\phi(y)dy d\xi.\end{aligned}\ ] ] hence , ( utilizing @xmath35 for any pseudo - differential operator @xmath14 ) @xmath36 in other words , both trace integrals in @xmath37 have kernels of the form @xmath38}e^{i\langle x - y,\xi\rangle}e^{ih_2(x,\xi)}e^{ih_1(x,\xi)}a(x,\xi)d\xi\end{aligned}\ ] ] with poly-@xmath39-homogeneous @xmath40 provided @xmath41 and @xmath42 are poly-@xmath39-homogeneous . in order to @xmath2-regularize these integrals , they need to be gauged
. one of the simplest and most convenient gauges is the @xmath43-gauge ( or mellin - gauge ; cf .
definition 2.10 in @xcite ) @xmath38}e^{i\langle x - y,\xi\rangle}e^{ih_2(x,\xi)}e^{ih_1(x,\xi)}a(x,\xi)\norm\xi_{\ell_2(n)}^z d\xi.\end{aligned}\ ] ] [ theorem - regularization ] let @xmath44 be a compact , orientable , @xmath45-dimensional riemannian @xmath46-manifold without boundary , @xmath42 polyhomogeneous , and @xmath47}e^{-i\sigma_h(x,\xi)}\sigma_\omega(x,\xi)\ d\xi\ d\vol_x(x ) \end{aligned}\ ] ] with @xmath48}:\ \sigma_h(x , r\eta)=h_2(x,\eta)r^2+h_1(x,\eta)r+h_0(x , r\eta ) \end{aligned}\ ] ] where @xmath49})$ ] , @xmath41 polyhomogeneous , and 1 .
either @xmath50 and @xmath51 is a non - degenerate phase function 2 . or @xmath52}:\ \abs{h_2(x,\eta)}>0 $ ] .
then , @xmath53 can be regularized using the generalized kontsevich - vishik trace .
since we can absorb @xmath54 into the amplitude @xmath55 , we obtain without loss of generality @xmath56}}\int_{\rn_{>0}}e^{-i\l(h_2(x,\eta)r^2+h_1(x,\eta)r\r)}\sigma(x , r,\eta)\ dr\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x ) . \end{aligned}\ ] ] `` ( i ) '' if @xmath50 and @xmath51 is a non - degenerate phase function , then @xmath53 is a fourier integral operator trace already and can , thus , be regularized using the generalized kontsevich - vishik trace . ``
( ii ) '' let @xmath52}:\ \abs{h_2(x,\eta)}>0 $ ] and @xmath57}\r\}. \end{aligned}\ ] ] then , we can split @xmath53 into two parts @xmath58}}\int_{(0,r)}e^{-i\l(h_2(x,\eta)r^2+h_1(x,\eta)r\r)}\sigma(x , r,\eta)\ dr\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x ) \end{aligned}\ ] ] and @xmath59}}\int_{\rn_{\ge r}}e^{-i\l(h_2(x,\eta)r^2+h_1(x,\eta)r\r)}\sigma(x , r,\eta)\ dr\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x ) .
\end{aligned}\ ] ] considering @xmath60 , we observe that @xmath61}}\int_{(0,r)}e^{-i\l(h_2(x,\eta)r^2+h_1(x,\eta)r\r)}\sigma(x , r,\eta)\ dr\ d\vol_{\d b_{\rn[n]}}(\eta ) \end{aligned}\ ] ] is the fourier transform of a compactly supported distribution and continuous in @xmath62 .
hence , @xmath60 is well - defined ( by schwartz s paley - wiener theorem ) . in other words , it suffices to show that we can find a fourier integral operator whose trace coincides with @xmath63 distributionally .
since @xmath64 , we obtain @xmath65 and , absorbing @xmath66 into @xmath55 and setting @xmath67 , @xmath68}}\int_{\rn_{\ge r}}e^{-ih_2(x,\eta)\l(r+\frac{h_1(x,\eta)}{2h_2(x,\eta)}\r)^2}\sigma(x , r,\eta)\ dr\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x)\\ = & \int_x\int_{\d b_{\rn[n]}}\int_{\rn_{\ge r+\frac{h_1(x,\eta)}{2h_2(x,\eta)}}}e^{-ih_2(x,\eta)s^2}\sigma_1\l(x , s,\eta\r)\ ds\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x)\\ = & \int_x\int_{\d b_{\rn[n]}}\int_{\rn_{\ge\l(r+\frac{h_1(x,\eta)}{2h_2(x,\eta)}\r)^2}}e^{-ih_2(x,\eta)t}\sigma_1\l(x,\sqrt t,\eta\r)\ \frac{dt}{2\sqrt t}\ d\vol_{\d b_{\rn[n]}}(\eta)\ d\vol_x(x ) .
\end{aligned}\ ] ] this shows that there exists a fourier integral operator with polyhomogeneous amplitude whose trace coincides with @xmath63 ( up to another fourier transform of a compactly supported distribution ) .
thus , we can write @xmath69 with meromorphic functions @xmath70 and @xmath71 . as we are interested in @xmath72 , there are a few cases to consider
. if both @xmath73 and @xmath74 are regular and at most one of them vanishes , then gauge independence ( cf .
lemma 2.6 in @xcite ) implies that @xmath75 is independent of the choice of gauge in @xmath45 and @xmath76 ( though it may diverge if @xmath74 vanishes ) .
if one of the limits diverges and the other is finite , then @xmath75 is either zero or divergent .
thus , the only interesting cases are if both tend to zero or diverge . in the @xmath77 case
the result depends on the order of the pole
. if the pole order of @xmath45 and @xmath76 are different , then the limits are trivial . if they are the same , then the limit is the quotient of the leading order residues , which again is gauge independent ( cf .
lemma 2.5 in @xcite ) .
hence , gauge dependence can only appear if @xmath45 or @xmath76 have vanishing leading laurent coefficient or we have the @xmath78 case .
in these cases gauge dependence is , in fact , to be expected .
furthermore , the free schwinger model ( section [ sec : free - schwinger ] ) is a @xmath78 case , i.e. , the choice of gauge is physically important . since gauging
the denominator is essentially changing physics by replacing the solution operator @xmath79 with some other operator @xmath80 , it seems sensible to apply this idea to the entire system .
in other words , we are considering the family of `` evolution operators '' @xmath80 .
then , we obtain the following new definition of our quantum theory .
[ def : reg - path - int ] let @xmath13 be the hamiltonian , @xmath8 an observable , and @xmath80 a gauged family of operators with @xmath81 then , we define the expectation value @xmath82 of @xmath8 as @xmath83 where @xmath84 denotes the meromorphic extension of a function @xmath85 .
in order to see how the proposed @xmath2-regularization works , let us consider the traditional entry level model ; the harmonic oscillator with hamiltonian @xmath86 where @xmath87 thus , @xmath88 imply @xmath89 note that @xmath90 is a polynomial of order @xmath91 in @xmath62 .
thus , we may treat @xmath62 the same way we treat @xmath92 and gauge with respect to @xmath62 , as well .
otherwise , we would have to compactify the @xmath62-domain and consider the limit @xmath62-domain@xmath93 .
gauging in @xmath62 and @xmath92 yields the regularized ground state energy @xmath94 theorem 8.7 in @xcite shows that these are trace integrals of hilbert - schmidt operators and regular .
although it is possible ( and tedious ) to compute this limit by hand , it is preferable to have a computer do the work ( especially once the model is not analytically solvable anymore ) .
implementing this limit in python2.7 is fairly straightforward ( using the fact that integration over @xmath95 is equivalent to taking the fourier transform and evaluating at zero ) .
.... import sympy as smp z1,z2,t = smp.symbols("z1,z2,t " ) x , xi = smp.symbols("x,xi",real=true ) m , hbar , omega = smp.symbols("m,hbar,omega",positive=true ) a = smp.sqrt(m*omega/(2*hbar ) ) * ( x + smp.i*hbar*xi/(m*omega ) ) a_dag = smp.sqrt(m*omega/(2*hbar ) ) * ( x - smp.i*hbar*xi/(m*omega ) ) h = hbar * omega * ( a_dag * a + smp.sympify(1)/2 ) exph = smp.exp(-smp.i * t * h / hbar )
g = smp.abs(xi)**z1 * smp.abs(x)**z2 num = smp.fourier_transform(h * exph * g , xi,0 ) num = smp.fourier_transform(num,x,0 ) den = smp.fourier_transform(exph * g , xi,0 ) den = smp.fourier_transform(den,x,0 ) l = smp.limit(num/den,z1,0 ) l = smp.limit(l,z2,0 ) print " < h > = " + str(smp.limit(l , t , smp.oo ) ) ....
this program correctly outputs the ground state energy @xmath96 .
similarly , we can consider the @xmath97-dimensional harmonic oscillator whose hamiltonian is given by @xmath98 choosing the gauge @xmath99 we obtain the ground state energy @xmath100 .
the topological oscillator ( a.k.a . quantum rotor ) models a particle of mass @xmath101 moving on a circle with radius @xmath102 .
thus , choosing the angle @xmath103 as the free coordinate of the position @xmath104 , we obtain the lagrangian @xmath105 with the moment of inertia @xmath106 .
the momentum is , then , given by @xmath107 and the hamiltonian @xmath108 hence , @xmath109 a characteristic value is the topological charge @xmath110 which counts the number of revolutions the rotor performs in the time - torus and an interesting observable is the topological susceptibility @xmath111 which is directly connected to the energy gap @xmath112 between the ground state and the first excited state @xmath113 again , we can implement this directly in python2.7 .... import sympy as smp z , t = smp.symbols("z,t " ) j , xi = smp.symbols("j,xi",real=true ) h = xi**2 / ( 2 * j ) q = t*xi/(2 * smp.pi * j ) g = smp.abs(xi)**z num = smp.fourier_transform(smp.exp(-smp.i * t * h ) * g * q**2,xi,0 ) den = smp.fourier_transform(smp.exp(-smp.i * t * h ) * g , xi,0 ) chi_top = smp.limit(smp.limit(num/(-smp.i * t * den),z,0),t , smp.oo ) energy_gap = 2 * smp.pi**2 * chi_top print " chi_top = " + str(chi_top ) print " energy gap = " + str(energy_gap ) .... and obtain the correct results @xmath114 and @xmath115 .
let us now consider the free massive schwinger model @xcite whose hamiltonian , in the zero - momentum frame using natural units @xmath116 , is given by ( cf . ,
e.g. , equation ( 2.2 ) in @xcite ) @xmath117 using the @xmath43-gauge and a cut - off function @xmath118 ( that is , to introduce a space - torus in order to compactify the spatial domain ) , the ground state energy is , then , given by @xmath119 theorem 8.7 in @xcite shows that these are trace integrals of hilbert - schmidt operators and regular .
implementing this limit in python2.7 is straightforward again ( the limit @xmath120 can be ignored since @xmath121 is independent of @xmath44 ) . ....
import sympy as smp z = smp.symbols("z
" ) x , xi = smp.symbols("x,xi",real=true ) m , t = smp.symbols("m,t",positive=true ) chi = smp.function("chi")(x ) h = smp.matrix([[m,xi],[xi,m ] ] ) eith = smp.exp(smp.i*t*h ) gauge = smp.abs(xi)**z num = smp.fourier_transform((h*eith*gauge).trace()*chi,xi,0 ) den = smp.fourier_transform((eith*gauge).trace()*chi,xi,0 ) num = smp.integrate(num,x ) den = smp.integrate(den,x ) print " < h_m > = " + str(smp.limit(smp.limit(num / den , z,0),t , smp.oo ) ) ....
this program outputs `
< h_m > = m ` . in other words , we have just correctly computed @xmath122 for the free massive schwinger model .
let us now step up to @xmath123 space - time dimensions and consider a free relativistic fermion of mass @xmath124 . then , using the pauli matrices @xmath125 , we obtain the hamiltonian ( einstein summation over spatial indices ) @xmath126 which yields @xmath127 and @xmath128 thus , @xmath129}\l(4m\cos\l(t\norm\xi_{\ell_2(3)}\r)-4i\norm\xi_{\ell_2(3)}\sin\l(t\norm\xi_{\ell_2(3)}\r)\r)\norm\xi_{\ell_2(3)}^zd\xi}{\int_{\rn[3]}4\cos\l(t\norm\xi_{\ell_2(3)}\r)\norm\xi_{\ell_2(3)}^zd\xi}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{-i\int_{\rn[3]}\norm\xi_{\ell_2(3)}^{z+1}\l(e^{it\norm\xi_{\ell_2(3)}}-e^{-it\norm\xi_{\ell_2(3)}}\r)d\xi}{\int_{\rn[3]}\norm\xi_{\ell_2(3)}^{z}\l(e^{it\norm\xi_{\ell_2(3)}}+e^{-it\norm\xi_{\ell_2(3)}}\r)d\xi}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{-i\vol\l(\d b_{\rn[3]}\r)\int_{\rn_{>0}}r^{z+3}\l(e^{itr}-e^{-itr}\r)dr}{\vol\l(\d b_{\rn[3]}\r)\int_{\rn_{>0}}r^{z+2}\l(e^{itr}+e^{-itr}\r)dr}\tag{$*$}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{\l(-e^{-i\frac{\pi ( z+3)}{2}}-e^{-3i\frac{\pi ( z+3)}{2}}\r)\gamma(z+4)t^{-z-4}}{i\l(-e^{-i\frac{\pi ( z+2)}{2}}+e^{-3i\frac{\pi ( z+2)}{2}}\r)\gamma(z+3)t^{-z-3}}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{\l(-e^{-i\frac{\pi ( z+3)}{2}}-e^{-i\frac{\pi ( z+3)}{2}}e^{-i\pi ( z+3)}\r)\gamma(z+4)t^{-z-4}}{i\l(-e^{-i\frac{\pi ( z+2)}{2}}+e^{-i\frac{\pi ( z+2)}{2}}e^{-i\pi ( z+2)}\r)\gamma(z+3)t^{-z-3}}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{e^{-i\frac{\pi ( z+3)}{2}}\l(-1-e^{-i\pi ( z+2)}e^{-i\pi}\r)(z+3)\gamma(z+3)t^{-z-4}}{ie^{-i\frac{\pi ( z+2)}{2}}\l(e^{-i\pi ( z+2)}-1\r)\gamma(z+3)t^{-z-3}}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{e^{-i\frac{\pi ( z+2)}{2}}e^{-i\frac{\pi}{2}}(z+3)}{ie^{-i\frac{\pi ( z+2)}{2}}t}\\ = & m+\lim_{t\to\infty}\lim_{z\to0}\frac{-z-3}{t}\\ = & m.\end{aligned}\ ] ] in other words , we have correctly computed @xmath130 again .
the calculation above highlights a number of properties which will be even more important in the case of the massive schwinger model ( section [ sec : schwinger - gauge - boson - mass ] ) . in the first summand ,
the `` observable '' @xmath124 depends on none of the variables which leads to many cancellations .
this will be paramount for the massive schwinger model since the part of the model that is not analytically solvable will vanish in one such cancellation .
the other important property can be seen in the latter summand .
since the argument of the laplace transform is homogeneous and the volume @xmath5 of the time - torus enters through the evaluation of the laplace transform , we obtain that the limit @xmath131 depends primarily on the asymptotic expansion of the observable .
thus , knowing the asymptotic behavior of the observable enables us to decide whether or not a term will vanish in the limit @xmath131 .
we should note that ( @xmath132 ) can be implemented just like the implementations above since it is the laplace transform @xmath133 which holds for @xmath134 and @xmath135 , and through analytic extension for @xmath136 and @xmath137 .
in particular , @xmath138 and @xmath139 furthermore , using stationary phase approximation ( cf .
chapter 8 in @xcite ) in the setting of theorem [ theorem - regularization ] , we can see that the regularization is given in terms of these laplace transforms only .
hence , the actual difficulty in computing @xmath140 are the integrals over @xmath141}$ ] and possibly computing the limits @xmath142 and @xmath131 . using these laplace transforms and the fact that the integrals over @xmath141}$ ] yield @xmath143}$ ] in this case , we obtain the following implementation for the ground state energy of a free relativistic fermion in @xmath45 spatial dimensions . ....
import sympy as smp z , t = smp.symbols("z,t " ) m , voldb , r = smp.symbols("m,voldb,r",positive=true ) n ,
k = smp.symbols("n,k",positive=true,integer=true ) f = voldb*k*r**(z+n-1)/(2*smp.pi)**n g = voldb*k*smp.i*m*r**(z+n)/(2*smp.pi)**n num = smp.laplace_transform(m*f,r,-smp.i*t)[0 ] num + = smp.laplace_transform(m*f,r,smp.i*t)[0 ] num -= smp.laplace_transform(g,r,-smp.i*t)[0 ] num -= smp.laplace_transform(g,r,smp.i*t)[0 ] den = smp.laplace_transform(f,r,-smp.i*t)[0 ] den + = smp.laplace_transform(f,r,smp.i*t)[0 ] lim = smp.limit(smp.limit(num/den,z,0),t,smp.oo ) print " < h_m > = " + str(smp.simplify(lim ) ) ....
at this point , we will return to the massive schwinger model but add an abelian vector gauge field .
thus , the model becomes fully interacting with a non - trivial dynamics leading to the confinement of the charges and , hence , bound states .
hence , applying the @xmath2-regularization of the generalized kontsevich - vishik trace constitutes a first highly non - trivial example of the proposed method . here , we will only provide a demonstration for the calculation of the gauge boson mass .
further observables could be computed in a similar way if required . here
, we have the fermionic hamiltonian in the temporal gauge @xmath144 as well as the self - interaction hamiltonian @xmath145 of the gauge field , where @xmath146 , @xmath147 ( @xmath148 is the temporal gauge ) , and @xmath149 . at this point
, it is important to address the space - time dependence of the gauge fields . due to the nature of the observable in question ( the gauge boson mass ) , it is convenient to choose the family @xmath150 as canonical coordinates . though the space torus @xmath44 needs to be formally introduced
( recall that the @xmath2-regularization needs a compact manifold ) , we will suppress it in the following since the limit @xmath151 is trivial .
more importantly , this setting implies that the time - dependence of the gauge fields is implicit while the space - dependence is still explicit .
thus , @xmath152 in @xcite , schwinger himself supplied us with the green s function of the abelian vector gauge field . from it , we can read off the observable @xmath8 for the squared mass of the gauge boson ; @xmath153 hence , the gauge boson mass @xmath154 is given by ( suppressing gauges for @xmath92 , @xmath62 , and @xmath14 ) @xmath155 this calculation highlights the cancellations and asymptotic properties we observed in section [ sec : free - relativistic - fermions ] again . here , the integrals with respect to @xmath14 are very difficult and not analytically solvable . however ,
due to the structure of the observable , these integrals cancel out to a factor of @xmath156 .
more importantly , even if they did not cancel , we would know that the second term had to vanish in the limit @xmath131 since we know the asymptotics of the observable in @xmath157 .
more precisely , having @xmath158 in the numerator and @xmath159 in the denominator ( and a phase function in terms of @xmath160 ) implies that the quotient is proportional to @xmath161 and , as such , vanishes for @xmath131 .
since spontaneous symmetry breaking is essential to the higgs mechanism , we will have a quick look at it here , as well . in the simplest relativistic case , we have scalar fields @xmath162 and the langrangian contains a potential term @xmath163 .
then , we are looking for constant fields @xmath164 which locally minimize @xmath165 .
these @xmath164 are the vacuum expectation values of the @xmath166 .
furthermore , the matrix @xmath167 is symmetric and its eigenvalues give the squared masses of the fields . in particular ,
if @xmath168 , we obtain the vacuum expectation values from @xmath169 and @xmath170 where @xmath171 is the mass of the field . however , in general , we will not be able to simply read off @xmath163 .
instead , we will consider the partition function as a function of @xmath103 and obtain an effective potential @xmath172 through the identity @xmath173 , i.e. , @xmath174 where we used the fact that we are looking for constant @xmath103 and introduced a space - time torus of volume @xmath175 .
consider the @xmath1 model whose hamiltonian is given by @xmath176 in this case , the minima are given by @xmath177 and the field mass is @xmath178 .
using the @xmath2-regularized partition function , we obtain @xmath179 which we may implement directly . ....
import sympy as smp z = smp.symbols("z " ) phi , p = smp.symbols("phi,p",real=true ) tx , mu , l = smp.symbols("tx,mu,l",positive=true ) h = p**2/2 - mu**2/2*phi**2 + l/24*phi**4 exph = smp.exp(-smp.i*tx*h ) gauge = smp.abs(p)**z z = smp.fourier_transform(exph * gauge , p,0).doit()/(2*smp.pi ) v = smp.ln(z)/(-smp.i*tx ) dv = smp.simplify(smp.diff(v,phi ) ) ddv = smp.diff(dv,phi ) # take limit z->0 dv = smp.limit(dv,z,0 ) ddv = smp.limit(ddv,z,0 ) extrema = smp.solve(dv,phi ) # check extrema for minima , in physical limit tx->\infty ddv = smp.limit(ddv,tx,smp.oo ) for i in range(len(extrema ) ) : extrema[i ] = smp.limit(extrema[i],tx,smp.oo ) minima = [ ] for phi0 in extrema : if ddv.subs(phi,phi0)>=0 : minima.append(phi0 ) print minima masses = [ ] for phi0 in minima : m = smp.sqrt(ddv.subs(phi,phi0 ) ) if m not in masses : masses.append(m ) print masses .... note that gauge independence of @xmath180 means this computation may only fail if @xmath181 has a pole in @xmath182 or @xmath183 since we can not take the logarithm in that case ( @xmath184 can be treated choosing an appropriate branch cut of @xmath185 ) .
here , neither of these cases occurs , i.e. , the results are independent of the chosen gauge and we correctly obtain the minima @xmath186 and the field mass @xmath178 . the situation will be more complex if we are not using the fact that the @xmath103 are ( spatially ) constant since the @xmath187 term will not vanish . in that case , it might be more appropriate to write the term as @xmath188 which is of the same form as the @xmath189 again , but the best choice will most likely depend on the specifics of the problem and observable in consideration
. it should be noted , however , that space - discretization ( i.e. , replacing @xmath103 by a vector @xmath190 $ ] ) can be very viable and the resulting path integral will be @xmath2-regularizable again .
we proposed a new definition of the path integral based on feynman s formulation ( definition [ def : reg - path - int ] ) . by construction
the proposed definition restricts to feynman s definition whenever it is well - defined ; e.g. , using wick rotations , lattice discretization , or trace - class observables .
we obtained the new definition by replacing feynman s path integral with its corresponding version as constructed using the generalized kontsevich - vishik trace .
more precisely , we replaced the time - evolution semi - group @xmath191 by a holomorphic family of operator families @xmath192 satisfying @xmath193 and for which feynman s path integral is well - defined if @xmath194 is sufficiently small .
if @xmath195 is chosen appropriately , we showed that the path integrals defined for @xmath196 can be extended meromorphically to @xmath197 ( cf .
theorem [ theorem - regularization ] and appendix [ sec : kv ] ) and defined the regularized path integral as the value of the meromorphic extension at @xmath198 ( provided it exists ; definition [ def : reg - path - int ] ) . furthermore , we considered a number of fundamental models , including the non - trivial cases of the massive schwinger model and a @xmath1 theory , as evidence for the validity of the proposed definition and provided implementations using symbolic arithmetic .
it is particularly important to note that the underlying regularization is a priori known which reduces the computational effort of evaluating these regularized path integrals to the evaluation of spherical integrals ( and possibly the limits @xmath199 and @xmath200 ) .
hence , continuum computations without wick rotations are possible with the new definition .
in particular , we can answer the questions we set out in the beginning .
* is it possible to @xmath2-regularize the partition function and expectation values of observables in minkowski space - time ?
yes , provided that the hamiltonian and observable satisfy certain homogeneity and positivity or non - degeneracy assumptions in the leading order terms ( cf .
theorem [ theorem - regularization ] ) .
* are the regularized partition functions and expectation values of observables independent of the choices made in the construction of the @xmath2-function ? almost always , the answer to this question is yes .
both are well - defined and choice independent if there are no critical degrees of homogeneity ( which depends only on the space - time dimension ) .
thus , the quotient is almost always well - defined and independent of the choices made ( though it might be infinite ) .
gauge dependence can only appear if the laurent coefficient of lowest possible order vanish ( which depends on the degrees of homogeneity and logarithmic degrees at critical degree of homogeneity ) .
in particular , if there are no critical degrees of homogeneity , then the partition function is gauge independent and the expectation value of the observable can only depend on the gauge if we are in the @xmath78 case .
case is gauge independent , i.e. , the @xmath78 case can not be removed through the choice of gauge . ] on the other hand , this case does appear in practice ; the ground state energy of the free relativistic fermion ( section [ sec : free - relativistic - fermions ] ) , for instance , is of this form . to overcome the problem that different choices of gauge in numerator and denominator can generate arbitrary results , we conjecture that choosing the same gauge should give a meaningful choice physically .
* does the regularization contain the known special cases of well - defined path integrals ( e.g. , wick rotated or space - time discretized ) ?
space - time discretization replaces the operators by matrices .
hence all traces are well - defined and the construction of the @xmath2-regularization coincides with the canonical trace on trace - class operators .
similarly , wick rotations yield pseudo - differential operators and there it is known that the @xmath2-regularization used in this context is the unique extension of the canonical trace .
* is the construction physically `` meaningful '' ?
this question can be interpreted in different ways .
on one hand , we may ask if the regularization can be interpreted physically . in this sense , choosing the same gauge for both numerator and denominator in the expectation value of observables is important .
this means that we replace the time - evolution of our system by a holomorphic family of time - evolutions . in other words ,
we consider a holomorphic family of physical systems and conjecture that the physical values of the system to be studied can be obtained by analytic continuation . on the other hand , we need to ask whether or not the regularized theory is physically correct . by construction , we know that the regularized theory coincides with the physical theory if we have trace - class operators to begin with . in case
the regularization is necessary , we have considered a number of physical models . in each of these models ,
the regularization not only recovered the known physical values but is also computable .
in fact , the dependence on the regularizing parameter is known explicitly and the remaining integrals are over compact manifolds .
in this appendix , we will give a non - technical overview of fourier integral operator @xmath2-functions and the ( generalized ) kontsevich - vishik trace .
for more detail , please refer to @xcite .
given a closed , compact , orientable , connected , finite dimensional riemannian manifold @xmath44 and a closed conic lagrangian submanifold @xmath201 of @xmath202 , we can consider the space @xmath203 of lagrangian distributions of order @xmath124 with microsupport in @xmath201 ( cf .
chapter 25 in @xcite ) .
integral operators with kernels in some @xmath203 are called fourier integral operators .
more precisely , we have the following definition .
let @xmath44 be a @xmath46 manifold , @xmath157 a ( complex ) vector bundle over @xmath44 , and @xmath204 a closed @xmath46 sub - manifold of @xmath44 .
then , the space @xmath205 of distribution sections of @xmath157 that are conormal to @xmath204 and of order less than or equal to @xmath124 is the set of all distributions @xmath206 such that @xmath207 for all @xmath208 and all first order differential operators @xmath209 between distribution sections of @xmath157 whose coefficients are @xmath46 tangential to y. here , @xmath210)$ ] denotes the usual besov space and , for @xmath211 $ ] open , we define @xmath212 as the set of distributions @xmath213 such that @xmath214)$ ] . this definition
can then be lifted to manifolds in the usual manner .
the definition of conormality can be extended to pseudo - differential operators from @xmath157 to @xmath157 with principal symbol vanishing on @xmath204 .
thus , it can be extended to lagrangian manifolds .
let @xmath44 be a @xmath46 manifold , @xmath157 a ( complex ) vector bundle over @xmath44 , and @xmath215 a closed , conic , @xmath46 , lagrangian sub - manifold .
then , the space @xmath216 of lagrangian distribution sections of @xmath157 of order less than or equal to @xmath124 is the set of all distributions @xmath206 such that @xmath207 for all @xmath208 and all properly supported first order pseudo - differential operators @xmath217 whose principal symbols vanish on @xmath201 .
it is common to denote @xmath201 in terms of a canonical relation @xmath218 ( cf . ,
e.g. , chapter 1 in @xcite ) which satisfies @xmath219 if @xmath220 is chosen to be a homogeneous canonical relation ( cf . ,
e.g. , chapter 1 in @xcite , theorem 2.4.1 in @xcite , and example 1 in @xcite ) , then the set of operators @xmath221 with kernels in @xmath222 forms an associative algebra .
furthermore , it can be shown that @xmath221 has a non - trivial intersection with the set of trace - class operators in @xmath223 ( cf . lemmata 1.12 and 1.13 in @xcite ) ; more precisely , if @xmath224 has kernel has kernel @xmath15 if and only if @xmath225 holds for all @xmath85 in the domain of @xmath14 . ] @xmath226 with @xmath124 sufficiently small , then @xmath14 is of trace - class , @xmath15 continuous , and @xmath227 in many applications ( like the feynman path integral ; cf .
section [ sec : feynman ] ) we would like to extend this trace to operators that are not of trace - class .
such an extension of the trace can be obtained using @xmath2-regularization .
let @xmath228 with kernel @xmath229 .
then , we consider a holomorphic family @xmath230 with kernels @xmath231 such that @xmath232 and @xmath233 , and define @xmath234 to be the maximal meromorphic extension of @xmath235 which is well - defined for @xmath194 sufficiently small , that is , @xmath196 . an important class of holomorphic families of fourier integral operators are gauged fourier integral operators with @xmath39-polyhomogeneous amplitudes .
these gauged fourier integral operators have kernels of the form @xmath236}e^{i\theta(x , y,\xi)}a(z)(x , y,\xi)d\xi\end{aligned}\ ] ] where @xmath237 is a phase function @xmath238 and @xmath239 where @xmath240)$ ] and @xmath241 holds with a number of additional properties making everything well - defined ( cf .
chapter 2 in @xcite ) .
we call @xmath242 the degree of homogeneity of @xmath243 and @xmath244 the logarithmic order
. if all @xmath244 vanish , then we call the amplitude polyhomogeneous .
fourier integral operator @xmath2-functions with polyhomogeneous amplitudes were shown to exist as meromorphic functions on @xmath197 and had their residues studied by guillemin @xcite .
using guillemin s approach and introducing the notion of gauged poly-@xmath39-homogeneous distributions , the author @xcite was able to compute the laurent expansion of @xmath234 for fourier integral operators with @xmath39-polyhomogeneous amplitudes , as well .
in particular , it can be shown that @xmath234 only has isolated poles of finite order .
the poles are located at @xmath245 and the maximal pole order is @xmath246 . while the residues of @xmath234 yield important traces ( cf .
, e.g. , @xcite ) , we are interested in the values @xmath247 provided none of the degrees of homogeneity satisfies @xmath248 .
then , @xmath234 is holomorphic in a neighborhood of zero and @xmath247 depends only on @xmath249 .
in fact , @xmath250 defines a trace provided the amplitude of @xmath251 has no critical degree of homogeneity @xmath248 ( cf .
chapter 7 in @xcite ) .
this trace is called the generalized kontsevich - vishik trace ( the original kontsevich - vishik trace is the special case of @xmath251 being a classical pseudo - differential operator ) and given by @xmath252}(0,1)}e^{i\theta(x , x,\xi)}a(0)(x , x,\xi)\ d\xi\ d\vol_x(x)\\ & + \int_{\rn_{\ge1}\times\d b_{\rn[n]}}\int_xe^{i\theta(x , x,\xi)}a_0(0)(x , x,\xi)\ d\vol_x(x)\ d\vol_{\rn_{\ge1}\times\d b_{\rn[n]}}(\xi)\\ & + \sum_{\iota\in i}\frac{(-1)^{l_\iota+1}l_\iota!\int_{x\times\d b_{\rn[n]}}e^{i\theta(x , x,\xi)}\tilde a_\iota(0)(x , x,\xi)\ d\vol_{x\times\d b_{\rn[n]}}(x , x,\xi)}{(n+d_\iota)^{l_\iota+1}}. \end{aligned}\end{aligned}\ ] ] by construction , the generalized kontsevich - vishik trace coincides with @xmath253 on trace - class operators .
furthermore , it was shown that the kontsevich - vishik trace is the only trace on the algebra of classical pseudo - differential operators that restricts to @xmath253 in @xmath223 @xcite .
these properties make the generalized kontsevich - vishik trace a prime candidate for path integral regularization as such a path integral regularization is consistent with respect to discretization ( turning operators into matrices and , thus , trace - class ) and wick rotations ( turning the path integral into pseudo - differential operator traces ) .
|
a fully regulated definition of feynman s path integral is presented here . the proposed re - formulation of the path integral coincides with the familiar formulation whenever the path integral is well - defined . in particular
, it is consistent with respect to lattice formulations and wick rotations , i.e. , it can be used in euclidean and minkowskian space - time .
the path integral regularization is introduced through the generalized kontsevich - vishik trace , that is , the extension of the classical trace to fourier integral operators .
physically , we are replacing the time - evolution semi - group by a holomorphic family of operator families such that the corresponding path integrals are well - defined in some half space of @xmath0 .
the regularized path integral is , thus , defined through analytic continuation .
this regularization can be performed by means of stationary phase approximation or computed analytically depending only on the hamiltonian and the observable ( i.e. , known a priori ) . in either case , the computational effort to evaluate path integrals or expectations of observables reduces to the evaluation of integrals over spheres .
furthermore , computations can be performed directly in the continuum and applications ( analytic computations and their implementations ) to a number of models including the non - trivial cases of the massive schwinger model and a @xmath1 theory .
|
in the electromagnetism , oppositely charged particles under the uniform electric field separate from each other to cancel out the background flux .
this is the well - known dielectric effect , and analogous phenomena can be observed in the type ii superstring theories and the m - theory@xcite .
let us illustrate the dielectric phenomenon in the type iia superstring theory with the case where a spherical d2-brane is located in the background of coincident ns5-branes@xcite .
ns5-branes carry the magnetic charge of the ns - ns 2-form potential , and its 3-from flux vertically penetrates three dimensional sphere @xmath0 which encloses ns5-branes in the transverse space .
the d2-brane is assumed to be spherical in the @xmath0 . in case
there is no magnetic flux on the d2-brane , it collapses because of its tension . on the other hand ,
if the magnetic flux exists on the d2-brane , the background ns - ns 3-form flux supports it against the collapse .
notice that the number of the magnetic flux on the d2-brane is less than that of ns5-branes .
this is the dielectric effect in the type iia superstring theory .
interestingly the dielectric phenomena in the type ii superstring theories admit the dual descriptions .
namely similar dielectric effects are observed from the dynamics of d0-branes , which are identical to the magnetic flux on the d2-brane@xcite .
the dynamical variables of @xmath1 coincident d0-branes are @xmath2 gauge field and nine adjoint scalars , and the latter represent the position of d0-branes when they are simultaneously diagonalized . in the background of ns5-branes , d0-branes form a fuzzy ( noncommutative ) geometry in the @xmath0 because of the support from the ns - ns 3-form flux@xcite .
in fact d0-branes compose a fuzzy sphere in the @xmath0 , which matches the d2-brane picture when the number of d0-branes or magnetic flux is sufficiently large .
this is called the myers effect@xcite . since the essence of the dielectric effect and the myers effect is the existence of the background flux , at first sight it seems impossible to observe these effects in purely gravitational backgrounds .
note , however , that the ns5-brane is related via t - duality to a kaluza - klein monopole which forms the purely gravitational geometry@xcite .
therefore it is natural to expect gravitational versions of the dielectric effect and the myers effect in the background of kaluza - klein monopoles . in this paper
, we examine the spherical d2-brane in the background of kaluza - klein monopoles in the type iia superstring theory .
the geometry of the 10 dimensional kaluza - klein monopoles is described by @xmath3 , where the @xmath4 is equivalent to @xmath5 but the @xmath6 vanishes at the origin .
the d2-brane is spherically wrapped around the origin of the @xmath7 and has momentum along the @xmath6 .
then we will see that the spherical d2-brane with some finite radius becomes the classical solution of the d2-brane action .
since the background is purely gravitational , this is exactly the case where the dielectric effect is induced via the gravity
. the above system is easily reinterpreted in the framework of the 11 dimensional m - theory .
the 10 dimensional kaluza - klein monopole is just lifted to a 11 dimensional kaluza - klein monopole , @xmath8 , and the d2-brane is done to a m2-brane .
of course the m2-brane moving along the @xmath6 in the @xmath4 is spherically stable around the origin of the @xmath7 .
this is the gravitational dielectric effect in the m - theory .
now we change the roles of the two @xmath6s in the above discussion , which is often called the 9 - 11 flip in the m - theory .
after the dimensional reduction to the 10 dimensional spacetime , we obtain spherical d2-d0 bound state in the background of d6-branes in the type iia superstring theory@xcite . as discussed before , this system will admit the dual description of d0-branes , that is , the myers effect .
we will see that the fuzzy sphere composed by d0-branes really becomes a classical solution of equations of motion in the background of d6-branes .
this is the gravitational myers effect in the type iia superstring theory .
the organization of this paper is as follows . in section 2
, we discuss the gravitational dielectric effect in the background of kaluza - klein monopoles and d6-branes in the type iia superstring theory . in section 3
, we observe the gravitational myers effect in the background of d6-branes .
the stability of the spherical configuration of the d2-d0 bound state in the background of d6-branes is examined in section 4 .
section 5 is devoted to conclusions and discussions .
in this section we study the dielectric effects which are induced not by the background flux but by the gravitational force . at first sight
it seems difficult to observe the gravitational dielectric phenomena in the absence of the background flux .
however , as mentioned in the introduction , the spherical d2-d0 bound state is stable in the background of ns5-branes , and by the t - duality transformation this background is related to the kaluza - klein monopoles which are purely gravitational objects . therefore it is interesting to study whether the gravitational dielectric effect can be observed in the background of kaluza - klein monopoles . in the following we investigate a single d2-brane with @xmath1 units of magnetic flux on its world - volume in the background of @xmath9 coincident kaluza - klein monopoles .
the kaluza - klein monopoles are purely gravitational objects and the metric for the @xmath9 coincident kaluza - klein monopoles is given by@xcite @xmath10 where @xmath11 and @xmath12 is the radius of the @xmath13 direction at @xmath14 . when the above solution is reduced to @xmath15 space , it describes @xmath9 coincident magnetic monopoles which are localized at @xmath16 .
dynamical degrees of freedom on the d2-brane are @xmath17 gauge field and scalar fields , which represent the position of the d2-brane . and the effective action for the d2-brane is described by the born - infeld action and the chern - simons action : @xmath18_{ab } + \lambda f_{ab } \big ) } + \mu_2 \int \big ( c^{(3 ) } + c^{(1)}\wedge \lambda f \big ) .
\label{eq : actd2}\end{aligned}\ ] ] here @xmath19 and @xmath20 are tension and charge of a d2-brane , respectively .
these are exspressed as @xmath21 by using the string length @xmath22 and the string coupling @xmath23 .
the notation @xmath24 $ ] represents the pullback on the world - volume of the d2-brane and @xmath25 are the world - volume coordinates .
@xmath26 is the @xmath17 gauge field strength on the d2-brane and @xmath27 is defined as @xmath28 .
we assume that the d2-brane is spherical and moving along the @xmath13 direction in the background of kaluza - klein monopoles .
so the world - volume coordinates @xmath29 are identified with @xmath30 , and @xmath31 . now the d2-brane is located in the background of kaluza - klein monopoles which are purely gravitational objects , the relevant part of the d2-brane action is only the born - infeld part .
it is straightforward to evaluate born - infeld part of the d2-brane action . by substituting these assumptions for the action ,
it becomes as @xmath32 in order to examine the stability of this system , it is convenient to switch to the hamiltonian formalism .
the canonical momentum conjugate to the variable @xmath13 is given by @xmath33 note that the @xmath13 direction is compactified with the circle , so the total momentum should be conserved as @xmath34 where @xmath1 is an integer .
then we obtain the hamiltonian of the form @xmath35 the interpretation of this hamiltonian ( [ eq : e2 ] ) is as follows .
the first square root part is the correction to the volume factor of the sphere @xmath36 , because of the nontrivial winding of @xmath37 along the compactified direction .
terms in the second square root are the sum of the mass squared of the spherical d2-brane and the momentum energy squared along compactified direction .
notice that the radii of sphere and compactified direction are affected by gravitational factor @xmath38 , respectively .
these modifications are expected from the form of the metric ( [ eq : kkmono ] ) .
let us examine the stability of the spherical d2-brane in the background of kaluza - klein monopoles . in the region of @xmath39 , the factor @xmath38
is approximated as @xmath40 , and the spherical d2-brane shrinks to a point because of its tension .
on the other hand , near the region of @xmath41 , the spherical d2-brane tends to expand because of the repulsive force between the momentum along the @xmath13 direction and the kaluza - klein monopoles .
thus we obtain the gravitational dielectric effect . in order to see this effect in detail , we focus on the neighborhood of the kaluza - klein monopoles . in this region , @xmath42 ,
the factor @xmath38 is approximated as @xmath43 , and the conservations of the momentum ( [ eq : conmom ] ) and the energy ( [ eq : e1 ] ) are written like @xmath44 where @xmath45 .
we also introduced @xmath46 and @xmath47 which are the complete elliptic integrals of the first kind and second kind , respectively . by combining these equations
, we see that the energy is given as a function of the radius @xmath48 .
the energy takes the minimum with finite redius @xmath49 .
( see fig .
[ fig : d6 ] . ) note that the analysis here is reliable if the parameters satisfy the condition of @xmath50 , and this holds actually when @xmath9 and @xmath1 are chosen as sufficiently large . in the previous subsection
, the gravitational dielectric phenomenon of the spherical d2-brane with the momentum in the background of kaluza - klein monopoles is observed in the type iia superstring theory .
this system is easily lifted to the 11 dimensional m - theory , and we obtain a dielectric effect of a spherical m2-brane with momentum along the compactified direction @xmath13 in the background of 11 dimensional kaluza - klein monopoles .
again we reduce the 11 dimensional spacetime to the 10 dimensional one , by identifying the @xmath13 direction with the reduced 10th one .
then the 11 dimensional kaluza - klein monopoles become d6-branes in the type iia superstring theory@xcite . and
the m2-brane with the momentum along the reduced @xmath13 direction turns into a bound state of a d2-brane and d0-branes .
consequently the dielectric effect of the spherical d2-d0 bound state in the background of d6-branes will be obtained in the type iia superstring theory . in this subsection
we will see this phenomenon from the viewpoint of the effective theory on the d2-brane , where d0-branes are identical to the magnetic flux on it .
the d6-brane is extending to @xmath51 dimensional spacetime and carries the magnetic charge of the r - r 1-form potential @xmath52 .
the classical solution of @xmath9 coincident d6-branes is given by@xcite @xmath53 where @xmath54 and @xmath55 is the r - r 2-form flux originating from the d6-branes .
as in the previous subsection , the d2-brane is embedded into the background of d6-branes spherically , so the world - volume coordinates on the d2-brane @xmath29 are identified with @xmath30 .
on the d2-brane we introduce electric field @xmath56 and magnetic field @xmath57 .
total magnetic flux on the spherical d2-brane is given by @xmath58 .
now it is straightforward to evaluate the d2-brane action , and the lagrangian density is given by @xmath59 the momentum conjugate to @xmath60 is defined as @xmath61 and we obtain the hamiltonian of the form @xmath62 terms in the second square root are mass squared of the spherical d2-brane and @xmath1 d0-branes , respectively .
both terms are modified by the gravitational factor @xmath38 , as compared with the flat case .
( see fig .
[ fig : d6 ] . )
( 300,0 ) ( 66,132)@xmath63 ( 234,12)@xmath48 ( 100,4)@xmath64 ( 57,36)@xmath65 let us examine the stability of the spherical d2-brane in the background of d6-branes .
far from the d6-branes the factor @xmath38 is approximated as @xmath40 , and the spherical d2-brane shrinks to reduce the total energy . on the other hand , near the d6-branes the factor @xmath38 is approximated as @xmath66 , and the d2-brane tends to expand to reduce the energy .
therefore the spherical d2-brane with some finite radius becomes the classical solution .
now we concentrate on the neighborhood of the d6-branes , where the factor @xmath38 is described as @xmath66 . in order to estimate the energy of the d2-brane , we employ the equations of motion for gauge fields , which say variables @xmath67 and @xmath68 only depend on the radius .
so we choose solutions as @xmath69 and @xmath70 , and the magnetic field is expressed as @xmath71 then total magnetic flux and the energy are given by @xmath72 the energy becomes a function of the radius , and it takes minimum @xmath65 with finite radius @xmath64 .
( see fig .
[ fig : d6 ] . )
this is the gravitational dielectric effect of the spherical d2-d0 bound state in the background of d6-branes .
the analyses in this paragraph are reliable if the parameters satisfy the condition of @xmath73 , and this is satisfied by choosing @xmath9 and @xmath1 as sufficiently large .
it is worth mentioning that above relations are obtained by setting @xmath74 , @xmath75 and @xmath76 in the previous subsection .
d0-branes become dynamical through the excitations of open strings ending on them . and the massless excitations correspond to the @xmath2 gauge field @xmath77 and nine adjoint scalars @xmath78 , where @xmath1 is the number of the d0-branes@xcite . when three of nine adjoint scalars do not commute simultaneously , d0-branes form a fuzzy surface . in particular , d0-branes compose a fuzzy sphere when three adjoint scalars satisfy the @xmath79 lie algebra . in the background of flat spacetime ,
the fuzzy sphere is unstable and shrinks to a point because of its tension . on the other hand ,
as discussed in the introduction , the fuzzy sphere becomes stable in the background of ns5-branes , because the tension balances with the dielectric force induced by the background ns - ns 3-form flux .
furthermore the fuzzy spherical configuration of d0-branes is also stable in the background of constant r - r 4-form flux . in both cases
the background flux plays an important role to lead to the myers effect . in this section
we consider another type of myers effect which is induced via the gravitational force . in the previous section we observed the gravitational dielectric phenomena in the background of kaluza - klein monopoles and d6-branes .
since the dielectric effect admits the dual description , i.e. , the myers effect , the gravitational myers effect will also be observed in these backgrounds .
the metric and r - r 1-form potential for @xmath9 coincident d6-branes are written as before . here
it is convenient to employ not polar coordinates but rectangular coordinates like @xmath80 where @xmath54 , which label the tangent directions to d6-branes and @xmath81 , which describe normal ones .
the radius @xmath48 is given by @xmath82 , and the constant term in @xmath38 is neglected because we only consider the region near the d6-branes .
the bosonic part of the effective action for @xmath1 d0-branes is described by the nonabelian born - infeld action and chern - simons action @xcite .
as explained in the above , this action possesses @xmath2 gauge symmetry and consists of @xmath2 gauge field @xmath77 and nine adjoint scalars @xmath78 which appear from massless excitations of open strings ending on d0-branes .
the explicit form of the action for @xmath1 d0-branes is written as @xmath83_{00 } \det { q^m}_n } \big ) + \mu_0 \int dt \ , \text{str } \big ( c^{(1)}_i d_0 x^i \big),\end{aligned}\ ] ] where @xmath84 is the mass of a d0-brane .
note that str represents the symmetrized trace prescription , and partial derivatives in the pullbacks should be replaced with covariant derivatives to ensure the @xmath2 gauge symmetry .
the field @xmath85 and @xmath86 matrices @xmath87 and @xmath88 are defined as @xmath89 e_{ln } , \notag\end{aligned}\ ] ] where the indices @xmath90 label the target spacetime directions . @xmath91 and @xmath92 are the background metric and the ns - ns 2-form field respectively
. now we trace the similar calculations as in ref .
@xcite . setting @xmath93 , which are parallel to d6-branes ,
the lagrangian for d0-branes evaluated in the background of d6-branes becomes as @xmath94 ^ 2 - f^2 (
d_0 x^i)^2 + \frac{f^3}{\lambda^2}(\epsilon_{ijk } \{d_0 x^i , [ x^j , x^k]\})^2 } \\ & \qquad\quad - \frac{n}{4r } \frac{x^3}{(x^1)^2 + ( x^2)^2 } \epsilon_{ij9 } \{x^i , d_0 x^j\ } \bigg ) .
\notag\end{aligned}\ ] ] note that the radius @xmath48 should be understood in terms of @xmath95 as @xmath96 , and symmetrized trace prescription is replaced with trace by choosing appropriate orderings .
furthermore we assume that @xmath95 and @xmath77 take following forms @xmath97 these forms are useful when we are considering axial asymmetric configurations around @xmath98 .
after some calculations we obtain @xmath99 where we defined @xmath100 when @xmath1 is sufficiently large we can take the continuous limit , and the above lagrangian becomes @xmath101 where @xmath102 , @xmath103 and @xmath104 .
this coincides with the action ( [ eq : d2lag ] ) after setting @xmath105 and doing coordinate transformation from @xmath106 to @xmath107 .
then we trace the same arguments as before , and obtain the gravitational myers effect .
in the section [ sec : d6 ] , the dielectric phenomenon in the background of d6-branes is observed by assuming the spherical configuration of the d2-brane . here
we examine the stability of this configuration .
but it is difficult to analyze the stability by adding small fluctuations , since the energy is given by the elliptic integral .
so in order to make discussions simple , let us consider another distribution of the magnetic flux on the d2-brane like @xmath108 of course this configuration is not a classical solution , but the energy ( [ eq : d2ene ] ) of the d2-d0 bound state is simply given by @xmath109 the original solution can be obtained by adding fluctuations around the above configuration , and it will be possible to understand the stability of the original one .
now we reexamine the energy ( [ eq : fakeene ] ) by adding a fluctuation @xmath110 around @xmath111 , where @xmath112 gives the minimum of the energy .
other fluctuations on @xmath37 and @xmath113 are neglected for simplicity . by setting @xmath114 and @xmath115 , the pullback metric on the d2-brane
is given by @xmath116_{ab } = \begin{pmatrix } -f^{-\frac{1}{2 } } & 0 & 0 \\
0 & f^{\frac{1}{2 } } r^2 + f^{\frac{1}{2 } } \bar{r}^2 \epsilon'^2 & 0 \\ 0 & 0 & f^{\frac{1}{2 } } r^2\sin^2\theta \end{pmatrix } , \label{eq : mflu}\end{aligned}\ ] ] where @xmath117 label @xmath118 , and @xmath119 is an abbreviation for @xmath120 .
note that the function @xmath38 depends on the angular directions , but @xmath121 is just the constant .
furthermore we add fluctuation of the gauge field @xmath122 around the background of @xmath123 .
then the component of the gauge field strengths on the d2-brane is described by @xmath124 the fluctuation @xmath125 is defined as @xmath126 and should satisfy the constraint @xmath127 which follows from the flux quantization condition @xmath128 . by substituting the pullback metric ( [ eq : mflu ] ) and the gauge field strength ( [ eq : gflu ] ) into the arguments in sec .
[ sec : d6 ] , we obtain the energy for the d2-d0 bound state of the form , @xmath129^{\frac{1}{2 } } \notag \\ & = \frac{\bar{\mathcal{h}}}{\pi } \int d\theta \big [ \frac{1}{3}(1+\epsilon)^2 + \frac{2}{3}\frac{(1+\delta)^2}{1+\epsilon } + \frac{1}{3 } { \epsilon'}^2 \big]^{\frac{1}{2 } } , \label{eq : d2pot}\end{aligned}\ ] ] where @xmath130 .
the last line is derived by using the relations @xmath131 .
note that the energy is expanded as @xmath132 the linear term on @xmath133 represents the instability of the configuration ( [ eq : fake ] ) .
if @xmath133 is a fluctuation around the original configuration ( [ eq : d2mag ] ) , the linear term vanishes after imposing the magnetic flux conservation .
but the original configuration is unstable against @xmath134 fluctuation .
the final state of this d2-d0 bound state after the perturbation is conjectured as the case where @xmath1 d0-branes are radiated away from @xmath9 d6-branes .
this configuration has the same quantum numbers as those of the spherical d2-d0 bound state , and the potential energy at @xmath135 is zero is neglected . ] . during this process
the d2-brane will be deformed and decay through the tachyon condensation process .
the numerical result on this deformation is shown in the appendix a. to avoid the above instability , we just prepare more d6-branes around the original ones properly .
d0-branes will meet the potential barrier before they go away from the original d6-branes .
then the dielectric d2-d0 configuration , which will depend on the locations of the d6-brnaes , becomes stable .
in this paper we observed the gravitational dielectric effect and myers effect in the type iia superstring theory . in the background of kaluza - klein monopoles ,
the spherical d2-brane with the momentum along the compactified circle becomes the classical solution .
this phenomenon is occurred because the tension of the spherical d2-brane balances with the repulsive force between the kaluza - klein monopoles and the momentum along the circle . as the kaluza - klein monopoles are related to the d6-branes via t - s - t duality
, it is expected that the gravitational dielectric effect occurs in the background of d6-branes .
in fact we find that the spherical d2-brane with magnetic flux on it becomes the classical solution with the finite radius .
furthermore the similar effect is obtained from the viewpoint of d0-branes , that is , d0-branes compose the fuzzy sphere in the background of d6-branes .
we called this phenomenon the gravitational myers effect . in the continuous limit
, the gravitational myers effect coincides with the gravitational dielectric effect .
the essence of both effects is the fact that the tension of the spherical d2-brane balances with the repulsive force between the d6-branes and the d0-branes .
it is worth emphasizing that the gravitational dielectric effect and myers effect are insensitive to the charges carried by kaluza - klein monopoles or d - branes .
consequently , for instance , a bound state of an anti - d2-brane and d0-branes spherically expands in the background of d6-branes or anti - d6-branes .
this is in contrast with the dielectric phenomena induced by the background flux . by examining the fluctuations around the spherical d2-d0 bound state
we see that this configuration is unstable and the final state is conjectured by employing the numerical analyses . to avoid this instability
, we need more d6-branes around the original ones
. then the dielectric d2-d0 bound state will be stable aginst collapse .
let us consider generalization of the gravitational dielectric effect to other d@xmath136-branes background .
in the case of d6-branes background , the gravitational dielectric effect is induced via the factor @xmath137 in front of @xmath138 in the potential energy ( [ eq : pot ] ) . in cases of a spherical d@xmath139-brane with magnetic flux on it in the background of d@xmath136-branes ,
this factor is modified as @xmath140 . here
the d@xmath139-brane is completely transverse to the d@xmath136-branes .
the first part of the exponent comes from the contribution of the dilaton field and the second one does from that of @xmath141 .
the third term appears from @xmath142 , where @xmath143 is the number of world - volume coordinates which are normal to both the background d@xmath136-branes and the sphere . from this
we see that the dielectric effect occurs if @xmath144 .
these are all obtained by t - duality transformations of the spherical d2-d0 bound state in the background of d6-branes . in this paper we focused on the single d2-d0 bound state wrapped around d6-branes .
but this system can be generalized to the system where multiple d2-d0 bound states are wrapped around each bunch of d6-branes . here
the bunches of d6-branes are separated each other . in case
some of them are sufficiently close , the spherical d2-d0 bound states which are wrapped around them will make a transition to a single d2-d0 bound state .
the configuration of the d2-brane or the fuzzy d0-branes will depend on the locations of the bunches of d6-branes , and it is interesting to analyze the dynamics from the viewpoint of tachyon condensation@xcite . after the tachyon condensation , the d0-branes will compose a general fuzzy surface and representations for the matrices @xmath95 will be described as in ref . @xcite .
i would like to thank members of komaba particle theory group , especially koji hashimoto and hidehiko shimada for useful discussions .
the work was supported in part by the grant - in - aid for jsps fellows .
numerical calculations about the potential energy ( [ eq : d2pot ] ) are shown in this appendix .
fluctuations @xmath134 and @xmath133 are expanded by legendre polynomials as @xmath145 the expansions are limited to @xmath146 , which is purely numerical reason .
the deformation of the d2-d0 bound state in the background of d6-branes is parametrized by @xmath147 , and @xmath148 for all @xmath149 .
we consider the following deformation : @xmath150 the figures of @xmath151 and @xmath152 at @xmath153 are drawn in fig .
[ fig : d01 ] . the potential energy along this deformation
is plotted in fig .
[ fig : d03 ] . from this
we see that the d2-brane is deformed and d0-branes are distributed around the southern hemisphere . after the tachyon condensation process , the d2-brane will vanish and d0-branes will be scattered to the infinity .
|
in this paper we study the gravitational dielectric phenomena of a d2-brane in the background of kaluza - klein monopoles and d6-branes . in both cases
the spherical d2-brane with nonzero radius becomes classical solution of the d2-brane action .
we also investigate the gravitational myers effect in the background of d6-branes .
this phenomenon occurs since the tension of the d2-brane balances with the repulsive force between d0-branes and d6-branes .
ou - het 462 + hep - th/0401026 + january 2004 + yoshifumi hyakutake +
|
a molecular communication network @xcite consists of multiple transmitters and receivers in a fluid medium . the transmitters
encode the messages by types , concentration or emission frequency of signalling molecules .
the signalling molecules diffuse freely in the medium .
when these signalling molecules reach the receiver , they trigger chemical reactions in the receiver to allow their presence to be detected .
there are a number of reasons why the study of molecular communication networks , both natural and synthetic , are important .
first , molecular communication is a fundamental ingredient of life on earth .
multi - cellular organisms make extensive use of molecular communication to regulate body functions @xcite .
second , synthetic molecular communication networks can be used as sensor networks for cancer detection and treatment @xcite , as well as many other applications @xcite . an important part of the research on molecular communication networks is to understand their performance . work done in this area includes the modelling of molecular communication networks @xcite , capacity analysis @xcite , characterising the noise of transmitters , channels and receivers @xcite , and receiver design @xcite .
this paper will focus on the receiver part of the network .
when signalling molecules arrive at the receiver , they react , via one or more chemical reactions , to produce a number of output molecules . the number ( or counts ) of output molecules in the receiver over time is regarded as the _ output signal _ of the receiver .
a way to detect the presence of signalling molecules at the receiver is to test whether the number of output molecules at the receiver is more than a threshold .
however , the output signal is _ noisy _ due to the stochastic nature of diffusion and chemical reactions .
an important part of performance analysis is to characterise the probability distribution , mean and variance of the receiver output signals .
the aim of this paper is to characterise two different types of receiver reaction mechanisms , namely _
reversible conversion _ and _ linear catalytic _ , and compare their performance .
the main contributions of this paper are : * we characterise the output signals of receivers that use reversible conversion in terms of probability distribution , mean , variance and frequency response . *
we derive the mean , variance and frequency response of the output signal for receivers that use the linear catalytic reaction mechanism . *
we investigate the properties of the output signals of these receivers . for reversible conversion , we find that the variance of the output signal is always smaller than the mean output signal
. however , the linear catalytic reaction magnifies the noise to create an output signal whose variance is greater than the mean .
we also find that , for both reaction mechanisms , an increase in mean output also leads to an increase in variance .
* we show that the diffusibility of the receiver membrane ( a parameter which does not appear to have been studied in the literature ) can be used to affect the properties of the receiver output signal .
the rest of the paper is organised as follows .
we present our modelling framework in section [ sec : model ] .
this is followed by performance analysis of the two different types of receiver reaction mechanisms in sections [ sec : sol ] and [ sec : sol2 ] .
section [ sec : num ] presents numerical results .
related work is discussed in section [ sec : related ] . finally , section [ sec : con ] concludes the paper .
this section presents a model for molecular communication networks .
the model assumes the medium ( or space ) is discretised into voxels while time is continuous .
features included in the model include geometric constraints , directional emissions of signalling molecules and others .
in particular , we will consider two different types of receivers based on different types of chemical reactions .
the proposed model will be analysed in sections [ sec : sol ] and [ sec : sol2 ] .
we will present the _
basic _ model in section [ sec : model_basic ] and various _ extensions _ will be discussed in section [ sec : model_ext ] .
a molecular communication network consists of multiple transmitters and receivers . in this paper , we limit ourselves to one transmitter and one receiver .
the model and the solution method can be readily generalised to the multiple transmitter and receiver scenario .
we assume the transmitter uses one type of signalling molecules @xmath0 .
generalisation to multiple types of non - interacting signalling molecules is straightforward .
we model the transmission medium as a three dimensional ( 3-d ) space with dimensions @xmath1 , where @xmath2 , @xmath3 and @xmath4 are integral multiples of length @xmath5 .
that is , there exist positive integers @xmath6 , @xmath7 and @xmath8 such that @xmath9 and @xmath10 , @xmath11 .
the 3-d volume can be partitioned into @xmath12 cubic _ voxels _ of volume @xmath13 .
figure [ fig : model ] shows an arrangement with @xmath14 and @xmath15 .
we refer to a voxel by a triple @xmath16 where @xmath17 , @xmath18 and @xmath19 are integers or by a single index @xmath20 $ ] where @xmath21 .
the indices for the voxels are shown in figure [ fig : model ] .
diffusion is modelled by molecules moving from one voxel to another .
diffusion from a voxel to a non - neighbouring voxel is always not allowed . the diffusion from a voxel to a neighbouring voxel may or may not be allowed .
this can be used to specify different modelling constraints .
we use a few examples in figure [ fig : model ] to explain this : 1 . for voxel 11 , the diffusion of signalling molecules @xmath0 is allowed in both directions , i.e. in and out of the voxel .
the two - way arrows are used to indicate this .
for voxel 1 , signalling molecules can only diffuse out of the voxel but not into it .
this is indicated by the one - way arrows .
3 . with the exception of voxel 4 , diffusion to the outside of the medium is not allowed . for voxel 4 , diffusion to the outside of the medium
is allowed for one surface as indicated by the one - way arrow .
our model can be used to capture standard boundary conditions such as reflecting and absorbing boundaries .
no molecules are allowed to flow in and out of voxel 8 ( the hatched voxel ) .
this can be used to model shadowing effects or regions that are impermeable to the signalling molecules .
the _ basic _ model assumes that the medium is homogeneous with the diffusion coefficient for @xmath0 in the medium is @xmath22 .
define @xmath23 .
if a molecule is allowed to diffuse from a voxel to another , it takes place at a rate of @xmath24 , i.e. within an infinitesimal time @xmath25 , the probability that a molecule diffuses to a neighbouring voxel is @xmath26 .
we assume the transmitter and the receiver each occupies a distinct voxel .
however , it is straightforward to generalise to the case where a transmitter or a receiver occupies multiple voxels .
the transmitter ( resp . receiver ) is assumed to be located at the @xmath27-th ( @xmath28-th ) voxel , where @xmath27 ( @xmath28 ) is an integer to indicate the position of the voxel . in figure
[ fig : model ] , we assume voxel 1 ( dark grey ) contains the transmitter , and voxel 11 ( light grey ) contains a receiver . hence @xmath29 and @xmath30 in this example .
we model a transmitter by a sequence which specifies the number of signalling molecules emitted by the transmitter at a certain time .
we assume that , at time @xmath31 ( where @xmath32 ) , the transmitter emits @xmath33 signalling molecules .
this means that : @xmath34 signalling molecules are added to the voxel containing the transmitter at time @xmath35 , @xmath36 signalling molecules are added at time @xmath37 etc .
note that in figure [ fig : model ] , we only allow signalling molecules to leave the transmitter ( voxel 1 ) but our model can also deal with one - way or two - way flow on any surface of a voxel .
when a signalling molecule @xmath0 arrives at a receiver , it may react , via one or more chemical reactions , to produce one or more _
output molecules_. we assume that these reactions can only take place within a receiver voxel , and the output molecules do not leave the receiver voxel .
in this paper , we consider the following two types of reactions at the receivers . 1 . the signalling molecules @xmath0 are converted to complexes @xmath38 , reversibly , at the receiver via the reaction : @xmath39{\text{k}_{+ } } c \label{cr : conversion } \end{aligned}\ ] ] where @xmath40 and @xmath41 are , respectively , the macroscopic rate constant for the forward and reverse reactions .
we will refer to this reaction as _ reversible conversion _
the output molecule is the complex @xmath38 .
the signalling molecule @xmath0 acts as a catalyst to produce the output molecule @xmath42 .
the molecule @xmath42 decays at a certain rate
. we will refer to this type of receiver as _ linear catalytic _ or cat .
the chemical formulas for cat are : @xmath43 l + e \label{cr : cat1 } \\ e & ->[\text{f}_{- } ] \emptyset } \label{cr : cat2 } \end{aligned}\ ] ] where @xmath44 and @xmath45 are macroscopic rate constants , and @xmath46 denotes a chemical species that we are not interested in . for both rc and cat ,
the output molecules are formed at a rate of @xmath47 or @xmath48 times the number of signalling molecules in the voxel . if @xmath49 and @xmath50 , then the relation between the mean number of signalling molecules in the voxel and mean number of output molecules is described by exactly the same first order ordinary differential equations for both receiver types .
however , as we shall later , the behaviour of the mean and variance of the number of output molecules , for the same transmitter emission pattern , are very different for the two types of receivers .
note that both rc and cat consists of _ linear reactions _ , which means the reaction rate is a linear function of the concentration of the reactants .
reactions and are examples of _ monomolecular _ reactions because there is exactly one chemical species on each side of the chemical formula .
it is appropriate to make a few remarks here .
first , rc and cat can be used to approximate more complex reactions .
for example , we show in @xcite that rc can approximate more complicated reactions such as michaelis - menten .
cat reactions have been studied in @xcite as a detection mechanism .
second , a receiver may consist of a chain of reactions but we know from data processing inequality @xcite that the capacity of the network is upper bounded by the output of the first reaction in the chain , therefore we limit our consideration to one reaction .
we define the state of the molecular communication network as the number of signalling molecules in each voxel , the number of complexes in the receiver and the total number of molecules that have left or degraded in the network .
let @xmath51 be the number of signalling molecules in voxel with index @xmath52 at time @xmath53 , @xmath54 be the number of signalling molecules that have either left the medium or degraded by time @xmath53 and @xmath55 be the number of output molecules in the receiver at time @xmath53 .
also let @xmath56 be the total number of voxels in the medium .
the state @xmath57 of the network is defined as : @xmath58^t \label{eqn : state}\end{aligned}\ ] ] where the superscript @xmath59 is used to denote matrix transpose .
note that we also use @xmath27 and its subscripted form @xmath60 to indicate the index of the transmitter voxel .
although the same symbol @xmath27 is used in different situations , its meaning can easily be deduced from its context .
we use the network depicted in figure [ fig : model ] as an example .
the unfilled and solid circles represent , respectively , signalling and output molecules .
the state vector @xmath61 has 5 non - zero elements : @xmath62 , @xmath63 , @xmath64 , @xmath65 and @xmath66 .
all other elements of @xmath61 are zero .
note that , for @xmath67 , the @xmath68-th element of @xmath61 is the number of signalling molecules in the @xmath68-th voxel .
a convenient way to represent the emission of molecules by the transmitter is to use the following notation : assume that at time @xmath53 , @xmath69 molecules are emitted by the transmitter , which is located at the @xmath27-th voxel , we can write : @xmath70 where @xmath71 and @xmath72 are just before and after time @xmath53 , and @xmath73 is the @xmath27-th standard basis vector .
for this reason , we also include unreachable voxels , such as voxel 6 in figure [ fig : model ] , in @xmath61 to simplify notation .
the state @xmath61 is in fact a vector - valued random process .
the state evolves because of diffusion and reactions .
we show in @xcite and @xcite that the probability distribution of @xmath61 evolves according to a reaction - diffusion master equation with exogenous input ( rdmex ) , which is a generalisation of the reaction - diffusion master equation ( rdme ) taking into consideration the emission of signalling molecules by the transmitters . in @xcite , we show how the mean of @xmath61 can be calculated for an infinite transmission medium . in @xcite , we show how the covariance of @xmath61 can be approximately calculated for an infinite medium .
we will derive the probability distribution of @xmath61 for rc receivers in section [ sec : sol ] .
for cat receivers , we derive the mean and variance of the number of output molecules in section [ sec : sol2 ] . [
sec : ext ] in this section , we present a number of extensions to the basic model presented earlier .
the solution method in sections [ sec : sol ] and [ sec : sol2 ] can still be applied to these extensions . 1 .
we assume that each voxel has the shape of a cube .
it is possible to use other shapes for the voxel too .
however , the rate of diffusion has to be appropriately adjusted according to the geometry , see @xcite on how this can be done .
we assume a homogenous medium . an heterogeneous medium with different diffusion constants at different surfaces of the voxels can easily be incorporated
. we will consider in later section how heterogeneity can be used to improve the communication performance .
we assume that no flows exist in the medium .
flows can be easy incorporated by adjusting the rate of flow to neighbouring voxels .
for example , let us assume there is a flow in the positive @xmath17 directions of speed @xmath74 , then the rate of flow of molecules from a voxel to its neighbour in the positive @xmath17 direction should be changed to @xmath75 .
4 . degradation of signalling molecules in the voxels can also be handled , in fact in the same way as a molecule leaving the medium .
multiple transmitters and multiple receivers can be used .
6 . instead of assuming
the number of molecules emitted by the transmitter at a time instance is deterministic , a stochastic emission model can also be handled .
in this section , we solve for the probability distribution of the state @xmath61 of a network consisting of one transmitter and one rc receiver .
the solution uses the method introduced in @xcite for solving a system of monomolecular reactions .
the reactions in the rc receiver are certainly monomolecular .
the diffusion of signalling molecules between the voxels can be viewed as monomolecular reactions if we identify the signalling molecules in each voxel as a distinct chemical species .
therefore , we can view our network as a system of monomolecular reactions .
note that the work in @xcite does _ not _ consider transmitters .
our contribution is to extend the method in @xcite so that it can used to analyse the performance of molecular communication networks .
an important result for a system of monomolecular reactions is that each molecule evolves * independently * @xcite .
this means we can analyse one molecule at a time and sum " the results up .
we begin with some background .
this section is divided into two parts .
we first consider markov chains with one molecule and then we present some background on multinomial distributions .
since each molecule in a rc receiver network evolves independently , we present a _ continuous - time _ markov chain representation of such networks with only _
one _ molecule .
if there is only one molecule in the network , the molecule can be a signalling molecule in a voxel , in an absorbing state or as a complex in the receiver .
we will refer to this markov chain as the single - molecule markov chain associated with the network or 1-mc for short .
the easiest way to describe 1-mc is via an example . for the network depicted in figure [ fig : model ] ,
the corresponding 1-mc is depicted in figure [ fig : mc ] . in the figure ,
each circle is a state of the markov chain .
each state is labelled by @xmath76 ( @xmath77 ) , @xmath78 or @xmath28 .
the molecule is in state @xmath76 if it is a signalling molecule in voxel @xmath68 . if the molecule in state @xmath78 , the molecule has either left the medium or degraded , and is in an absorbing state . if the molecule is in @xmath28 , then it is a complex in the receiver .
all the state transitions given in solid lines are due to diffusion and occur at a rate of @xmath24 for the basic model which assumes homogeneous medium .
these rates are not indicated in the figure to avoid cluttering .
the conversion rates ( @xmath79 and @xmath41 ) between a signalling molecule and a complex in a receiver are shown in the diagram .
we will order the states of the 1-mc in the same order as that in the network state vector @xmath61 in equation .
let @xmath80 , @xmath81 and @xmath82 be the probability that the molecule is , respectively , a signalling molecule in @xmath68-th voxel , in absorbing state and in the receiver as a complex . define the state probability vector @xmath83 as : @xmath84^t \label{eqn : qvec}\end{aligned}\ ] ] the probability vector @xmath83 of 1-mc evolves according to : @xmath85 where @xmath86 is the infinitesimal generator of the 1-mc .
the matrix @xmath86 can be transcribed from the state transition diagram such as figure [ fig : mc ] . note that , when there is only one molecule in the network , there is a one - to - one correspondence between the states in the 1-mc and the network state @xmath61 .
in particular , if the molecule is in the @xmath87-th state of the 1-mc at time @xmath53 , then @xmath88 and @xmath89 .
we will make extensive use of multinomial distributions .
let @xmath90 $ ] be a vector of non - negative integers and @xmath91 $ ] be a real vector with @xmath92 and @xmath93 , the multinomial distribution @xmath94 is defined as : @xmath95 where @xmath96 and @xmath97 are respectively the 1-norm of @xmath98 and @xmath17 .
note that in the above expression , we assume that @xmath99 .
a standard result that we will make use of is that the marginalization of a multinomial distribution is again a multinomial distribution .
let @xmath100 $ ] and @xmath101 $ ] , then the marginalization of @xmath94 over the first @xmath69 random variables is : @xmath102 finally , note that if @xmath103 is a scalar , the distribution is in fact binomial .
we have the following theorem on the probability distribution of the state @xmath61 .
consider a network with a transmitter and a rc receiver .
the initial network state @xmath61 is zero .
the transmitter , located at voxel @xmath27 , emits @xmath33 molecules at time @xmath31 where @xmath104 and @xmath105 .
the probability that the state vector @xmath106 is : @xmath107 where @xmath108 denotes convolution . for @xmath109
, @xmath110 is the state probability of the associated 1-mc and is the solution to : @xmath111 where @xmath86 is the generator of the 1-mc ; if @xmath112 , then @xmath113 . @xmath114
[ thm : rc ] * proof : * we will establish the result step by step .
let us assume that the transmitter emits only once , and this takes place at time @xmath35 and the number of molecules emitted is @xmath34 .
since the network state is zero initially , it means the network state remains zero until time @xmath35 . at this time , @xmath115 which means @xmath34 molecules in the transmitter voxel .
since the network can be viewed as a collection of monomolecular reactions , we can now apply ( * ? ? ?
* proposition 1 ) to show that the network state @xmath106 is @xmath116 where @xmath117 is the solution to .
note that the time shift is needed in the multinomial distribution because the signalling molecules are emitted at time @xmath35 .
let us now assume that the transmitter emits only twice : @xmath118 and @xmath119 molecules at , respectively , time @xmath35 and @xmath37 .
since the network consists of a collection of monomolecular reactions , the molecules in the network evolves independently .
we can consider the @xmath34 molecules emitted at time @xmath120 separately from those @xmath36 molecules emitted at time @xmath121 .
let @xmath122 and @xmath123 denote the network state of the molecules emitted at time @xmath120 and @xmath121 .
following on from the result in the last paragraph , the probability distributions of these two state vectors are respectively @xmath124 and @xmath125 .
the network state @xmath61 is the sum of two independent random variables @xmath126 .
therefore , the probability distribution for @xmath61 is the convolution of the distributions of @xmath122 and @xmath123 . in other words , the probability
that @xmath106 is : @xmath127 equation is obtained by repeating the above argument .
we remark that : ( 1 ) theorem [ thm : rc ] can also be proved by substituting the result into the rdmex ( * ? ? ?
* equation 4 ) ( * ? ? ?
* equation 9 ) , but such proof would be tedious and un - intuitive ; ( 2 ) the above results can be extended to non - zero initial condition , multiple transmitters , multiple receivers and non - deterministic emission patterns of transmitters .
let @xmath128 denote the number of output molecules in the rc receiver at time @xmath53 .
note that @xmath128 is a random process .
the probability distribution of @xmath128 can be obtained by marginalising over the first @xmath129 elements of @xmath61 given in theorem [ thm : rc ] .
since marginalization of a multinomial distribution is again multinomial , and marginalization and convolution are interchangeable , we have the probability that @xmath130 is @xmath131 where @xmath132 is the element in @xmath83 related to the output molecule , see . from the discussion in section
[ sec : rc : bg:1mc ] , we know that @xmath132 is the probability that a molecule circulating in the network taking the form of an output molecule in the receiver .
note that the individual distributions in equation are in fact binomial . by using probability generating functions
, it can be shown that the mean number of output molecules @xmath133 in the rc receiver is equal to @xmath134 where @xmath135 and @xmath136 is the dirac delta function .
furthermore , we show in the appendix [ app : rc ] that the laplace transform of @xmath133 is : @xmath137 here , @xmath138 is the laplace transform of the transmitter emission pattern @xmath139 and @xmath140 is the transfer function of the rc reaction .
the laplace transforms @xmath141 and @xmath142 are related to a network with its receiver reaction mechanisms _
removed_. this is the same as assuming @xmath79 and @xmath41 are both zero .
this means that a signalling molecule in the network can only diffuse from voxel to voxel , leave the medium or decay , but _ can not _ be converted to a complex .
to explain what @xmath141 is , we consider a network with receiver reaction mechanism removed .
we assume that there is only one signalling molecule in the network at time zero and it is located in the _ transmitter _
let @xmath143 denote the probability that this signalling molecule is in the receiver voxel @xmath28 at time @xmath53 .
the @xmath141 is the laplace transform of @xmath143 .
in fact , we can view @xmath141 as the transfer function from the transmitter voxel to the receiver voxel , hence the subscript @xmath144 in the notation .
furthermore , in the absence of receiver reaction mechanisms , the mean number of signalling molecules in the receiver voxel is @xmath145 . for @xmath142
, we again consider a network with receiver reaction mechanism removed .
we assume that there is only one signalling molecule in the network at time zero and it is located in the _ receiver _ voxel .
let @xmath146 denote the probability that this signalling molecule is in the receiver voxel @xmath28 at time @xmath53 .
@xmath142 is the laplace transform of @xmath146 .
equation can be interpreted as a system with feedback , see figure [ fig : block_rc ] .
note that an expression similar to has also been derived in our earlier work @xcite assuming an homogeneous infinite transmission medium .
we now present an example to demonstrate some of the properties of rc receivers . consider a molecular communication network with 2 voxels in figure [ fig:2voxel ] .
voxel 1 is the transmitter and voxel 2 is the receiver .
the environment is heterogeneous where @xmath147 and @xmath148 are , respectively , the transition rates of signalling molecules leaving voxels 1 and 2 . here
, we assume the receiver is of rc type .
the corresponding 1-mc is shown in figure [ fig:2voxel_1mc ] .
let @xmath149^t$ ] be the state probability vector .
the generator @xmath86 of the 1-mc is : @xmath150 \label{eqn:2voxel_rc_h}\end{aligned}\ ] ] if the transmitter emits @xmath151 molecules intially and none afterwards . by solving @xmath152 , the mean number of complexes @xmath153 ( note
: we drop @xmath154 for brevity ) at steady state is : @xmath155 the above result can also be derived by applying the final value theorem to and noting that as @xmath156 , we have @xmath157 and @xmath158 .
equation shows that one can increase the mean number of complexes by having a smaller @xmath148 , i.e. making it harder for the signalling molecules to leave the receiver voxel .
we will show in section [ sec : num ] that such heterogeneity in the permeability can be used to improve the performance of the receiver .
the distribution of the number of output molecules is binomial .
therefore , the variance of the number of molecules @xmath159 at steady state is @xmath160 .
we have @xmath161 for rc receiver . @xmath114
[ ex : rc ] we can derive variance and covariance results from theorem [ thm : rc ] .
for example , @xmath162 equation gives the variance of the number of output molecules in the rc receiver .
the covariance of the number of signalling molecules in voxels @xmath68 and @xmath87 is given in equation .
these results can be derived by using the generating function of the probability distribution of @xmath61 .
this section presents results on the mean and variance of the number of output molecules in a cat receiver .
a special feature of the cat reaction mechanism is that the signalling molecule @xmath0 appears on both sides of the chemical formula .
this means that the dynamics of the signalling molecules is independent of that of the output molecules @xmath42 .
in fact , we can derive the probability distribution of the signalling molecule in the cat network by using theorem [ thm : rc ] by removing the rc receiver .
let @xmath163 and @xmath164 be vectors containing the first @xmath165 elements of the state vector @xmath61 ( see ) and @xmath83 ( see ) . in other words ,
@xmath163 and @xmath164 count only signalling molecules . assuming that the transmitter emits @xmath33 molecules at time @xmath31 ( @xmath32 ) , and the @xmath166 . in a cat network
, we have the probability that @xmath167 is : @xmath168 the vector @xmath164 is the solution to @xmath169 the matrix @xmath170 is the generator of the 1-mc with the receiver removed , i.e. the matrix @xmath170 accounts only for the diffusion of signalling molecules in the network . as an example , for the network in figure [ fig : model ] , we have drawn its 1-mc with its receiver removed in figure [ fig : mc_sm ] .
the matrix @xmath170 is the generator for such types of 1-mc .
since we have the probability distribution of @xmath163 , we can show that the laplace transform of the mean number of signalling molecules in the receiver voxel is @xmath145 where @xmath141 and @xmath138 are as defined in section [ sec : rc_mean ] .
in particular , @xmath141 is the laplace transform of @xmath143 , which is the probability that a signalling molecule emitted at the transmitter at time zero will be at the receiver voxel at time @xmath53 .
we can also obtain results on the variance and covariance of the number of signalling molecules in the voxels as in section [ sec : rc_var ] .
the following theorem is on the mean number of output molecules in the cat receiver .
let @xmath171 denote the mean number of output molecules in the cat receiver and @xmath172 be its laplace transform .
we have @xmath173 where @xmath174 with @xmath175 and @xmath45 defined in section [ sec : model ] . @xmath114
[ th : cat_mean ] the proof of this theorem can be found in appendix [ app : cat ] .
equation can be depicted by the block diagram in figure [ fig : block_cat ] .
the transmitter emission pattern @xmath138 is first transformed by @xmath141 to obtain @xmath176 which is the laplace transform of the mean number of signalling molecules in the receiver voxel .
the latter is then subsequently transformed by @xmath177 to obtain @xmath172 .
although the transfer functions of rc and cat receivers have the same form , their behaviours are different . in a rc receiver , signalling molecules are converted into complexes and this creates a feedback , see figure [ fig : block_rc ] . in a cat receiver , the signalling molecule @xmath0 appears on both sides of the chemical formula , hence the absence of feedback .
the following theorem presents results on the variance of the number of output molecules .
let @xmath178 denote the variance of the number of output molecules in the cat receiver and @xmath179 be its laplace transform .
we have @xmath180 where @xmath181 is a @xmath165-dimensional vector and is the laplace transform of the covariance vector @xmath182 , i.e. the covariance between @xmath163 and the number of signalling molecules @xmath183 in the receiver voxel . @xmath114 [ th : cat_var ] the proof of this theorem can be found in appendix [ app : cat ] . note that @xmath182 can be computed because we know the distribution of @xmath163 .
we will now use the 2-voxel network in figure [ fig:2voxel ] to illustrate some of the properties of the cat receiver .
consider the two voxel network in figure [ fig:2voxel ] .
the receiver is now of the cat type .
the transmitter again emits @xmath151 molecules in the beginning .
we will present expressions for steady state mean and variance of the number of output molecules .
for convenience , we drop the @xmath154 for the time argument .
the mean number of signalling molecules @xmath184 in the receiver voxel is @xmath185 .
the mean number of output molecules @xmath186 is @xmath187 by comparing and , we can see that the cat receiver has a higher mean output compared to the rc receiver if @xmath188 and @xmath189 .
the steady state variance @xmath190 of the number of output molecules is : @xmath191 where @xmath192 is the variance of the number of signalling molecules in the receive voxel and is equal to @xmath193 .
the second term on the right - hand side of deserves some explanation .
the quantities @xmath183 and @xmath194 are related by a stochastic differential equation of the form : @xmath195 where @xmath196 is a noise term .
if @xmath196 is independent of @xmath183 and @xmath194 , then standard results on the response of linear time - invariant systems to noise @xcite says that the variance in @xmath197 due to the variance of @xmath198 is @xmath199 .
the noise amplification factor in is in fact smaller .
this is due to the fact that the noise @xmath196 is correlated with both @xmath183 and @xmath194 in a chemical reaction , see @xcite .
equation shows that we can increase @xmath186 by decreasing @xmath148 , but this will also increase the variance @xmath200 .
the mean output @xmath186 can also be increased by adjusting the reaction constants @xmath175 and @xmath45 but a fundamental tradeoff is that an increase in mean also leads to an increase in variance , or noise .
we will make a comparison between the rc and cat receivers .
note that the same input has been used in example [ ex : rc ] and in this example .
we further assume that @xmath188 and @xmath189 . under these conditions
, we have @xmath201 .
we can view the mean and variance of the output as , respectively , the signal and noise strength .
this means a rc receiver has a lower signal strength and a lower noise while a cat receiver has a higher signal strength and a higher noise .
we will further compare the performance of these two receivers in section [ sec : num ] .
@xmath114 [ ex : cat ]
in this section , we present numerical examples to illustrate the properties of the rc and cat receivers . before presenting the main findings
, we first use a network of 75 voxels to verify the analytical solutions for computing the mean and variance of the number of output molecules .
the results for rc and cat receivers are presented in figures [ fig : verify_rc ] and [ fig : verify_cat ] respectively .
each figure presents the mean and variance computed by analytical methods as well as by simulation .
we used the @xmath202-leaping @xcite , which is a commonly used method in chemistry to simulate diffusion and reactions of discrete molecules .
the simulation results shown are the average of 1500 runs .
the figures clearly show that the analytical expressions are accurate .
we consider a medium of 10@xmath203 m @xmath204 5 @xmath203 m @xmath204 1 @xmath203 m .
we assume a voxel size of ( @xmath205@xmath203m)@xmath206 ( i.e. @xmath207 @xmath203 m ) , creating an array of @xmath208 voxels .
the transmitter and receiver are located at ( 2.5,2.5,0.5 ) and ( 7.5,2.5,0.5 ) ( in @xmath203 m ) in the medium . the voxel co - ordinates are ( 7,7,3 ) and ( 22,7,3 ) .
we assume the diffusion coefficient @xmath22 of the medium is 1 @xmath203m@xmath209s@xmath210 . in all the comparisons below
, we assume @xmath211 and @xmath50 , i.e. the rc and cat receivers use the same parameters .
we use two different values of @xmath40 : 0.2 to 0.8 @xmath203m@xmath212s@xmath210 .
the value of @xmath41 is 0.1 s@xmath210 .
these values are similar to those used in @xcite and are realistic for biological systems .
we assume an absorbing boundary for the medium and the signalling molecules escape from the boundary voxel surface at a rate of @xmath213 .
the outflow permeability of the receiver voxel surface ( i.e. the rate at which signalling molecules leave the receiver ) is assumed to be @xmath24 unless otherwise stated .
the transmitter emits at times 0 , 1 and 2 seconds .
the number of signalling molecules emitted at each time is 1000 .
figure [ fig : env2_mean_2k ] compares the mean output of rc and cat receivers for two different values of @xmath214 0.2 to 0.8 .
two observations can be made . for smaller value of @xmath40 ,
the mean output of both receiver types is almost the same .
however , for larger value of @xmath40 , the mean output of cat is higher than that of rc .
this can be explained by the form of transfer functions for rc and cat receivers in equations and respectively . for small @xmath40 ,
the denominator in is almost equal to 1 , so the two receivers have the same output . for large @xmath40 , the denominator in is large and results in rc having a lower mean output .
figure [ fig : rc_mv_2k ] shows for the mean and variance of the number of output molecules for the rc receiver for @xmath214 0.2 to 0.8 .
it shows that for the rc receivers , the variance of the output signal is smaller than the mean of the signal .
figure [ fig : cat_mv_2k ] shows for mean and variance of the number of output molecules for the cat receiver for @xmath214 0.2 to 0.8 . for cat receivers , the variance of the receiver output is higher than the mean because cat magnifies the fluctuations in the number of signalling molecules as discussed in example [ ex : cat ] .
we investigate the impact of receiver outflow permeability on the mean and variance of the output signal . for the rc receivers , we compare the effect of a lower outflow permeability of @xmath215 against the normal value of @xmath24 .
the lower value means signalling molecules leave the receiver at a lower rate of @xmath215 .
we keep @xmath216 in this study .
figure [ fig : rc_mv_2p ] shows the mean and variance of the number of output molecules for the normal and lower permeabilities .
it can be seen that a lower permeability magnifies the mean , and consequently , also the variance of the output signal . for cat receivers
, we compare the effect of a lower outflow permeability of @xmath217 against the normal value of @xmath24 .
figure [ fig : cat_mv_2p ] shows the mean and variance of the output signals of the cat receiver .
the same observations can be made : a lower receiver outflow permeability increases both the mean and variance of the output .
if we use the same lower value of receiver outflow permeability , we find that the magnification in mean in variance for the cat receivers is higher than that for the rc receivers .
we can draw a few observations from these studies .
( 1 ) mean response for cat receivers is higher than rc receivers if @xmath40 is large ; otherwise they are almost the same .
( 2 ) the variance of rc receivers is always lower than the mean .
( 3 ) the variance of cat receivers is higher than the mean .
( 4 ) the receiver outflow permeability can be used to increase the mean response of the receivers .
we see in the above numerical results and in example [ ex : cat ] that for the same input signal and if @xmath218 and @xmath219 , then we have @xmath220 .
this means a rc receiver has a lower signal strength and a lower noise while a cat receiver has a higher signal strength and a higher noise .
it is not obvious which receiver has a better communication performance . in this section ,
we study the communication capacity of the two receivers as discrete memoryless channels .
we assume the transmitter encodes the different messages by emitting a different number of molecules .
this is comparable to pulse amplitude modulation .
we assume that the transmitter can emit 10 , 20 , 30 , ... , 1000 molecules at a time .
this gives 100 input symbols .
we use the steady state number of output molecules as the receiver output and we want to see how many different input levels the receiver can distinguish .
we do that by using the blahut algorithm @xcite to calculate the capacity of the discrete memoryless channel .
we assume that consecutive symbols are well separated in time so that inter - symbol interference can be neglected . in order to use the blahut algorithm ,
we need the probability distribution of the number of output molecules for each input symbol .
this is available for the rc receiver but not for the cat receiver . for the cat receiver , we use the mean and variance of the output and assume that the probability is gaussian distributed . since gaussian distribution has the maximum entropy , the estimated capacity is a lower bound of the true capacity @xcite .
we assume a reflecting boundary in this study .
this means the molecules can not leave the medium .
we use two different values of receiver outflow permeability : normal value of @xmath24 and a lower value of @xmath217 .
we vary the @xmath40 value from 0.05 to 0.6 .
figure [ fig : cap_kp ] shows the capacity for both receivers under two different receiver outflow permeabilities . for small @xmath40 , rc and
cat receivers have almost the same capacity .
however , for larger @xmath40 , rc has a higher capacity compared to cat for both receiver outflow permeabilities .
this is due to cat having a higher level of noise .
we have observed earlier that a lower receiver outflow permeability can lead to a higher mean and variance of the output signals for both the rc and cat receivers .
figure [ fig : cap_kp ] shows that a lower receiver outflow permeability can lead to a higher channel capacity .
molecular communication plays a fundamental role in living organisms and have been widely studied in biology @xcite .
the study of molecular communication in the communication theory literature has been growing in the past decade . for recent review of this area ,
see @xcite .
molecules in a molecular communication network can be propagated by active transport or diffusion .
the former class of networks has been studied in @xcite while the majority of the work assumes that molecules diffuse freely in the medium .
this paper also assumes diffusion .
diffusion is a major source of noise in molecular communication networks .
the work in @xcite considers using the emission time of a molecule for encoding .
it shows that diffusion causes the deviation from expected arrival time to have an inverse gaussian distribution .
the authors of @xcite derive the probability distribution of the number of molecules arriving at a time at the molecular receiver and use the distribution to design a decoder .
both pieces of work consider only diffusion and do not consider the reaction mechanisms at the receiver .
the authors in @xcite model the noise at the molecular receivers by tracking the particle dynamics of the molecules .
the authors in @xcite assumes the receiver consists of a michaelis - menten catalytic process and shows that the receiver noise has a poisson distribution . in our earlier work in @xcite , we propose an alternative approach of using master equations to study the noise in molecular communication network in a homogeneous infinite medium . in this work
, we consider a finite medium and study the impact of heterogeneity on the performance of molecular receivers .
the work in @xcite presents results on the mean and variance of the number of output molecules in a rc receiver . in this work ,
we derive its probability distribution .
receiver design is an important topic in communication theory .
there is much recent work on decoder design for molecular communication , see @xcite .
these pieces of work present decoders that can be used to decode a received signal .
the receiver reaction mechanisms in these papers have been chosen beforehand . in this work ,
we compare two receiver reaction mechanisms to characterise their performance in terms of mean and variance .
the capacity of diffusion based molecular communication network has been studied in @xcite . both papers model diffusion by means of diffusion partial differential equation .
this paper takes a different approach and models the network using voxels .
this modelling approach allows us to vary the permeability of the membrane of the receiver and study its impact on the receiver performance .
such type of heterogeneity does not appear be have been considered in earlier work .
reaction mechanisms similar to rc and cat have also been studied in biophysics literature @xcite .
however , the study considers only the reaction mechanism itself , and does not consider transmitters and diffusion .
this paper investigates the impact of different receiver reaction mechanisms on the performance of molecular communication networks .
two receiver mechanisms , reversible conversion and linear catalytic , have been studied .
we derive analytical expressions for the mean and variance of the output signals for these two reaction mechanisms . for reversible conversion
, we have also derived the probability distribution of the output signal .
we find that linear catalytic magnifies the noise and results in a noisier output signal .
this ultimately leads to linear catalytic having a lower communication performance compared to reversible conversion .
the analytical expressions that we have derived also allow us to study the impact of the diffusibility of the receiver membrane on the mean and variance of the receiver output signal .
we find that a selective receiver membrane , which makes it harder for the signalling molecules to leave the receiver , can be used to improve the communication performance .
the use of a selective membrane to improve communication performance does not appear to have studied before .
given that @xmath221 , we have @xmath222 where @xmath223 is the laplace transform of @xmath132 . we will now derive an expression for @xmath223 .
we first partition @xmath83 in as : @xmath224^t \end{aligned}\ ] ] where @xmath164 contains all but the last element in @xmath83 . according to theorem [ thm : rc ] , @xmath83 is the solution to @xmath225 where @xmath86 is the generator of the corresponding 1-mc .
the matrix @xmath86 has a specific block structure , which allows us to write as : @xmath226 = \left [ \begin{array}{c|c } \tilde{h } - \mathds{1}_r \mathds{1}_r^{t } k_+ & \mathds{1}_r k_- \\
\hline \mathds{1}_r^t k_+ & - k_- \end{array } \right ] \left [ \begin{array}{c } \tilde{q}(t ) \\
q_r(t ) \end{array } \right ] \mbox { with } \tilde{q}(0 ) = \mathds{1}_t \mbox { and } q_r(0 ) = 0 \label{eqn : hblock2 } \end{aligned}\ ] ] where @xmath28 is the index for the receiver voxel and @xmath227 is the @xmath28-th standard basis vector .
the matrix @xmath170 contains only diffusion parameters and does not contain any parameters from the receiver reaction mechanism .
in fact , @xmath170 is the generator of the 1-mc of a network with its receiver reaction mechanisms _
removed_. for example , for the network in figure [ fig : model ] , the 1-mc of the network with removed receiver is depicted in figure [ fig : mc_sm ] . by taking the laplace transform of and noting that @xmath228 where @xmath229 is the the probability that the signalling molecule is in the receiver voxel , we have , after some manipulations : @xmath230 where @xmath231 and @xmath232 denote , respectively , the laplace transform of @xmath229 and the identity matrix .
equation can now be obtained after eliminating @xmath231 from the above two equations .
the transfer function @xmath141 is the laplace transform of @xmath143 which is the solution of @xmath233 this means that @xmath143 is the probability that the signalling molecule is in the receiver voxel @xmath28 at time @xmath53 given that it is initially at the transmitter voxel .
since @xmath170 is used , this applies to the network with its receiver reaction mechanisms removed .
the interpretation of @xmath142 is similar .
the state vector @xmath61 of the cat receiver network can be written as : @xmath234\end{aligned}\ ] ] where @xmath163 is as defined in section [ sec : sol2 ] and @xmath194 is the number of output molecules in the cat receiver at time @xmath53 .
we denote their means by @xmath235 and @xmath171 .
the cat receiver network can be analysed by the rdmex model in @xcite , we have @xmath236 = \left [ \begin{array}{c|c } \tilde{h } & 0 \\ \hline \mathds{1}_r^t f_+ & f_- \end{array } \right ] \left [ \begin{array}{c } \tilde{n}(t ) \\ \hline \bar{n}_{cat}(t ) \end{array } \right ] + \mathds{1}_t \underbrace{\sum_{i=1}^\infty k_i \delta(t - t_i)}_{= u(t ) } \label{eqn : rdemx_cat_mean } \end{aligned}\ ] ] where @xmath170 is the generator matrix of the 1-mc with the receiver removed as discussed in section [ sec : sol2 ] .
let @xmath238 be the covariance of the state vector @xmath61 .
we partition @xmath238 into a 2@xmath2042 block matrix conformal to the partitioning of @xmath61 above . by applying the results in @xcite on covariance of the rdmex model to a cat receiver network
, the covariance matrix @xmath238 satisfies the following differential equation : @xmath239 = & \left [ \begin{array}{c|c } \tilde{h } & 0 \\ \hline \mathds{1}_r^t f_+ & -f_- \end{array } \right ] \left [ \begin{array}{c|c } \sigma_{11 } & \sigma_{12 } \\ \hline \sigma_{21 } & \sigma_{22 } \end{array } \right ] + \left [ \begin{array}{c|c } \sigma_{11 } & \sigma_{12 } \\ \hline \sigma_{21 } & \sigma_{22 } \end{array } \right ]
\left [ \begin{array}{c|c } \tilde{h}^t & \mathds{1}_r f_+ \\ \hline 0 & -f_- \end{array } \right ] + \nonumber \\ & \left [ \begin{array}{c|c } * & 0 \\ \hline 0 & f_+ \bar{n}_{l , r}(t ) + f_- \bar{n}_{cat}(t ) \end{array } \right ] \label{eqn : lyap}\end{aligned}\ ] ] where @xmath240 denotes a matrix whose details are not important for this derivation . our goal is to obtain the variance @xmath241 which is equal to @xmath242 . let @xmath243 which is the @xmath28-th column of the covariance matrix @xmath244 .
since @xmath238 is the covariance of the state vector @xmath61 , @xmath244 is the covariance of @xmath163 .
therefore the @xmath28-th column of @xmath244 is @xmath245 .
we can write the ( 1,2 ) and ( 2,2 ) blocks of as @xmath246 by taking the laplace transform of the above two equations and after eliminating @xmath247 , we arrive at the results of theorem [ th : cat_var ] .
|
in a molecular communication network , transmitters and receivers communicate by using signalling molecules . at the receivers ,
the signalling molecules react , via a chain of chemical reactions , to produce output molecules .
the counts of output molecules over time is considered to be the output signal of the receiver .
this output signal is used to detect the presence of signalling molecules at the receiver .
the output signal is noisy due to the stochastic nature of diffusion and chemical reactions .
the aim of this paper is to characterise the properties of the output signals for two types of receivers , which are based on two different types of reaction mechanisms .
we derive analytical expressions for the mean , variance and frequency properties of these two types of receivers .
these expressions allow us to study the properties of these two types of receivers .
in addition , our model allows us to study the effect of the diffusibility of the receiver membrane on the performance of the receivers .
* keywords : * molecular communication networks ; molecular receivers ; performance analysis ; stochastic models ; master equations ; receiver membrane ; noise
|
grs 1915 + 105 is a low mass x - ray binary , which appears to be the most distinct galactic microquasar @xcite .
the system contains a black hole and a low mass k
m iii giant donor @xcite , and it has a long period of @xmath2 d @xcite .
the donor , which got a variable - star name v1387 aql , fills its roche lobe and supplies the matter accreted by the black hole .
the first determination of the mass of the black hole based on a measurement of the radial velocity amplitude , @xmath3 , was by @xcite , who , observing with the vlt , obtained @xmath4 km s@xmath5 .
this indicated that the system contains one of the heaviest galactic black holes with the mass of @xmath6 ( assuming the binary inclination of @xmath7 of @xcite ) .
this measurement was later refined by @xcite , who also used the vlt but at a higher spectral resolution , and obtained @xmath8 km s@xmath5 , which , given that @xmath9 , results in a reduction of this mass .
the exact value depends on the inclination , which , for grs 1915 + 105 , is usually determined from the jet kinematics , which , in turn , depends on the distance to the source @xcite .
after the successful radio parallax measurement by @xcite , the distance is determined to be @xmath10 kpc , which then yields @xmath11 @xcite .
this distance determination is consistent with an independent estimate of @xmath12 kpc by @xcite based on considering the jet kinetic power .
while the mass of the black hole may be now considered as relatively precisely known , it is less so for the mass of the donor .
the first rough estimate implied by the spectral classification ( k m iii ) was proposed by @xcite as @xmath13@xmath14 . however
, this is a range typical for isolated giants of these classes @xcite , while the donor in grs 1915 + 105 could have lost a large fraction of its mass .
this has been confirmed by measurements of the rotational broadening .
first , @xcite found it as @xmath15 km s@xmath5 .
this quantity combined with @xmath3 yields the binary mass ratio , @xmath16 .
@xcite used the value of @xmath3 of @xcite , which yields @xmath17 .
the current best measurement of the rotational broadening is that of @xcite , @xmath18 km s@xmath5 .
this value combined with their @xmath3 measurement yields @xmath19 . for the distance - dependent black - hole mass estimate of @xcite , this gives @xmath20 .
thus , the donor has the mass substantially lower than that of an isolated star of the same spectral class , i.e. , it is a stripped giant ( e.g. , @xcite , hereafter wrs83 ) .
we note that @xcite gave a rather conservative estimate on the error of @xmath21 . from the bottom panel of their fig .
4 we would infer a standard deviation of @xmath223 km s@xmath5 , which would translate into a standard deviation of @xmath23 of @xmath220.017 , and that on the standard deviation of @xmath24 of @xmath25 . even then , the uncertainty on the donor mass is rather large .
we explore here another possibility to constrain the donor mass based on modelling of its internal structure .
since the donor is not a main sequence star but rather an evolved low - mass giant , its structure is relatively simple and easy to model . making use of the observational constraints on the luminosity and the radius of v1387 aql
, we construct an evolutionary model of this star and attempt to constrain its mass .
for that , we also need the effective temperature of the donor .
the currently most accurate nir observations of this system are those of @xcite , who matched their spectra to those of k0 , k1 , k2 , k5 iii and m0 iii template stars .
thus , we adopt here the possible range of the spectral classes , which appears to be due to a typo . ] from k0 iii to m0 iii .
from the third kepler law , we have the separation of the components as @xmath26 where @xmath27 is the total mass .
the relation between the radius of the optical component ( which must be equal to the radius of its roche lobe ) and the separation of the components for @xmath28 is @xcite @xmath29 from equations ( [ kepler][roche ] ) we have @xmath30 let us note that generally equation ( [ r_k ] ) makes radius of the donor filling its roche lobe one of the most accurately determined parameters of that star .
the orbital period is usually known with the high precision and the dependence on @xmath24 is weak .
moreover , equation ( [ r_k ] ) does not depend on the distance to the binary system .
for discussion of the accuracy of equation ( 3 ) see , e.g. , @xcite .
given the small uncertainty on the period , its contribution to the error is negligible .
the roche - lobe radius for a given donor mass of equation ( [ r_k ] ) , @xmath31 , is , in fact , identical to that implied by the rotational broadening , @xmath32 for @xmath33 substituted in the former . here
@xmath23 follows from the standard rotational - broadening relationship ( e.g. , @xcite ) , which solution can be found as @xmath34 ^ 2}{3 r(y)},\quad r(y)=2^{-1/3}\left[2 + 27y+3^{3/2}\sqrt{y(4 + 27y)}\right]^{1/3}\geq 1,\nonumber \\ & y=\left(\frac{3^{4/3 } v\sin i}{2 k_2}\right)^3\!,\label{qratio}\end{aligned}\ ] ] while the black - hole mass is @xmath35 equations ( [ rad_rot][qratio ] ) assume corotation , which is very likely in grs 1915 + 105 .
we can relate the inclination to the distance by assuming the inner jet has the same direction as the binary axis , which yields @xmath36 $ ] , where @xmath37 and @xmath38 are , respectively , the angular velocities of the approaching and receding jet .
@xcite used the weighted average of the inclination resulting from the observations by @xcite and @xcite . on the other hand
, @xcite argued that the jet direction changes between the projected distance from the centre of @xmath39 ( observed by @xcite ) and @xmath40 @xcite , and thus the results of the former are more relevant for estimating the binary plane orientation , resulting in @xmath41 .
we show the @xmath42 relationship in fig .
[ i_d ] .
is also the binary inclination .
the solid curve gives the best estimate , and the dotted curves give the range corresponding to the measurement errors .
the range of @xmath43 determined from the radio parallax by @xcite is shown by the vertical thin lines . ]
@xcite argued that the alignment is very likely , and this assumption has indeed been universally used in the mass estimates for grs 1915 + 105 . on the other hand , an also likely and widely accepted , model for the low - frequency qpos / breaks in the power spectra of black - hole binaries
is the lense - thirring precession of an inner hot part of the accretion flow ( e.g. , @xcite ) .
this model requires a misalignment between the black - hole spin and binary axes , though the minimum required misalignment appears not to be specified .
grs 1915 + 105 does show low - frequency qpos ( e.g. , @xcite ) , which appear very similar to those in other black - hole binaries , and the above model can apply to it .
also , the black - hole binary gro j165540 has an accurate determination of the orbital axis inclination of @xmath44 @xcite , which is significantly different from the jet axis inclination of @xmath45 @xcite .
if the axes are different in grs 1915 + 105 as well , the alignment - based estimates would become inaccurate , and , e.g. , the marked difference between the black - hole mass of grs 1915 + 105 of @xmath11 @xcite and those typical for accreting low - mass black - hole binaries of @xmath46 @xcite may disappear .
( we note here that the above average may need to be updated , e.g. , taking into account the revision of the black - hole mass in nova muscae from @xmath47 to @xmath48 , @xcite . ) given this uncertainty , we show the constraints on the radius and mass of the donor as functions of both the inclination and the distance , with the latter assuming the alignment .
[ r2_i ] shows the @xmath49 and @xmath50 dependencies , which do not assume the alignment .
we show there , though , the range of the inclinations obtained by @xcite ( assuming the alignment ) .
then , fig .
[ r2_d ] shows the @xmath51 and @xmath52 dependencies , which assume the alignment . .
the range of @xmath53 compatible with the parallax distance , assuming the alignment of the jet with the binary axis , and using the results of @xcite is @xmath54@xmath55 @xcite , as shown by the vertical thin lines .
the horizontal blue dot - dashed lines show @xmath56 and the corresponding radius , which are the minimum possible values found from evolutionary stellar models in section [ models ] .
the ranges allowed by both the observations and the models are those below the upper dotted curve and above both the dot - dashed and lower dotted curves . ] .
the range of @xmath53 compatible with the parallax distance , assuming the alignment of the jet with the binary axis , and using the results of @xcite is @xmath54@xmath55 @xcite , as shown by the vertical thin lines .
the horizontal blue dot - dashed lines show @xmath56 and the corresponding radius , which are the minimum possible values found from evolutionary stellar models in section [ models ] .
the ranges allowed by both the observations and the models are those below the upper dotted curve and above both the dot - dashed and lower dotted curves . ] from the parallax @xcite is shown by the vertical thin lines .
the red dashed curves enclose the range allowed by the observed nir flux ( approximately independent of @xmath53 ) .
the horizontal blue dot - dashed lines show @xmath56 and the corresponding radius , which are the minimum possible values based on evolutionary consideration .
the allowed parameter region is within the innermost dotted , dashed and dot - dashed curves and the vertical lines . ] from the parallax @xcite is shown by the vertical thin lines .
the red dashed curves enclose the range allowed by the observed nir flux ( approximately independent of @xmath53 ) .
the horizontal blue dot - dashed lines show @xmath56 and the corresponding radius , which are the minimum possible values based on evolutionary consideration .
the allowed parameter region is within the innermost dotted , dashed and dot - dashed curves and the vertical lines . ]
we can also constrain the size and mass of the donor vs. the distance by using the observed nir flux of the donor ( as first done by @xcite ) . the unveiled donor k magnitude ( at @xmath57 m ) and the extinction towards the system have been estimated as 14.515.0 and @xmath58 by @xcite and @xcite , respectively .
this gives the extinction - corrected magnitude of k @xmath5913.1 , or the flux per unit wavelength at @xmath60 m as ( 2.2@xmath61 erg s@xmath5 @xmath62 . by approximating the stellar spectrum as a blackbody at the effective temperature , @xmath63 , i.e. , @xmath64 , we
can then obtain the stellar radius as a function of the distance .
the results for the temperatures within the range of @xmath654660 k @xcite , corresponding to the adopted range of the spectral classes of k0 iii
m0 iii , are shown by the dashed red curves in fig .
these results are basically equivalent , and almost the same as those obtained using the barnes - evans relation @xcite , as given for the k magnitude by equation ( 1a ) of @xcite , and for the range of the surface brightness of @xmath663.86 ( which approximately corresponds to k0m0 giants , see fig . 2 of @xcite ) .
combining the above constraints for the allowed range of the distance , we find @xmath67 and @xmath68 .
the upper limit is larger than that of @xcite because they gave that from propagation of errors , while here we give it for the entire allowed range of @xmath43 . at a given distance ,
the constraints are more stringent , as shown on fig .
@xmath69 diagram for @xmath70 , 0.28 , 0.3 0.5 and @xmath71 for the grey , blue , magenta , red and green curves , respectively .
the evolution proceeds ( from left to right ) at the constant total mass , during which the h - burning shell is moving outwards .
this increases the mass of the he core and decreases the mass of the h - rich envelope .
the horizontal lines ( in respective colours ) show the roche - lobe radii for ( from the top ) @xmath72 , 0.5 , 0.3 , 0.28 and @xmath73 .
possible solutions for v1387 aql are given by the intersections between a horizontal line and the corresponding evolutionary track .
( a ) the full studied parameter space .
( b ) a magnification of the intersection region .
the possible solutions are marked with the letters : a and a@xmath74 for @xmath75 , b and b@xmath74 for @xmath76 , and c and d for 0.5 and @xmath71 , respectively .
the solutions a@xmath74 and b@xmath74 are unphysical as they do not assure a continuous mass transfer . ]
0.2 cm @xmath69 diagram for @xmath70 , 0.28 , 0.3 0.5 and @xmath71 for the grey , blue , magenta , red and green curves , respectively .
the evolution proceeds ( from left to right ) at the constant total mass , during which the h - burning shell is moving outwards .
this increases the mass of the he core and decreases the mass of the h - rich envelope .
the horizontal lines ( in respective colours ) show the roche - lobe radii for ( from the top ) @xmath72 , 0.5 , 0.3 , 0.28 and @xmath73 .
possible solutions for v1387 aql are given by the intersections between a horizontal line and the corresponding evolutionary track .
( a ) the full studied parameter space .
( b ) a magnification of the intersection region .
the possible solutions are marked with the letters : a and a@xmath74 for @xmath75 , b and b@xmath74 for @xmath76 , and c and d for 0.5 and @xmath71 , respectively .
the solutions a@xmath74 and b@xmath74 are unphysical as they do not assure a continuous mass transfer . ] in order to calculate evolutionary models of stripped giants , we used the warsaw stellar - evolution code ( described in @xcite ) .
the code was calibrated to reproduce the sun at the solar age .
this calibration resulted in the chemical composition of the h mass fraction of @xmath77 , the metallicity of @xmath78 , and the mixing length parameter of @xmath79 . to reproduce the present state of v1387 aql
, we followed the evolution of a @xmath80 star , which was maintained at a constant mass until hydrogen was nearly exhausted in its centre .
then , the mass removal from the surface started and continued until the donor star reached four different values of the mass , 0.26 , 0.28 , 0.3 , 0.5 and @xmath81 .
we applied two different ( arbitrary ) rates of the mass removal , @xmath82 and @xmath83/yr .
we found that the structure of the remnants was identical in both cases .
our motivation was to obtain thermal - equilibrium stars of a given mass .
since the structure of such stars does not depend on their evolutionary history , the prescription for mass removal from the surface was unimportant .
the large amount of the mass from the donor accreted by black hole would then help to explain its claimed large spin .
however , @xcite assumed fully conservative mass transfer throughout the system history , which included epochs of highly super - eddington accretion , during which strong outflows most likely took place , questioning that assumption . ] .
until then , the he core had not formed .
the further evolution was followed at a constant total mass .
the h - burning shell moves outwards , increasing the mass of the he core , @xmath84 , and decreasing that of the h - rich envelope .
generally , this causes an increase of the radius of the stripped giant .
however , when the mass of the remaining envelope gets sufficiently low , the giant starts to shrink , as shown in fig .
[ rad_mc ] .
the results of our calculations are presented in fig .
[ rad_mc ] , which show the evolutionary tracks in the core mass radius diagram .
the tracks are shown for the stripped giants of the masses equal 0.26 , 0.28 , 0.3 , 0.5 and @xmath71 .
the stars evolve at constant mass and the driving mechanism is the progress of the h - burning shell moving outwards .
the radii of the partially stripped giants generally increase with @xmath84 , except for the shrinking when the masses of their envelopes become very low . [ solutions ] .the parameters of the physical evolutionary models .
the mass loss rate is in the unit of @xmath85/y .
[ cols="^,^,^,^,^,^,^",options="header " , ] fig .
[ rad_mc ] show also values of the roche - lobe radius of the donor calculated with the formula ( [ r_k ] ) for the assumed masses .
possible solutions that we consider as models for v1387 aql are given by the intersections between a horizontal line ( for a given mass ) and the corresponding evolutionary track .
from fig .
[ rad_mc ] , we immediately see that a stripped giant of the mass @xmath73 can not provide a solution since during its evolution it never attains a sufficiently large radius .
we have checked that the same is true for the mass of @xmath86 .
so , we are left with the remnants of the mass @xmath87 , which minimum value and its corresponding radius we show in figs .
[ r2_i ] and [ r2_d ] . in further discussion
, we shall consider remnants of the mass 0.28 , 0.3 , 0.5 and @xmath71 .
magnified portions of the relevant tracks and horizontal lines from fig .
[ rad_mc](a ) are shown in fig .
[ rad_mc](b ) .
the possible solutions given by the intersections between the horizontal line ( for a given mass ) and the evolutionary track ( for the same mass ) are marked with consecutive capital letters , a , b , c and d for @xmath56 , 0.3 , 0.5 and @xmath81 , respectively , while the primed letters indicate the intersections during the final evolution stages .
we point out that the solutions a@xmath74 and b@xmath74 are unphysical as they do not assure a continuous mass transfer between the components of the binary system since solutions lie on the declining parts of the evolutionary tracks . during this evolutionary phase ,
the star shrinks with the growing mass of the core .
therefore , any mass outflow would be quickly stopped .
the parameters of the physical solutions are given in table 1 , including their effective temperature and luminosity .
we note that the dependence of the stellar radius on the core mass can be considered for three different situations .
first , we can consider the evolution of isolated giants at constant mass , as in fig .
[ rad_mc ] .
while this evolution does depend on the stellar mass , this dependence is relatively weak , and we can provide a fitting formula for the main dependence of @xmath88 averaged over @xmath24 , as also done by wrs83 . then , we can consider @xmath88 during the mass transfer via roche - lobe overflow , i.e. , with a simultaneous decrease of the donor mass .
this gives usually steeper dependencies .
it corresponds to considering also the second term in equation ( 1 ) for @xmath89 of wrs83 .
finally , we can consider @xmath88 for our solutions , imposing the stellar radius equal to that of the roche lobe of a given mass .
this does not correspond to any evolutionary sequence , but just parametrises our results for @xmath90 d. this , in turn , gives a flatter dependence .
we compare our radius vs. the core mass dependencies with the results of wrs83 ( and of @xcite , who retained only the first order in their formulae ) .
they correspond to the first case above , i.e. , for evolution of an isolated star , and averaging over neglecting the dependence on @xmath24 .
given that we use different chemical composition and more contemporary physics ( especially opacities ) , our results , while qualitatively similar , are quantitatively different .
while wrs83 obtained @xmath91 , we find on average for @xmath92 , @xmath93^{4.3},\label{r_mcore}\ ] ] for the evolution of a giant at a constant mass .
we do see some dependence on @xmath24 in fig .
[ rad_mc ] , especially for low masses ( see also @xcite ) .
also , this dependence excludes ranges of @xmath84 close to @xmath24 , when the star starts to shrink . if we consider only the radii equal to the roche - lobe radii at the period of grs 1915 + 105 ( the third case above ) , we find @xmath94 .
so far , we have made use only of the value of the radius of v1387 aql ( which can be determined precisely for a given total mass using equation [ r_k ] ) .
a second parameter , luminosity , can be estimated only less precisely , given the range of the allowed effective temperatures , @xmath95 , where @xmath96 is the stefan - boltzmann constant , and we use the values corresponding to giants of the spectral classes k0 iii
m0 iii , @xmath654660 k @xcite .
this uncertainty leads to the corresponding uncertainty of the luminosity estimate .
the range of @xmath97 at a given @xmath69 corresponding to the adopted range of @xmath98 is shown in fig .
[ r_l ] .
our evolutionary model predicts the luminosity at any stage , and we compare the values corresponding to our solutions , see table 1 , with the above constraints in fig .
we have found that all our solutions lie in middle of the allowed range , indicating a good agreement of our evolutionary calculations with the standard parameters of giants of @xcite , in spite of the reduction of the mass due to accretion mass loss , which also agrees with above luminosity constraints .
it was this solution that @xcite advocated as the best model .
unfortunately , it was not noted at that time that this model did not assure a continuous mass transfer since it corresponded to the radius decreasing with time , and hence was unphysical . ] .
all of the solutions have the temperatures of @xmath994160 k , which correspond to the spectral class of k5 iii @xcite .
we thus find that the evolutionary considerations do not provide any significant upper limit on the mass and radius .
for example , we have also calculated models with the final mass of @xmath100 ( solution e ) , @xmath101 ( solution f ) , and @xmath102 ( solution g ) , with the last one being well above the mass range allowed by the observations , and found that those solutions also lie in the middle of the range of @xmath97 allowed by the adopted range of @xmath98 .
we show their parameters in table 1 .
@xmath97 plane .
the heavy solid lines show the uncertainty range of @xmath97 due to the uncertainty of @xmath98 , and the vertical thin lines show the minimum radius of @xmath103 ( from our evolutionary calculations ) and the maximum radius found to be allowed in section [ radius ] .
the blue points marked show the positions of the solutions a g ( in order of increasing @xmath69 ) .
the red dotted line shows the fit to the cases with @xmath104 ( solutions c
one comment that should be made about our models concerns the luminosity of a star losing mass through roche lobe overflow .
if the mass outflow is very rapid , the surface luminosity of the star may become significantly lower than that without outflow ( as noted by @xcite ) .
it is known that the black hole in the system is accreting matter roughly at the eddington rate , @xmath105/yr , though the rate of the mass loss from v1387 aql is much lower , see section [ outflow ] . however , we have found that even the eddington rate is not high enough to decrease substantially the stellar surface luminosity .
we superimposed a mass outflow at about eddington rate on our model a , and found that the internal radiation flux in the outer layers started to decrease toward the surface , indicating that these layers were departing from thermal equilibrium . however , this departure was very small .
the surface luminosity of the model with the outflow was smaller by less than 1 per cent compared with the model without outflow .
the obtained values of the luminosity show a similar steep dependence ( excluding @xmath56 and @xmath106 ) on the core mass to that of wrs83 of @xmath107 , but our normalization is @xmath10850 per cent higher , @xmath109^{8.15 } , \label{l_mcore}\ ] ] with the relative error @xmath1103 per cent . combining it with the dependence of @xmath69 assuming the roche - lobe radius for the observed @xmath111 ( the third case at the end of section [ core_radius ] ) , we find @xmath112 , which is shown in fig .
having constructed the models describing the internal structure of v1387 aql , we can calculate the rate of the mass transfer between the components of the binary system implied by our models .
to do so , let us locate each of our models in a binary with @xmath90 d and a @xmath113 black hole .
we assume the conservative mode of the mass transfer ( conservation of the total mass and of the total orbital angular momentum ) .
then , we calculate numerically at which rate of the mass outflow from the star the changes of the stellar radius will follow the changes of the roche lobe around it . the resulting rates are given in table 1 and shown in fig .
[ mdot_m2 ] , where we see a strong dependence of @xmath114 on @xmath24 .
the results can be compared with the theoretical expression for @xmath114 separated into three physical terms , @xmath115 where @xmath116 is the rate of the roche - lobe changes with the changing mass , given by equation ( 12 ) of wrs83 , with @xmath117 at @xmath28 , and @xmath118 is the derivative along the solution with the mass transfer , i.e. , that for the second case discussed at the end of section [ core_radius ] . we find @xmath119 along the solutions with the mass transfer , with , e.g. , @xmath120 at @xmath121 ( and @xmath1224.88 for @xmath123@xmath102 ) .
the time derivative of the core mass is linked to the luminosity and the efficiency of hydrogen burning ( accounting for neutrino losses ) , @xmath124 erg / g ( as given by wrs83 ) , @xmath125 . here , @xmath126 is the h content immediately above the h - burning shell , which , due to the evolutionary h burning , can be substantially lower than the intial @xmath127 ( which still approximately equal that at the stellar surface ) .
for example , @xmath128 for our case with @xmath129 ( and @xmath1300.63 for @xmath123@xmath102 ) .
we find the values of @xmath114 obtained from equation ( [ mdot ] ) using the obtained values of @xmath131 and @xmath126 to agree very well with our numerical results , shown in table 1 and fig .
[ mdot_m2 ] .
we note that @xmath132 at the exact minimum possible mass , because then @xmath133 .
thus , arbitrarily low values of @xmath114 can be obtained when the minimum allowed mass is approached , but this requires significant fine - tuning , see fig . [ rad_mc ] .
we thus see that for the range of @xmath24 we have found to be allowed for grs 1915 + 105 , from 0.28 to @xmath80 , @xmath134 . ,
equation ( [ mdot_w ] ) .
the dashed line shows the dependence of equation ( 25a ) of wrs83 .
the mass range allowed for grs 1915 + 105 is @xmath135 . ]
we can compare our values with expressions given in literature .
the rate given by equation ( 15 ) of wrs83 in terms of @xmath24 and @xmath84 ( and with a weak dependence on @xmath136 ) is completely equivalent to our equation ( [ mdot ] ) , but we have now different numerical coefficients appearing in that formula . the formula ( 25a ) in wrs83 , giving @xmath137 in terms of @xmath24 and @xmath111 ( and @xmath136 ) is shown by the dashed line in fig .
[ mdot_m2 ] .
we see it has a different slope from that shown by our values .
we have fitted the values of @xmath137 for @xmath92 , and obtained @xmath138 where the dependence on @xmath111 is kept unchanged with respect to wrs83 .
this is shown in fig .
[ mdot_m2 ] by the solid line .
the disagreement at the lowest masses is a boundary effect , caused by @xmath132 at the exact minimum possible mass ( as discussed above ) .
we can compare the obtained mass transfer rates with the average accretion rate required to power the outburst of grs 1915 + 105 .
we used for that the results of @xcite . from their fig .
5 , we infer the average observed bolometric flux corresponding to @xmath1083/4 of the eddington luminosity for @xmath139 , pure h and at @xmath140 kpc assumed by them , which is @xmath141 erg @xmath142 s@xmath5 .
this corresponds to @xmath143 erg s@xmath5 , and the accretion rate of @xmath144y .
we can compare our estimate of @xmath145 with that of @xcite , which gives @xmath146 erg s@xmath5 , a very similar value .
we thus adopt @xmath147 erg s@xmath5 at the observational best - fit donor mass value of @xmath148 and at our corresponding value of @xmath114 ( table 1 ) , the duty cycle is about 5 per cent .
the accreted mass during the 25 y of the outburst is @xmath149 .
these values can be four times higher at the maximum allowed mass of @xmath80 , or much lower at the lowest allowed donor masses .
see @xcite for estimates of the disc mass in grs 1915 + 105 .
we can compare our results with those of @xcite .
he used the fits of wrs83 and considered the donor mass of @xcite as @xmath150 .
thus , his best fit value of @xmath151 is @xmath152y , which is comparable to our values at @xmath1530.9 .
a potentially important effect we have neglected is the irradiation of the donor by the x - ray source ( pointed out by @xcite ) . at the parameters at @xmath154 , the donor subtends a solid angle of @xmath155 ( see equation [ roche ] ) , and
the ratio of the irradiating luminosity ( neglecting shielding and assuming isotropy ) to @xmath156 ( emitted by the hemisphere facing the black hole ) is 33 . at the estimated duty cycle
, the average irradiating luminosity is larger than the intrinsic one , which may affect our estimates .
we have studied constraints on the present mass and radius of v1387 aql by constructing evolutionary models of the internal structure of this star .
we have compared the resulting radii and luminosity of our physical models to observational constraints .
we find solutions with stripped giants of the mass of @xmath157 and of the spectral class k5 iii , independent of the distance to the system , and a distance - dependent upper limit , @xmath158 .
we have also obtained the constraints from the observed nir magnitude of the companion , which are found to be consistent with the other constraints . finally , we have estimated the mass transfer rate , which we found to be @xmath159 . comparing to the average luminosity of grs 1915 + 105 ,
we obtain the duty cycle of @xmath160 per cent .
these two numbers are 4 times lower at @xmath148 , which is the observationally - determined best - fit mass .
we thank joanna mikoajewska and tomaso belloni for valuable discussions . this research has been supported in part by the polish national science centre grants
2013/10/m / st9/00729 and 2015/18/a / st9/00746 .
|
v1387 aql ( the donor in the microquasar grs 1915 + 105 ) is a low - mass giant .
such a star consists of a degenerate helium core and a hydrogen - rich envelope .
both components are separated by an hydrogen burning shell .
the structure of such an object is relatively simple and easy to model . making use of the observational constraints on the luminosity and the radius of v1387 aql , we constrain the mass of this star with evolutionary models . we find a very good agreement between the constraints from those models and from the observed rotational broadening and the nir magnitude . combining the constraints , we find solutions with stripped giants of the mass of @xmath0 and of the spectral class k5 iii , independent of the distance to the system , and a distance - dependent upper limit , @xmath1 .
we also calculate the average mass transfer rate and the duty cycle of the system as a function of the donor mass .
[ firstpage ] binaries : general stars : evolution stars : individual : v1387 aql x - rays : binaries x - rays : individual : grs 1915 + 105 .
|
the relevance of spatial and other kinds of heterogeneity in the design of immunization strategies has been widely addressed in the epidemic modeling of infectious diseases @xcite . in particular , it has been pointed out that population inhomogeneities can substantially enhance the spread of diseases , making them harder to eradicate and calling for specific immunization strategies .
this issue assumes the greatest importance in a wide range of natural interconnected systems such as food - webs , communication and social networks , metabolic and neural systems @xcite .
the complexity of these networks resides in the small average path lengths among any two nodes ( small - world property ) , along with a large degree of local clustering . in other words , some special nodes of the structure
develop a larger probability to establish connections pointing to other nodes .
this feature has dramatic consequences in the topology of scale - free ( sf ) networks @xcite which exhibit a power - law distribution @xmath0 for the probability that any node has @xmath1 connections to other nodes . for exponents in the range
@xmath2 , this connectivity distribution implies that , for large network sizes , the nodes have a statistically significant probability of having a very large number of connections compared to the average connectivity , @xmath3 .
this feature contrasts with what is found for homogeneous networks ( local or nonlocal ) in which each node has approximately the same number of links , @xmath4 @xcite .
the extreme heterogeneity of sf networks finds the most stunning examples in two artificial systems , the world - wide - web ( www ) @xcite and the internet @xcite . along with these technological networks , it has also been pointed out that sexual partnership networks are often extremely heterogeneous @xcite , and it has been recently observed that the network of sexual human contacts possesses a well - defined scale - free nature @xcite . in homogeneous networks , an epidemic occurs only if the rate of infection of `` healthy '' individuals connected to infected ones exceeds the so - called _ epidemic threshold _ ; in other words , if the disease can not transmit itself faster than the time of cure , it dies out @xcite . in heterogeneous networks , on the other hand , it is well - known that the epidemic threshold decreases with the standard deviation of the connectivity distribution @xcite .
this feature is paradoxically amplified in scale - free networks which have diverging connectivity fluctuations .
in fact , as it was first noted in refs .
@xcite , epidemic processes in sf networks do not possess , in the limit of an infinite network , an epidemic threshold below which diseases can not produce a major epidemic outbreak or the inset of an endemic state . sf networks are therefore prone to the spreading and the persistence of infections , whatever virulence the infective agent might possess . in view of this weakness , it becomes a major task to find optimal immunization strategies oriented to minimize the risk of epidemic outbreaks on sf networks , task with immediate practical and economical implications .
this paper presents a parallel comparison of the effect of different immunization schemes in the case of two different complex networks : the watts - strogatz model @xcite and the barabsi and albert model @xcite .
the first is a homogeneous network exhibiting small - world properties , while the second one is the prototype example of sf network . by studying the susceptible - infected - susceptible model @xcite in presence of progressively greater immunization rates , we find that uniformly applied immunization strategies are effective only in complex networks with bounded connectivity fluctuations . on the contrary , in sf networks
the infection is not eradicated even in the presence of an unrealistically high fraction of immunized individuals .
actually , sf systems do not have any critical fraction of immunized individuals and only the total immunization of the network achieves the infection s eradication . in order to overcome these difficulties we define optimal immunization strategies that rely on the particular sf structure of the network .
the developed strategies allow us to achieve the total protection of the network even for extremely low fractions of successfully immunized individuals .
in order to estimate the effect of an increasing density of immune individuals in complex networks , we will investigate the standard susceptible - infected - susceptible ( sis ) model @xcite .
this model relies on a coarse - grained description of individuals in the population .
namely , each node of the graph represents an individual and each link is a connection along which the infection can spread . each susceptible ( healthy ) node is infected with rate @xmath5 if it is connected to one or more infected nodes .
infected nodes are cured and become again susceptible with rate @xmath6 , defining an effective spreading rate @xmath7 ( without lack of generality , we set @xmath8 ) .
the sis model does not take into account the possibility of individuals removal due to death or acquired immunization @xcite , and thus individuals run stochastically through the cycle susceptible @xmath9 infected @xmath9 susceptible .
this model is generally used to study infections leading to endemic states with a stationary average density of infected individuals .
a wide class of network models @xcite have exponentially bounded connectivity fluctuations .
a paradigmatic example of this kind of networks that has recently attracted a great deal of attention is the watts - strogatz ( ws ) model @xcite , which is constructed as follows : the starting point is a ring with @xmath10 nodes , in which each node is symmetrically connected with its @xmath11 nearest neighbors .
then , for every node each link connected to a clockwise neighbor ( thus @xmath12 links for each node ) is kept as originating from the original node and rewired to a randomly chosen target node with probability @xmath13 .
this procedure generates a random graph with a connectivity distributed exponentially for large @xmath1 , and an average connectivity @xmath14 .
it is worth remarking that even in the case @xmath15 the network keeps the memory of the construction algorithm and is not equivalent to a random graph .
in fact , by definition each node emanates at least the @xmath12 links which have been rewired from the clockwise neighbors to randomly chosen nodes ; a property that affects also the clustering properties of the graph ( for details see ref .
@xcite ) . for the class of exponentially bounded networks , one can generally consider that each node has roughly the same number of links , @xmath16 , and therefore we can consider them as fairly homogeneous in their connectivity properties . at a mean - field level ,
the equation describing the time evolution of the average density of infected individuals @xmath17 ( prevalence ) is @xmath18 .
\label{eq : ws}\ ] ] the mean - field character of this equation stems from the fact that we have neglected the density correlations among the different nodes , independently of their respective connectivities . the first term on the r.h.s . in eq .
( [ eq : ws ] ) considers infected nodes becoming healthy with unit rate .
the second term represents the average density of newly infected nodes generated by each active node .
this is proportional to the infection spreading rate @xmath19 , the number of links emanating from each node @xmath16 , and the probability that a given link points to a healthy node , @xmath20 $ ] . after imposing the stationary condition @xmath21 , the most significant and general result is the existence of a nonzero epidemic threshold @xmath22 @xcite such that @xmath23 in other words , if the value of @xmath19 is above the threshold , @xmath24 , the infection spreads and becomes endemic .
below it , @xmath25 , the infection dies out exponentially fast .
the existence of an epidemic threshold is a general results in epidemic modeling , present also in different models such as the susceptible - infected - removed ( sir ) model @xcite . in analogy with critical phenomena
@xcite , this kind of behavior can be identified as an absorbing - state phase transition , in which @xmath26 plays the role of the order parameter in the phase transition and @xmath19 is the tuning parameter , recovering the usual mean - field behavior @xcite .
this standard framework is radically changed in the class of sf networks @xcite , in which the probability distribution that a node has @xmath1 connections has the form @xmath27 and the connectivity fluctuations , @xmath28 , diverge in infinite networks for any value @xmath29 .
the paradigmatic example of sf network is the barabsi and albert ( ba ) model @xcite .
the construction of the ba graph starts from a small number @xmath30 of disconnected nodes ; every time step a new vertex is added , with @xmath31 links that are connected to an old node @xmath32 with probability @xmath33 , where @xmath34 is the connectivity of the @xmath32-th node .
after iterating this scheme a sufficient number of times , we obtain a network composed by @xmath10 nodes with connectivity distribution @xmath35 and average connectivity @xmath36 . for this class of graphs ,
the absence of a characteristic scale for the connectivity makes highly connected nodes statistically significant , and induces strong fluctuations in the connectivity distribution which can not be neglected . in order to take into account these fluctuations
, we have to relax the homogeneity assumption used for homogeneous networks , and consider the relative density @xmath37 of infected nodes with given connectivity @xmath1 ; i.e. , the probability that a node with @xmath1 links is infected .
the dynamical mean - field equations can thus be written as @xcite @xmath38 \theta(\rho(t ) ) , \label{mfk}\ ] ] where also in this case we have considered a unity recovery rate .
the creation term considers the probability that a node with @xmath1 links is healthy @xmath39 $ ] and gets the infection via a connected node .
the probability of this last event is proportional to the infection rate @xmath19 , the number of connections @xmath1 , and the probability @xmath40 that any given link points to an infected node .
the probability that a link points to a node with @xmath41 links is proportional to @xmath42 .
in other words , a randomly chosen link is more likely to be connected to an infected node with high connectivity , yielding @xmath43 where @xmath44 is identical to @xmath45 by definition . in the stationary state
[ @xmath46 , eq .
( [ mfk ] ) yields the following infected node density form @xmath47 by inserting the above expression for @xmath48 in eq .
( [ first ] ) , we obtain the self - consistency equation @xmath49 where @xmath50 is now a function of @xmath19 alone @xcite .
the solution @xmath51 is always satisfying the consistency equation .
a non - zero stationary prevalence ( @xmath52 ) is obtained when the r.h.s . and the l.h.s . of eq .
( [ cons ] ) , expressed as function of @xmath50 , cross in the interval @xmath53 , allowing a nontrivial solution .
it is easy to realize that this corresponds to the inequality @xmath54 being satisfied . the value of @xmath19 yielding the equality in eq .
( [ eq : critpunt ] ) defines the critical epidemic threshold @xmath55 , that is given by @xmath56 this results implies that in sf networks with connectivity exponent @xmath29 , for which @xmath57 , we have @xmath58 .
this fact implies in turn that for any positive value of @xmath19 the infection can pervade the system with a finite prevalence , in a sufficiently large network @xcite .
for small @xmath19 it is possible to solve explicitly eq .
( [ cons ] ) for sf networks and calculate the prevalence in the endemic state as @xmath59 as shown in refs .
calculations can be carried out by using the continuous @xmath1 approximation , valid for large @xmath1 @xcite , that assumes @xmath60 , where @xmath31 is the minimum number of connections of any node and @xmath61 is a properly defined probability density of connections .
for the particular case of the ba network we have @xmath62 @xcite , that in the limit of an infinitely large network yields the prevalence @xcite @xmath63 obviously , @xmath64 assumes a bounded value in finite size networks , defining an effective threshold @xmath65 due to finite size effects , as customarily encountered in nonequilibrium phase transitions @xcite .
this epidemic threshold , however , is not an _ intrinsic _ quantity as in exponential networks and it is vanishing for increasing network sizes ; i.e. in the thermodynamic limit . since real networks have always a finite size , however , it is interesting to calculate how the epidemic threshold scales with the system size @xcite . by considering the continuous @xmath1 approximation , it is possible to calculate the finite size distribution moments as @xmath66 , where @xmath67 is the largest connectivity present in the finite network . for networks composed by @xmath10 nodes , @xmath67 is obviously an increasing function of @xmath10 . in the particular case of the ba model ,
we readily obtain @xmath68 and @xmath69 as @xmath70 . substituting this values in the eq .
( [ thr ] ) we obtain a threshold @xmath71 . in order to find the size dependence of @xmath55
, we have to relate the maximum connectivity @xmath67 with the network size @xmath10 .
this relation is given by @xmath72 @xcite , yielding finally a threshold @xmath73 this result can be generalized to sf networks with an arbitrary connectivity distribution , which show an epidemic threshold vanishing as a power law behavior in @xmath10 with an exponent depending on the connectivity exponent @xmath74 @xcite .
the simplest immunization procedure one can consider consists in the random introduction of immune individuals in the population @xcite , in order to get a uniform immunization density .
immune nodes can not become infected and thus do not transmit the infection to their neighbors . in this case , for a fixed spreading rate @xmath19 , the relevant control parameter is the immunity @xmath75 , defined as the fraction of immune nodes present in the network . at the mean - field level , the presence of uniform immunity will effectively reduce the spreading rate @xmath19 by a factor @xmath76 ; i.e. the probability of finding and infecting a susceptible and non - immune node . by substituting @xmath77 in eqs .
( [ eq : ws ] ) and ( [ mfk ] ) we obtain the prevalence behavior for progressively larger immunization rates . in homogeneous networks , such as the ws model , it is easy to show that in the case of a constant @xmath19 , the stationary prevalence obtained from eq .
( [ eq : ws ] ) is given by @xmath78 here , @xmath79 is the critical immunization value above which the density of infected individuals in the stationary state is null and depends on @xmath19 as @xmath80 thus , the critical immunization which achieves eradication is related to the spreading rate and the epidemic threshold of the infection .
( [ eq : gcws ] ) is obviously valid only for @xmath81 , and it implies that the critical immunization allowing the complete protection of the network ( null prevalence ) is increasing with the spreading rate @xmath19 . on the contrary , uniform immunization strategies on sf networks are totally ineffective .
the presence of immunization depresses the infection s prevalence too slowly , and it is impossible to find any critical fraction of immunized individuals that ensures the infection eradication .
the absence of an epidemic threshold ( @xmath58 ) in the thermodynamic limit implies that whatever rescaling @xmath77 of the spreading rate does not eradicate the infection except the case @xmath82 .
in fact , by using eq .
( [ thr ] ) we have that the immunization threshold is given by @xmath83 in sf networks with @xmath84 only a complete immunization of the network ( i.e. , @xmath85 ensures an infection - free stationary state . the fact that uniform immunization strategies are less effective has been noted in several cases of spatial heterogeneity @xcite . in sf networks we face a limiting case due to the extremely high ( virtually infinite )
heterogeneity in connectivity properties .
also in this case finite networks present an effective threshold @xmath86 depending on the number of nodes @xmath10 . as for the epidemic threshold , however , we are not in presence of an intrinsic quantity and we have that @xmath87 in the thermodynamic limit @xmath88 . in the case of the ba model , inserting the expression ( [ eq : lambdan ] ) into eq .
( [ eq : gcdef ] ) , we observe that the immunization threshold scales as @xmath89 also in this case it is possible to generalize this result for arbitrary connectivity exponents @xmath74 @xcite . in order to provide further support to the present mean - field ( sometimes called the deterministic approximation ) description ,
we study by means of numerical simulations the behavior of the sis model on the ws and the ba networks .
in these systems , because of the nonlocal connectivity , mean - field predictions are expected to correctly depict the model s behavior . in the present work we consider the parameters @xmath90 and maximal disorder @xmath15 for the ws network , and @xmath91 and @xmath92 in the case of the ba network . in the presence of uniform immunization
, we can study the system by looking at the infection s prevalence in the stationary regime ( endemic state ) as a function of the immunity @xmath75 . the uniform immunization is implemented by randomly selecting and immunizing @xmath93 nodes on a network of fixed size @xmath10 .
our simulations are implemented at a fixed spreading rate @xmath94 .
the number of nodes range from @xmath95 to @xmath96 .
we analyze the stationary properties of the density of infected nodes @xmath97 ( the infection prevalence ) for different values of the immunization @xmath75 .
initially we infect half of the susceptible nodes in the network , and iterate the rules of the sis model with parallel updating .
the prevalence is computed averaging over at least @xmath98 different starting configurations , performed on at least @xmath99 different realizations of the network . in fig .
1(a ) , we show the behavior of the reduced prevalence @xmath100 ( where @xmath101 is the prevalence without immunization ) as a function of the uniform immunization @xmath75 in the ws network .
we observe that the prevalence of infected nodes decays drastically for increasing immunization densities ( see fig .
1(b ) ) . in particular
, we observe the presence of a sharp immunization threshold @xmath102 , in fair agreement with the estimate @xmath103 from eq .
( [ eq : gcws ] ) with the values @xmath94 and the estimate @xmath104 from ref . @xcite . in the biological case ,
this effect motivates the use of global vaccination campaigns in homogeneous populations in order to reach a density of immune individuals that secures from major outbreaks or endemic states . on the contrary , the results for the sf network , depicted in fig .
2(a ) , show a strikingly different behavior .
namely , the density of infected individuals decays slowly with increasing immunization , and it would be null only for the complete immunization of the whole network ( @xmath82 ) .
specifically , it follows from eq .
( [ op ] ) that the sis model on the ba network shows for @xmath105 and any @xmath19 the prevalence @xmath106 we have checked this prediction in fig .
in other words , the infection always reaches an endemic state if the network size is enough large ( see fig .
this points out the absence of an immunization threshold ; sf networks are weak in face of infections , also after massive uniform vaccination campaigns .
when fighting an epidemic in an heterogeneous population with a uniform vaccination scheme , it is necessary to vaccinate a fraction of the population larger than the estimate given by a simple ( homogeneous ) assumption @xcite . in this case
, it can be proved @xcite that _ optimal _ vaccination programs can eradicate the disease vaccinating a smaller number of individuals .
sf networks can be considered as a limiting case of heterogeneous systems and it is natural to look for specifically devised immunization strategies . a straightforward way to reintroduce an intrinsic immunization threshold in sf networks consists in using different fractions of immunized individuals according to their connectivity .
let us define @xmath107 as the fraction of immune individuals with a given connectivity @xmath1 .
if we impose the condition @xmath108 we observe that eqs .
( [ mfk ] ) become identical and decoupled , defining effectively an homogeneous system .
the density of infected individuals is the same for all connectivities @xmath1 , and an epidemic threshold @xmath109 is reintroduced in the system .
this condition requires that @xmath110 is constant for all groups of connectivity @xmath1 at the threshold , implying that @xmath111 ; i.e. , a larger portion of individuals must be immunized in groups with larger connectivity . in this scheme
the total density of immunized individuals can be easily calculated by averaging @xmath107 over the various connectivity classes .
the fraction of non - immunized individuals @xmath112 can not be larger than one , thus we focus only on classes with connectivity such that the reproductive number @xmath113 . to eradicate the infection , we need that @xmath114 in all classes with connectivity @xmath115 , defining the critical fraction of immunized individuals as @xmath116 in order to perform an explicit calculation for the ba model , we use again the continuous @xmath1 approximation @xcite . in this case we obtain that @xmath117 this result can be readily extended to sf networks with arbitrary @xmath74 values , and it is worth remarking that this recipe is along the lines of that introduced in the immunization of heterogeneously populated groups @xcite .
recently , a similar strategy has been put forward in ref .
@xcite by proposing to cure with proportionally higher rates the most connected nodes .
while proportional immunization schemes are effective in finally introducing an well - defined immunization threshold , the very peculiar nature of sf networks allows to define more efficient strategies based on the nodes hierarchy .
in particular , it has been shown that sf networks posses a noticeable resilience to random connection failures @xcite , which implies that the network can resist a high level of damage ( disconnected links ) , without loosing its global connectivity properties ; i.e. , the possibility to find a connected path between almost any two nodes in the system .
at the same time , sf networks are strongly affected by selective damage ; if a few of the most connected nodes are removed , the network suffers a dramatic reduction of its ability to carry information @xcite . applying this argument to the case of epidemic spreading
, we can devise a _ targeted _ immunization scheme in which we progressively make immune the most highly connected nodes , i.e. , the ones more likely to spread the disease . while this strategy is the simplest solution to the optimal immunization problem in heterogeneous populations @xcite ,
its efficiency is comparable to the uniform strategies in networks with finite connectivity variance . in sf networks , on the contrary
, it produces an arresting increase of the network tolerance to infections at the price of a tiny fraction of immune individuals .
let us consider the situation in which a fraction @xmath75 of the individuals with the highest connectivity are successfully immunized .
this corresponds , in the limit of a large network , to the introduction an upper threshold @xmath118 , such that all nodes with connectivity @xmath119 are immune .
the fraction of immunized individuals is then given by @xmath120 relation that renders @xmath118 an implicit function of @xmath75 .
the presence of the cut - off @xmath121 defines the new average quantities @xmath122 and @xmath123 , which are on their turn function of @xmath75 . at the same time , all links emanating from immunized individuals can be considered as if they were removed . the probability @xmath124 that any link will lead to an immunized individual is then given by @xmath125 and if we consider that this fraction @xmath124 of links are effectively removed , the new connectivity distribution after the immunization of a fraction @xmath75 of the most connected individuals is @xcite : @xmath126 the new distribution ( after cut - off introduction and link removal ) yields the first two moments @xmath127 and @xmath128 @xcite . by recalling eq .
( [ thr ] ) , the critical fraction @xmath79 of immune individuals needed to eradicate the infection will be given by the relation @xmath129 an explicit calculation for the ba network in the continuous @xmath1 approximation yields that the density of immunized nodes is related to the connectivity threshold as @xmath130 by inverting this relation we obtain that the connectivity threshold is @xmath131 , yielding that @xmath132 as well , we can obtain @xmath133 and @xmath134 as @xmath135 . by inserting these values into eq .
( [ th_targ ] ) we obtain the approximate solution for the immunization threshold in the case of targeted immunization as @xmath136 this clearly indicates that the targeted immunization program is extremely convenient in sf networks where the critical immunization is exponentially small in a wide range of spreading rates @xmath19 .
also in this case , the present result can be generalized for sf networks with arbitrary connectivity exponent @xmath74 . in order to test the targeted immunization scheme
we have implemented numerical simulations of the sis model on the ws and ba networks by immunizing the @xmath137 nodes with the highest connectivity .
note that , for a given network , this method is essentially deterministic : once we identify the hierarchy in the node s connectivity distribution , we proceed to protect those nodes on top of the list .
simulations are performed at a fixed spreading rate @xmath94 . in fig .
1(a ) we report the behavior of the prevalence of infected nodes for the ws network with targeted immunization ; the results corresponding to the ba graph are plotted in fig .
in the case of the ws network , the behavior of the prevalence as a function of @xmath75 is equivalent in the uniform and targeted immunization procedures .
the connectivity fluctuations are small , and the immunization of the most connected nodes is equivalent to the random choice of immune nodes .
this confirms that targeted strategies do not have a particular efficiency in systems with limited heterogeneity . on the contrary ,
in the case of the ba network , we observe a drastic variation in the prevalence behavior .
in particular , the prevalence suffers a very sharp drop and exhibits the onset of an immunization threshold above which no endemic state is possible ( zero infected individuals ) .
a linear extrapolation from the largest values of @xmath75 yields an estimate of the very convenient threshold @xmath138 .
this definitely shows that sf networks are highly sensitive to the targeted immunization of a small fraction of the most connected nodes ( see fig .
2(a ) and fig .
3(b ) ) . while these networks are particularly weak in face of infections , the good news consist in the possibility to devise immunization strategies which are extremely effective .
the present results indicate that the sf networks susceptibility to epidemic spreading is reflected also in an intrinsic difficulty in protecting them with local uniform immunization . on a global level ,
uniform immunization policies are not satisfactory and , in analogy with disease spreading in heterogenous media , only targeted immunization procedures achieve the desired lowering of epidemic outbreaks and prevalence .
this evidence radically changes the usual perspective of the regular epidemiological framework . spreading of infectious or polluting agents on sf networks , such as food or social webs ,
might be contrasted only by a careful choice of the immunization procedure . in particular , these procedures
should rely on the identification of the most connected individuals .
the protection of just a tiny fraction of these individuals raises dramatically the tolerance to infections of the whole population a practical example is provided by the spreading of viruses in the internet @xcite .
the sf nature of this network is the outcome of a connectivity redundancy which is very welcome because it ensures a greater error tolerance than in less connected networks . on the other hand , despite the large use of anti - virus software which is available on the market within days or weeks after the first virus incident report , the average lifetime of digital epidemics is impressively large ( 10 - 14 months ) @xcite .
numerical simulations of the sis model on real maps of the internet can provide further support to our picture .
the sis model is , in fact , well suited to describe dns - cache computer viruses @xcite ( the so - called `` natural computer viruses '' ) , and different digital viruses can be modeled by considering the random neighbor version of the model @xcite ) ; i.e. infected e - mails can be sent to different nodes which are not nearest neighbors .
the map considered here , provided by the national laboratory for applied network research ( nlanr ) and available at the web site http://www.moat.nlanr.net/routing/rawdata/ , contains @xmath139 nodes and @xmath140 links , corresponding to an average connectivity @xmath141 .
the connectivity distribution is scale - free , with a characteristic exponent @xmath142 @xcite .
our simulations are performed at a fixed spreading rate @xmath94 , averaging over at least @xmath143 different starting configurations .
we implement both the uniform and the targeted immunization procedures .
the results obtained clearly indicate that the behavior is completely analogous to that found on the ba network .
4 illustrates that , while uniform immunization does not allow any drastic reduction of the infection prevalence
the immunization of @xmath144 of the nodes reduces by less than a factor @xmath145 the relative prevalence
the targeted immunization drastically removes the occurrence of endemic states even at very low value of the immunization parameter .
the fact that sf networks can be properly secured only by a selective immunization , points out that an optimized immunization of the internet can be reached only through a global immunization organization that secures a small set of selected high - traffic routers or internet domains .
unfortunately , the self - organized nature of the internet does not allow to easily figure out how such an organization should operate .
the present results also appear to have potentially interesting implications in the case of human sexual disease control @xcite .
most sexually transmitted diseases can not be characterized without including the noticeable differences of sexual activity within a given population .
epidemic modeling is thus based on partitioning population groups by the number of sexual partners per unit time @xcite .
this implicitly corresponds to the knowledge of the probability distribution function @xmath61 that gives the fraction of the population within the @xmath1 class .
the recent observation that the web of human sexual contacts exhibits scale - free features @xcite points out that also sexually transmitted diseases are eventually spreading in a network with virtually infinite heterogeneity .
it follows that concepts such as the mean number of sexual partners or its variance are not good indicators in this case .
as well , the definition of a core group of `` super - spreader '' individuals could be a non well - defined concept because of the lack of precisely defined thresholds or characteristic magnitudes in the scale - free distribution of sexual contacts .
nevertheless , the striking effectiveness of targeted immunization indicates that control and prevention campaigns should be strongly focused at the most promiscuous individuals .
these represent the most connected nodes of the network and are thus the key individuals in the spreading of the infection .
while the simple sis model is very instructive , many other ingredients should be considered in a more realistic representation of real epidemics @xcite .
one would also want to add simple rules defining the temporal patterns of networks such as the frequency of forming new connections , the actual length of time that a connection exists , or different types of connections .
these dynamical features are highly valuable experimental inputs which are necessary ingredients in the use of complex networks theory in epidemic modeling .
this work has been partially supported by the european network contract no .
the nlanr project is supported by nsf .
rp - s also acknowledges support from the grant cicyt pb97 - 0693 .
we thank a. barrat , m. mezard , m .- c .
miguel , r. v. sol , s. visintin , and r. zecchina for helpful comments and discussions .
we are grateful to a. l. lloyd and r. m. may for enlightening suggestions and for pointing out to us some fundamental references .
|
complex networks such as the sexual partnership web or the internet often show a high degree of redundancy and heterogeneity in their connectivity properties .
this peculiar connectivity provides an ideal environment for the spreading of infective agents .
here we show that the random uniform immunization of individuals does not lead to the eradication of infections in all complex networks .
namely , networks with scale - free properties do not acquire global immunity from major epidemic outbreaks even in the presence of unrealistically high densities of randomly immunized individuals .
the absence of any critical immunization threshold is due to the unbounded connectivity fluctuations of scale - free networks .
successful immunization strategies can be developed only by taking into account the inhomogeneous connectivity properties of scale - free networks .
in particular , _ targeted _ immunization schemes , based on the nodes connectivity hierarchy , sharply lower the network s vulnerability to epidemic attacks .
|
the astrod ( astrodynamical space test of relativity using optical devices ) mission series aims at high precision measurements in the interplanetary space for the determination of several quantities in the areas of general relativity and fundamental physics as well as solar and solar system research @xcite .
the main goals are improvements in the determination of the relativistic parameterized - post - newtonian ( ppn ) parameters @xmath0 and @xmath1 ( eddington parameter ) , the solar and solar system parameters and the variation of the gravitational constant .
gravitational wave detection is a further goal for the multi - spacecraft missions astrod - gw @xcite and super - astrod @xcite . in the frame of this overview article on the single - spacecraft astrod
i mission concept the specific mission goals , the basic mission concept as well as several technical aspects of the mission are presented .
more detailed descriptions of the different mission aspects are given in refs . .
astrod i is a single spacecraft operated in drag - free - mode with a laser link to the earth ( ground based laser stations ) using pulsed laser ranging with state of the art measurement techniques .
the main payload components are a two color pulsed laser system , a laser pulse receiving unit , an event timer with a corresponding atomic clock for precise timing measurements with a timing accuracy of 3 ps and a high performance drag free control system containig an inertial sensor and several micro - newton thrusters for a precise attitude and orbit control . to achieve the mission goals the spacecraft needs to follow a pure gravitational orbit ( geodesic ) which means that all disturbing forces have to be compensated by the drag free system to a high precision .
the ideal orbit of astrod i can be reached with the help of venus encounters using the swing - by - technique .
especially the period when the spacecraft is located on the opposite of the sun ( viewed from the earth - fig .
[ f1 ] ) is very important for the measurement of the shapiro time delay which can be used for the determination of the eddington parameter @xmath1 . due to
the special orbit with venus encounters this leads to a minimum sun distance of less than 0.5 astronomical units ( au ) which is a challenge for the mission design in terms of thermal load for the spacecraft and in terms of optical setup and sunlight shielding / filtering .
the most recent simulations for possible orbits with corresponding venus encounters show that there are several launch windows for astrod i. detailed calculations have been carried through with a launch date in 2021 and a mission duration of 1200 days@xcite . in principle
there are many possible launch windows due to the fact that the orbital periods of earth and venus are short - especially compared to the orbital periods of the planets in the outer solar system .
therefore there is not a strong constraint on the year of launch . in 2010 astrod
i was proposed for the esa cosmic vision 2015 - 2025 m - class program by an international researcher team .
astrod i was selected as one of the final 14 candidates out of nearly 50 proposals which shows the genaral interest in the mission concept .
finally astrod i was not selected yet .
there are three main scientific goals .
the first goal is to test relativistic gravity and the fundamental laws of spacetime with an improvement of three orders of magnitude in sensitivity , specifically , to measure the ppn parameters @xmath1 ( eddington light - deflection parameter ; for general relativity , it is 1 ) via measurements of the shapiro time delay to 3 @xmath2 @xmath3 , @xmath0 ( relativistic nonlinear - gravity parameter ; for general relativity , it is 1 ) to 6 @xmath2 @xmath4 and others with significant improvement ; and to measure the fractional time rate of change of the gravitational constant ( dg / dt)/g with two orders of magnitude improvement .
the pioneer anomaly could meanwhile be explained as an anisotropic thermal radiation effect @xcite .
therefor a further investigation on the pioneer anomaly is not of strong interest anymore .
however , relativistic mond ( modified newtonian dynamics ) theories @xcite are of current interests as they are alternatives to dark matter theories .
mond theories have effects similar to the pioneer effect in the solar system .
testing mond using lisa pathfinder @xcite has been considered and analyzed @xcite .
with astrod i precision , mond effects on astrod i need to be studied , and possible tests on mond theories would then be evaluated .
the second goal is to initiate a revolution of astrodynamics with laser ranging in the solar system . with mm precision , the astrodynamics and ephemeris of the solar - system bodies
would be improved by orders of magnitude .
the third goal is to increase the sensitivity of the determination of solar , planetary and asteroid parameters by 1 - 3 orders of magnitude . in this context
, the measurement of the solar quadrupole moment parameter @xmath5 will be improved by two orders of magnitude , i.e. , to @xmath6 .
the mission goals for astrod i are listed in table [ ta1 ] ) .
these goals are based on orbit simulation and assumptions for a timing uncertainty of 3 pico - seconds ( ps ) and a drag - free uncertainty of 3 @xmath2 @xmath7 ms@xmath8 @xmath9 at 0.1 mhz .
the timing uncertainty has been achieved in t2l2 on board jason 2 @xcite .
the drag - free requirement is comparable to the actual goal of lisa pathfinder @xcite .
see section 5 for more discussions . in the area of solar and galactic physics astrod
i can also be used for measuring solar energetic particles ( seps ) and galactic cosmic rays ( gcrs ) , with corresponding applications to space weather @xcite .
the radiation detectors onboard the spacecraft for monitoring test - mass charging of the inertial drag - free sensor can be used for this purpose . like llr ,
the scientific data are ranges between astrod i and laser stations on earth as functions of time .
all goals can ( must ) be extracted from these ranges with fittings . no payload system could be omitted without affecting the range measurement except the possibility of omitting the radiation monitor which is only a small part of the astrod i payload related to the inertial sensor and which is good for space weather applications .
besides the technical details of astrod i ( spacecraft and payload ) the success of the mission depends on the selection of an appropriate orbit to reach the mission goals
. some effort has been made to simulate possible orbits including planetary swing - bys in order to find adequate solutions @xcite .
the swing - bys are intended to be geodesic and can be used for mapping venus gravity .
the fuel requirement is minimal with only drag - free thrusters needed .
one simulation with a launch in 2021 and a venus encounter 112 days after launch and three oppositions ( spacecraft , sun and earth in one line of sight ) - after 250 , 613 and 976 days after launch - gives a good example for such an orbit @xcite .
for the choice of the best orbit a trade - off is needed between minimising the time to reach the far side of the sun because of the risk to the mission from aging of equipment . the simulated launch-2021 orbit for the astrod i spacecraft is shown in fig . [ f2 ] , the corresponding orbit in the sun - earth fixed frame in fig .
the venus swing - by strategy is also useful for astrod - gw mission design and optimization @xcite .
astrod i is planned as a cylindrical spacecraft with a diameter of 2.5 meters and a hight of 2 meters . the total mass including a 100 kg apogee motor with liquid fuel is 490 kg . the spacecraft is stabilized around 3 axes by a drag free system .
the total electrical power ( including payload ) is estimated to be around 390 w. in order to protect the payload from thermal load of sunlight - especially at close distances to the sun - the spacecraft will have to be equipped with passive thermal insulation systems - e.g. : multi - layer - insulation ( mli ) , and perhaps even with active thermal control devices like thermel louvers .
the minimum requirement for the thermal stability of the payload is a temperature variation of less than @xmath10 1 k. due to the variation of the distance between the spacecraft and the sun ( 0.4 to 0.8 au ) the thermal influx varies by a factor of 4 @xcite .
this will have to be taken into account for the detailed spacecraft design in terms of thermal control .
the needed electrical power of 390 w will be generated by solar cells .
one possibility is to cover the spacecraft surface with the needed area of solar cells .
this could create a thermal problem due to the limited efficency of solar cells .
assuming an efficency of 0.2 , most of the lost power will be transformed into heat . in order to prevent the spacecraft from heating up , the solar cells could also be placed on solar panels ( see fig .
[ f4 ] ) . for the radio communication with the earth
the spacecraft is equipped with a 1.3 m diameter x - band antenna and low gain s - band antennas . during the whole mission the astrod i spacecraft will have to be pointed towards the earth to keep the laser link running .
therefore the high gain antenna does not need to be tracked as long as the high gain antenna is pointed parallel to the laser optics which simplifies the setup .
the astrod i payload consists of the following components : 300 mm cassegrain telescope optical bench two color pulsed nd : yag - laser system ( 532 nm , 1064 nm ) single photon detecting photodiodes event timer cesium / rubidium clock drag - free inertial sensor drag - free micro - newton - thrusters the cassegrain telescope is used for transmitting and receiving the laser pulses from / to earth / spacecraft .
the optical bench consits of several optical components for processing the laser light inside the spacecraft .
the main purpose of the optical bench is to seperate the incoming / outgoing laser pulses by polarisation- and color - filtering .
[ f5 ] shows the preliminary design of the optical bench .
detailed descriptions are given in refs . .
the laser system onboard the spacecraft generates laser pulses with two wavelengths ( 532 and 1064 nm ) .
the reason for this two - color setup is the need for a correction of the effect of the air column of the atmosphere . with two lasers with sufficiently different wavelengths it is possible to measure and substract the effect @xcite .
it is sufficient as demonstrated already by slr and llr .
the pulse width is 50 ps , the repetition rate 100 hz and the pulse energy around 10 mj . the specific type of the nd : yag laser system is not yet fixed .
the choice depends on the mass and performance and on the availability of space qualified units .
single photon detectors are intended to be used for the laser pulse reception onboard the spacecraft . for the required 3 ps timing accuracy
a precise event timer is needed .
the 3 ps timing accuracy is already achieved by the t2l2 ( time transfer by laser link ) event timer onboard jason 2 @xcite .
the emitting times and receiving times will be recorded by a cesium atomic clock . for a ranging uncertainty of 0.9 mm in a distance of 2.55 @xmath2 @xmath11 m ( 1.7 au ) , the laser / clock frequency needs to be known to one part in @xmath12 by comparison with ground clocks over a period of time . stability to 6 @xmath2 @xmath7 in 1,000
s ( round - trip time ) is required .
interplanetary pulse laser ranging(both up and down ) was demonstrated by messenger , using its laser altimeter in 2005 @xcite .
the technologies needed for a dedicated mission using interplanetary pulse laser ranging with millimetre accuracy are already mature .
a drag - free sensor with a free falling test mass is necessary as a reference for the pure gravitational motion of the spacecraft and micro - newton thrusters are needed to compensate disturbing forces acting on the spacecraft in a closed loop control system .
the drag - free requirement for astrod i is relaxed by one order of magnitude as compared with that of lisa ( laser interferometer space antenna for gravitational wave detection ) @xcite .
the drag - free technologies under development for lisa pathfinder @xcite will largely meet the requirement of astrod i. initially the feep technology ( field - emission electric propulsion ) seemed to be the best choice for micro - newton thrusters for missions like lisa pathfinder , microscope and the astrod missions . due to technical problems during the development of the feep technology
the cold gas technology has also been taken into acount for these missions .
the gaia mission will carry cold gas thrusters for the aocs ( attitude and orbit control system)@xcite .
microscope and lisa pathfinder will be equipped with cold gas thrusters based on the gaia thrusters .
the main disadvantage of cold gas thrusters compared to feeps is the higher mass per @xmath13v .
the total mission duration is limited by the amount of propellant stored in the tanks
. therefore the feep technology would be preferred if it is available soon enough for astrod i. the successful functioning of the accelerometer on board goce @xcite reassured the development for achieving improved inertial sensors .
astrod - gw , decigo pathfinder and decigo also require drag - free technologies for their performances @xcite . sunlight shielding is a common technology which needs to be adopted to the special requirements for astrod i ( see next section ) .
the overall technology readiness level of the astrod i payload components will be improved during the further development .
for the measurement of the post - newtonian parameter ( eddington light deflection parameter ) near the opposition , i.e. , 1.7 au = 255 million km away from earth , the spacecraft telescope will pick up only about @xmath14 of the power emitted . for the wavelength of 1064 nm that would mean a pulse of @xmath15 j or 5,000 photons @xcite . since avalanche photodiodes ( apd )
are used for single photon detection , a beam attenuation may be needed before it enters the detector . at the same time
, the sunlight will , at opposition , shine with 400 w into the spacecraft telescope and extreme care must be taken to reduce that sunlight to a level that the laser signal is dominant .
three different measures are planned for the reduction .
( i ) spatial filtering by placing a pinhole plate in the focal plane of the telescope , ( ii ) spectral filtering with narrow bandwidth dielectric filters and ( iii ) temporal filtering ( 10 ns window ) .
the remaining solar photons will be sufficiently less than the laser photons at the photo detector and the laser pulse signals can be detected .
especially the spectral narrow band filtering for two wavelengths ( in one filter ) needs some more investigation .
when there is a phase a study , laboratory implementation of the sunlight filtering system needs to be demonstrated .
optis is a satellite - based test of special and general relativity @xcite .
the tests are based on ultrastable optical cavities , lasers , an atomic clock and a frequency comb generator .
optis projects a michelson - morley test , a kennedy - thorndike test and a test of the universality of the gravitational redshift by comparison of an atomic clock with an optical clock . for optis ,
a laser link to the ground for comparison with ground clocks is much desired . in the early version of astrod i , interferometric ranging is included . for the interferometric ranging ,
the frequency of the laser offset - phase - locked to the incoming light can be measured by comparison with a harmonic frequency generated by an optic comb using a standard input frequency from the cs clock on the spacecraft .
it would be natural to combine these two versions of astrod i and optis ( as to acronym , it could be astroptis , asop , ) and go to deep space .
michelson - morley experiments could gain much sensitivity from a deep space mission .
redshift comparison would be good also .
kennedy - thorndike tests may lose some sensitivity .
however , deep space interferometric ranging will be tested .
more study regarding to costs is needed .
the field of the experimental gravitational physics stands to be revolutionized by the advancements in several critical technologies , over the next few years .
these technologies include deep space drag - free navigation and interplanetary laser ranging .
a combination of these serves as a technology base for astrod i. astrod i is a solar - system gravity survey mission to test relativistic gravity with an improvement in sensitivity of three orders of magnitude , improving our understanding of gravity and aiding the development of a new quantum gravity theory ; to measure key solar system parameters with increased accuracy , advancing solar physics and our knowledge of the solar system ; and to measure the time rate of change of the gravitational constant with an order of magnitude improvement and probing dark matter and dark energy gravitationally .
this will be the beginning of a series of precise space experiments on the dynamics of gravity .
the techniques are becoming mature for such experiments .
b. braxmaier and the astrod i team , astrodynamical space test of relativity using optical devices i ( astrod i ) - a class - m fundamental physics mission proposal for cosmic vision 20152025 : 2010 update , _ exp .
* 34 * ( 2012 ) p. 181 . ; and references therein .
t. appouchaux,_et al _ , astrodynamical space test of relativity using optical devices i ( astrod i ) - a class - m fundamental physics mission proposal for cosmic vision 20152025 , _ exp .
_ * 23 * ( 2009 ) p. 491 .
p. exertier , t2l2/jason-2 , first year of processing activities , in _ proceedings of the socit francaise dastronomie et dastrophysique ( sf2a ) 2009 _ , eds .
m. heydari - malayeri , c. reyl , r. samadi , p. 111
( 2009 ) c. grimani , implications of galactic and solar particle measurements on board interferometers for gravitational wave detection in space , _ int .
* 22 * , xxx ( 2013 ) ; and references therein .
|
astrod i is the first planned space mission in a series of astrod missions for testing relativity in space using optical devices .
the main aims are : ( i ) to test general relativity with an improvement of three orders of magnitude compared to current results , ( ii ) to measure solar and solar system parameters with improved accuracy , ( iii ) to test the constancy of the gravitational constant and in general to get a deeper understanding of gravity .
the first ideas for the astrod missions go back to the last century when new technologies in the area of laser physics and time measurement began to appear on the horizon .
astrod is a mission concept that is supported by a broad international community covering the areas of space technology , fundamental physics , high performance laser and clock technology and drag free control . while astrod i is a single - spacecraft concept that performes measurements with pulsed laser ranging between the spacecraft and earthbound laser ranging stations , astrod - gw
is planned to be a three spacecraft mission with inter - spacecraft laser ranging .
astrod - gw would be able to detect gravitational waves at frequencies below the elisa / ngo bandwidth . as a third step super - astrod with larger orbits could even probe primordial gravitational waves .
this article gives an overview on the basic principles especially for astrod i.
|
ideal crystals are absent in nature , and most of their physical properties , such as plasticity , melting , growth , etc . , are defined by defects of the crystalline structure .
therefore , a study of defects is a topical scientific question of importance for applications in the first place . at present
, a fundamental theory of defects is absent in spite of the existence of dozens of monographs and thousands of articles .
one of the most promising approaches to the theory of defects is based on riemann cartan geometry , which involves nontrivial metric and torsion . in this approach ,
a crystal is considered as a continuous elastic medium with a spin structure .
if the displacement vector field is a smooth function , then there are only elastic stresses corresponding to diffeomorphisms of the euclidean space .
if the displacement vector field has discontinuities , then we are saying that there are defects in the elastic structure .
defects in the elastic structure are called dislocations and lead to the appearance of nontrivial geometry . precisely , they correspond to a nonzero torsion tensor , equal to the surface density of the burgers vector .
defects in the spin structure are called disclinations .
they correspond to nonzero curvature tensor , curvature tensor being the surface density of the frank vector .
the idea to relate torsion to dislocations appeared in the 1950s [ 14 ] .
this approach is still being successfully developed ( note reviews [ 511 ] ) , and is often called the gauge theory of dislocations .
some time ago we proposed the geometrical theory of defects [ 1214 ] .
our approach is essentially different from others in two respects .
firstly , we do not have the displacement and rotational angle vector fields as independent variables because , in general , they are not continuous . instead , the triad field and @xmath0-connection are considered as independent variables .
if defects are absent , then the triad and @xmath0-connection reduce to partial derivatives of the displacement and rotational angle vector fields . in this case the latter can be reconstructed .
secondly , the set of equilibrium equations is different .
we proposed purely geometric set which coincides with that of euclidean three dimensional gravity with torsion .
the nonlinear elasticity equations and principal chiral @xmath0 model for the spin structure enter the model through the elastic and lorentz gauge conditions [ 1416 ] which allow to reconstruct the displacement and rotational angle vector fields in the absence of dislocations in full agreement with classical models .
the advantage of the geometric theory of defects is that it allows one to describe single defects as well as their continuous distributions . in the present paper ,
we consider propagation of torsional elastic waves in double wall tube with cylindrical dislocation .
this defect was first described in @xcite .
the schrdinger equation for the double wall tube was solved in @xcite and applied to double wall nanotubes .
a similar problem was also solved for the cylindrical waveguide with wedge dislocation @xcite .
let us describe double wall tube with cylindrical dislocation in the framework of the geometric theory of defects .
we consider cylindrical coordinates @xmath1 , @xmath2 in tree dimensional euclidean space @xmath3 .
let there be two thick tubes @xmath4 and @xmath5 of elastic media , each axis coinciding with the @xmath6 axis .
we suppose that @xmath7 ( see fig.[ftubed],_a _ , where a section @xmath8 is shown ) .
section @xmath8 of double wall tube before ( _ a _ ) and after ( _ b _ ) defect creation.,title="fig : " ] section @xmath8 of double wall tube before ( _ a _ ) and after ( _ b _ ) defect creation.,title="fig : " ] now we make one tube with the inside cylindrical dislocation in the following manner .
we stretch symmetrically the inner tube and compress the outer one . then glue together the external surface of the inner tube with the internal surface of the outer tube .
afterwards the media comes to some equilibrium state . due to rotational and translational symmetry
we obtain one tube @xmath9 with the axis which coincides with the @xmath6 axis ( see fig.[ftubed],_b _ ) .
radii of cylinders constituting tube surfaces are mapped as follows @xmath10 the gluing is performed along the cylinder @xmath11 , and there is cylindrical defect ( dislocation ) because part of the media between tubes is removed .
the obtained double wall tube with cylindrical dislocation is rotationally and translationally symmetric .
the constructed model of the tube with cylindrical dislocation can be considered as continuous model of double wall nanotube ( for a general review , see @xcite ) .
consider double wall nanotube having two atomic layers .
suppose the inner layer has 18 and outer layer has 20 atoms which are shown in fig.[ftubed],_b _ by points .
natural length measure here is the interatomic distance .
then the length of a circle has a jump when one goes from inner to outer layer . in the geometric theory of defects , it means that the metric component @xmath12 is not continuous in cylindrical coordinates .
the corresponding model will be described below .
to find radii @xmath13 , @xmath11 , and @xmath14 we have to solve the classical elasticity problem .
let us define the displacement vector field by @xmath15 , @xmath16 , @xmath17 where @xmath18 and @xmath19 are coordinates of a point before and after deformation respectively .
we consider the displacement field as a vector function on points of media after deformation and gluing .
this is more adequate because the resulting media after gluing is a connected manifold ( before the gluing procedure , each tube represents a connected component ) . in equilibrium state ,
the vector displacement field satisfies equation @xmath20 where @xmath21 is the poisson ratio and @xmath22 is the laplacian . for convenience , we consider components of the displacement vector field with respect to the orthonormal basis @xmath23 where @xmath24 we denote indices with respect to the orthonormal basis by hat : @xmath25 the latin indices referred to an orthonormal basis are rased and lowered by kronecker symbol : @xmath26 .
the divergence and laplacian have the following form in cylindrical coordinates @xmath27 from the symmetry of the problem , we deduce that only radial component of the displacement field differs from zero , and it does not depend on the angle @xmath28 and @xmath6 coordinates : @xmath29 equation ( [ qbhfyk ] ) for zero @xmath30 and @xmath31 components are automatically satisfied .
it is easy to check that the radial derivative of the divergence , @xmath32 coincides with the laplacian @xmath33 therefore the radial component of eq.([qbhfyk ] ) takes the form @xmath34 a general solution of this equation depends on two integration constants : @xmath35 note that the equilibrium equation ( [ qbncgd ] ) does not depend on the poisson ratio @xmath21 .
this means that the cylindrical dislocation is the geometrical defect .
boundary conditions have to be imposed to fix the integration constants .
let us introduce notation for inner and outer tubes : @xmath36 now boundary conditions are to be imposed .
we assume that the surface of two wall nanotube is free , i.e. the deformation tensor is zero on the boundary : @xmath37 we assume also that the media is in equilibrium .
it means that elastic forces on the gluing surface must be equal to zero : @xmath38 each of boundary conditions ( [ qnbdtg ] ) defines one integration constant for internal and external solutions : @xmath39 signs of the integration constants @xmath40 and @xmath41 are chosen in such a way that displacement vector is positive and negative for inner and outer tubes respectively .
this in agreement with the imposed problem . substitution of obtained solutions ( [ qbndtr ] ) into the gluing condition ( [ qbdjii ] ) defines the ratio of integration constants : @xmath42 the entirety condition for the media is @xmath43 these equations allow to find the distance between initial tubes which characterize the cylindrical dislocation : @xmath44 where expression for @xmath41 ( [ qbkkos ] ) is used .
afterwards we find the integration constants : @xmath45 thus we find the displacement vector field @xmath46 for double wall tube where constants @xmath40 and @xmath41 are given by eqs.([qnhydh ] ) .
qualitative behaviour of this vector field is shown in fig.[ftubeu],_a_. differentiation of this vector field in domains @xmath47 , @xmath48 and its extension to the point @xmath11 by continuity yields the function @xmath49 which is depicted in fig.[ftubeu],_b_. qualitative behaviour of the radial component of the displacement vector field for the double wall tube ( _ a _ ) .
the derivative of the displacement vector field ( _ b_).,title="fig : " ] qualitative behaviour of the radial component of the displacement vector field for the double wall tube ( _ a _ ) .
the derivative of the displacement vector field ( _ b_).,title="fig : " ] note that media entirety condition ( [ qfaelk ] ) leads to the jump of the displacement vector field at the point @xmath11 : @xmath50 since vector field @xmath51 has discontinuity at the point @xmath11 , the formal derivative of @xmath51 contains @xmath52 .
this @xmath53-function is thrown away in the geometric theory of defects @xcite .
double wall tube with cylindrical dislocation is parameterized by four constants @xmath54 , @xmath55 , @xmath56 , @xmath57 or @xmath13 , @xmath11 , @xmath14 , @xmath58 .
formulae ( [ qnhydh ] ) define constants @xmath40 , @xmath41 and , consequently , the displacement vector field through the second set of parameters .
it follows from definition ( [ qcasew ] ) that there is one - to - one correspondence between two sets of parameters .
now we calculate the metric induced in double wall tube . by definition
, it has the form @xmath59 where @xmath60 is the euclidean metric in cylindrical coordinates . the relation between coordinates before and after defect creation
is given by equality ( [ qcasew ] ) .
using explicit form of displacement vector field ( [ qbvxgt ] ) we find the induced metric in the double wall tube @xmath61 the component @xmath62 of this metric is continuous function but its derivative has a jump at @xmath11 .
the component @xmath63 is discontinuous at the point @xmath64 .
the volum element for metric ( [ qbcnct ] ) is @xmath65 the right hand side of this relation is positive because both multipliers are positive .
the second multiplier @xmath66 is positive by construction .
the first multiplier is also positive .
indeed , the function @xmath67 has the maximum at @xmath64 . at this point
, the following inequality holds @xmath68 where expression ( [ qfaelk ] ) for @xmath40 is used .
the circumference is a geometric invariant .
it is equal to @xmath69 for metric ( [ qbnctf ] ) .
when we go from the inner tube to the outer one it has the jump @xmath70 where @xmath58 is the distance between tubes before the dislocation is made .
this observation agrees with the continuous model of double wall tube .
here we consider torsional waves in the double wall tube with cylindrical dislocation described in the previous section .
we denote the displacement vector field by the new letter @xmath71 because the letter @xmath51 was used in the previous section for the displacements corresponding to the defect creation .
the total displacement vector field is equal to the sum @xmath72 where @xmath51 corresponds to the defect creation and @xmath71 describes oscillations in the double wall tube with cylindrical dislocation . by definition ,
the displacement vector field @xmath71 satisfies the wave equation @xmath73 where the covariant derivative @xmath74 and laplace beltrami operator @xmath75 are defined by metric ( [ qbcnct ] ) of double wall tube .
this equation is covariant with respect to changing of coordinate systems .
it can be solved in cylindrical coordinates @xmath76 after the defect creation with metric ( [ qbcnct ] ) .
however it is easier to follow another way .
we solve the wave equation in cylindrical coordinates @xmath77 where @xmath78 denotes the old radial coordinate before the defect creation @xmath79 and afterwards impose necessary boundary conditions .
it is easier because the metric is euclidean in the initial coordinate system .
let us consider torsional waves
. in this case only angular component of the displacement vector field differs from zero : @xmath80 from symmetry consideration , the angular component @xmath81 does not depend on @xmath28 and @xmath6 . for this vector field
the @xmath82 and @xmath83 components of eq.([qnbhfy ] ) are automatically satisfied .
it follows that the dilation for torsional waves is equal to zero @xmath84 and consequently torsional waves take place without media compression .
we look for solution of eq.([qnbhfy ] ) in the plain wave form @xmath85,\ ] ] where @xmath86 is the amplitude , @xmath87 is the wave vector , and @xmath88 is the frequency of the wave .
then wave equation ( [ qnbhfy ] ) in cylindrical coordinates reduces to the bessel equation @xmath89 where @xmath90 a general solution of this equation depends on two integration constants .
therefore general solutions for inner and outer tubes are @xmath91 where @xmath92 is the bessel function of the first kind and first order , @xmath93 is the neumann function of the first order ( see , i.e. @xcite ) , and @xmath94 are integration constants . to find the integration constants we impose boundary conditions .
the boundary surfaces are assumed to be free , i.e. the deformation tensor on the boundary must be zero @xmath95 because @xmath96 and @xmath97 , these equalities in the initial coordinates take the form @xmath98 where @xmath99 and prime denotes differentiation with respect to the argument @xmath6 .
these equalities define two integration constants : @xmath100 on the gluing surface , we impose two conditions : entirety and equality of stresses , @xmath101 as a result we get two equations @xmath102-c_3\big[j_1(z_2)-k_3n_1(z_2)\big]=0 , \\ c_1\big[j'_1(z_1)-k_0n'_1(z_1)\big]-c_3\big[j'_1(z_2)-k_3n'_1(z_2)\big]=0 . \end{split}\ ] ] the necessary and sufficient condition for this system to have a nontrivial solution is the equality of its determinant to zero : @xmath103\big[j'_1(z_2)-k_3n'_1(z_2)\big]- \\ -\big[j_1(z_2)-k_3n_1(z_2)\big]\big[j'_1(z_1)-k_0n'_1(z_1)\big]=0.\end{gathered}\ ] ] for given parameters of the double wall tube @xmath54 , @xmath55 , @xmath56 @xmath57 , the obtained relation is the equation for the constant @xmath104 . let @xmath104 be a root of eq .
( [ qbsioo ] ) , then the equality @xmath105 defines the dispersion relation .
the phase velocity of torsional waves @xmath106 is easily found from dispersion relation ( [ qnjuxp ] ) : @xmath107 the group velocity is also easily found @xmath108 we see that the phase velocity of torsional waves is always greater then the velocity of transverse waves , and group velocity is smaller .
dispersion relation ( [ qnjuxp ] ) does depend on double wall parameters through eq.([qbsioo ] ) .
we found the induced metric in double wall tube with cylindrical dislocation in the framework of the geometrical theory of defect .
though components of this metric are not continuous functions , the three dimensional einstein equations are well defined ( @xcite ) .
afterwards propagation of torsional waves in double wall tube is described .
the presence of the cylindrical dislocation inside the double wall tube leads to changing of the dispersion relation .
i. a. kunin and b. i. kunin .
gauge theories in mechanics . in _ trends in application of pure mathematics to mechanics .
lecture notes in physics , v.249 .
_ , pages 246249 , berlin heidelberg , 1986
. springer
verlag .
g. de berredo - peixoto , m. o. katanaev , e. konstantinova , and i. l. shapiro .
schrdinder equation in the space with cylindrical geometric defect and possible application to multi - wall nanotubes .
, 125b(8):915931 , 2010 . arxiv:1010.2913[cond - mat.mes - hall ] .
|
we describe the double wall tube with cylindrical dislocation in the framework of the geometric theory of defects .
the induced metric is found .
the dispersion relation is obtained for the propagation of torsional elastic waves in the double wall tube .
|
the notion of supernova fallback , where some of the ejecta from a core - collapse supernova ends up captured by the newly formed neutron star @xcite and may have sufficient angular momentum to form a disk @xcite , has been a general prediction of supernova models ( e.g. , * ? ? ?
fallback can have profound effects on the final state of the neutron star , forming disks that can give rise to planets @xcite , and possibly even causing the neutron star to collapse into a black hole .
fallback disks should manifest as a thermal infrared excess @xcite , but previous searches for such excesses around neutron stars were unsuccessful ( * ? ? ? * and references therein ) .
several years ago , though , we discovered the mid - infrared ( mid - ir ; here , @xmath6 and @xmath7 ) counterpart to the magnetar 4u 0142 + 61 ( * ? ? ? * hereafter ) .
the combined optical / ir spectrum of this magnetar suggests that the optical and ir data arise from two different spectral components . while the optical component is demonstrably of magnetospheric origin
@xcite , we showed that the ir component may arise from a passive ( i.e. , not accreting , but see @xcite ) , dust disk irradiated by x - rays from the magnetar (;
@xcite ) . the inferred spectral shape , while not well constrained , is remarkably similar to those of protoplanetary disks around young stars @xcite . the disk s survival lifetime ( @xmath8 yr ) significantly exceeds the pulsar s spin - down age ( @xmath9 yr ) , consistent with a supernova fallback origin . this dust disk model , while intriguing , is not definitive .
in particular , we have yet to firmly establish whether the mid - ir flux in 4u 0142 + 61 comes from a disk or from the magnetosphere ( via any of a number of mechanisms , e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) .
variability at a single wavelength or across wavelengths can be a powerful discriminant , depending on the timescale .
this was illustrated by @xcite , who found that the pulsed fraction in the optical exceeded that in soft x - rays , making it impossible for the optical to result from reprocessed x - rays ( also see @xcite ) . in the infrared the situation
is less clear .
the significant variability seen by @xcite at @xmath10 with no related change in soft x - rays provides a stiff test for a disk model , but is not definitive ( see @xcite ) . the unknown behavior at higher energies , the complicated behavior across the optical / near - infrared ( near - ir ) , and the lack of pulsations at @xmath10 ( @xmath11% pulsed at 90% confidence ; * ? ? ?
* ) all complicate the matter . as we attempt to determine the origin of the mid - ir emission , we must equally attempt to understand how ubiquitous mid - ir counterparts to young neutron stars are .
suggestively , all four of the magnetars with confirmed quiescent near - ir counterparts have the same near - ir / x - ray flux ratio of @xmath12 @xcite . in the disk scenario
, this ratio should be determined largely by geometry , so mid - ir observations of other magnetars down to this fractional level may help establish whether disks are present , but similar behavior might also be expected from magnetospheric emission . shallow mid - ir searches of three other magnetars were not very constraining @xcite , but deep a @xmath0 upper limit for 1e 1048.1@xmath135937 following an x - ray flare concluded that it did not have a mid - ir counterpart similar to that of 4u 0142 + 61 despite its similar near - ir / x - ray flux ratio , thus casting some doubt on the disk interpretation @xcite . here , we present the results of a deep search with the _ spitzer space telescope _ for a mid - ir counterpart to the magnetar 1e 2259 + 586 . was identified as an x - ray point source with coherent @xmath14s pulsations in the center of the supernova remnant ( snr ) ctb 109 @xcite .
the association leads to a distance of @xmath15 kpc based on interactions between the snr and regions with measured distances @xcite , although @xcite determined a distance to 1e 2259 + 586 of @xmath16kpc using the `` red clump '' method .
@xcite found the column density to 1e 2259 + 586 to be @xmath17 , corresponding to a visual extinction of @xmath18mag , from fitting of x - ray absorption edges ( consistent with earlier measurements ; @xcite ) .
the near - ir ( @xmath19-band , or @xmath10 ) counterpart to 1e 2259 + 586 was identified by @xcite .
uniquely among magnetars , the near - ir flux was observed to vary in concert with the x - ray flux , with both declining following a major series of x - ray bursts @xcite .
this led to the suggestion of x - ray heating of a disk ( as in * ? ? ? * for a low - mass x - ray binary ) , making this source an especially attractive target for mid - ir searches .
our were observations of 1e 2259 + 586 with the infrared array camera ( irac ; @xcite ) onboard the _
spitzer space telescope _
@xcite on 2007 august 8 .
we observed 1e 2259 + 586 using two of the four possible irac channels : @xmath0 and @xmath7 ( irac channels 2 and 4 ) .
the observation consisted of 50 dithered exposures of @xmath20s for a total integration of @xmath21minutes .
we started with the basic calibrated data ( bcd , from pipeline version s16.1.0 ) , discarding the first exposure at each wavelength as recommended .
we then processed the bcd images though an artifact mitigation pipeline ( 2005 october 13 version ) . with the ` mopex ` ( mosaicker and point source extractor , ver.16.3.7 ) pipeline , we mosaiced the individual exposures together by interpolating the data onto a common grid and rejecting radiation hits .
the final images are shown in figure [ fig : images ] .
our analysis requires precise relative astrometry , and the angular resolution of the irac images results in many blended objects . to aid in interpreting our mid - ir data and in particular to serve as an improved astrometric reference
, we analyzed an archival near - ir ( @xmath22-band , or @xmath10 ) observation taken with the near - infrared imager ( niri ; @xcite ) on the 8 m gemini north telescope at mauna kea , hawaii .
the data consist of 50 exposures with the @xmath23/6 camera taken on 2003 may 27 ( and previously published by @xcite ) , each with @xmath24s integrations , for a total exposure of @xmath25minutes .
we used an ` iraf ` package available from the niri web site to reduce the data .
we proceeded through the steps of this package , flatfielding the data , subtracting the sky , shifting the images , and adding them together .
we referenced the astrometry of the final image to the two micron all sky survey ( 2mass ; @xcite ) , finding 57 stars that were not saturated or badly blended , and getting rms residuals of @xmath26 in each coordinate .
this image is also shown in figure [ fig : images ] .
we searched the irac images in figure [ fig : images ] for a counterpart to 1e 2259 + 586 by looking for a point source at the position corresponding to star 1 ( which is 1e 2259 + 586 ; @xcite ) in the niri image . at this position
, we see a faint point source in the irac @xmath0 image ( hereafter the `` irac source '' )
. however , one must be careful , as the position of star 2 only differs from that of the magnetar by @xmath27 ; this is @xmath281 irac pixel width , and only half of the angular resolution of the image ( @xmath29 fwhm ) .
we performed photometry on the @xmath0 irac image to find the positions and flux densities of all of the point sources .
we used the ` apex ` ( astronomical point source extraction , part of ` mopex ` ) software to identify and perform point - response function ( prf ) fitting for the photometry using the routine ` apex_1frame.pl ` with the _
spitzer_-supplied prf .
the prf is the traditional point - spread function ( psf ) convolved with the pixel - response function to determine how a source actually appears in the data .
part of this analysis identified blended objects using simultaneous prf fitting , but the irac counterpart was consistent with a single point source ( the @xmath30 per degree of freedom for the prf fit was @xmath31 , consistent with other point sources in the field which have reduced @xmath30 of 0.13 ) with flux density @xmath1 at @xmath0 .
the uncertainty here includes the standard term from the prf fitting but is dominated by substantial contributions from difficulty in robustly measuring the background and identifying which pixels make up the source . at @xmath7
there is no detection , and we estimate a limit of @xmath32 ( 95% confidence , limited largely by the varying diffuse background ) based on the ` apex ` detections in that region . in determining whether the irac source is indeed the counterpart to star 1 , we considered three issues .
first , we were concerned with whether the position of the irac source is consistent with that of star 1 or 2 .
we took the astrometry ( using both ` apex ` and standard centroiding ; both gave consistent results ) of the eight reference sources ( some of which are labeled in figure [ fig : images ] ) , along with the niri astrometry of those objects and stars 1 and 2 .
we used the reference stars to refine the astrometry of the irac image relative to the niri image ( we did not use a proper reference frame as an intermediary , as this would have introduced additional uncertainties ) . fitting only for an offset (
the position angles of both observations were referenced independently to 2mass ) , we found a small shift ( @xmath33 ) with an rms of @xmath34 . with this shift ,
the position of the irac source differs from that of star 1 by @xmath35 , and from that of star 2 by @xmath36 ( see figure [ fig : astrom ] ) .
there is an intrinsic centroiding uncertainty for the irac source ( the potential counterpart ) of @xmath37 in each coordinate ( 90% confidence radius of @xmath38 ) .
given this , the irac source is largely consistent with the position of star 1 ( the probability of chance alignment is 78% ) and is inconsistent with the position of star 2 ( the probability of chance alignment is 99.998% ) .
the situation is also illustrated in the bottom right panel of figure [ fig : images ] , which shows the extent of the prf of the irac counterpart , the measured centroid , as well as the positions of stars 1 and 2 in the same reference frame .
second , we compared the ir colors of the irac source to all of the other objects in the field .
we show a color - color diagram in figure [ fig : cc ] . for the irac source
, we used the near - ir ( @xmath39- and @xmath22-band ) photometry of star 1 given by @xcite ; for star 2 , we also used the photometry from @xcite ; for the field stars , we used 2mass for the near - ir . assuming ,
as deduced from the above astrometry , that the irac source is associated with star 1 and not with star 2 , we see that star 2 is consistent with the bulk of the field population , i.e. , reddened main - sequence stars . on the other hand ,
the irac source is significantly redder in both @xmath40 and @xmath41 $ ] and clearly stands out from the population .
it is quite close in colors to 4u 0142 + 61 , for the same extinction .
uncertainties in the visual extinctions of both 1e 2259 + 586 and 4u 0142 + 61 amount to @xmath42mag @xcite and do not change the general agreement because of the small extinctions at near- and mid - ir bands ( @xmath43 , @xmath44 , @xmath45 ; also see figure [ fig : sed ] ) ) .
finally , we examined the spectral energy distributions ( seds ) of the irac source and star 2 ( figure [ fig : sed ] ) .
first , we estimated the spectral type of star 2 . with only two data points ( @xmath46 , @xmath47 )
we could not discriminate between later stellar type and higher reddening ( the @xmath48- and @xmath49-band upper limits from @xcite were not constraining ) , but we find reasonable solutions for k / m dwarf stars with a few magnitudes of extinction and distances of a few kpc ( based on @xcite ) . for stars of those spectral types , we would expect the sed to peak in the near - ir region and decline in the irac bands , with typical flux densities of @xmath50 expected .
even brown dwarfs , which can be redder in @xmath41 $ ] , have @xmath51 @xcite
. there are classes of objects such as young stellar objects or massive stars with dusty winds that could produce such red colors ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , but they are relatively rare ( although this is a complicated region with many objects along the line of sight ; @xcite ) , often brighter than star 2 ( spectral types earlier than about f0v would have to be well outside the galaxy to have the observed @xmath19 magnitude ) , and would likely have been identified at this position in other wavelengths ( x - ray , near - ir , molecular line , or continuum radio , etc . ; see @xcite ) but were not .
in contrast , the @xmath0 flux density of the irac source is a factor of 3 higher than the @xmath19-band flux densities of star 1 or 2 , and an order of magnitude higher than the expectation for a normal star at @xmath0 . taken together with the astrometry ,
this is strong support for not associating the irac source with star 2 , but instead with star 1 .
it is extremely unlikely that the irac source is associated with neither object , as probability of chance coincidence with star 1 is only 0.3% , and this would also imply even more extreme colors ( figure [ fig : cc ] ) that would be clearly non - stellar .
even with several tens of magnitudes of extinction it would be difficult to move the irac source onto the stellar locus , and examining the niri image we see no other sources with @xmath52>2 $ ] . from @xcite , an extinction of only @xmath53mag is reached at @xmath54kpc , and the molecular gas maps of colden find a maximum extinction of @xmath55mag , so having very high extinction outside of very local extinctions seems unlikely .
we therefore conclude that the irac source is indeed the mid - ir counterpart to star 1/1e 2259 + 586 .
we have found a @xmath0 counterpart to the magnetar 1e 2259 + 586 , with flux density @xmath1 and a limit of @xmath32 at @xmath7 .
the absorption - corrected mid - ir / x - ray @xcite , although this is only over the 210@xmath56kev range and does not include softer emission which is uncertain due to the moderate absorption .
] flux ratio for this source is @xmath57 , very similar to the ratio for 4u 0142 + 61 , so the common x - ray to @xmath10 flux ratios @xcite seem to extend further into the infrared .
we note , though , that one must be careful with _ any _ model or interpretation for the ir emission since the @xmath10 flux density of 1e 2259 + 586 ( and of 4u 0142 + 61 for that matter ; @xcite ; @xcite ) is known to vary . the @xmath10 flux density we plot in figure [
fig : sed ] is the faintest measured and is assumed to be close to the baseline level @xcite and should not be affected by flares , but it was measured two years before the irac observations . for 1e 2259 + 586 ,
the ir variability had been seen along with x - ray flaring @xcite in the past . since then , though , no transient x - ray behavior for 1e 2259 + 586 was reported and the x - ray flux is now near the quiescent level ( although with a slightly different spectrum ; @xcite ) , so our assumption of a baseline flux seems safe , but x - ray monitoring is not very regular and there could potentially be ir variability independent of x - ray activity . with the limited data we have , a wide range of model fits are possible . here
we discuss two general categories : a power law and an irradiated disk model .
a simple power law ( @xmath58 ) can fit the two detected ir flux points and is also below the upper limits at @xmath59 and @xmath4 ( figure [ fig : sed ] ) .
such a power law spectrum could arise from the pulsar magnetosphere ( as for crab pulsar ; see * ? ? ?
* ) , which could also possibly produce correlated x - ray and infrared flux changes , although this has not been studied in detail .
however , unlike the crab pulsar , this power - law rises further into the mid - ir band , and it would be interesting to obtain longer - wavelength data on 1e 2259 + 586 to see if the sed keeps rising past @xmath0 , but source confusion makes this very difficult .
the rising spectrum is problematic for some detailed magnetospheric models ( like those of @xcite and @xcite ) , but a generic magnetospheric origin is certainly possible .
however , as with 4u 0142 + 61 , we can also fit the 1e 2259 + 586 data with an irradiated passive disk model ( also see @xcite ; other spectral shapes are also possible , but the fits are too unconstrained and without specific motivation we will not address them ) . without more data , we can not make conclusions about the presence or absence of a disk , and indeed other observations suggest that a disk interpretation is problematic @xcite .
however , it is still possible , and we find it worthwhile to expand on the discussion of and further explore the implications such a disk would have , noting in particular where 1e 2259 + 586 differs from 4u 0142 + 61 .
this model fit is qualitatively similar to what we obtained in for 4u 0142 + 61 , with a small inner radius and an outer radius a factor of a few to 10 larger .
the details of both fits depend on assumptions about the distances , extinctions , and inclinations , but the rough shapes of the seds going from the near- to mid - ir are similar , with a factor of 3 in flux density increase from @xmath60 to @xmath0 , after correcting for extinction ( unlike what @xcite found for 1e 1048.1@xmath135937 ) .
the inner radius is constrained by the need to have the @xmath19-band measurement lie below the @xmath0 measurement , while the outer radius is less constrained by the @xmath7 upper limit .
note that the nominal inner radius of @xmath61 ( for a distance of 3kpc and an inclination of @xmath62 ) is smaller than the light cylinder radius ( @xmath63 ) which would imply interaction between the disk and the magnetar s spin ( e.g. , * ? ? ?
* ) possibly contradicted by its observed steady spin - down @xcite .
this could either point to an inconsistency in our model , or simply that we must choose the other model parameters ( distance , inclination , etc . )
such that the inner radius is @xmath64 ; as an example , a larger distance such as that determined by @xcite has an inner radius more than twice the light cylinder radius .
the binding energy of the putative disk , following to estimate the disk mass , is @xmath65 for the upper end of the mass estimate .
this is large enough that we do not have to worry that such a disk would have been unbound by the large x - ray bursts ( a total fluence of @xmath66 , or total isotropic energy of @xmath67 ) found by @xcite , and the physical state and structure of the disk may not be altered dramatically , as this energy is equivalent to only @xmath68s of normal x - ray activity .
even a disk several order of magnitude lower in mass would remain bound . only a truly giant flare , such as those seen from several soft @xmath69-ray repeaters @xcite , would be energetic enough to disrupt a disk like this , although the details of that process are difficult ( also see @xcite ) , but we stress that our mass estimate is an upper limit , and at the low end of the allowed mass range the disk would not be bound .
again , this could indicate either a problem with the disk model , or that objects with disks can not have had large flares . among magnetars , the correlated near - ir and x - ray variability found by @xcite for 1e 2259
+ 586 is unique .
such a correlation is a natural ( but not necessarily unique ) consequence of the disk model : a change in the x - ray flux would produce an accompanying change in the infrared flux from the disk , and while the total reprocessed flux is a constant fraction of the x - ray flux , the ir flux at a given wavelength can change faster or slower as annuli of a given temperature move and change area ( cf .
we find that the @xmath10 flux increases seen by @xcite , which were of comparable amplitude to the x - ray flux increases , are roughly consistent with a disk like that considered here , although this assumes that 100% of the @xmath10 flux comes from the disk and ignores the possibility of physical changes to the disk such as movement of the inner radius following the flux changes .
however , correlated changes were not seen in 4u 0142 + 61 @xcite , who instead saw relatively rapid , significant changes at @xmath10 without any changes in x - rays .
any model must be able to accommodate this wide range in behavior ( see @xcite ) .
additional observations at longer wavelengths where the decomposition is less ambiguous could again provide important constraints .
we thank an anonymous referee for helpful comments .
this work is based on observations made with the _
spitzer space telescope _
, which is operated by the jet propulsion laboratory , california institute of technology under a contract with nasa .
support for this work was provided by nasa through an award issued by jpl / caltech .
support for d.l.k . was provided by nasa through hubble fellowship grant 01207.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 and by the national science foundation under grant phy05 - 51164 . guest user , canadian astronomy data centre , which is operated by the herzberg institute of astrophysics , national research council of canada . , z. , chakrabarty , d. , & kaplan , d. l. 2008b , in american institute of physics conference series , vol .
983 , 40 years of pulsars : millisecond pulsars , magnetars and more , ed .
c. g. bassa , z. wang , a. cumming , & v. m. kaspi ( new york : aip ) , 274
|
we report the discovery of a @xmath0 counterpart to the anomalous x - ray pulsar ( magnetar ) 1e 2259 + 586 with the _ spitzer space telescope_. the mid - infrared flux density is @xmath1 at @xmath0 and @xmath220@xmath3 ( at 95% confidence ) at @xmath4 , or @xmath5% of the 210kev x - ray flux ( corrected for extinction ) . combining our _ spitzer _ measurements with previously published near - infrared data
, we show that the overall infrared emission from 1e 2259 + 586 is qualitatively similar to that from the magnetar 4u 0142 + 61 .
therefore , the passive x - ray - heated dust disk model originally developed for 4u 0142 + 61 might also apply to 1e 2259 + 586 .
however , the ir data from this source can also be fitted by a simple power - law spectrum as might be expected from magnetospheric emission .
|
electron tunneling through a thin insulating layer ( a potential barrier ) in a metal - insulator - metal ( mim ) multilayered structure is one of the most fundamental research topics in condensed matter physics .
it also lies at the heart of numerous solid - state devices , such as tunnel diodes , @xcite coulomb blockade thermometers , @xcite josephson junctions , @xcite and memory elements based on magnetic tunnel junctions .
@xcite the functionality of a tunnel device relies heavily on the material properties of the intermediate thin insulating layer .
usually , a weak insulating - like temperature dependence of the zero - bias junction conductance , @xmath5_{v\,\rightarrow\,0}$ ] , as described by the simmons model @xcite is used to ascertain the quality and reliability of the insulating layer .
@xcite in the absence of any pinholes in the insulating layer [ see the schematic diagram depicted in fig .
[ fig_1](a ) ] , the simmons model predicts a quadratic temperature dependent @xmath6 law from low temperatures up to room temperature . in practice ,
a precise control of the material properties of the thin insulating layer during the junction fabrication process is extremely difficult .
often , the junction yield varies sensitively with the oxide layer thickness .
@xcite furthermore , the thickness distribution of the oxide layer may be substantial even for those junctions grown under similar conditions .
@xcite as such , the @xmath3 behavior might differ substantially from the simmons law in certain cases .
@xcite the key reason to explain these large variations in the mim quality can be ascribed to the common formation of junction - barrier interfacial roughness .
@xcite because the electron transmission probability increases exponentially with decreasing barrier width , the presence of any notable interfacial roughness thus can significantly affect the overall junction transport properties . in an mim junction with marked interfacial roughness as schematically depicted in fig .
[ fig_1](b ) , the sharp points of the closest approaches ( called hot spots " hereafter @xcite ) between the two conducting regimes will no doubt play a predominant role in the electron tunneling .
[ we note that a hot spot is a precursor of a fully connected
pinhole , " see fig .
[ fig_1](c ) . ]
the current flowing through the hot spots between two large metal grains is theoretically found to depend sensitively not only on the applied bias voltage @xmath7 but also on the thermal voltage @xmath8 across the hot spots , where @xmath9 , with @xmath10 being the boltzmann constant , and @xmath11 being the capacitance of the hot spots .
@xcite ( the plus / minus sign in @xmath8 stresses the fact that @xmath8 can be in the same / opposite direction to @xmath7 . ) owing to the smallness of the hot spot , the magnitude of @xmath11 is tiny and thus @xmath8 is notable .
the resulting effect of the total voltage @xmath12 gives rise to the so - called thermally fluctuation - induced tunneling conduction ( fitc ) process through the junction .
the fitc model predicts unique functional forms for the zero - bias @xmath4 versus @xmath13 dependence [ eq .
( [ fitc_rt ] ) ] as well as finite - bias current - voltage ( @xmath14-@xmath15 ) characteristics [ eq . ( [ fitc_iv ] ) ] over a wide range of @xmath13 from liquid - helium temperatures to room temperature .
while this phenomenological fitc model has reasonably well described the @xmath3 behavior and @xmath14-@xmath15 curves in a good number of experiments , such as nanowire contacts , @xcite the extracted barrier height is often too small ( e.g. , @xmath16 few mev ) and barrier width too large ( e.g. , @xmath17 few tens nm ) . in order to resolve this puzzle ,
xie and sheng have recently retreated this problem by microscopically calculating the electronic wave transmission through atomic - scale fine structures ( short nanoconstrictions " ) connecting two large metallic grains .
@xcite in particular , the relevant conduction channels they theoretically considered are those with a transverse dimension slightly smaller than the half of the fermi wavelength of the tunneling electron . by employing the landauer formula , they obtained the @xmath3 attributes and @xmath14-@xmath15 characteristics very similar to the original phenomenological fitc predictions .
in other words , short nanoconstrictions with a transverse dimension smaller than the cutoff wavelength ( of a waveguide ) can host thermally fluctuation induced tunneling conduction process at finite temperatures .
@xcite the extracted potential barrier height and width according to this microscopic model are predicted to be much more realistic .
( a)(c ) schematic diagrams depicting three types of insulating barrier : ( a ) a comparatively uniform oxide barrier , ( b ) an oxide barrier with two hot spots ( close approaches ) but no pinholes , and ( c ) an oxide barrier with two pinholes .
the red dashed or solid curves in each case schematically depicts the electron transport process .
( d ) schematic diagram showing the four - probe measurement method on a tunnel junction . ] on the experimental side , it is surprising that the possible occurrence of the fitc - like mechanism in mim multilayered structures has never been reported , even though the formation of hot spots is very likely in certain junctions . in this paper , we present our observations of the fitc process governed @xmath3 features and @xmath14-@xmath15 characteristics in micrometer - sized al / alo@xmath0/y tunnel junctions in a wide temperature range 1.5300 k. the extracted parameters are examined in terms of the presence of hot spots in the alo@xmath0 layer , which mimics the short nanoconstrictions in the microscopic fitc model .
our results are presented below .
the fabrication method of our planar tunnel junctions was described previously .
@xcite the designed geometrical junction area @xmath18 ( @xmath1 20@xmath1920 @xmath20m@xmath21 ) was defined by the electron - beam lithography technique . both the bottom al ( @xmath1 30 nm thickness ) and the top y ( @xmath1 100 nm thickness ) electrodes were deposited by thermal evaporation . prior to the deposition of the top y electrode , an insulating alo@xmath0 layer ( @xmath1 1.52 nm thickness on average )
was grown on the top surface of the al base electrode by the o@xmath2 glow discharge .
@xcite low angle x - ray diffraction studies revealed an amorphous structure of the oxide layer .
the temperature dependent zero - bias resistance @xmath22 and the @xmath14-@xmath15 curves of the junctions were measured by the four - probe method [ fig .
[ fig_1](d ) ] , using a keithley k220 current source and a k182 voltmeter .
a standard @xmath23he cryostat equipped with a calibrated si diode thermometer was utilized for @xmath24 and @xmath14-@xmath15 curve measurements between 1.5 and 300 k. the details of our sample configuration and measurement method had previously been discussed in ref . .
we first discern the three possible types of junction - barrier interfacial roughness that can emerge in a fabricated mim tunnel junction .
figure [ fig_1](a ) depicts the normal case where the interfacial roughness is moderate .
electron tunneling occurs more or less evenly over the entire geometrical area @xmath18 , and the @xmath13 dependence of the junction conductance @xmath4 obeys the well - established simmons law . @xcite in the opposite case [ fig .
[ fig_1](c ) ] , pinholes may exist in the barrier , shorting the two electrodes , and cause overall metallic features of @xmath3 .
what is more interesting and will be the focus of this paper is the intermediate case [ fig .
[ fig_1](b ) ] , where the interfacial roughness is notable but pinholes are just about to fully develop .
electron tunneling then occurs predominantly at the sharp points of the closest approaches of the top and bottom electrodes .
experimentally , the locations of such hot spots ( incomplete pinholes ) may be identified by mapping the tunnel current over the lateral area @xmath18 , e.g. , from conducting atomic force microscopy study at room temperature .
@xcite alternatively , the occurrence of such hot spots can substantiate the fitc process in a wide @xmath13 range from liquid - helium temperatures to room temperature , @xcite as we demonstrate below in micrometer - sized al / alo@xmath0/y tunnel junctions .
according to the fitc model , @xcite the small effective junction area @xmath25
[ i.e. , the size of a hot spot depicted in fig . [ fig_1](b ) , where @xmath26 can lead to large random thermal voltages @xmath9 that fluctuate across the narrow gap , which would in turn effectively lower and narrow the shape of the potential barrier . as a consequence , electron tunneling
would be significantly influenced , causing @xmath3 to increase exponentially with increasing @xmath13 .
( in the opposite limit of @xmath27 , @xmath4 becomes constant , i.e. , recovering the elastic tunneling behavior , due to the gradually vanishing @xmath8 with @xmath13 . ) previously , the fitc model has been applied to explain the @xmath3 behavior observed in a good number of conductor - dielectric composite systems ( granular films ) @xcite and nanowire contacts .
@xcite surprisingly , although the interfacial roughness must be very common in artificially fabricated mim tunnel junctions , @xcite the possible manifestation of the fitc process through hot spots in the oxide layer has never been reported in the literature . in this work ,
a dozen of al / alo@xmath0/y junctions have been fabricated and studied . in the following ,
we report the electrical - transport properties of four representative junctions .
two of them show the standard simmons behavior , while the other two manifest the fitc mechanism .
figure [ fig_2](a ) displays the @xmath13 dependence of @xmath28 for the four junctions a to d , as indicated .
in the two higher resistance junctions a and b , @xmath28 increases only slightly with decreasing @xmath13 and it progressively saturates at @xmath13@xmath16100 k. in sharp contrast , in the two lower resistance junctions c and d , @xmath28 increases exponentially with decreasing @xmath13 .
the relative resistance ratios [ @xmath29@xmath30@xmath31/@xmath32 in the junctions c and d are larger than those in the junctions a and b. ( a ) zero - bias resistance @xmath28 as a function of @xmath13 and ( b ) the corresponding conductance @xmath4 as a function of @xmath33 for the junctions a to d , as indicated .
the straight lines drawn through junctions a and b in the top panel of ( b ) are least - squares fits to eq .
( [ simmons ] ) . ] in the simmons model , the zero - bias conductance , @xmath34 , due to direct ( elastic ) tunneling of electrons through a potential barrier possesses a @xmath35 temperature dependence as given by @xcite @xmath36~,% \label{simmons}%\ ] ] where @xmath37 and @xmath38 are temperature independent parameters . as shown in the top panel of fig .
[ fig_2](b ) , the @xmath39 law clearly holds for the junctions a and b in the wide temperature range 1.5300 k. on the other hand , the bottom panel of fig .
[ fig_2](b ) unambiguously indicates that the conductances of the junctions c and d significantly deviate from the simmons law . in the phenomenological fitc theory , at small bias voltages ( i.e. , the ohmic @xmath14-@xmath15 regime ) ,
the @xmath24 of a small junction is described by @xcite @xmath40 where the parameter @xmath41 depends only weakly on @xmath13 , and the characteristic temperatures @xmath42 and @xmath43 where @xmath44 is the barrier height , @xmath45 is the barrier width , @xmath46 is the permittivity of vacuum , @xmath47 is the dielectric constant of the insulating barrier ( for al@xmath2o@xmath48 , @xmath49 4.58.4 , see ref . ) , @xmath50 is the planck s constant , and @xmath51 is the electronic mass .
the characteristic thermal energy @xmath52 can be regarded as a measure of the energy required for an electron to cross the barrier , and @xmath53 is the temperature well below which fluctuation effects become insignificant .
( a ) log - log plot of @xmath28 versus @xmath13 for junctions c and d , as indicated . the solid curves are least - squares fits to eq .
( [ fitc_rt ] ) .
the dashed curve through junction c is the theoretical prediction of eq .
( [ fitc_rt ] ) but is plotted by directly substituting the @xmath54 and @xmath53 values extracted from @xmath55 in eq .
( [ fitc_iv ] ) .
( b ) log(@xmath56 ) versus @xmath57 for junctions c and d between 25 and 300 k. note that there exists no linear regime in any temperature regime . ] figure [ fig_3](a ) shows a log - log plot of the measured @xmath28 as a function of @xmath13 for the junctions c and d. the symbols are the experimental data and the solid curves are the least - squares fits to eq .
( [ fitc_rt ] ) .
the fitted @xmath41 , @xmath54 , and @xmath53 values are listed in table [ table_1 ] .
the predictions of eq .
( [ fitc_rt ] ) are seen to well describe the data .
( the dashed curve drawn through the junction c will be discussed below . )
.[table_1]parameters for junctions c and d. the values of @xmath45 and @xmath44 were extracted by using the geometrical area @xmath18 ( @xmath58 @xmath59 @xmath20m@xmath21 ) , and the dielectric constant @xmath47(al@xmath2o@xmath48 ) @xmath1 6 . [ cols="^,^,^,^,^,^,^ " , ] it might be conjectured that , if @xmath44 is small , the exponential increase of @xmath28 with decreasing @xmath13 in the junctions c and d could arise from the thermionic emission of electrons over the barrier .
@xcite if this were the case , one would expect the total conductance to be given by the sum of the simmons conductance and the thermionic - emission conductance @xmath60 , i.e. , @xmath3@xmath61@xmath34@xmath62@xmath60 , where @xmath63 , with @xmath64 and @xmath65 being two temperature independent parameters . at low temperatures ,
@xmath34 should dominate over @xmath60 . as @xmath13 increases
, @xmath60 would become progressively important and might eventually dominate over @xmath34 .
figure [ fig_3](b ) shows a plot of log@xmath66 versus @xmath57 for the junctions c and d. clearly , there does not exist any linear regime between 25 and 300 k. therefore , the thermionic - emission conduction process can not account for our measured @xmath4 versus @xmath13 dependence . figures [ fig_4](a ) shows the measured nonlinear @xmath14-@xmath15 curves for the junction c at four @xmath13 values .
the symbols are the experimental data and the solid curves are the least - squares fits to eq .
( [ fitc_iv ] ) .
a good agreement between the theory and experiment is evident .
our extracted values of @xmath74 as a function of @xmath13 are plotted in fig .
[ fig_4](b ) , together with the least - squares fits ( solid curves ) to the expression @xmath73 .
the @xmath54 and @xmath53 values thus inferred are listed in table [ table_1 ] .
the insets of fig [ fig_4](b ) plot the variations of our fitted @xmath69 and @xmath70 values with @xmath13 . note that they are essentially temperature independent , as predicted by the fitc theory .
current - voltage characteristics of junction c : ( a ) nonlinear @xmath14-@xmath15 curves at four @xmath13 values .
the solid curves are least - squares fits to eq .
( [ fitc_iv ] ) . for clarity ,
the data for 15 , 30 , and 45 k have been shifted up by multiplying factors of 1.2 , 1.4 , and 1.6 , respectively .
inset : the @xmath14-@xmath15 curve at 1.5 k in double - linear scales .
( b ) the parameter @xmath74 in eq .
( [ fitc_iv ] ) as a function of @xmath13 .
the solid curve is the least - squares fits to @xmath54/(@xmath53+@xmath13 ) .
the dashed curve is the theoretical prediction of @xmath54/(@xmath53+@xmath13 ) but is plotted by directly substituting the @xmath54 and @xmath53 values extracted from the @xmath24 fit to eq .
( [ fitc_rt ] ) .
insets : ( top ) variation of @xmath69 and ( bottom ) variation of @xmath70 with @xmath13 . ]
we now substitute those @xmath54 and @xmath53 values extracted from the @xmath24 fit [ eq .
( [ fitc_rt ] ) ] into the expression @xmath73 to compute @xmath55 for the junction c. the calculated result ( dashed curve ) is plotted in fig .
[ fig_4](b ) , which is seen to describe the @xmath55 behavior reasonably well . in turn , in fig .
[ fig_3](a ) we plot the prediction of eq .
( [ fitc_rt ] ) by directly substituting those @xmath54 and @xmath53 values inferred from the @xmath55 fit through eq .
( [ fitc_iv ] ) ( while allowing @xmath41 as the sole adjustable parameter , whose value is listed in table [ table_1 ] ) .
the result ( dashed curve ) is clearly in excellent agreement with the experimental data .
numerically , our @xmath54 and @xmath53 values extracted from the two methods differ by an amount of @xmath110% , which is very satisfactory and notably smaller than those ( @xmath13040% ) reported in previous studies . @xcite
this close consistency provides a strong support for the pivotal relevance of the fitc mechanism occurring in our micrometer - sized tunnel junctions .
we would like to comment on our inferred values of the barrier parameters @xmath44 and @xmath45 .
inspection of table [ table_1 ] indicates that the @xmath44 values are relatively small ( @xmath16 0.05 mev ) , while the @xmath45 values are relatively large ( @xmath17 20 nm ) .
these seemingly nonphysical values can be reconciled by taking into account the possible formation of hot spots as those depicted in fig .
[ fig_1](b ) . since the hot - spot area @xmath25@xmath75@xmath18 , our using the geometrical area @xmath18 to extract @xmath44 and @xmath45 from the inferred @xmath54 and @xmath53 values through eqs .
( [ t1 ] ) and ( [ t0 ] ) hence led to huge underestimate ( overestimate ) of @xmath44 ( @xmath45 ) .
although the actual area @xmath25 of a hot spot can not be precisely known for a given tunnel junction , it has previously been pointed out that the typical size of a hot spot can be about a few nanometers .
@xcite then , taking @xmath25@xmath1 2@xmath192 nm@xmath76 , one obtains values of @xmath44 @xmath1 80 ( 10 ) mev and @xmath45 @xmath1 0.9 ( 0.6 ) nm for the junction c ( d ) .
these values are much acceptable .
in fact , the relevant hot - spot area could even be reduced to the sub - nanometer scale . in order to search for a microscopic insight and to obtain meaningful parameter values ,
xie and sheng @xcite have recently considered the electron tunneling through finite segment(s ) of nanoconstrictions whose transverse dimension is less than the half of the electronic fermi wavelength @xmath77 . in this new theoretical proposal ,
the insulating gap in the original fitc model @xcite is replaced by short and very narrow constrictions ( which behave like waveguides with transverse dimensions slightly shorter than the cutoff wavelength @xmath78 ) . by applying the landauer formula and taking into account the effects of
thermally induced voltage fluctuations across the nanoconstrictions , xie and sheng obtained a @xmath4 versus @xmath13 dependence very similar to the original fitc prediction [ eq . ( [ fitc_rt ] ) ] but with more realistic parameter values . following this new theoretical idea and taking @xmath25@xmath1 [email protected]
nm@xmath76 [ note that @xmath77(al ) @xmath1 0.36 nm ] , we obtain @xmath44 @xmath1 0.5 ev and @xmath45 @xmath1 0.4 nm for the junction c. these values are fairly realistic . therefore , the manifestation of the fitc mechanism in micrometer - sized tunnel junctions implies the existence of such fine structures ( hot spots or short nanoconstrictions ) in the oxide layer .
electron tunneling through these fine structures governs the overall @xmath3 behavior .
[ there could be a few such fine structures coexisting in a given mim junction .
however , the measured @xmath3 might be predominated by only one of them , owing to the exponential dependence of the electron transmissivity on the fine - structure geometries . ] in a series of ballistic electron emission microscopy ( beem ) and scanning tunneling spectroscopy ( sts ) studies , buhrman and coworkers @xcite have systematically shown that in an ultrathin ( @xmath16 1 nm ) and not fully oxidized alo@xmath79 layer ( which was not covered with a top electrode ) , there could possibly exist low - energy electron channels that could provide low - voltage leakage " through the barrier . in the present study ,
our alo@xmath79 layers are relatively thicker ( 1.52 nm ) and fully oxidized by the o@xmath80 glow discharge , as mentioned in sec .
ii . therefore ,
we think that the features of the fitc conduction process observed in our al / alo@xmath79/y junctions are not associated with the possible presence of low - energy electron channels that extended through the thin disordered oxide layer .
theoretically , the crossover from a tunneling regime to an incoherent thermal - activation regime in a correlated barrier has been investigated by freericks and coauthors .
@xcite they have shown generically that the crossover takes place around a characteristic temperature which approximately equals a generalized thouless energy for the barrier . under certain parameter values ,
the predicted temperature behavior of the junction resistance ( or , the specific resistance @xmath81 ) looks nominally similar to the temperature dependence shown in fig . [
fig_3](a ) .
unfortunately , the theory of freericks does not provide a functional form for comparison with experiment .
this issue deserves further studies .
we have observed the fluctuation - induced tunneling conduction mechanism in micrometer - sized al / alo@xmath0/y tunnel junctions in a wide temperature range 1.5300 k. the manifestation of the fitc process at the micrometer scale reflects the existence of large junction - barrier interfacial roughness in the thin oxide layer , where electron tunneling occurs at the sharp points of the closest approaches of the top and bottom electrodes .
the formation of a few hot spots , but not pinholes , predominantly governs the conductance versus temperature behavior as well as the current - voltage characteristics .
our results can have important bearing on the reliability and functionality of solid - state tunnel devices .
a. c. perrella , w. h. rippard , p. g. mather , m. j. plisch , and r. a. buhrman , phys .
b * 65 * , 201403(r ) ( 2002 ) ; e. tan , p. g. mather , a. c. perrella , j. c. read , and r. a. buhrman , phys . rev .
b * 71 * , 161401(r ) ( 2005 ) ; p. g. mather , a. c. perrella , e. tan , j. c. read , and r. a. buhrman , appl . phys
86 * , 242504 ( 2005 ) .
|
micrometer - sized al / alo@xmath0/y tunnel junctions were fabricated by the electron - beam lithography technique . the thin ( @xmath1 1.52 nm thickness ) insulating alo@xmath0 layer was grown on top of the al base electrode by o@xmath2 glow discharge .
the zero - bias conductances @xmath3 and the current - voltage characteristics of the junctions were measured in a wide temperature range 1.5300 k. in addition to the direct tunneling conduction mechanism observed in low-@xmath4 junctions , high-@xmath4 junctions reveal a distinct charge transport process which manifests the thermally fluctuation - induced tunneling conduction ( fitc ) through short nanoconstrictions .
we ascribe the experimental realization of the fitc mechanism to originating from the formations of hot spots " ( incomplete pinholes ) in the alo@xmath0 layer owing to large junction - barrier interfacial roughness .
|
the scalar field theory with a @xmath1 interaction has been widely studied for a variety of problems .
for instance , it underlies the physics of the anti - ferromagnetic phase transition in statistical physics and also is the starting point for the higgs mechanism of the standard model in particle physics . from another viewpoint
it has been used as a toy model in four dimensions to examine fundamental ideas in quantum field theory .
one such example is understanding perturbation theory at high orders .
( for a review of these points see , for example , @xcite . ) in particular the fundamental functions of the renormalization group like the @xmath7-function are known to five loops in the @xmath0 scheme , @xcite . in other field theories of interest in particle physics such as gauge theories
four loop results have only become available in the last few years .
for instance the four loop @xmath7-functions of qed and qcd , which describes the strong interactions , were computed in @xcite respectively .
such calculations are necessary in qcd primarily because it is asymptotically free and therefore it describes the physics of high energy collisions involving quarks .
as these deep inelastic experiments are becoming increasingly more accurate the theoretical input must be refined accordingly .
therefore current perturbative calculations in qcd have been focussed on the contribution of the anomalous dimensions and coefficient functions of the twist-@xmath3 flavour non - singlet and singlet operators that arise in the operator product expansion .
these were originally computed at one and two loops in @xcite as a function of the moment of the operator @xmath2 .
recently the first few moments of the three loop dimensions were determined in @xcite . however , the programme is not yet complete as one requires the explicit @xmath2-dependence at three loops .
this is required in order to compute the dglap splitting functions , @xcite , which are given by the inverse mellin transform with respect to @xmath2 . in @xcite
an approximate fitting procedure was developed to obtain these functions .
this was achieved by using general information on the ( expected ) form of the function in bjorken @xmath8 , which is the variable conjugate to @xmath2 with @xmath9 @xmath10 @xmath8 @xmath10 @xmath11 .
also the information from the explicit exact result for the first few moments was used to constrain the fitting parameters . in the absence of a full @xmath2-dependent result for the operator dimensions this strategy has to suffice for present .
clearly there are several potential problems in such an exercise . for example , the first few moments would only be expected to give a fairly accurate approximation to the splitting functions in the bulk of the @xmath8-range .
however , the low @xmath8 region which is presently of physical interest would not be well covered since there the higher moments give important contributions .
so it would be useful to understand to what extent such a fitting procedure can be relied upon . whilst the complete calculation in qcd will be a formidable computational task one could at least address this and other issues in a toy model .
this is the primary motivation for this paper .
we will compute the anomalous dimensions of the analogous operators in @xmath1 theory to four loops in the coupling constant expansion in the @xmath0 scheme .
although this would appear to be one order beyond that currently of interest in qcd , it turns out that due to the form of the model there can be no one loop contribution .
we view this theory as a toy for several reasons .
first , the nature of the interaction is simpler than in a non - abelian gauge theory and hence there will be substantially fewer diagrams to consider and also no tensorial complication in the numerator of the feynman diagrams which occurs when one has fermion and derivative couplings .
consequently although we do not expect the results to be as involved as those which will undoubtably occur when the full qcd result is realised , those functions of @xmath2 , such as finite sums in @xmath2 , which will occur will at least mimic analogous structures .
indeed as will be seen , it is these forms which will drive the strategy of the fitting procedure . as we will obtain the full @xmath12-loop results as a function of @xmath2 we will be in a position to investigate and gain insight into the approximation procedure for the splitting functions .
this leads us to our second reason why we will regard our work as a laboratory for testing ideas . clearly unlike qcd @xmath1 theory in four spacetime dimensions is not asymptotically free and therefore it does not sense to think of the high energy limit describing particle collisions .
however we stress that the aim is a _ mathematical _ one to endeavour to understand the relation of functions of the operator moment @xmath2 to their mellin transform which in @xmath1 we will regard as being on the same footing as the dglap functions .
in this situation asymptotic freedom plays a minimal role .
finally , our second motivation for this study is in relation to understanding the content of the operator product expansion of @xmath1 in @xmath13-dimensions .
as noted earlier the model relates to the perturbation theory of the heisenberg model which equivalently , in the sense of critical phenomena , can be described by the @xmath14 non - linear @xmath15 model in @xmath13 @xmath16 @xmath17 @xmath18 @xmath19 dimensions .
( see , for example , @xcite . )
the operator product expansion in that model has received wide attention in recent years both perturbatively and in the @xmath20 expansion , @xcite .
therefore the computation of the anomalous dimension of a set of operators involving derivatives or gradients at high order in the related @xmath1 theory is important for providing at least a cross check on future perturbative calculations in this area .
the paper is organised as follows . in section @xmath3
, we review basic features of the renormalization of composite operators in quantum field theories and then perform the two and three loop computations for the anomalous dimensions of the operators in question .
this calculation is extended to the fourth order in section @xmath21 where we also discuss the technical details of the determination of the divergent part of some of the underlying feynman integrals .
these results are then used to deduce the corresponding splitting functions in section @xmath12 where the approximate fitting procedure is analysed .
our conclusions are given in section @xmath22 and intermediate results in the four loop calculations are given in an appendix .
we begin our discussion by reviewing the background to the computation of anomalous dimensions of composite operators .
we will concentrate on one particular set of operators in this paper , @xmath23 which are twist-@xmath3 from the phenomenology point of view and symmetric and traceless .
the field @xmath24 is the basic field of the @xmath1 theory which has the lagrangian @xmath25 where @xmath26 is the coupling constant which is dimensionless in @xmath12-dimensions . as we will be applying standard techniques of renormalization such as dimensional regularization in @xmath13 @xmath16 @xmath12 @xmath27 @xmath28 dimensions and using the modified minimal subtraction scheme , ( @xmath0 )
, we must derive the momentum space version of the operator .
its divergent structure is deduced from the divergence structure of @xmath3-point green s functions where @xmath29 has been inserted .
therefore we write @xmath30 where @xmath31 is a symmetric traceless tensor .
it is more convenient to swamp the lorentz indices by contracting with a constant vector @xmath32 and define the object @xmath33 given the fact that @xmath34 is traceless one can derive the form of @xmath35 .
however , in higher order calculations we will exploit the first two terms in the construction of recurrence relations for subintegrals .
therefore we record , @xmath36 in qcd calculations @xcite the constant vector is usually taken to be a null vector in which case ( [ tdefn ] ) terminates at the first term .
the anomalous dimension we are aiming to compute is @xmath37 . in terms of renormalization constants
this is defined to be @xmath38 where @xmath39 is the mass scale introduced to ensure @xmath26 remains dimensionless in @xmath13-dimensions .
the renormalization constant @xmath40 corresponds to the renormalization of the field @xmath24 and is known at five loops in the more general theory with an internal @xmath14 symmetry @xcite .
the lower order calculations were performed in @xcite . in our notation defining @xmath41 then , for @xmath42 @xmath16 @xmath11 , @xmath43 the remaining renormalization constant @xmath44 defines the renormalization group function @xmath45 via @xmath46 which corresponds to the renormalization of the operator insertion in the one particle irreducible @xmath3-point green s function . from the nature of the quartic interaction @xmath45 will have the @xmath47 term as the first non - zero contribution . in the analogous calculation in qcd
we note that the twist-@xmath3 operators are physical operators and therefore their anomalous dimensions are gauge independent .
however the renormalization group functions corresponding to @xmath48 and @xmath45 are each gauge dependent .
therefore although there is no gauge symmetry in @xmath1 it is @xmath37 which we will regard as being on the same footing as the gauge independent dimensions in qcd .
we now detail the computation of the two and three loop contributions to ( [ zo ] ) .
these results will play an important role in the four loop calculation where they will enter multiplied by vertex counterterms for example . although strictly it is their @xmath49-expansion to @xmath50 and @xmath51 respectively for two and three loops which we will need it is in fact possible to compute the relevant integrals exactly as a function of @xmath13 .
the two loop contribution to ( [ zo ] ) arises from the first graph of fig .
1 where the line with a dot on it represents the insertion of the operator @xmath35 .
there is a flow of momenta @xmath52 through the external legs of the green s function . as we are interested only in the ultra - violet divergence of the graphs we compute with massless propagators which in turn allows us to exploit massless integration techniques such as uniqueness @xcite .
however for the lower order cases one needs only to apply , for instance , the elementary chain rule of @xcite .
therefore the exact result for the _ value _ the @xmath3-loop graph of fig .
1 is @xmath53 where @xmath13 @xmath16 @xmath54 @xmath16 @xmath12 @xmath27 @xmath28 , @xmath55 @xmath56 @xmath16 @xmath57 and @xmath58 @xmath16 @xmath59 .
the @xmath52 and @xmath60 dependence is easy to reconstruct from the integral dimension .
performing the @xmath49-expansion of ( [ twoloop ] ) and including the symmetry factor of @xmath61 gives the contribution to ( [ zo ] ) as @xmath62 the three loop calculation requires the calculation of the second graph of fig .
1 . as there are two graphs with the same topology we have included both in the same diagram with their respective names , @xmath63 and @xmath64 .
again it is straightforward to compute each integral exactly using the chain rule of @xcite and we find that their values are @xmath65 however each graph has at least one subgraph divergence arising when either one or both loops can be enclosed by a box .
the divergence arising from these subgraphs are cancelled by multiplying the two loop graphs of fig . 1 by the relevant ( vertex ) counterterm .
in this and the four loop case we include in our expressions for the final value of the divergent part of the graphs the subtraction of these subgraph divergences .
( further background to this procedure is well documented in , for example , @xcite . )
however it is appropriate to record that for @xmath64 the subtraction is @xmath66 hence the full @xmath0 contributions to ( [ zo ] ) at @xmath67 are @xmath68 where the finite sums @xmath69 are defined as @xmath69 @xmath16 @xmath70 . with these values
we deduce that @xmath71
we now turn to the renormalization of the operator at the four loop level .
there are four basic topologies to be considered which are illustrated in fig .
2 . as in fig .
1 we have indicated the location of the operator insertion by a dot and denoted the corresponding graph by the symbol beside each insertion . of these four topologies sets @xmath72 and @xmath73 represent integrals whose divergence structure and subgraph subtractions can be computed directly by elementary chain integrals involving ( [ nudefns ] ) and we will not discuss them further .
likewise integral @xmath74 is trivial to compute .
the remaining graphs required some ingenuity and we detail the techniques for several as an aid to the interested reader and a potential method for future similar problems .
first , we consider the integral @xmath75 and detail its calculation .
it can quickly be reduced to the @xmath3-loop integral illustrated in the first graph of fig .
3 where the factor arising from the integrations is @xmath76 in fig .
3 we denote the power of the propagator beside each line if it is not unity and the location of the numerator @xmath77 by the symbol @xmath78 and an arrow on the line beside the propagator involving @xmath79 .
we have used the coordinate space representation of the integrals so that the variables of integration are the location of the vertices and _ not _ the flow of momentum around a loop , @xcite . on dimensional grounds
the value of this integral is @xmath80 and the aim is to deduce @xmath81 to the term linear in @xmath49 .
this is because ( [ a1fact ] ) is @xmath82 .
the first step is to relate @xmath81 to an integral with all exponents unity or @xmath50 which is achieved by the integration by parts rule given in @xcite .
the result is the set of integrals on the right side of the first equation of fig .
the three terms with a line missing are easy to compute exactly and give the contribution to @xmath81 of @xmath83\ ] ] the remaining two diagrams have similar structure . if we denote by @xmath84 the value of the first graph on the right side of fig .
3 then we can obtain a recurrence relation for @xmath84 with respect to @xmath2 which can be solved .
one way of achieving this is to consider the integral without the @xmath60 contraction and decompose the numerator structure @xmath85 into a sum of symmetric and traceless objects @xmath86 defined in @xcite . as
such objects are independent the resulting integral must be a sum of the set @xmath87 with @xmath88 @xmath10 @xmath2 where the balance of indices is made up by including the appropriate number of @xmath89 tensors .
their coefficients are functions of @xmath2 with the leading term corresponding to @xmath84 .
this is deduced by eliminating the next to leading order term between the equations obtained by first contracting with @xmath90 and applying the differential operator @xmath91 and second by contracting with , say , @xmath92 .
this results in the relation @xmath93 \label{a1rr}\ ] ] where @xmath94 is defined in the second equation of fig . 3 and @xmath95 clearly the solution to ( [ a1rr ] ) is @xmath96
the integral @xmath97 has been computed explicitly in @xcite and has the value @xmath98 in our notation . for @xmath94
the second term is @xmath99 whilst the first and third are similar in form where , for example , it is simple to derive the integral representation for the first graph of the second equation of fig . 3 as @xmath100^{n}\ ] ] since we are only interested in the divergent part of @xmath75 it is straightforward to expand the integral in powers of @xmath49 to the @xmath50 term and obtain for it alone @xmath101^{n } \nonumber \\ & & = ~ \frac{1}{(n+1 ) } ~-~ \frac{s_1(n+2)}{(n+1)(n+2 ) } \nonumber \\ & & + ~ \frac{\epsilon}{(n+1 ) } \left[3 + s_1(n+2 ) + \frac{3 s_1(n+2)}{(n+2 ) } - \frac{2 s_1 ^ 2(n+2)}{(n+2 ) } - \frac{5 s_2(n+2)}{(n+2 ) } \right ] + o(\epsilon^2 ) \label{exparrow}\end{aligned}\ ] ] hence we have @xmath102 \label{bn}\end{aligned}\ ] ] assemblying all the relevant contributions and factors like ( [ a1fact ] ) and expanding in powers of @xmath49 we deduce that the divergent part of @xmath75 itself is @xmath103 performing the simple operation of subtraction of subgraph divergences yields the result for @xmath75 which we have recorded in eqn ( [ appa1 ] ) .
another technique was also used to compute several of the other graphs of fig . 2 and we detail this for graph @xmath104 .
after performing the two elementary loop integrations we obtain a simple two loop graph . rather than considering this graph
we examine the more general graph illustrated in fig . 4 where we have introduced an extra parameter @xmath105 and the ( analytic ) regularization @xmath106 . denoting the value of this graph by @xmath107
we record that for the problem in hand we require @xmath108 .
the presence of a general @xmath105 and @xmath106 will allow us more freedom in using various calculational techniques such as uniqueness to obtain expressions to determine the coefficients in the power series . the first step is to apply the operator @xmath109 to the diagram to yield the two graphs @xmath110 and @xmath111 of fig .
4 where @xmath112 \label{b1defn}\ ] ] whilst the former is finite with respect to @xmath49 the second is divergent .
it is determined by integrating by parts on the integral with the same exponent structure on each line except that @xmath11 is subtracted from the top right line and added to the bottom right . substituting the result of this operation into ( [ b1defn ] ) we obtain @xmath113 \label{b1eqn}\ ] ] where @xmath114 is integral given in fig . 4 and @xmath115 represents the sum of the remaining graphs which are finite with respect to @xmath49 . as @xmath116 is a set of elementary chain integrals all that remains is the calculation of @xmath117
whose @xmath50 term is needed due to the pole in @xmath49 in ( [ b1eqn ] ) .
this is achieved by expanding in powers of @xmath105 and @xmath49 @xmath118 and then determining the constant coefficients @xmath119 . rewriting @xmath115 from ( [ b1eqn ] ) as @xmath120
then it can be evaluated for special cases of @xmath105 to give @xmath121 in the explicit evaluation of the integrals , however , there is a potential problem when using uniqueness to integrate the right triangle of @xmath122 . to avoid the appearance of factors such as @xmath123 in both numerator and denominator one needs the temporary regulator @xmath106 and so ( [ fdefn ] ) ought to be replaced by @xmath124_{\delta \rightarrow 0}\ ] ] as an aid we record the results of these intermediate steps as @xmath125 with these expressions it is a simple exercise to verify that the result ( [ appb1 ] ) is finally obtained for graph @xmath104 .
we have used either of these algebraic techniques outlined here to evaluate the remaining divergent parts of the integrals . for completeness
the results for each are recorded in the appendix where the subgraph divergences have been removed . in several of the results the additional finite sum @xmath126
@xmath16 @xmath127 occurs . in ( [ appb3 ] ) , for instance , it arises in the computation of an intermediate two loop integral where the exponents on the lines diagonally opposite each other are @xmath3 and @xmath11
@xmath27 @xmath49 respectively and the central line has zero exponent but an @xmath78 insertion .
however , its @xmath49-expansion is elementary to compute using the rule for chains after the application of the binomial expansion .
also in the graph @xmath128 after completing the first two elementary loop integrals one is left with a two loop integral multiplied by @xmath129 .
this integral , which is finite with respect to @xmath49 , has been evaluated explicitly in @xcite and involves @xmath130 .
finally it is a simple exercise to assemble all the contributions to the four loop result and allowing for the symmetry factors of the graphs which are given in table 1 , we find the @xmath131 term of ( [ 3loop ] ) is @xmath132 we summarize the results of this and the previous section by recording our @xmath12-loop @xmath0 value for @xmath37 as @xmath133 one non - trivial check on this result is that it vanishes for @xmath2 @xmath16 @xmath3 . in that case
the original operator corresponds to the energy momentum tensor which is conserved in the quantum theory .
it is well established that the anomalous dimension of non - anomalous conserved physical currents vanish to all orders in perturbation theory .
( see , for example , @xcite . )
we now apply the results of the previous sections to the problem outlined in the introduction .
again we emphasise that the aim of the exercise is to investigate the mathematics underlying the fitting to the mellin transform of the anomalous dimensions given only knowledge of the first few moments . in qcd
it is this transform which corresponds to the dglap splitting functions , @xcite .
these are a measure of the probability that a parton decomposes into other partons ie quarks and gluons .
the variable conjugate to the moment @xmath2 is bjorken @xmath8 which represents the momentum fraction carried by the parton in the nucleon
. therefore it is restricted to lie in the unit interval .
though in general the domain of @xmath8 is the half - line . to approximate the three loop splitting functions in qcd
the authors of @xcite took a set of trial functions of @xmath8 whose properties were consistent with expected general behaviour of the splitting functions in the unit interval and fixed the unknown coefficients using the available exact @xmath21-loop moments .
for example , for the non - singlet case these expressions are known for @xmath2 @xmath16 @xmath3 , @xmath12 , @xmath134 , @xmath135 and @xmath136 .
of course such an exercise is unnecessary if the full @xmath2-dependent result was available . in @xcite as a prelude to exploring the third order correction the two loop result was examined . that had the advantage of knowledge of the exact result of the operator dimension for all @xmath2 and hence the splitting functions for all @xmath8 .
one disadvantage is that the second order result has a simpler structure of functions of @xmath8 than those which would arise at next order .
also in this second order case the result is unlikely to be unambiguous as only a few moments are sufficient to isolate which of the set of trial functions play the important role in , say , the small @xmath8 behaviour .
having summarized the status of the qcd calculation we can use our results at four loops in @xmath1 to try and gain a more detailed insight into improvements to the fitting .
first , some elementary remarks are in order .
it is important to note that we are not claiming there is a parton interpretation in @xmath1 theory .
the lack of asymptotic freedom counts against this .
second , although the qcd calculations are at third order and have computed @xmath37 to four loops , it is in effect a third order calculation since there is no one loop contribution .
therefore it is this fourth order result which we will regard as being on the same footing as the three loop qcd result although we are not claiming that it has as complex a form .
we begin by defining the mellin transform of a function @xmath137 as @xmath138 ~=~ \int_0 ^ 1 \ , dx \ , x^{n-1 } f(x ) \label{meldefn}\ ] ] where we restrict the function to exist on the unit interval . in qcd
the operator dimension is gauge independent though the field dimension and the renormalization of the bare operator are gauge dependent .
as there is no gauge symmetry in @xmath1 theory rather than take the full dimension @xmath37 we concentrate on the part @xmath45 .
this is justified by the fact that @xmath48 is @xmath2-independent and therefore its contribution to the @xmath8-dependence will be purely @xmath139 which is uninteresting for our study . with the definition ( [ meldefn ] ) it is straightforward to verify that the splitting functions are given by @xmath140 where @xmath141 in these expressions @xmath142 is the polylogarithm function
whose properties are well known , @xcite , and @xmath143 is the logarithmic derivative of the euler @xmath144-function . to gain some idea into the form of these functions we have plotted them in figs 5 - 7 . as a preliminary to the four loop analysis
we first summarize the method for the three loop case since the two loop example is clearly trivial .
the basic idea is to postulate a basis set of trial functions of @xmath8 with arbitrary coefficients .
these are fixed by using ( [ meldefn ] ) to obtain a function of @xmath2 which can be evaluated for the first few moments and compared with the analogous value of ( [ 3loop ] ) .
once the coefficients have been fixed one can compare the approximate function with the exact result to see how well it covers the function in the unit interval .
it is worth pointing out that the trial set of functions ought to contain a few which have logarithmic dependence given the nature of the exact result and its form in fig .
therefore some guidance on how good the fit is , is that it should be better than taking a trial set which gives a polyniomial in @xmath8 of the appropriate degree . given this criterion we discovered that the set @xmath145 gave a very good approximation .
we have plotted this approximation with the exact result in fig .
8 where the straight line is the exact result and the dots represent the approximation .
interestingly the discrepancy at @xmath8 @xmath16 @xmath9 is about @xmath146 . clearly to achieve this accuracy
we have used only five moments . however , comparing the functions of ( [ trial3 ] ) with the exact function one observes that @xmath147 is absent .
this illustrates a potential pitfall in fitting approximations to exact results .
one feature of the three loop case to emerge was the necessity of having functions in the trial set which reproduced some of the functions of @xmath8 which appeared in the exact result .
we repeated this exercise with a variety of sets either with fewer functions and therefore fewer moments or replaced several of the entries in ( [ trial3 ] ) with other functions . however
the result of fig .
8 represents the best fit .
we have repeated this analysis for the @xmath12-loop case . from the form of the exact result it turns out to be a harder exercise especially in covering the small @xmath8 region well . again with the criterion that a fit must be better than a polynomial , we managed to gain reasonable approximations with two trial sets . to achieve this we needed to extend the set to seven elements and include higher powers of the logarithm .
even with this , however , an answer as accurate as three loops could not be achieved .
these sets were @xmath148 @xmath149 and we have plotted each beside the exact results respectively in figs 9 and 10 .
although set ( [ trial42 ] ) appears to deviate in the middle @xmath8 range it is clearly more accurate for small @xmath8 than ( [ trial41 ] ) .
this is due to the alternating form of the expansion of the dilogarithm for small @xmath8 .
the respective discrepancies at @xmath8 @xmath16 @xmath9 from the exact and approximate answers were @xmath150 and @xmath151 .
we have constructed the @xmath0 @xmath12-loop anomalous dimensions for a set of twist-@xmath3 operators which are analogous to operators which occur in the operator product expansion used in deep inelastic scattering in qcd . in the course of examining the inverse mellin
transform for the four loop case we found that it was important to isolate those functions of @xmath8 and include them in the fit that reproduced similar forms to the basic finite sums that appear in the exact answer . in the absence of a full exact answer for qcd
it would therefore seem to us that one could endeavour to improve the fits of the anomalous dimensions , @xcite , in the low @xmath8-region by trying to deduce the type of finite sums which would occur at three loops .
an insight into this could be obtained by generalizing some of the series that already occur in the exact two loop results .
for instance , the sums @xmath69 occur at @xmath152-loops for @xmath152 @xmath16 @xmath11 and @xmath3 and it would appear odd if @xmath153 did not occur in the @xmath21-loop result .
therefore another structure to study might arise from generalizing those finite sums which only occur in the two loop result .
the other feature which we observed in our analysis was that to maintain the accuracy one would require several more exact moments for the anomalous dimensions .
* acknowledgements . *
the authors thank the organisers of the conference ` renormalization group 96 ' where this work was initiated .
jag is supported by pparc through an advanced fellowship , anm is supported by grant 97 - 01 - 01152 of the russian fond for fundamental research and sed by intas grant 932492ext .
part of this work was carried out with the aid of the computer algebra packages reduce , @xcite , and maple versions @xmath154 .
in this section we record the values of the divergent parts of the integrals which arise in the computation of the @xmath12-loop anomalous dimension in @xmath0 .
the label refers to the diagrams of fig . 2 though the values given here correspond to those with the subgraph divergences removed .
we found 99 j. zinn - justin , `` quantum field theory and critical phenomena '' ( clarendon press , oxford , 1989 ) .
e. brzin , j.c .
le guillou , j. zinn - justin & b.g .
nickel , phys .
a44 * ( 1973 ) , 227 ; a.a .
vladimirov , d.i .
kazakov & o.v .
tarasov , sov .
jetp * 50 * ( 1979 ) , 521 ; f.m .
dittes , yu.a .
kubyshin & o.v .
tarasov , theor .
* 37 * ( 1978 ) , 879 .
chetyrkin , a.l .
kataev & f.v .
tkachov , phys .
* b99 * ( 1981 ) , 147 ; * b101 * ( 1981 ) , 457(e ) ; k.g .
chetyrkin , s.g .
gorishniy , s.a .
larin & f.v .
tkachov , phys .
b132 * ( 1983 ) , 351 ; h. kleinert , j. neu , v. schulte - frohlinde , k.g .
chetyrkin & s.a .
larin , phys . lett .
* b272 * ( 1991 ) , 39 ; * b319 * ( 1993 ) , 545(e ) .
chetyrkin , a.l .
kataev & f.v .
tkachov , nucl .
* b174 * ( 1980 ) , 345 ; a.a .
vladimirov , teor . mat .
* 43 * ( 1980 ) , 210 ; s.g .
gorishniy , a.l .
kataev , s.a .
larin & l.r .
surguladze , phys . lett .
* b256 * ( 1991 ) , 81 .
tarasov , a.a .
vladimirov & a.yu .
zharkov , phys . lett . * 93b * ( 1980 ) , 429 ; s.a .
larin & j.a.m .
vermaseren , phys . lett .
* b303 * ( 1993 ) , 334 ; t. van ritbergen , j.a.m .
vermaseren & s.a .
larin , hep - ph/9701390 .
h. georgi & h.d .
politzer , phys . rev .
* d8 * ( 1974 ) , 416 ; d.j .
gross & f.j .
wilczek , phys . rev .
* d8 * ( 1973 ) 3633 .
e.g. floratos , d.a .
ross & c.t .
sachrajda , nucl . phys .
* b129 * ( 1977 ) , 66 ; * b139 * ( 1978 ) , 545(e ) ; nucl . phys . *
b152 * ( 1979 ) , 493 .
larin , t. van ritbergen and j.a.m .
vermaseren , nucl . phys .
* b427 * ( 1994 ) , 41 .
larin , p. nogueira , t. van ritbergen and j.a.m .
vermaseren , hep - ph/9605317 .
g. altarelli and g. parisi , nucl . phys . *
b126 * ( 1977 ) , 298 ; yu.l .
dokshitzer , sov .
jetp * 46 * ( 1977 ) , 641 ; l.n .
lipatov , sov .
. phys . * 20 * ( 1975 ) , 95 ; v.n .
gribov & l.n .
lipatov , sov .
. phys . * 15 * ( 1972 ) , 438 .
f. wegner , z. phys . * b78 * ( 1990 ) , 33 ; s.k .
kehrein , f. wegner & yu.m .
pismak , nucl . phys .
* b402 * ( 1993 ) , 669 ; k. lang & w. rhl , z. phys .
* c63 * ( 1994 ) , 531 .
vasilev , yu.m .
pismak & j.r .
honkonen , theor . math
. phys . * 47 * ( 1981 ) , 291 .
kazakov , theor .
phys . * 58 * ( 1984 ) , 223 ; phys . lett .
* b133 * ( 1983 ) , 406 .
callan jr , in `` methods in field theory '' ed .
r. balian & j. zinn - justin ( north - holland , 1976 ) .
k.g . chetyrkin & f.v .
tkachov , nucl . phys . *
192 * ( 1981 ) , 159 .
kazakov , theor .
* 62 * ( 1985 ) , 84 .
kazakov & a.v .
kotikov , theor .
* 73 * ( 1988 ) , 1264 .
collins , `` renormalization '' ( cup , 1984 ) .
l. lewin , `` dilogarithms and associated functions '' ( macdonald , london , 1958 ) .
hearn , `` reduce users manual '' version 3.4 , rand publication cp78 , ( 1991 ) .
fig . 1 . : : two and three loop graphs contributing to @xmath37 .
2 . : : four loop graphs contributing to @xmath37 .
3 . : : intermediate steps in the calculation of @xmath75 . fig .
4 . : : intermediate integrals in the calculation of @xmath104 . fig .
5 . : : two loop splitting function .
: : three loop splitting function .
: : four loop splitting function .
: : approximation to the three loop splitting function . fig .
: : approximation to the four loop splitting function with first trial set .
: : approximation to the four loop splitting function with second trial set .
( 410.00,200.00)(510.00,400.00)(610.00,200.00 ) ( 410.00,200.00)(460.00,270.00)(510.00,200.00 ) ( 410.00,200.00)(460.00,130.00)(510.00,200.00 ) ( 510.00,200.00)(560.00,270.00)(610.00,200.00 ) ( 510.00,200.00)(560.00,130.00)(610.00,200.00 ) ( 400,200)(1,0)10 ( 610,200)(1,0)10 ( 460.00,165.00 ) ( 510.00,300.00 ) ( 500.00,320.00)@xmath159 ( 450.00,120.00)@xmath160 ( 510.00,400.00 ) ( 110.00,200.00)(200,200)[t ] ( 10.00,200.00)(3,2)100 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 60.00,233.33 ) ( 45.00,253.33)@xmath161 ( 110.00,300 ) ( 92.00,320)@xmath162 ( 80.00,200 ) ( 90.00,190)@xmath163 ( 110.00,266.66)(50.00,200.00)(110.00,133.33 ) ( 110.00,266.66)(170.00,200.00)(110.00,133.33 ) ( 00.00,200.00)(1,0)10 ( 210.00,200.00)(1,0)10 ( 910.00,200.00)(-3,2)100 ( 810.00,266.66)(0,-1)133.33 ( 710.00,200.00)(3,-2)100 ( 750.00,246.33 ) ( 740.00,270.33)@xmath164 ( 760.00,166.66 ) ( 740.00,130.66)@xmath165 ( 810.00,200.00 ) ( 820.00,200.00)@xmath166 ( 710.00,200.00)(740.00,263.00)(810.00,266.66 ) ( 710.00,200.00)(780.00,203.00)(810.00,266.66 ) ( 910.00,200.00)(880.00,136.00)(810.00,133.33 ) ( 910.00,200.00)(840.00,199.00)(810.00,133.33 ) ( 700.00,200.00)(1,0)10 ( 910.00,200.00)(1,0)10 ( 10.00,-200.00)(160.00,0.00)(310.00,-200.00 ) ( 10.00,-200.00)(60.00,-130.00)(110.00,-200.00 ) ( 10.00,-200.00)(60.00,-270.00)(110.00,-200.00 ) ( 110.00,-200.00)(160.00,-130.00)(210.00,-200.00 ) ( 110.00,-200.00)(160.00,-270.00)(210.00,-200.00 ) ( 210.00,-200.00)(260.00,-130.00)(310.00,-200.00 ) ( 210.00,-200.00)(260.00,-270.00)(310.00,-200.00 ) ( 0,-200)(1,0)10 ( 310,-200)(1,0)10 ( 60.00,-235.00 ) ( 160.00,-235.00 ) ( 160.00,-100.00 ) ( 150.00,-80.00)@xmath167 ( 45.00,-280.00)@xmath168 ( 145.00,-280.00)@xmath169 ( 610.00,-200.00)(760.00,0.00)(910.00,-200.00 ) ( 610.00,-200.00)(760.00,-400.00)(910.00,-200.00 ) ( 760.00,-200.00 ) ( 600.00,-200.00)(1,0)10 ( 910.00,-200.00)(1,0)10 ( 610.00,-200.00)(1,0)300 ( 660.00,-200.00 ) ( 760.00,-152.00 ) ( 760.00,-100.00 ) ( 750.00,-80.00)@xmath170 ( 650.00,-180.00)@xmath171 ( 750.00,-135.00)@xmath172 ( 210.00,400.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 50.00,250.00)@xmath173 ( 60.00,210.00)@xmath174 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,400.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 50.00,250.00)@xmath173 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,400.00 ) ( 110.00,200.00)(200,200)[t ] ( 100.00,320.00)@xmath178 ( 80.00,300)(1,0)30 ( 30.00,250.00)@xmath179 ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,400.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 50.00,250.00)@xmath173 ( 60.00,210.00)@xmath174 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,400.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 50.00,250.00)@xmath173 ( 60.00,210.00)@xmath186 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,400.00 ) ( 28.00,250.00)@xmath179 ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 60.00,210.00)@xmath186 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 110.00,266.66)(0,-1)133.33 ( 120.00,200.00)@xmath175 ( 210.00,300.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 10.00,240.00)@xmath189 ( 50.00,210.00)@xmath173 ( 110.00,266.66)(0,-1)133.33 ( 160.00,240.00)@xmath190 ( 160.00,150.00)@xmath191 ( -160.00,195.00)@xmath192 ( 210.00,300.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 110.00,133.33)(3,2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 10.00,240.00)@xmath189 ( 50.00,210.00)@xmath173 ( 110.00,266.66)(0,-1)133.33 ( 160.00,240.00)@xmath190 ( 160.00,150.00)@xmath193 ( -160.00,195.00)@xmath194 ( 210.00,300.00 ) ( 10.00,200.00)(3,2)100 ( 10.00,200.00)(3,2)50 ( 110.00,266.66)(3,-2)50 ( 210.00,200.00)(-3,2)100 ( 10.00,200.00)(3,-2)100 ( 210.00,200.00)(-3,-2)100 ( 10.00,240.00)@xmath189 ( 50.00,210.00)@xmath173 ( 110.00,266.66)(0,-1)133.33 ( 160.00,240.00)@xmath195 ( 160.00,150.00)@xmath191 ( -160.00,195.00)@xmath196 ( 210.00,300.00 ) ( 10.00,100.00)(3,2)100 ( 10.00,100.00)(3,2)50 ( 210.00,100.00)(-3,2)100 ( 10.00,100.00)(3,-2)100 ( 210.00,100.00)(-3,-2)100 ( 70.00,120.00)@xmath173 ( 120.00,120.00)@xmath197 ( 110.00,166.66)(3,-2)50.00 ( 15.00,150.00)@xmath183 ( 160.00,140.00)@xmath195 ( 160.00,50.00)@xmath193 ( -160.00,95.00)@xmath198
|
we compute the anomalous dimensions of a set of composite operators which involve derivatives at four loops in @xmath0 in @xmath1 theory as a function of the operator moment @xmath2 .
these operators are similar to the twist-@xmath3 operators which arise in qcd in the operator product expansion in deep inelastic scattering . by regarding their inverse mellin
transform as being equivalent to the dglap splitting functions we explore to what extent taking a restricted set of operator moments can give a good approximation to the _ exact _ four loop result .
spbu
ip9707 lth391 + [ 8 mm ] s. .
derkachov@xmath4 , j.a .
gracey@xmath5 & a.n .
manashov@xmath6 + [ 3 mm ] * department of mathematics , st petersburg technology institute , + sankt petersburg , russia . * theoretical physics division , department of mathematical sciences , + university of liverpool , liverpool , l69 7zf , united kingdom .
* department of theoretical physics , state university of st petersburg , + sankt petersburg , 198904 russia .
|
the composition of dark halos around galaxies is a difficult problem . many of the baryons in the universe are dark and at least some of these dark baryons could be in galactic halos in the form of very massive objects ( vmos ) , with masses above @xmath9 m@xmath1 .
astrophysically motivated candidates include massive compact halo objects ( machos ) and black holes either of intermediate mass ( imbhs ; @xmath10 to @xmath11 m@xmath1 ) or massive ( @xmath12 m@xmath1 ) .
imbhs are an intringuing possibility as they could contribute , in principle , to all the baryonic dark matter and may be the engines behind ultraluminous x - ray sources recently discovered in nearby galaxies . a successful model in which vmos are the dominant component of dark matter halos could resolve some long - standing problems @xcite . if the halos of dsphs are comprised by black holes of masses between @xmath5 and @xmath13 m@xmath1 , they evolve towards a shallower inner profile in less than a hubble time , providing an explanation for the origin of dark matter cores in dwarf galaxies , and the orbits of globular clusters ( gcs ) do not shrink to the center by dynamical friction @xcite .
very few observational limits on vmos in dsph halos have been derived so far .
this letter is aimed at constraining the mass and abundance of vmos in the halos of dsphs by the disruptive effects they would have on gcs and cold long - lived substructures .
fornax is a dark dominated dsph galaxy with an unusually high gc frequency for its dynamical mass . in order to place dynamical constraints on the mass of vmos by requiring that not too many gcs
are disrupted , the density of vmos @xmath14 with mass @xmath15 along the orbits of the gcs should be known .
since the three - dimensional distances of the gcs to the center of fornax are unknown , the density of vmos at a mean distance of @xmath16 kpc will be adopted .
by @xmath17 we will denote the halo mass fraction in vmos , i.e. @xmath18 , where @xmath19 is the halo density ; @xmath20@xmath21 m@xmath1 pc@xmath3 at @xmath22 kpc @xcite .
@xcite give the `` survival diagram '' of gcs , considering different encounter histories for @xmath23 m@xmath1 pc@xmath24 and a velocity dispersion of halo particles @xmath25 km s@xmath26 .
the survival diagram establishes the range of mass and concentration such that gcs with central densities between @xmath0 m@xmath1 pc@xmath3 and @xmath27 m@xmath1 pc@xmath3 ( or , equivalently , core radii between @xmath7 and @xmath28 pc ) have a probability of less than @xmath29 to survive after @xmath30 gyr .
since the dissolution timescale for a certain gc ( and a given @xmath15 ) scales as @xmath31 , and @xmath32 km / s in fornax , it will be a factor of @xmath33@xmath34 less in fornax as compared to the galactic case considered by @xcite .
therefore , the survival diagram for fornax s gcs can be derived at once ( see fig . [
fig : klessen ] ) .
gcs in the region above the thick line have a probability of less than @xmath35 to survive in a dark halo with @xmath36 m@xmath37 pc@xmath3 .
for such a @xmath38 value and if the present parameters of the gcs are representative of their parameters at the time they formed , we would be observing the lucky survivors of an initial population of @xmath39 gcs of @xmath40 m@xmath1 , which turns out to be very unrealistic for a galaxy with a v - band luminosity of @xmath41 l@xmath1 .
it is possible to estimate the probability that we are observing only the survivors of a larger original population that are in the process of quick disruption @xcite .
the distribution of the dissolution age of gcs is expected to follow a scale - free power law @xmath42 , where @xmath43 is the characteristic dissolution timescale and @xmath44 is the age of the cluster population @xcite .
[ fig : klessen ] shows that at least @xmath45 gcs are above the thick line ; this indicates that their present disruption timescales are @xmath46 .
if the exponent of distribution of lifetimes is @xmath47 , as derived for galactic gcs @xcite , the probability to have @xmath45 out of @xmath48 gcs with lifetimes less than @xmath49 is @xmath50 , whereas the probability that the lifetimes of all the gcs are less then @xmath51 is @xmath52 .
hence the probability that the whole dark halo is comprised of objects with masses @xmath53 m@xmath1 is less than @xmath29 .
if only a fraction @xmath17 of the dark mass is in compact objects of mass @xmath15 , then @xmath54 m@xmath55 .
localized regions with enhanced stellar density and , where data permit , extremely cold kinematics have been detected in some dsphs ( e.g. , @xcite , hereafter k03 ; @xcite ) .
in particular , umi dsph has received the most attention .
collecting the velocity of stars in @xmath56 radius aperture , k03 found that a two - gaussian populations , one representing the underlying @xmath57 km / s gaussian and the other with velocity dispersion @xmath58 km / s , representing a subpopulation of fraction @xmath59 , is @xmath60 times more likely than the default @xmath57 km / s .
the best - fit @xmath61 is ill - determined as it is much smaller than the median velocity errors ( @xmath48 km / s ) .
nevertheless , even with these fiducial errors , we can be certain that the velocity dispersion is @xmath62 km / s at @xmath63 confidence level .
the stars which form the secondary density peak are not distinguished in colour and magnitude from the remainder of the umi population @xcite .
in fact , umi star formation history indicates that its stars have been formed in a single burst earlier than @xmath30 gyr ago @xcite .
although umi has long been suspected of experiencing ongoing tidal disruption , regions with enhanced volume density and _ cold kinematics _ can not be the result of tidal interactions because the coarse - grained phase - density , @xmath64 , in collisionless sytems must be constant or even decrease , thus implying that overdensity regions should appear dynamically hotter .
this suggests that the clump is long - lived .
an alternative explanation is that the density peak is a projection effect and that what we are seeing is a cold , low - density tidal tail .
however , numerical experiments have shown that this scenario is very unlikely @xcite .
the most plausible interpretation is that the clump is a disrupted stellar cluster , now surviving in phase - space because the underlying gravitational potential is harmonic ( k03 ) . within this
potential , gravitational encounters with the hypothetical vmos will dominate the orbital diffusion of the stars once they become unbound from the progenitor cluster .
the integrity of the cold clump may impose useful upper limits on the mass of vmos .
the fact that the subpopulation is orbiting within the dark matter core of umi will greatly simplify its dynamical description .
clump s stars will undergo a random walk in momentum space by the collisions with the population of vmos .
here we are interested in the velocity change induced in a star relative to the clump center of mass .
the mean - square velocity change of a star in an encounter with a vmo of mass @xmath15 , and impact parameter @xmath65 , with @xmath66 the clump s median radius , in the impulsive approximation is : @xmath67 where @xmath68 is the maximum relative velocity between the clump and the perturber and @xmath69 is the mean - square position of the stars in the clump @xcite . in the opposite case of a head - on collision ( @xmath70 ) ,
the mean change is comparable to that predicted by the tidal approximation when @xmath71 .
the usual way to proceed is to integrate @xmath72 given in eq .
( [ eq : spitzer ] ) for impact parameters @xmath73 from @xmath74 to infinity and correct for the encounters in which the tidal approximation fails by a factor @xmath75 ( e.g. , @xcite ) .
doing so , and for a distribution of clumps and vmos with a relative one - dimensional velocity dispersion @xmath76 , we obtain : @xmath77 for umi , the persistence of the clump for a large fraction of a hubble time indicates a core of the dark halo of at least @xmath28-@xmath78 times the size of the orbit of the clump , which is @xmath79 pc . in terms of the stellar core radius ( @xmath80 pc ) , this makes a halo core @xmath81@xmath28 times the stellar core and , consequently , the velocity dispersion of the halo particles in the core is , at least @xmath82@xmath28 times the stellar velocity dispersion , corresponding to @xmath83@xmath84 km / s .
the impulsive approximation is valid for @xmath85 , where @xmath86 and @xmath61 is the internal one - dimensional velocity dispersion of the subpopulation .
for the encounters with @xmath87 , responsible for most of the velocity impulse , this condition is well satisfied for @xmath88 .
therefore , for @xmath89 km / s , this requirement is fulfilled within the isothermal dark core of umi .
since stars in the clump are unbound , the self - gravity of the clump in a first approximation can be ignored considering only orbit diffusion in the large - scale harmonic potential of the parent galaxy . in a one - dimensional harmonic potential , a velocity impulse
@xmath90 produces a change in the velocity dispersion @xmath91 , where the brakets @xmath92 refer to the mean value after averaging over one orbit . combining this relation with eq .
( [ eq : deltav ] ) , we find the change of @xmath61 in a time @xmath93 : @xmath94 where @xmath95 is assumed , since the population of clumps is expected to have a velocity dispersion similar to the stellar background , @xmath96 km / s in umi .
the ratio @xmath97 depends on the model : for a plummer cluster @xmath98 , whereas @xmath99 for both a king profile with a dimensionless potential depth of @xmath100 ( e.g. , @xcite ) and a gaussian density distribution . in order to take a conservative value and to facilitate comparison with photometric and theoretical analysis that assume gaussian models , we will adopt @xmath99 constant in time . by @xmath101
we will indicate the time required for a very cold group of unbound stars @xmath102 , to acquire a velocity dispersion of @xmath103 km / s . from eq .
( [ eq : heating ] ) with @xmath104 we then obtain : @xmath105 since many of the recent dynamical models suggest mean dark matter densities within the stellar core radius of umi @xmath106 m@xmath1 pc@xmath3 , corresponding to a central mass - to - light ratio of @xmath107 m@xmath108l@xmath1 @xcite , @xmath109 m@xmath1 pc@xmath3 will be accepted as a conservative reference value . the mean - square radius of the subpopulation increases in time with @xmath61 , according to the relation @xmath110 , where @xmath111 is the orbital frequency in the constant - density core .
if the core is dominated by the dark matter component @xmath112 , then @xmath113 if @xmath114 km / s , the mean radius @xmath115 pc .
this corresponds to a @xmath116 radius of @xmath117 pc for a gaussian density profile , implying an angular size of almost @xmath118 at the distance of umi ( @xmath119 kpc ) .
this value is comparable to but slightly larger than the observed @xmath116 radius of the secondary peak @xmath120 . as a consequence @xmath121 km / s ; otherwise , the stellar subpopulation would appear more extended and diffuse than it is observed .
this upper value is in agreement with the observations of k03 .
our estimates for the size and velocity dispersion of the subpopulation are independent of the eccentricity of the orbit because there is little variation of the macroscopic properties of the halo , @xmath14 , @xmath122 or @xmath111 , within the core where the clump is orbiting .
nevertheless , the dark halo could have suffered significant evolution due to two - body processes and tidal stirring . for @xmath123
m@xmath1 relaxation processes induce an insignificant change in the internal properties of the halo ( e.g. , @xcite ) .
tidal heating can lead to a reduction of both the density of dark matter particles and its velocity dispersion ( e.g. , @xcite ) .
since the phase - space density , @xmath124 , for collisionless systems is nearly constant or decreases with time , the rate of energy gained by the clump due to encounters with vmos should have been more intense in the past .
the inclusion of evolution of halo properties by tidal effects would lead to a stringent upper mass limit . if the stellar progenitor cluster became unbound immediately after formation when supernovae expel the gas content ( @xcite ; k03 ) , @xmath101 should be greater than the age of the cluster @xmath125 gyr .
this requirement combined with eq .
( [ eq : t25 ] ) implies : @xmath126 if initially the stellar cluster is gravitationally bound , the increase in the internal energy gained by encounters with vmos will eventually exceed its binding energy after a time @xmath127 .
let us estimate @xmath127 , the time at which the cluster becomes unbound .
k03 infer a total mass of the cluster , @xmath128 , of @xmath129 m@xmath1 . if this cluster followed the recently observed relation between radius and mass of @xcite , @xmath130 pc . adopting the reference values of the dark matter halo ( @xmath109 m@xmath1 pc@xmath3 and @xmath131 km / s ) and rescaling the survival diagram of @xcite ( their figure 11 ) for the parameters of umi
, we infer that more than @xmath132 of the clusters with mass @xmath129 m@xmath1 will become unbound after @xmath133 gyr if @xmath134 m@xmath1 . therefore , in order to have a dynamically cold subpopulation with @xmath135 km / s , as that observed in umi at the present time @xmath125 gyr , we need @xmath136 , which implies the following upper limit for @xmath15 @xmath137 other corrosive effects such as mass loss by stellar evolution or tidal heating may also accelerate the dissolution of the cluster
. therefore , our estimates for @xmath127 and , hence , for @xmath15 , are upper limits .
the survival probability after collisions with vmos increases for progenitors that are more compact .
for instance , if the progenitor were a supercluster with a core radius @xmath138 pc and central density @xmath139 m@xmath1 pc@xmath3 , the probability of its remaining gravitationally bound after @xmath140 gyr is @xmath141 , for @xmath142 m@xmath1 . hence , there may be a non - negligible probability that such a supercluster has survived bound for @xmath140 gyr and that during the subsequent @xmath45 gyr it is dynamically heated by @xmath143 m@xmath1 vmos to reach @xmath114 km / s at the present time .
unfortunately , the evaporation time for this supercluster , setting the scale for dynamical dissolution by internal processes , is very short .
in fact , for such a stellar cluster , the evaporation time is @xmath144 , with @xmath145 the half - mass relaxation time @xcite .
the resulting evaporation time is @xmath146 gyr for an average stellar mass @xmath147 m@xmath1 and , hence , internal processes would have produced a fast desintegration of such a cluster .
we conclude that the upper limit given in eq .
( [ eq : mh ] ) is realistic and robust .
our approximations break down when the halo only contains a few vmos ; at least @xmath48 objects within a radius of @xmath148 pc are required , implying that our analysis is restricted to masses @xmath149 m@xmath1 . in fig .
[ fig : massfraction ] , the observational limits on vmos over a wide range of masses and dark matter fractions are shown .
the analysis of the survival of fornax s gcs rules out the mass range that would be interesting for explaining the origin of dark matter cores in dwarf galaxies ( k03 ; @xcite ) , because the relaxation timescale for vmos of mass @xmath150 m@xmath1 exceeds the hubble time .
moreover , it was found that the integrity of cold small - scale clustering seen in some dsphs imposes more stringent constraints on the mass of vmos .
a source of uncertainty is the mean density of dark matter within the core of dsphs . in the particular case of umi and
according to the scaling relations compiled in @xcite , the corresponding central dark matter density is @xmath2 m@xmath1 pc@xmath3 .
a slightly larger value has been derived from its internal dynamics @xcite . at a density @xmath151
m@xmath1 pc@xmath3 , eq .
( [ eq : mh ] ) implies that @xmath152 m@xmath1 .
we strongly encourage better determinations of the velocity dispersion of cold density aggregates ( bound or unbound ) in dsphs . for instance
, if the preliminary quoted value of @xmath7 km / s for the secondary peak of umi were confirmed , our upper limit for @xmath15 would be immediately reduced by a factor of @xmath153 , implying a very tight bound @xmath154@xmath155 m@xmath1 , depending on the adopted dark matter density ( @xmath156 m@xmath1 pc@xmath3 to @xmath157 m@xmath1 pc@xmath3 ) . in a unified scheme such as the ` stirring scenario ' by @xcite
, the composition of the dark halos of low - surface brightness and dsph galaxies should be the same .
@xcite found an upper limit of @xmath27 m@xmath1 by examining the dynamical heating of the stellar disk of gr 8 .
@xcite warn about the weakness of the argument of @xcite because the rotation curve of gr 8 decays in a keplerian fashion suggesting that gr 8 does not host a halo of dark matter , as expected if the dark halo has evaporated by two - body collisions . however , there exist dwarf galaxies with flat rotation curves . from the sample of @xcite , we have selected dwarf galaxies with high inclination angles and estimated the maximum permitted value of @xmath15 consistent with the thickness of the old stellar disk .
perhaps one of the most pristine cases is the edge - on galaxy ugca 442 .
this galaxy shows the typical flat rotation curve , implying the existence of a halo of dark matter with a central density of @xmath158 m@xmath1 pc@xmath3 and a velocity dispersion @xmath159 km / s @xcite .
the vertical stellar velocity dispersion has been derived using vertical hydrostatic equilibrium @xmath160 , with @xmath161 the total central surface density of the disk @xmath162 m@xmath1 pc@xmath163 and @xmath164 the scale height of the old disk derived photometrically to be @xmath165 pc .
this yields @xmath166 km / s . by requiring that @xmath167 for @xmath168 gyr , which is the characteristic age of the old stellar population
, the disk - heating argument establishes @xmath169 m@xmath1 @xcite .
[ fig : massfraction ] contains the most relevant dynamical constraints on vmos . in the conservative case @xmath114
km / s , the most stringent bound for masses between @xmath170 m@xmath1 and @xmath0 m@xmath1 comes from studies of the distribution of separations of wide stellar binaries in the galactic halo @xcite .
still , masses in that range are permitted if halo objects are slightly extended ( sizes @xmath171 pc ) .
other sources of uncertainty in this approach are the orbital distribution of the binaries and the duration of perturbations they are subjected to @xcite .
binney , j. , & tremaine , s. 1987 , galactic dynamics , ( princeton university press : new york ) bissantz , n. , englmaier , p. , & gerhard , o. 2003 , , 340 , 949 carrera , r. , aparicio , a. , martnez - delgado , d. , & alonso - garca , j. 2002 , , 123 , 3199 coleman , m. , da costa , g. s. , bland - hawthorn , j. , martnez - delgado , d. , freeman , k. c. , & malin , d. 2004 , , 127 , 832 ct , s. , carignan , c. , & freeman , k. c. 2000 , , 120 , 3027 gieles , m. et al .
2006 , , 371 , 793 gnedin , o. y. , & ostriker , j. p. 1997 , apj , 474 , 223 goerdt , t. , moore , b. , read , j. i. , joachim , s. , & zemp , m. 2006 , , 368 , 1073 goodwin , s. p. 1997
, , 286 , 669 hidalgo - gmez , a. m. 2004 , revmexaa , 40 , 37 irwin , m. , & hatzidimitriou , d. 1995 , , 277 , 1354 jin , s. , ostriker , j. p. , & wilkinson , m. i. 2005 , , 359 , 104 klessen , r. , & burkert , a. 1996 , , 280 , 735 kleyna , j. t. , geller , m. j. , kurtz , m. j. , & thorstensen , j. r. 1998 , , 115 , 2359 kleyna , j. t. , wilkinson , m. i. , gilmore , g. , & evans , n. w. 2003 , , 588 , l21 kormendy , j. , & freeman , k. c. 2004 , in dark matter in galaxies , ed .
s. d. ryder et al . , iau symp . 220
( san francisco : asp ) , 377 lacey , c. g. , & ostriker , j. p. 1985
, , 299 , 633 lake , g. 1990 , , 244 , 701 larsen , s. s. 2004 , , 416 , 537 mackey , a. d. , & gilmore , g. f. 2003 , , 340 , 175 mateo , m. 1998 , , 36 , 435 mayer , l. et al .
2001 , , 559 , 754 olszewski , w. w. , & aronson , m. 1985 , , 90 , 2221 pryor , c. , & kormendy , j. 1990 , , 100 , 127 read , j. i. , wilkinson , m. i. , evans , n. w. , gilmore , g. , & kleyna , j. t. 2006 , , 367 , 387 rix , h .- w . , & lake , g. 1993 , , 417 , l1 snchez - salcedo , f. j. 1999 , , 303 , 755 snchez - salcedo , f. j. , reyes - iturbide , j. , & hernandez , x. 2006 , , 370 , 1829 spitzer , l. jr . 1958 , , 127 , 17 tremaine , s. , & ostriker , j. p. 1999
, , 306 , 662 walker , m. g. , mateo , m. , olszewiski , e. w. , bernstein , r. a. , wang , x. , & woodroofe , m. 2006a , aj , 131 , 2114 walker , m. g. , mateo , m. , olszewiski , e. w. , pal , j. k. , sen , b. , & woodroofe , m. 2006b , apj , 642 , l41 wilkinson , m. i. et al .
2006 , in mass profiles and shapes of cosmological structures , eds .
mamon , f. combes , c. deffayet , b. fort , ( eas publications series ) , volume 20 , 105 yoo , j. , chanam , j. , & gould , a. 2004 , apj , 601 , 311
|
we establish new dynamical constraints on the mass and abundance of compact objects in the halo of dwarf spheroidal galaxies . in order to preserve kinematically cold the second peak of the ursa minor dwarf spheroidal ( umi dsph ) against gravitational scattering , we place upper limits on the density of compact objects as a function of their assumed mass . the mass of the dark matter constituents can not be larger than @xmath0 m@xmath1 at a halo density in umi s core of @xmath2 m@xmath1 pc@xmath3 .
this constraint rules out a scenario in which dark halo cores are formed by two - body relaxation processes .
our bounds on the fraction of dark matter in compact objects with masses @xmath4 m@xmath1 improve those based on dynamical arguments in the galactic halo . in particular , objects with masses @xmath5 m@xmath1 can comprise no more than a halo mass fraction @xmath6 .
better determinations of the velocity dispersion of old overdense regions in dsphs may result in more stringent constraints on the mass of halo objects . for illustration , if the preliminary value of @xmath7 km / s for the secondary peak of umi is confirmed , compact objects with masses above
@xmath8 m@xmath1 could be excluded from comprising all its dark matter halo .
|
conformal field theories ( cft ) in two space - time dimensions have been the object of intensive work since the mid eighties , after the fundamental paper @xcite .
these theories have very diverse and important physical applications , from the description of critical behavior in statistical mechanics and solid state physics in low dimension , to the worldsheet description of string theory .
furthermore , their remarkable analytical and algebraic structures and their connections with many other domains of mathematics make them outstanding laboratories of new techniques and ideas .
a particularly striking feature is the possibility of classifying large classes of cft through the study of representations of the virasoro algebra of conformal transformations @xcite .
the classification of cft partition functions , leading to the ade scheme , follows from exploiting the additional symmetry under modular transformations , the discrete coordinate changes that leave invariant the double - periodic finite - size geometry of the torus .
modular invariance has a precursor in ordinary statistical mechanics of lattice models , like the ising model @xcite .
consider a finite @xmath0 square lattice in two dimensions as in fig.[fig - tm ] : it is common to use the transfer matrix formalism to compute the partition function and other quantities of physical relevance in the model .
each configuration of degrees of freedom on a row of the lattice is associated with a state vector in a hilbert space .
the transfer matrix @xmath1 is an operator acting on a state vector and manufacturing the state vector of the next row .
it may thus be regarded as a discrete version of a time evolution operator , for time flowing vertically , see fig .
the partition function of a system with @xmath2 rows and periodic boundary conditions in the time direction is then @xmath3 . now
the model also admits a transfer matrix @xmath4 in the orthogonal direction , ( a column - to - column transfer matrix ) and if periodic boundary conditions are also imposed in that direction , one has an alternative expression of the partition function : @xmath5 .
therefore , in the doubly - periodic toroidal geometry there are two ways of writing the partition function .
imposing their equality may give some useful informations on the model , but in general these are too weak to constraint its operator content , spectrum , etc .
in the case of conformal field theories , there are special features that allow for a complete solution of the modular invariance conditions @xcite : * the spectrum of states in the theory , is organized in families , the `` conformal towers '' , with an infinite dimensional algebra acting as a spectrum generating algebra , ( the virasoro algebra or one of its extensions @xcite ) @xcite .
the structure of these conformal towers is given by the representation theory of the `` chiral '' algebra parameterized by the virasoro central charge @xmath6 and other quantum numbers .
* the hamiltonian can be expressed in terms of the virasoro generator of dilatations @xmath7 and its conjugate @xmath8 , whose eigenvalues give the conformal weights ( scale dimensions ) .
the lowest - energy state in each tower , i.e. in each virasoro representation , is called the `` highest weight vector '' with conformal weight @xmath9 and the excited states are obtained by acting on it with algebra generators .
( note the mismatch of terminology : the highest weight vector is actually the lowest energy state . )
the structure of each tower is encoded in the character of the representation , @xmath10 , which is the generating function of the dimensions of eigenspaces of given energy ( see below eq .
( [ chara ] ) ) . * the partition function @xmath11 can be rewritten in terms of the hamiltonian and thus of the virasoro generators @xmath12 , ( eq . ( [ evol ] )
hereafter ) . as a consequence ,
it involves a sum of bilinears of characters with undetermined non negative integer multiplicities of representations in the theory , see eq .
( [ toruspf ] ) . *
the requirement of modular invariance constrains these multiplicities and allows for the complete classification of partition functions in some theories for which that character expansion of @xmath11 is finite . in this contribution
, we describe this classification program in the simplest classes of conformal theories , namely the virasoro minimal models , having central charge @xmath13 , and the models with the affine lie algebra @xmath14 as an extended symmetry .
the main result is that the modular invariant partition functions are in one - to - one correspondence with the dynkin diagrams of the a , d and e types @xcite .
each partition function defines an independent theory with specific hilbert space and field content .
two - dimensional conformal field theories are quantum field theories enjoying covariance properties under conformal , _ i.e. _ local scale , transformations @xcite .
it is postulated that such theories exist and are consistent on any two - dimensional riemann surface @xcite . in the case of the plane with complex coordinate @xmath15
, one shows that @xmath15 and its complex conjugate @xmath16 may be treated as independent variables , called holomorphic and anti - holomorphic coordinates .
infinitesimal conformal transformations are generated by two copies of the virasoro algebra @xcite , acting on @xmath15 and @xmath16 and called right and left chiral algebras , respectively .
more generally , one may consider theories with an extended chiral algebra @xmath17 containing virasoro , such as the affine lie ( kac moody ) algebras @xcite , the superconformal algebras , etc .
the hilbert space @xmath18 of the cft decomposes onto pairs of representations @xmath19 of the left and right copies of the virasoro algebra or of @xmath17 . for `` rational '' conformal field theories ( rcft ) ,
the number of such irreducible representations @xmath19 is finite .
we label by @xmath20 and @xmath21 the left and right irreducible representations , respectively .
these pairs of representations are in one - to - one correspondence with the `` primary conformal fields '' @xmath22 @xcite . the finite decomposition of the rcft hilbert space can be written : @xmath23 where @xmath24 are non negative integer multiplicities .
these are subjected to consistency constraints , due to the fact that the rcft must exist and be consistent on any riemann surface @xcite . in particular ,
a crucial condition on the torus is the modular invariance of the partition function : this requirement determines the @xmath24 as described hereafter .
we start by considering the theory defined on a cylinder ( see fig.[fig - map ] ) of perimeter @xmath25 with a coordinate @xmath26 ; points of coordinates @xmath26 and @xmath27 are identified .
this cylinder is equivalent to the plane punctured at the origin , equipped with a complex coordinate @xmath15 , by means of the conformal mapping @xmath28 . on the cylinder ,
it is natural to think of the hamiltonian as the operator of translation along its axis ( the imaginary axis in @xmath26 ) , or more generally along any helix , defined by its complex period @xmath29 in the @xmath26 plane , with @xmath30 : @xmath31 in the particular case of fig.[fig - tm ] , while the general case is depicted in fig . [ fig2b ] .
the virasoro generators of translations in @xmath26 and @xmath32 ( regarded as independent variables ) are identified with @xmath33 and @xmath34 , therefore @xmath35 , with @xmath36 the complex conjugate of @xmath37 . through the conformal mapping @xmath38 ,
it is clear that these translations correspond to dilatations and rotations in the plane . indeed , using the transformation law of the energy - momentum tensor , one finds @xcite : [ cyltopl ] l_-1^cyl = -2il ( l_0 -c24 ) , where @xmath7 belongs to the virasoro algebra of the plane and the term @xmath39 comes from the conformal anomaly . in the continuum formulation ,
the transfer matrix is the exponential of the hamiltonian , @xmath40 and therefore , the evolution operator on a cylinder of length @xmath41 is given by : [ evol ] ( -h^cyl l ) = .
we now introduce the partition function of the theory on a torus @xmath42 : this is expressed by the trace of the evolution operator , owing to the identification of the two ends of the cylinder , [ ztorus ] z()=
_ e^2i [ ( l_0-c24 ) -|(|l_0-c24 ) ] .
each irreducible representation @xmath19 of vir ( or of @xmath17 ) is graded for the action of the virasoro generator @xmath7 : the spectrum of @xmath7 in @xmath19 is of the form @xmath43 , with non - trivial multiplicities @xmath44(subspace of eigenvalue @xmath45 ) .
it is thus natural to introduce a generating function of these multiplicities , i.e. a function of a dummy variable @xmath46 , called the character of the representation @xmath19 : [ chara ] _
j(q)=__j q^l_0-c24=q^h_j - c24 _
n=0^m_n q^n .
these functions have been computed for several classes of chiral algebras @xcite . using ( [ hilbert ] ) and the definition of characters , the trace in ( [ ztorus ] )
may be written as : [ toruspf ] z()=_j n_j _ j(q ) _ ( |q ) , q = e^2i , |q = e^-2i| .
let s stress that in these expressions , @xmath47 is the complex conjugate of @xmath46 , and therefore , @xmath11 is a sesquilinear form in the characters . equation ( [ toruspf ] ) shows that the torus partition function provides a convenient way of encoding the structure of the rcft hilbert space ( [ hilbert ] ) .
the geometry of the torus @xmath42 has been specified in ( [ toruspf ] ) through the modular parameter @xmath37 , @xmath30 ; the two periods are @xmath48 and @xmath37 , up to a global , irrelevant dilatation by @xmath25 .
equivalently , the torus may be regarded as the quotient of the complex @xmath26-plane by the lattice generated by the two numbers @xmath48 and @xmath37 : [ torus ] = / ( ) , in the sense that points are identified according to @xmath49 , @xmath50 .
this shows that there is a redundancy in the description of the torus : the modular parameters @xmath37 and @xmath51 describe the same torus , for any transformation of the modular group : @xmath52 , with @xmath53 defined up to a global sign and satisfying @xmath54 .
the group is generated by the two transformations : @xmath55 and @xmath56 @xcite : the first one is depicted in fig.[fig - tor ] and the second exchanges the two periods of the torus generalizing the case in fig.[fig - tm ] . the partition function must be uniquely defined on the torus , independently of coordinate choices , and thus be invariant under modular transformations : z ( - ) = z ( + 1 ) = z ( ) . [ z - inv ] these conditions , together with the expression ( [ toruspf ] ) of @xmath11 as a sesquilinear form in the characters , were introduced by cardy @xcite . as we shall show in the following , they open a route to the classification of rcfts . in rcfts ,
the finite set of characters @xmath57 , @xmath58 , transforms under modular transformations according to a unitary linear representation .
it suffices to give the action of the two generators of the modular group acting on @xmath37 , hence on @xmath59 , @xmath60 @xmath2 and @xmath61 are unitary matrices @xcite .
it follows that rcft partition functions ( [ toruspf ] ) are modular invariant if the ( non - negative ) integer multiplicity matrices @xmath62 obey the conditions : sns^=tnt^=n .
[ n - cond ] when explicit computations of the partition function ( path integral ) are possible , the results should automatically be modular invariant . in case of the strongly - interacting rcfts ,
such derivations are not usually available : one can consider the algebraic construction described here , starting from the representations of the chiral algebra , and impose the modular invariance condition .
let us stress that in this approach , modular invariance of @xmath11 is just one necessary condition for the consistency of the theory : there are additional requirements on other quantities , such as the crossing symmetry of the four - point function on the plane @xcite @xcite ( see @xcite ) .
we thus are led to the following : + * classification problem*. _ for a given chiral algebra , find all possible hilbert spaces ( [ hilbert ] ) , i.e. all possible sesquilinear forms ( [ toruspf ] ) with non negative integer coefficients that are modular invariant , and such that @xmath63 .
_ the label @xmath48 refers to the identity representation ( with @xmath64 ) and the condition @xmath63 expresses the unicity of the ground state . as mentioned above , the finite set of characters of any rcft , labelled by @xmath65 , supports a unitary representation of the modular group .
this implies that the diagonal combination of characters , @xmath66 , is always modular invariant for any rcft .
finding the general solutions for @xmath11 in any rcft is a well posed , although involved , algebraic problem .
complete results are known in a few cases , in particular in several families of rcft s based on the @xmath67 algebra , for which a special feature , the ade classification , emerges .
@xmath68 } \end{array } & e_7 \\
k=28 & \displaystyle { \left\vert \chi_1 + \chi_{11 } + \chi_{19 } + \chi_{29 } \right\vert^2 + \left\vert \chi_7 + \chi_{13 } + \chi_{17 } + \chi_{23 } \right\vert^2 } & e_8 \\
\hline \end{array}\ ] ] for @xmath70 , the @xmath67 affine kac - moody algebra at a non negative integer value of the `` level '' @xmath71 ( its central charge ) , the set of possible `` integrable '' representations is labelled by an integer or half - integer @xmath72 ( the spin of the representation of the `` horizontal algebra '' ) ,
subject to @xmath73 , or more conveniently by the integer @xmath74 , @xmath75 @xcite .
the matrix of modular transformations in ( [ smod ] ) reads : s_=()^1/2 ( ) , the virasoro central charge and the conformal weights are , c= , h_= .
we already introduced the ade dynkin diagrams that arise in the classification of simple lie algebras by killing and cartan @xcite . in the case of crystallographic groups ,
the dynkin diagrams describe the geometry of the hyperplanes of reflection and of the simple root vectors orthogonal to them .
the product of the reflections , @xmath76 , over all positive simple roots @xmath77 defines the coxeter element , unique up to conjugation , whose eigenvalues are @xmath78 , with @xmath79 running over the exponents defined above in section 3.1 ( table [ ade - table ] ) .
finite subgroups of @xmath80 form two infinite series and three exceptional cases : the cyclic groups @xmath81 , the binary dihedral groups @xmath82 , the binary tetrahedral group @xmath83 , the binary octahedral group @xmath84 and the binary icosahedral group @xmath65 .
it is natural to label them by ade as in table [ k - table ] .
a related classification is that of the regular solids in three dimensions : this may be the oldest ade classification , since it goes back to the school of plato ; here , @xmath85 is associated with the symmetry group of the tetrahedron , @xmath86 with the group of the octahedron or of the cube , and @xmath87 with the group of the dodecahedron or of the icosahedron .
the cyclic and dihedral groups may be thought of respectively as the rotation invariance group of a pyramid and of a prismus of base a regular @xmath88-gon , but those are not regular platonic solids .
the mckay correspondence provides a direct relation between subgroups and dynkin diagrams @xcite @xcite .
kleinian singularities ( table [ k - table ] ) : let @xmath89 be a finite subgroup of @xmath80 .
it acts on @xmath90 .
the algebra of @xmath89-invariant polynomials in @xmath91 is generated by three polynomials @xmath92 subject to one relation @xmath93
. the quotient variety @xmath94 is parameterized by these polynomials @xmath92 and is thus embedded into the hypersurface @xmath95 , @xmath96 .
this variety has an isolated singularity at the origin @xcite . in many cases ,
the classification follows from the spectral condition for symmetric matrices discussed in section 4 . in some others , however , the key point is the determination of triplets of integers @xmath97 such that : + 1q+1r>1 .
these are : note that the ade list above includes all the solutions except @xmath98 .
note also that for the @xmath99 and @xmath100 cases , these integers give the length ( plus one ) of the three branches of the dynkin diagram counted from the vertex of valency three . * the construction by ocneanu of topological invariants la turaev - reshetikhin with the methods of operator algebras , their ade classification , and their connection with jones classification of subfactors @xcite ; * the discussion of the minimal topological ( `` cohomological '' ) field theories , that may be regarded as twisted versions of the @xmath101 superconformal theories mentioned above : in these theories , the ( genus 0 ) three - point functions satisfy the so - called witten - dijkgraaf - verlinde - verlinde equations @xcite @xcite , for which dubrovin has shown the appearance of monodromy groups generated by reflections @xcite .
finally , the ade scheme not only controls the spectrum of the minimal conformal theories , but also contains some important information about their operator product algebra ( opa ) .
it can be proved that the structure constants of the opa of all the minimal theories may be determined from those of the diagonal theories in terms of the eigenvectors of the ade dynkin diagrams @xcite @xcite .
our first thought goes to a great friend , claude itzykson .
we would like to thank all colleagues that shared their insight and enthusiasm on this subject with us ; in particular , michel bauer , roger behrend , denis bernard , john cardy , philippe di francesco , terry gannon , lachezar georgiev , doron gepner , victor kac , ivan kostov , vincent pasquier , paul pearce , valentina petkova , hubert saleur , ivan todorov and guillermo zemba .
a. litvinov and a. b. zamolodchikov , _ conformal field theories in two dimensions _ , in scholarpedia .
j. l. cardy , _ boundary conformal field theory _ , in scholarpedia .
l p. kadanoff , _ the ising model _ , in scholarpedia .
, in scholarpedia .
e. verlinde , _ verlinde algebra _ , in scholarpedia .
m. virasoro , _ the virasoro algebra _ , in scholarpedia .
, in scholarpedia . , in scholarpedia .
a. a. belavin , a. m. polyakov and a. b. zamolodchikov , _ infinite conformal symmetry in two - dimensional quantum field theory , _ nucl .
b * 241 * ( 1984 ) 333 . j.e .
humphreys , _ introduction to lie algebras and representation theory _ , springer , new york , 1972 .
m. i. knopp , _ modular functions in analytic number theory _ , markham , chicago , 1970 . p. di francesco , p. mathieu and d. snchal , _ conformal field theory _ , springer , berlin , 1997 .
j. l. cardy , _ operator content of two - dimensional conformally invariant theories , _ nucl .
b * 270 * ( 1986 ) 186 .
g. w. moore and n. seiberg , _ naturality in conformal field theory , _ nucl .
b * 313 * ( 1989 ) 16 ; _ classical and quantum conformal field theory _ ,
* 123 * ( 1989 ) 177 .
a. cappelli , c. itzykson and j. b. zuber , _ modular invariant partition functions in two dimensions , _ nucl .
b * 280 * ( 1987 ) 445 .
a. cappelli , c. itzykson and j. b. zuber , _ the ade classification of minimal and a1(1 ) conformal invariant theories , _ commun .
* 113 * ( 1987 ) 1 .
a. kato , _ classification of modular invariant partition functions in two dimensions , _ mod .
a * 2 * ( 1987 ) 585 .
d. gepner , _ on the spectrum of 2d conformal field theories _ , nucl . phys .
b * 287 * ( 1987 ) 111 .
d. gepner and z. qiu , _ modular invariant partition functions for parafermionic field theories , _ nucl .
b * 285 * ( 1987 ) 423 .
t. gannon , _ the cappelli - itzykson - zuber a - d - e classification _ , rev .
* 12 * ( 2000 ) 739 .
p. goddard , a. kent and d. i. olive , _ unitary representations of the virasoro and supervirasoro algebras , _ commun .
* 103 * ( 1986 ) 105 .
g. e. andrews , r. j. baxter and p. j. forrester , _ eight vertex sos model and generalized rogers - ramanujan type identities , _ j. stat
* 35 * ( 1984 ) 193 .
d. a. huse , _ exact exponents for infinitely many new multicritical points , _ phys .
b * 30 * ( 1984 ) 3908 .
v. pasquier , _ operator content of the ade lattice models ,
_ j. phys .
a * 20 * ( 1987 ) 5707 ; _ continuum limit of lattice models built on quantum groups _ ,
b * 295 * ( 1988 ) 491 .
a. cappelli , _ modular invariant partition functions of superconformal theories , _ phys .
b * 185 * ( 1987 ) 82 .
t. gannon , _
u(1)**m modular invariants , n = 2 minimal models , and the quantum hall effect , _ nucl .
b * 491 * ( 1997 ) 659 .
t. gannon , _ the classification of affine su(3 ) modular invariant partition functions , _ commun .
* 161 * ( 1994 ) 233 ; _ the classification of su(3 ) modular invariants revisited , _ ann .
* 65 * ( 1996 ) 15 .
i. kostov , _ free field representation of the @xmath102 coset models on the torus , _ nucl .
b * 300 * [ fs22 ] ( 1988 ) 559 .
francesco and j. b. zuber , _
su(n ) lattice integrable models associated with graphs , _ nucl . phys .
b * 338 * ( 1990 ) 602 .
t. gannon , _ the monstruous moonshine and the classification of cft _ , arxiv : math/9906167 .
behrend , p.a .
pearce , v.b .
petkova and j .- b .
zuber , _ boundary conditions in rcft _ , nucl . phys .
b * 579 * ( 2000 ) 707 .
e. p. verlinde , _ fusion rules and modular transformations in 2d conformal field theory , _ nucl .
b * 300 * ( 1988 ) 360 .
j. l. cardy , _ boundary conditions , fusion rules and the verlinde formula _ , nucl .
b * 324 * ( 1989 ) 581 .
g. pradisi , a. sagnotti and ya .
s. stanev , _ completeness conditions for boundary operators in 2d conformal field theory _ , phys .
b * 381 * ( 1996 ) 97 .
f. m. goodman , p. de la harpe and v. f. r. jones , _ coxeter dynkin diagrams and towers of algebras _ , vol .
* 14 * , msri publications , springer , berlin , 1989 .
i. kostov , _ strings with discrete target space _ , nucl.phys .
b * 376 * ( 1992 ) 539 . ph .
roche , _ on the construction of integrable dilute ade models _ , phys .
b * 285 * ( 1992 ) 49 .
s. o. warnaar , b. nienhuis and k. a. seaton , _ new construction of solvable lattice models including an ising model in a field _
* 69 * ( 1992 ) 710 .
w. lerche , c. vafa and n. p. warner , _ chiral rings in n=2 superconformal theories _ , nucl .
b * 324 * ( 1989 ) 427 .
e. j. martinec , _ algebraic geometry and effective lagrangians _ , phys . lett .
b * 217 * ( 1989 ) 431 .
arnold , s.m .
gusein - zaide and a.n .
varchenko , _ singularities of differential maps _ , birkhuser ,
basel , 1985 .
a. b. zamolodchikov , _ conformal symmetry and multicritical points in two - dimensional quantum field theory _ , sov .
* 44 * ( 1986 ) 529 .
a. w. w. ludwig and j. l. cardy , _ perturbative evaluation of the conformal anomaly at new critical points with applications to random systems , _ nucl .
b * 285 * ( 1987 ) 687 .
a. cappelli , g. dappollonio and m. zabzine , _ landau - ginzburg description of boundary critical phenomena in two dimensions _
, jhep * 0404 * ( 2004 ) 010 .
a. m. tsvelik , _ quantum field theory in condensed matter physics _ , cambridge univ .
press , cambridge , 1995 .
x. g. wen , _ quantum field theory of many - body systems _ , oxford univ .
press , oxford , 2007 .
a. cappelli , l. s. georgiev and i. t. todorov , _ parafermion hall states from coset projections of abelian conformal theories _ , nucl . phys .
* b 599 * ( 2001 ) 499 .
a. cappelli , g. viola and g. r. zemba , _ chiral partition functions of quantum hall droplets _ , arxiv:0909.3588 , ann .
phys . in press . c. nayak , s. h. simon , a. stern , m. freedman and s. das sarma , _ non - abelian anyons and topological quantum computation _ , rev .
* 80 * ( 2008 ) 1083 .
humphreys , _ reflection groups and coxeter groups _ , cambridge univ . press , cambridge , 1990 .
p. slodowy , _ platonic solids , kleinian singularities and lie groups _ , in _ algebraic geometry _ ,
notes in math . *
1008 * springer , berlin , 1983 .
p. gabriel , _
berichtigungen zu unzerlegbare darstellungen i _ , manusc .
* 6 * ( 1972 ) 71 .
e. hille , _ ordinary differential equations in the complex domain _ , wiley , new york , 1976 .
index for subfactors _ , invent . math . * 72 * ( 1983 ) 1 .
j. mckay , _ graphs , singularities and finite groups , _ _ proc .
pure math . _ * 37 * ( 1980 ) 183 .
j. b. zuber , _ cft , bcft , ade and all that _ , lectures given at school , _ quantum symmetries in theoretical physics and mathematics _ , san carlos de bariloche , argentina , 10 - 24 jan 2000 , arxiv : hep - th/0006151 .
a. ocneanu , _ paths on coxeter diagrams : from platonic solids and singularities to minimal models and subfactors _ , in : rajarama bhat et al .
( eds . ) , _ lectures on operator theory _ , fields inst .
monographies , ams , 1999 .
e. witten , _ on the structure of the topological phase of two - dimensional gravity _
b * 340 * ( 1990 ) 281 ; r. dijkgraaf , h. l. verlinde and e. p. verlinde , _ topological strings in @xmath103 , _ nucl . phys .
b * 352 * ( 1991 ) 59 .
b. dubrovin , _ integrable systems in topological field theory _ , nucl .
b * 379 * ( 1992 ) 627 ; _ geometry of 2d topological field theories , _
springer lect .
notes in math . * 1620 * ( 1996 ) 120 .
petkova and j .- b .
zuber , _ on structure constants of sl(2 ) theories _ , nucl .
b * 438 * ( 1995 ) 347 ; _ the many faces of ocneanu cells , _ nucl . phys .
b * 603 * ( 2001 ) 449 .
|
the ade classification scheme is encountered in many areas of mathematics , most notably in the study of lie algebras . here
such a scheme is shown to describe families of two - dimensional conformal field theories .
1.5 cm * a - d - e classification * .3 cm * of conformal field theories * .5 cm andrea cappelli + _ infn + via g. sansone 1 , 50019 sesto fiorentino ( firenze ) , italy _ + .3 cm jean - bernard zuber + _ lpthe , tour 24 - 25 , 5me tage , + universit pierre et marie curie -
paris 6 , + 4 place jussieu , f 75252 paris cedex 5 , france _ .5 cm .5 cm _ review article to appear in scholapedia , http://www.scholarpedia.org/_
|
astronomers in the last decade have made major progress in understanding the properties and evolution of galaxies in the distant universe .
galaxies up to redshifts @xmath4 , and perhaps even higher , have been discovered in large numbers allowing statistically significant population characteristics to be derived ( e.g. * ? ? ?
* ; * ? ? ?
we in fact now have a good understanding of basic galaxy quantities , such as the luminosity function , as well as how scaling relations , such as the tully - fisher relation evolve up to , at least , @xmath5 .
we have also begun to trace the stellar mass evolution of galaxies , as well as the star formation rate , determining when stars and stellar mass was put into place in the modern galaxy population .
stellar masses are now becoming a standard measure for galaxies , and are being used to trace the evolution of the galaxy population in terms of star formation rates and morphologies ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . however , stellar masses only trace one aspect of the masses of galaxies , and ideally and ultimately , we want to be able to measure galaxy total masses , that include contributions from stars , gas , and dark matter .
galaxies are believed to be hosted by massive dark matter haloes that make up more than 85% of their total masses , and thus clearly tracing the co - evolution of galaxies and their haloes is a major and important goal .
measuring dynamical or total masses for galaxies is , however , very difficult , as it requires observations that are very challenging to obtain , or requires unusual and rare circumstances such as gravitational lensing .
for example , dynamical masses can be measured through rotation curves , or internal velocity dispersions , but this becomes difficult at higher redshifts , with reliable internal velocities existing for only a handful of galaxies at @xmath6 and above .
furthermore , it is difficult to know whether these kinematic measures are tracing the total potential , or just the inner parts of galaxies . likewise
, using gravitational lensing to measure galaxy total masses is difficult , as it requires lensed background galaxies , and such instances are very rare .
it is also not yet clear if these galaxies are representative , as they may have special mass profiles that are conducive to lensing .
the weak galaxy - galaxy lensing technique provides another way to estimate the total masses of galaxies .
no unusual circumstances are required , but individual measurements are extremely difficult to achieve .
stacking techniques give reliable results , but involve combining galaxies in haloes of different masses , which complicates the interpretation ( e.g. * ? ? ?
stacked satellite kinematics can also be used to probe the total masses of galaxies , but face the same problems as the weak galaxy - galaxy lensing method @xcite .
therefore , obtaining total mass estimates for galaxies is difficult , and very few measures , or even estimates , have been produced for galaxies outside the local universe .
one very powerful method for measuring the total masses of galaxies is to measure their clustering .
clustering measurements are independent of photometric properties , and as such they can be used to highlight fundamental properties of galaxy populations without assumptions concerning stellar populations or mass profiles .
previously , it has been shown that galaxies with higher stellar masses cluster more than systems with lower stellar mass , with a very strong clustering above the characteristic stellar mass @xmath7 of the schechter mass function @xcite . in the halo model of galaxy formation
, the large - scale distribution of galaxies is determined by the distribution of dark matter haloes @xcite .
therefore , halo clustering is a strong function of halo mass , with more massive haloes more strongly clustered , providing a means to study the relationship between galaxy properties , and dark matter halo masses .
in fact , basic calculations allow one to convert a clustering strength for a sample of galaxies into a corresponding halo mass in which galaxies reside ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) .
clustering measurements have been performed on galaxy samples in both the nearby and distant universe . in the local universe , studies have established that clustering depends on the type of galaxy under consideration ; for example , early - type , red , galaxies are more clustered than late - type , blue galaxies ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and luminous galaxies are more clustered than faint galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
moreover , distant massive galaxies selected by their extremely red colours have been shown to strongly cluster , and therefore are thought to inhabit massive dark matter haloes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in this paper , we present the first general study of the clustering properties for a stellar mass selected sample of galaxies up to @xmath8 .
we carry this out by measuring the correlation length and amplitude for galaxies selected with stellar masses @xmath9within the palomar observatory wide - field infrared survey ( powir ) @xcite , which covers the fields where spectroscopy and other multi - wavelength data are available through the deep extragalactic evolutionary probe 2 ( deep2 ) survey and all - wavelength extended groth strip international survey ( aegis ) @xcite . in total
we examine the clustering strength for 18,482 galaxies selected by stellar mass within 0.7 deg@xmath10 .
we derive an estimate of the total masses for these galaxies , and study in detail their stellar - mass - to - total - mass ratio .
this paper is presented as follows : our data - sets , catalogues and our photometric redshifts and stellar mass estimations are described in section [ sec : data ] ; in section [ sec : cf ] we describe the methods used in measuring the clustering properties of our samples ; section [ sec : dmm ] is dedicated to the methods we use to derive the masses of dark matter haloes for our samples ; in section [ sec : ratio ] we compare our results with the literature and models ; and finally we summerise our conclusions in section [ sec : summ ] . throughout the paper ,
we assume a @xmath11cdm cosmology with @xmath12 , @xmath13 , @xmath14 km s@xmath15 mpc@xmath15 . to ease comparisons with previous work
, we use a concordance model with fiducial values of @xmath16 and @xmath17 . to determine stellar masses throughout this paper
, we use the initial mass function ( imf ) from @xcite and assume a hubble constant of @xmath18 km s@xmath15 mpc@xmath15 .
all of the galaxies in this paper are found within three of the four fields covered by the palomar observatory wide - field infrared survey ( powir , table 1 ; * ? ? ?
the powir survey was designed to obtain deep @xmath19-band and @xmath20-band data over a significant area ( @xmath211.5 deg@xmath22 ) .
observations were carried out between september 2002 and october 2005 over a total of @xmath23 nights .
this survey covers the extended groth strip @xcite , and three other fields that the deep2 team has observed with the deimos spectrograph @xcite .
the total area imaged in the @xmath19-band is 4920 arcmin@xmath10=1.37 deg@xmath10 , with half of this area imaged in the @xmath20-band .
the goal depth was @xmath24 , although not all fields are covered to this depth , therefore we select the fields which have 5@xmath25 depths between @xmath26 for point sources , measured in a 2 diameter aperture . for our purposes
we abbreviate the fields covered as : egs ( extended groth strip ) , field 2 , field 3 , and field 4 ( table [ tab1 ] ) . in the following study we use data in the egs field only to train our photometric redshifts (
see section [ sec : data - mass - z ] ) , while only the other three fields are used to perform our clustering analysis ( see section [ sec : cf - acf ] ) . for extensive information on this survey , and the data products we use from it ,
see @xcite , @xcite and @xcite .
all of the @xmath27-band data were acquired utilising the wide field infrared camera ( wirc ) on the palomar 5 meter telescope .
wirc has an effective field of view of @xmath28 , with a pixel scale of 0.25pixel@xmath15 .
the total survey contains 75 wirc pointings . during the @xmath27-band observations we used 30 second integrations , with four exposures per pointing .
the @xmath20-band observations were taken with 120 second exposures per pointing .
typical total exposure times were between one and two hours for both bands .
the reduction procedure follows a standard method for combining near - infrared ( nir ) ground - based imaging , and is described in more detail in @xcite .
the resulting seeing fwhm in the @xmath27-band imaging ranges from 0.8 to 1.2 , and is typically 1.0 .
photometric calibration was carried out by referencing persson standard stars during photometric conditions , which were later cross calibrated with 2mass stars @xcite .
.the palomar fields and wirc pointings areas .
the egs field is used in this study for training the photometric redshifts , and the other three for measuring clustering properties . [
cols="<,^,^,^,^ " , ] the final nir images were made by combining individual mosaics obtained over different nights .
the @xmath27-band mosaics are comprised of coadditions of @xmath29 seconds exposures dithered over a non - repeating 7.0 pattern .
the images were processed using a double - pass reduction pipeline developed specifically for wirc .
the detection and photometry of our galaxies was performed using the sextractor package @xcite .
false artifacts are removed through sextractor flags which identify sources that do not have normal galaxy or stellar profiles . from this
we built a @xmath19-selected sample and then cross - referenced these sources with the deep2 redshift catalogue .
optical imaging from the canada - france - hawaii telescope ( cfht ) , over all fields , is used to estimate photometric redshifts and stellar masses , with the help of spectroscopy from the deimos spectrograph on the keck ii telescope @xcite .
this optical imaging comes from the cfht 3.6-m , and consists of data in the @xmath30- , @xmath31- and @xmath32-bands taken with the cfh12k camera - a 12,288 @xmath33 8,192 pixels ccd mosaic with a pixel scale of 0.21 .
the integration times for these observations are 1 hour in @xmath30 and @xmath31 , and 2 hours in @xmath32 , per pointing , with a @xmath31-band 5@xmath25 depth of @xmath34 , and similar depths in @xmath30 and @xmath32 @xcite . from this imaging data
a @xmath35 magnitude limit was used for determining targets for the deep2 spectroscopy .
the seeing for the optical imaging is roughly the same as that for the nir imaging , and we measure photometry consistently , using a 2diameter aperture . the keck spectra were acquired with the deimos spectrograph as part of the deep2 redshift survey @xcite .
the selection of targets for the deep2 spectroscopy was based on the optical properties of the galaxies detected in the cfht photometry , with the basic selection criteria @xmath36 .
objects in fields 2 , 3 and 4 were selected for spectroscopy based on their position in @xmath37 vs. @xmath38 colour space to focus on galaxies at redshifts @xmath39 .
spectroscopy in the egs was acquired using the magnitude limit , but to avoid the survey being completely dominated by lower redshift galaxies , sources with colours suggesting they are at @xmath40 were down - weighted in favour of sources with @xmath39 .
the total survey targeted over 30,000 galaxies , with about a third of these in the egs field . in all fields
the sampling rate for galaxies that meet the selection criteria is 60% .
the deimos spectroscopy was obtained using the 1200 line / mm grating , with a resolution r @xmath41 covering the wavelength range 6500 - 9100 .
redshifts were measured by comparing templates to data , and we only utilise those redshifts determined by the identification of two or more lines , providing very secure measurements . since accurate photometry , and photometric errors are important for the purposes of this paper we therefore give some details about our methods of measuring magnitudes .
the photometric errors , and the detection limit of each @xmath19-band image , are estimated by randomly inserting simulated galaxies of known magnitude , surface brightness profile , and size into each image , and then recovering these simulated objects with the same detection parameters used for real objects ( see * ? ? ?
* ; * ? ? ?
* ) . to determine the detection and photometric fidelity of the images in more detail , two sets of 10,000 mock galaxies were created , each with an intrinsic exponential and de vaucouleurs profile ( see * ? ? ?
systematic errors in the measurement of magnitudes , due to the detection method , were also estimated using the same simulations .
the recovery fraction is found to be essentially 100% at the magnitudes of our sample galaxies .
this was determined through simulating galaxies down to @xmath42 , as detailed in @xcite .
however , all of the galaxies in our sample are at least a magnitude brighter than this , with all galaxies within our stellar mass limits having magnitudes @xmath43 as shown in @xcite .
@xcite and @xcite furthermore carried out extensive simulations to show that at this limit we are nearly 100% complete and are retrieving nearly all of the light from these galaxies .
however , depending on the sizes of the galaxies , as much as 0.2 mag in the recovered @xmath19-band light could be missed , although for any single galaxy it is unlikely that this amount of light is missed due to the surface brightness profiles of our objects @xcite .
this issue is addressed in great detail in @xcite , but in summary we account for this uncertainty in the error budget .
we calculate photometric redshifts for our @xmath19-selected galaxies which do not have deep2 spectroscopy .
these photometric redshifts are based on the optical+near infrared imaging , in the @xmath44 ( or @xmath45 for half the sample ) bands , and are fit in two ways , depending on the brightness of a galaxy in the optical . for galaxies that meet the spectroscopic criteria , @xmath36 , we utilise a neural network photometric redshift technique to take advantage of the large number of secure redshifts with similar photometric data .
most of the @xmath36 sources not targeted for spectroscopy should be within @xmath46 , as very few objects this bright are found at higher redshifts ( e.g. * ? ? ?
the neural network fitting is done through the use of the annz @xcite method and code . to train the code , we use the @xmath41 redshifts in the egs field , whose
galaxies are selected with a magnitude limit of @xmath36 and span the redshift range @xmath47 ( see section [ sec : data - gen ] ) .
we then use this training to calculate the photometric redshifts for galaxies with @xmath36 in all fields .
the overall agreement between our photometric redshifts and our annz spectroscopic redshifts is very good using this technique , with @xmath48 out to @xmath49 .
the agreement is even better for the @xmath50galaxies where we find @xmath51 across all of our four fields .
the photometry we use for our photometric redshift estimates are measured within a 2 diameter aperture . for galaxies fainter than @xmath35 we estimate photometric redshifts using bayesian techniques based on the software from @xcite . for an object to have a photometric redshift
we require that it be detected at the 5@xmath25 level in all optical and near - infrared bands ( @xmath44 ) , which in the @xmath31-band reaches @xmath34 @xcite .
we optimise our results , and correct for systematics , through a comparison with spectroscopic redshifts , resulting in a redshift accuracy of @xmath52 for @xmath53 systems based on comparisons to redshifts from @xcite .
these @xmath53 galaxies are however only a very small part of our sample . up to @xmath49 only 6 ( 2.6% ) of our @xmath50galaxies
are in this regime , while 412 ( 9% ) of our @xmath54galaxies have an @xmath31-band magnitude this faint , with a similar fraction for galaxies down to m@xmath55 .
furthermore , all of these systems are at @xmath56 . at @xmath57
all of our sample galaxies are measured through the @xcite method due to a lack of training redshifts .
we also compared the results at low redshifts between the @xcite and @xcite methods and their agreement is very good , at the same level than the comparison with spectroscopic redshifts . a thorough explanation , including error budgets and uncertainties , for our redshifts are included in @xcite and @xcite .
we match our @xmath19-band selected catalogues to the cfht optical data to obtain spectral energy distributions ( seds ) for all of our sources , resulting in measured @xmath44 magnitudes . from these
we compute stellar masses based on the methods and results outlined in @xcite , @xcite and @xcite .
the basic mass fitting method consists of fitting a grid of model seds constructed from @xcite ( bc03 ) stellar population synthesis models , with different star formation histories . we use exponentially declining models to characterise the star formation history , with various ages , metallicities and dust contents included .
these models are parameterised by an age , and an e - folding time for parameterising the star formation history , where sfr @xmath58 .
the values of @xmath59 are randomly selected from a range between 0.01 and 10 gyr , while the age of the onset of star formation ranges from 0 to 10 gyr .
the metallicity ranges from 0.0001 to 0.05 ( bc03 ) , and the dust content is parametrised by @xmath60 , the effective v - band optical depth for which we use values @xmath61 .
although we vary several parameters , the resulting stellar masses from our fits do not depend strongly on the various selection criteria used to characterise the age and the metallicity of the stellar population .
we match magnitudes derived from these model star formation histories to the actual data to obtain a measurement of stellar mass using a bayesian approach .
we calculate the likely stellar mass , age , and absolute magnitudes for each galaxy at all star formation histories , and determine stellar masses based on this distribution .
distributions with larger ranges of stellar masses have larger resulting uncertainties .
typical errors for our stellar masses are 0.2 dex from the width of the probability distributions . there are also uncertainties from the choice of the initial mass function ( imf ) .
our stellar masses utilise the @xcite imf , which can be converted to salpeter imf stellar masses by dividing by a factor @xmath62 ( see * ? ? ?
there are additional random uncertainties due to photometric errors .
the resulting stellar masses thus have a total random error of 0.2 - 0.3 dex , roughly a factor of two .
the details behind these mass measurements and their uncertainties , including the problem of thermal - pulsating agb stars , is described in @xcite and @xcite .
we examine the changes in our stellar masses by using bruzual & charlot models updated in 2007 to include tp - agb stars , where we find at most a decrease of 0.07 dex , or stellar masses which are 20% less than what we calculate using models without tp - agb stars .
furthermore , @xcite have recently shown that stellar mass estimates , from stellar population synthesis , suffer systematic uncertainties mainly due to key phases of stellar evolution and the initial mass function .
however we are consistent in our analysis as we are comparing results from different studies based on the same stellar population models to determine stellar masses . throughout this work ,
stellar mass is given in unit of @xmath63 , having been computed assuming a hubble constant of @xmath18 km s@xmath15 mpc@xmath15 . in this study
we only examine massive galaxies with stellar masses @xmath0 .
in fact , we examine the clustering properties of galaxies within the following four mass selected samples @xmath64@xmath65 , @xmath66@xmath67 , @xmath68@xmath69and @xmath68@xmath70 .
we also examine these stellar mass cuts within four different redshift bins : @xmath71 , @xmath72 , @xmath73 and @xmath74 .
as discussed in @xcite , our lower - mass samples are largely incomplete in the higher redshift bins , and the volume we are sampling at lower redshift does not allow us to examine a large enough sample of massive galaxies .
the abundances of our different samples are reported in table [ tab : results1 ] .
we also note that our final two stellar - mass bins are not independent , as the third one overlaps completely with the fourth one .
we decided to keep the bins this way for statistical reasons , as otherwise it would have been impossible to conduct studies on an independent higher stellar mass bin within the highest redshift bins .
as we will see later , the results for our two higher mass bins are very similar , and this division does not effect our analysis . in order to better constrain the errors introduced in our clustering and abundance analysis due to possible inaccurate photometric redshifts , we built a set of five supplementary simulated catalogues based on a series of monte carlo simulations .
we perform this simulation by altering the redshifts of each of the galaxies in our sample , randomly within our well calibrated and known photometric redshift errors , as based on comparisons to spectroscopic redshifts .
the new stellar masses are then estimated with these altered redshifts for each object . after recalculating these masses from our simulated redshifts
, we then re - select the galaxies within our different redshift and stellar masses bins as in our original analysis , but by using these new simulated galaxy catalogues .
the abundances of our different samples reported in table [ tab : results1 ] , are an average over all the catalogues ( the original one and the five altered ones ) , and an average over our three fields .
the error budget listed in this table takes into account the errors introduced by the photometric redshift , and the resulting stellar mass errors , as accounted for by the rms over the six catalogues , plus the errors introduced by the cosmic variance as measured from the field - to - field variance over our three fields . within this paper
we systematically estimate our errors by accounting for these two effects .
this is likely an over - estimation of the total error , as the error on the photometric redshifts are partially included in the field - to - field variance .
we also use these catalogues to estimate the fraction of interlopers that are contaminating each of our redshift bins . as the determination of stellar mass
is directly linked with the photometric redshift determination , it is very difficult to conduct an interloper analysis for both selections at the same time .
we therefore focus on the interlopers between the different redshift bins , which is likely to have the highest impact on our clustering analyses .
furthermore , our measurements are made on three independent fields , all of which give similar results .
@xcite also performed a monte - carlo simulation in order to quantify the effect of stellar mass errors on the density and mass function of our samples .
they found that the systematic effect is minor compared to the effect of the cosmic variance , which we also find , and for which we account for as well .
the interloper fraction within the different redshift bins is determined by identifying objects that are within the same redshift bin in all our simulated catalogues .
then the number of potential interlopers is identified in each catalogue based on the difference between this standard set and the additional number of galaxies seen in each . however , this method of accounting for the photometric redshift errors increases artificially the number of interlopers , from the most populated redshift bins to the least populated . by comparing our five simulated catalogues with our original one , we find that this effects account for @xmath75 of the galaxies in the simulated catalogues .
taking into account this effect , we estimate that the maximum possible interloper fraction is : for @xmath76 , @xmath77 , for @xmath78 , @xmath79 , for @xmath80 , @xmath81 , and for @xmath82 , @xmath83 . our interloper fraction is therefore at most between 20% and 30% depending on the redshift bin .
given the errors on our photometric redshift determination , as quoted in section [ sec : data - mass - z ] , this is of the order we expect .
furthermore , we note that this fraction of interlopers is largely due to galaxies near the boundaries of our strict stellar mass and redshift cuts entering other bins
. these galaxies typically have just slightly different redshifts and stellar masses before they were simulated , creating a slightly different population within each simulated redshift bin which is not significantly different from the original one .
regardless , we fully account for this effect within our measured clustering strengths error budget .
the primary goal of this work is to study the clustering properties of galaxies selected according to their stellar mass .
a clustering analysis is a study of the distribution of galaxies at all scales .
given its peculiar geometry , the egs field is not ideally suited for such a study .
indeed , as galaxies are distributed along a strip in the egs , the clustering properties at the largest scales are difficult to establish given the small numbers of objects we have in each stellar mass and redshift bin in this field .
for the other three fields we mask out overlap regions , and other regions which do not have reliable photometry .
this masking of regions empty of galaxies is then taken into account in the following analysis .
overall we performed our analyses over a total area of 0.7 deg@xmath22 . within our clustering analysis
we first measure the 2-point angular correlation function @xmath84 for our sample using the @xcite estimator : @xmath85 where the dd , dr and rr terms refer to the number of data - data , data - random and random - random galaxy pairs having angular separations between @xmath86 and @xmath87 .
figure [ fig : cf ] shows the correlation function derived from our mass - selected samples in different redshift bins for one of our fields .
the best fit for the angular correlation is assumed to be a power - law of the form @xcite : @xmath88 with @xmath89 the amplitude at 1 degree , @xmath90 the slope , and @xmath91 the integral constraint due to the limited area of the survey . for statistical reasons we fix the slope to the commonly used fiducial value of @xmath92 @xcite .
we also estimate the integral constraint ( @xmath91 ) as follows @xcite , @xmath93 where @xmath94 is the area subtended by the survey field . to determine @xmath91
we numerically integrate this expression over each field , excluding masked regions . assuming a slope of @xmath92 , the unmasked areas ( @xmath95 ) covered by the fields , and the corresponding integral constraint values are : @xmath96 arcmin@xmath22 and @xmath97 , @xmath98 arcmin@xmath22 and @xmath99 , @xmath100 arcmin@xmath22 and @xmath101 , for fields 2 , 3 and 4 respectively . in order to fit the clustering reliably ,
we avoid the small - scale excesses due to possible multiple galaxy occupation of a single dark matter halo , as has been shown to exist in recent studies ( e.g. * ? ? ?
. the lower bound on which we fit our correlation was chosen to correspond to @xmath102 , the observed one - halo term limit established by @xcite .
these lower bounds are shown by the vertical dotted lines on figure [ fig : cf ] , and it is worth noting that no apparent excess is displayed at the smallest scales .
the error - bars shown on figure [ fig : cf ] are estimated from a jackknife monte - carlo method .
the error estimation for each scale is made following the bootstrap method . in each sub - sample
, objects are randomly removed and duplicated from our position catalogue , and the two - point correlation function is measured again .
the value of the correlation function at each scale is estimated from the mean of these bootstrap catalogues , and errors are derived from the variance .
we then derive the best fit amplitude for each of our fields at 1 degree using a marquardt least - linear method , that take into account our bootstrap errors , and provide an estimate of the error on the fitting .
we then derive a mean value from our six catalogues and over each of our three fields .
table [ tab : results1 ] summarises the values measured for our samples . as explained in section [ sec : data - mass - err ] , the variance over the six catalogues is used as an estimate for the systematic errors in the photometric redshift determination and the variance over the three field measurements is used as an estimate of the cosmic variance of our samples .
our total error budget is a quadratic sum of the fitting errors , photometric redshift errors , and cosmic variance errors . as discussed in @xcite , using models from @xcite , the cosmic variance on the number counts in the redshift range we use ( @xmath103 ) range from @xmath104 for our less massive samples , up to @xmath105 for our most massive ones .
this is comparable with the uncertainties we derived based on our field - to - field measurements on the clustering analysis . despite the accuracy of our photometric redshifts , as described in section [ sec : data - mass - err ] , our samples , selected in different redshift bins , are possibly contaminated by galaxies at other redshifts .
any contamination will dilute the original clustering signal , and therefore our measurements of the amplitude of the correlation function could be underestimated . from our monte - carlo simulations , we estimate the maximum contamination at a level from 20% to 30% , mainly due to galaxies from adjacent lower redshift bins ( see section [ sec : data - mass - err ] for a full description of this ) . this contamination is stronger at higher redshifts
however , this effect does not significantly alter our results . in the worst case , if the contaminating population is uncorrelated , the value of @xmath89 is diluted by a factor of @xmath106 ( with @xmath107 the contamination fraction ) , corresponding to an underestimation of our correlation amplitude by a factor between 1.56 and 2.04 ( for 20% and 30% contamination by other redshift galaxies ) .
these factors are upper limits , and the true value of @xmath89 is likely to remain within the error - bars quoted in table [ tab : results1 ] . in any case , this contamination will only reduce the strength of the clustering , and again our measures are therefore at worst lower limits .
this means that any errors in redshifts and/or stellar masses only dilute our signal , which would make the real clustering strengths even higher than the ones we observe .
furthermore such an effect is included in our error budget as we are taking the average of measurements over six catalogues with perturbed redshift distributions , as described in section [ sec : data - mass - err ] .
this method should account for the errors due to interlopers .
moreover , our samples , divided into different stellar mass bins , are also subject to possible contamination from other stellar mass bins .
given the mass uncertainties of 0.2 - 0.3 dex ( see section [ sec : data - mass - mass ] ) , it is possible that galaxies more or less massive , and therefore more or less clustered , contaminate our stellar mass cuts . as mentioned in the section [ sec : data - mass - err ]
, these effects are more difficult to constrain with our monte - carlo simulations .
however within a given redshift bin , the measurements of the clustering for samples in adjacent stellar mass bins are very similar and overlap within their errorbars .
the cross - contamination would therefore only have a minor impact , and in the sense of increasing the segregation . in order to compare the clustering of galaxy populations at different redshifts we derive the spatial correlations for our galaxy samples .
we do this by using the spatial correlation function ( @xmath108 ) @xcite to derive @xmath109 , the co - moving galaxy correlation length , as defined by : @xmath110 where @xmath111 .
the redshift dependence is included in the co - moving correlation length @xmath112 . in general the larger the correlation length ( @xmath109 ) , the more clustered the galaxies are . by measuring the redshift distribution for our samples
we can also derive the correlation length @xmath109 from the amplitude of the angular correlation @xmath89 using the relativistic limber equation @xcite .
we estimate correlation lengths assuming that our galaxies are distributed as top - hat functions in each of our narrow redshift bins . given the derived statistics of our samples and the narrow redshift bins we are using here , and despite the accuracy of our photometric redshifts , changing the shape of the redshift distribution has a negligible impact on the results . * 9c z bin@xmath113 & @xmath114 bin@xmath115 & @xmath116@xmath117 & @xmath118@xmath119 & @xmath120 & @xmath121@xmath122 & @xmath89@xmath123 & @xmath109@xmath124 & @xmath125@xmath126 + & & & & & @xmath127 & @xmath128 & & + 0.40.8 & 10.010.5 & @xmath129 & @xmath130 & @xmath131 & @xmath132 & @xmath133 & @xmath134 & @xmath135 + 0.81.2 & 10.511.0 & @xmath136 & @xmath137 & @xmath138 & @xmath139 & @xmath140 & @xmath141 & @xmath142 + & 11.011.5 & @xmath143 & @xmath144 & @xmath145 & @xmath146 & @xmath147 & @xmath148 & @xmath149 + & 11.012.0 & @xmath143 & @xmath150 & @xmath151 & @xmath152 & @xmath153 & @xmath154 & @xmath155 + 1.21.6 & 10.511.0 & @xmath156 & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 & @xmath162 + & 11.011.5 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & @xmath168 & @xmath169 + & 11.012.0 & @xmath163 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath174 & @xmath175 + 1.62.0 & 11.012.0 & @xmath176 & @xmath177 & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 + _ @xmath113redshift bin ; @xmath115stellar mass bin in log@xmath183 ; @xmath117mean redshift ; @xmath119mean stellar mass in log@xmath183 ; @xmath184density in arcmin@xmath185 ; @xmath122abundance in @xmath186mpc@xmath187 ; @xmath123correlation amplitude at 1 degree ; @xmath124correlation length in ; @xmath126bias .
the correlation length is derived from the amplitude of the angular correlation function in each of our redshift and stellar mass bins , using each of the six catalogues ( the original and the simulated versions ) for each of our three fields .
the mean correlation length is then computed , as well as its associated error budget , as described in section [ sec : cf - acf ] . in summary
, we find that correlation lengths vary from @xmath188to @xmath189for galaxies selected by stellar masses with @xmath0 .
the highest correlation lengths are for the most massive galaxies at the highest redshifts @xmath190 ( figure [ fig : r0z - nz](a ) ) . in general , at lower stellar masses and redshifts , the correlation length becomes smaller .
table [ tab : results1 ] summarises the values we calculate for the correlation length @xmath109 measured within our stellar mass selected samples . in figure
[ fig : r0z - nz ] we plot the correlation length , and the galaxy number density , as a function of redshift for our stellar - mass - selected galaxy samples .
we observe an apparent decrease in the correlation length with decreasing redshift , as is also observed in other previous studies . here
the main effect is the well - known luminosity - segregation , with brighter galaxies ( less abundant ) having a larger correlation length . as a first rough approximation ,
this can be directly linked to a mass - segregation effect , with more massive galaxies more clustered .
previously , showed that bright red galaxies , that are likely very massive , have clustering lengths that are almost invariant with redshift since @xmath191 , while less luminous ( likely less massive ) galaxies have correlation lengths that decrease with fainter magnitude at all redshifts .
as shown in figure [ fig : r0z - nz](a ) , we compare our results with , who have derived correlation lengths from spatial correlation functions using the vimos - vlt deep survey ( vvds ) data - sets .
the mean redshift of their sample is @xmath192 , and for galaxies with masses log @xmath193)=10.22 $ ] they derived @xmath194 , and for galaxies with masses log @xmath193=10.70 $ ] , @xmath195 .
we note that these results are in good agreement within the error - bars of our measurements .
as analysed in section [ sec : cf - acf ] a contamination of our different samples by galaxies in adjacent redshift and stellar mass bins will artificially decrease the measured value of their correlation length .
this effect is more likely to occur within the higher redshift samples , and thus the values summarised in table [ tab : results1 ] can be considered as lower limits of the real intrinsic correlation length .
such an effect has bin identified in previous analysis as well ( e.g. * ? ? ?
* ; * ? ? ?
the net effect of this is that our estimates of the total masses for these galaxies are also lower limits .
however , even with a contamination fraction at the upper limit of 20% and 30% , as quoted in section [ sec : cf - acf ] , the values of the correlation lengths we quote in table [ tab : results1 ] are only under - estimated by a factor @xmath196 and @xmath197 respectively .
the relative uncertainties on our measured correlation length are of the same order as this correction ( table [ tab : results1 ] ) .
this is an upper limit of the correction that could be applied , and furthermore our correlation length measurements and errors take into account these potential effects from interlopers as described in sections [ sec : data - mass - err ] and [ sec : cf - acf ] .
we also note that @xcite have found that the errors on photometric redshifts can artificially broaden the redshift distribution which leads to an overestimate of the correlation length @xmath109 .
however given the narrow bins we are using in the current analyses , our redshift distributions are more perturbed by objects scattered from one redshift bin to another , than from a broadening effect due to photometric redshift errors .
in addition , we derive the linear bias of our sample , which relates the clustering of galaxies to that of the overall dark matter distribution @xcite .
the rms density fluctuations of haloes ( @xmath198 ) are linked to the rms density fluctuations of the underlying mass ( @xmath199 ) by the bias ( @xmath200 ) following : @xmath201 where @xmath202 is the variance in @xmath203spheres , assuming that dark matter behaves as predicted in linear theory , renormalised to the fiducial value @xmath204 ( fixed to @xmath17 in this study ) .
if the galaxy correlation function is fit as a power law , then it can be integrated to give the relative variance in @xmath203spheres : @xmath205 @xcite .
we find for our sample that the bias varies from @xmath206 to @xmath207 for galaxies selected by stellar masses @xmath0 .
the most massive galaxies at the highest redshifts @xmath190 are more biased , in good agreement with biased galaxy formation models @xcite .
table [ tab : results1 ] summarises the values of the bias @xmath125 measured for our samples . * 10c z bin@xmath113 & @xmath114 bin@xmath115 & @xmath116@xmath117 & @xmath118@xmath119 & @xmath208@xmath184 & @xmath209@xmath122 & @xmath210@xmath123 & @xmath211@xmath124 & @xmath212@xmath126 & @xmath213@xmath214 + & & & & & & & & & @xmath215 + 0.81.2 & 11.011.5 & @xmath143 & @xmath144 & @xmath216 & @xmath217 & @xmath218 & @xmath219 & @xmath220 & @xmath221 + & 11.012.0 & @xmath143 & @xmath150 & @xmath222 & @xmath223 & @xmath224 & @xmath225 & @xmath226 & @xmath227 + 1.21.6 & 10.511.0 & @xmath156 & @xmath157 & @xmath228 & @xmath229 & @xmath230 & @xmath231 & @xmath232 & @xmath233 + & 11.011.5 & @xmath163 & @xmath164 & @xmath234 & @xmath235 & @xmath236 & @xmath237 & @xmath238 & @xmath239 + & 11.012.0 & @xmath163 & @xmath170 & @xmath240 & @xmath241 & @xmath242 & @xmath231 & @xmath243 & @xmath244 + 1.62.0 & 11.012.0 & @xmath176 & @xmath177 & @xmath245 & @xmath246 & @xmath247 & @xmath248 & @xmath249 & @xmath250 + _ @xmath113redshift bin ; @xmath115stellar mass bin in log@xmath183 ; @xmath117mean redshift ; @xmath119mean stellar mass in log@xmath183 ; @xmath184minimum mass of the dark matter halo estimated from the correlation length in log@xmath251 $ ] ; @xmath122correction factor for accounting of the halo distribution effect ( estimated from correlation length ) ; @xmath123minimum mass of the dark matter halo estimated from the correlation length in log@xmath251 $ ] ; @xmath124correction factor for accounting of the halo distribution effect ( estimated from abundance ) ; @xmath126mass of the dark matter halo as estimated from the correlation length and after correction for the halo occupation in log@xmath251 $ ] ; @xmath214stellar - mass - to - dark - matter - mass ratio in units of @xmath252 . _
the standard cdm model predicts that at any redshift more massive dark matter haloes are on average more clustered than lower mass systems . to model this effect quantitatively ,
we use the predicted effective bias , and the abundance evolution , as derived from the @xcite formalism for different minimum dark matter halo ( dmh ) mass thresholds .
this formalism is based on modelling the effective bias , which relates the mass fluctuations of haloes , to the mass fluctuations in spheres that contain an average mass @xmath254 of underlying dark matter , according to equation [ eq : bias ] . for this purpose
, we use the halo abundance distribution as a function of mass and redshift determined by @xcite , derived from fits to large n - body simulations , and the linear halo bias for a given mass at a given redshift , calculated using the function of @xcite .
we directly derive the effective bias by integrating the ratio of these functions on masses above a given minimum mass @xmath255 .
a modelling of the correlation length evolution for dmhs of a given minimum mass @xmath255 at a given redshift @xmath256 can be directly derived from the effective bias using the recipes described in @xcite .
we use also these recipes in the computation of the linear bias for our sample of galaxies , as described in section [ sec : cf - r0 ] .
details of this method can be found in the cited papers .
the predicted correlation lengths and abundances evolutions as a function of redshift are shown in figures [ fig : r0z - nz](a ) and ( b ) , for different values of the minimum mass of dmhs , @xmath255 .
we note that this model assumes that there is only one galaxy per dmh .
such an assumption can and often must be incorrect , especially in the case of the less massive galaxies in the lowest redshift bins . if satellite galaxies are taken into account , using the same galaxy bias factor measured from the data , a different @xmath255 would be inferred
in order to take into account the halo occupation distribution ( hod ) effect we follow the simple recipe provided in @xcite , by summing a step function for central galaxies and a power law for satellite galaxies , @xmath257 for @xmath258 . as inferred in @xcite , and following galaxy formation model predictions ( e.g. * ? ? ? * ) and hod modelling results ( e.g. * ? ? ?
* ) , we use @xmath259 .
the modified correlation lengths and abundance evolutions are shown in figures [ fig : r0z - nz](a ) and ( b ) with dotted lines , for different values of the minimum mass @xmath255 . in figures [ fig : r0z - nz](a ) and ( b ) we show the predicted relation , based on dark matter halo models and on halo occupation distribution , between the correlation lengths and the number densities of dark matter haloes , and how this relation evolves with redshift .
these predictions are in good agreement with our observations . as expected , the mass segregation effect is present , with the correlation lengths derived for galaxies with higher stellar mass agreeing well with the predicted correlation lengths of haloes with the highest dark matter masses .
we go a step further than this and associate the observed correlation lengths of our stellar mass selected samples with the predicted value of their host halo masses , thereby providing an estimate of the typical halo mass for each stellar mass selected bin . to do this
, we make the assumption that only one galaxy is hosted per halo .
later we apply a correction to take into account the halo occupation effect .
finally , we investigate the stellar masses of all galaxies within these haloes to obtain a measurement of the total stellar to halo mass within dark matter haloes .
this total mass measurement from our clustering analysis gives us a measure of how massive the haloes that host massive galaxies are . to obtain these mass measures , we interpolate masses from a grid of predictions in correlation length / redshift space , from the models described in section [ sec : dmm - cf ] .
each observed correlation length at a given redshift is then assigned to a model with a given dmh minimum mass .
this method has been used previously in papers such as @xcite .
the errors on the correlation lengths are propagated to obtain errors on the dmh minimum mass in the same way .
a similar approach is used to estimate the mass of dmhs , by comparing model results with the abundances .
table [ tab : results2 ] summarises the values derived for the dmh minimum mass @xmath255 from their correlation length values and from their abundances
. our assumption of a single galaxy per halo can result in overestimating the true value of the dmh minimum mass .
following @xcite , we estimate the effect of halo occupation using a simple prediction described in section [ sec : dmm - cf ] .
however as shown by @xcite , the effect of the `` one galaxy per halo '' assumption on the true measure of the dmh mass can be very important at the lower bias levels found in the lower redshift part of our catalogue .
the simple correction we apply is unlikely to be sufficient to correct for this effect , so we later discard these sub - samples from our following analysis , as shown in table [ tab : results2 ] .
more detailed models , such as hod models ( e.g. * ? ? ?
* ; * ? ? ?
* ) , are likely more appropriate to determine the mass of dmhs associated with a sample of galaxies .
for instance , in order to have a reliable estimate of the mass of the dmhs , such halo occupation models should fit directly from the angular correlation function determined for the samples in the lower redshift and stellar masses bins .
however , for our higher redshift samples with a strong bias , we have estimated our correlation amplitude on scales @xmath260 , where the effects of any extra halo terms is avoided ( see section [ sec : cf - acf ] ) . @xcite
have shown that one - halo term clustering is only observable on scales smaller than this limit .
therefore , our lower mass galaxies likely reside in the same haloes , but our measurements are made on scales where galaxies are in separate haloes .
this should reduce the effect of multiple halo occupation on our estimation of the dmh mass even for the lower mass samples .
one advantage in measuring clustering over large areas , such as in this study , is the ability to avoid the one - halo clustering regime , unlike in previous studies using smaller areas . as shown in figures [ fig : r0z - nz](a ) and ( b ) , the minimum mass of dark matter haloes , for a given sample of galaxies , can differ when estimated from the correlation length measurements or through using abundance measurements .
@xcite raised a similar issue in their analyses , and showed that this discrepancy could also come from an erroneous a priori assumption of the one - parameter halo bias model for a one - to - one correspondence between dark matter haloes and astrophysical sources .
when we correct for this effect the two estimations are broadly in agreement .
the masses of dmhs derived from the comparisons between the dmh model ( without halo occupation effects corrected ) and the correlation lengths and abundances are summarised in the table [ tab : results2 ] .
figure [ fig : mdmr0ab ] shows the mass of dmh inferred by the abundance measurements compared with the one inferred by the correlation length measurements , with both quantities corrected to account for halo distribution effects . the mild residual discrepancy between the masses estimated by comparing the models with the correlation length or the abundances ,
can be explained by several , likely combined , effects .
first , and most likely , the contamination effects between the different bins of our sample could be at the origin of the discrepancy seen in figure [ fig : mdmr0ab ] . as seen in section [ sec : cf - r0 ] , the correlation lengths we are using here are lower limits , due to possible contamination of the samples by interlopers .
therefore the masses of the dmhs we measure from the correlation lengths are also lower limits
. however , this is what we want as the models provide the minimum mass of the dmhs , and using the lower - limits is consistent with this .
also , interlopers will affect the measurements of the abundances of our different samples .
it is more likely that we are overestimating the number of objects in our less populated samples , which are the more massive systems . in this case , the masses of dmhs derived from the abundances of our most massive stellar samples are likely underestimated .
the combination of these two effects is likely the origin of the discrepancy between our two estimates , as shown in figure [ fig : mdmr0ab ] .
these contamination effects are included in our error budget as we discuss in section [ sec : cf - r0 ] , and , given the difficulty in obtaining a precise evaluation of the contamination , we do not correct our measurements directly for these effects .
a second possible explanation , would be incorrect redshift distributions for our samples .
@xcite found a similar discrepancy between the clustering and the abundance of their colour selected sample , and invoked an incorrect redshift distribution as a likely origin of such discrepancy . in our case as we are using simple top - hat functions , given the narrow size of our redshift bins , this effect relates directly to the contamination .
furthermore , @xcite claim that such a discrepancy could originate from cosmic variance and incorrect models of the halo mass function and halo bias function . in our case
, we take into account cosmic variance in our error budget , through the field - to - field variance , so we can partially reject this explanation . finally
, our models may be incorrect , despite being extensively tested .
we recompute the model using a slightly modified halo bias function from the original function of @xcite , as proposed by @xcite . by changing the halo bias function
we only modify the masses estimated from the correlation lengths .
as shown in figure [ fig : mdmr0ab ] , the masses inferred by this new model are not in better agreement , and the measurements remain almost within our original errorbars .
we decided to keep the measurements made using the @xcite function , as it is more commonly used .
however other inputs in our models , such as the halo mass function , may also be at the origin of the discrepancy .
@xcite have also shown that the clustering of dmhs depends not only on their halo mass but also on their assembly history , and especially on their formation time .
this `` halo assembly bias '' implies that , at a given mass , dmhs which assembled earlier are more clustered than dmhs that assembled later .
this effect will increase the discrepancy between masses measured from the correlation length matching , and from the abundance matching .
@xcite have shown that the bias induced by the environmental dependence of halo formation history at a fixed halo mass is stronger for faint red central galaxies and weaker for bright red galaxies .
the assembly bias enhances the two - point correlation by @xmath75 for bright galaxies , and suppresses it by the same amount for fainter galaxies .
however the relative strength in clustering between our more massive and less massive samples is @xmath261 , much higher than the effect of the assembly bias itself .
furthermore , @xcite report that this effect is weakest for dmhs with masses @xmath262which are within our range .
the masses of the dmhs calculated by matching correlation lengths and abundances to the models both suffer from uncertainties .
however as shown in figure [ fig : mdmr0ab ] , the two estimates are broadly in agreement within our errorbars .
we decided therefore to use the mass estimated from the correlation lengths corrected from the hod effect , as we have a more complete error budget in this case .
since we know the median stellar mass within each of our sample cuts , and the clustering and hence total masses of these same galaxies , we derive directly the stellar mass fraction as a function of total mass , and as a function of redshift for our sample .
we report these values in table [ tab : results2 ] , and display the results on figures [ fig : ratio - lita ] and [ fig : ratio - litb ] . in order to study the link between luminous and dark matter we make several assumptions .
first we assume that the minimum mass of the dmhs , determined as explained in section [ sec : dmm - mass ] , is an accurate approximation of the median mass of these dmhs ( @xmath264 ) .
as we are mainly studying dmhs with masses @xmath265 , and the mass function of dmhs drops rapidly above this limit , we conclude that this assumption is fair @xcite .
our second assumption , for comparison to other work , is that the virial mass of a system is equivalent to the total mass in the dmh , which has been confirmed by @xcite using a large n - body simulation . the third assumption we make , in order to compare our measurements with literature values , is that the mass in dark matter is equivalent to the total mass of the system .
indeed , the @xcite models we use to estimate the mass of the dmhs are based on the behaviour of dark matter , and do not take into account presence of baryonic matter . given that baryonic matter accounts for @xmath266 of the total density in the universe @xcite , this last assumption is justified , as @xmath267 of the density of the system is made up of dark matter . as explained in section [ sec : dmm - mass ] , we corrected our halo masses for multi - occupation of the haloes .
we also focus our analysis on the highly biased samples , those most likely to have a one - to - one correspondence between the galaxy and its dmh , sans any outer satellite galaxies , which is an issue we address later .
we can therefore assume that we are looking at the stellar mass fraction of the central galaxies , and are mostly free of any effect due to satellite galaxies in the present analysis . to compare our results to previous work , we collected measurements of stellar and total masses from the literature based on various different methods to estimate the masses of various dmhs .
these include total masses of disk - like galaxies based on rotation curves at @xmath268 , from @xcite , as well as at @xmath269 , from .
the stellar masses used in these studies are estimated from sed fits based on spectroscopically confirmed redshifts , and a wide range of photometry , similar to that described in section [ sec : data - mass - mass ] , but by using a salpeter imf .
we converted all stellar masses into a chabrier imf by dividing by a factor of @xmath270 @xcite .
we also added measurements of nearby early - type ( et ) and late - type ( lt ) galaxies from galaxy - galaxy lensing studies in the sloan digital sky survey ( sdss ) from @xcite .
stellar masses for these sdss galaxies are obtained from the method described in @xcite , based on the strength of the @xmath271 - break and the balmer absorption index h@xmath272 , for a @xcite imf ( which give estimates very close to the @xcite imf ) .
finally , we use the publicly available catalogue of groups in the sdss from @xcite to estimate the mass ratio for galaxy groups in the local universe .
total masses of these groups are determined by ranking the completeness - corrected group luminosity , and by assigning to them the mass of the dmhs , ranked according to their abundances predicted by models from @xcite .
the central galaxies of each group are assumed to be the most luminous galaxy in the group .
the stellar masses for this sample are determined using the method described in @xcite .
we add as a continuous line on figures [ fig : ratio - lita ] and [ fig : ratio - litb ] the value of the mean baryonic fraction in the universe , estimated from wmap5 @xcite , which is @xmath273 . from a study of galaxies in the local universe , @xcite measure the stellar mass density to be @xmath274 for a salpeter imf .
if we scale this value to a @xcite imf , and combine it with the cosmological parameters from wmap5 @xcite , we derive a mean stellar mass fraction in the universe of @xmath275 , which is shown in figures [ fig : ratio - lita ] and [ fig : ratio - litb ] as a dashed line .
the remaining part of the baryonic fraction that is not included in stars is possibly present in the form of hot and warm gas , as seen in clusters ( e.g. * ? ? ?
* ) . figure [ fig : ratio - lita ] shows that there is a good agreement between these very different mass measurements , especially at the massive end .
the @xmath213 ratio , even when estimated from completely different methods , shows a clear decrease at higher halo masses in all studies .
however , estimating accurate dmh masses is difficult , especially in the less massive cases .
dynamical estimates ( rotation curves , e.g. ; velocity dispersions , e.g. @xcite ) trace the innermost region of the galaxy , requiring extrapolation to estimate the total mass of the system . weak - lensing requires the use of stacking methods to provide an accurate estimate of the total mass , and tend to be difficult to achieve in the case of less massive dmhs ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in our case
, galaxies in less massive dark matter haloes are not strongly biased , making the assumption of a one - to - one correspondence between the galaxy and the dmh less appropriate , as mentioned earlier .
however our method , based on clustering properties at large scales , is efficient enough to determine the luminous - to - dark matter mass ratio for massive galaxies , and has the advantage of allowing us to estimate masses at higher redshifts , where for instance lensing can not be used .
overall , we find that more massive systems have a lower ratio of stellar - to - halo - mass , as shown in figure [ fig : ratio - lita ] .
this behaviour is seen in all of our measurements at @xmath276 , as well as in the local universe with the sdss , and thus appears to be independent of redshift , and is perhaps universal .
this relation implies that , at fixed total mass , the relationship between the mass of the central galaxy and its dmh does not evolve strongly with cosmic time .
physically this may imply that very massive galaxies struggle to increase their stellar mass in very clustered environments , resulting in a departure in the luminous - to - dark - matter mass ratio in massive dmhs compared to lower mass dmhs , which is below the average value in the local universe .
these observations also imply that there is a limit to how much stellar mass a galaxy can have , with a cut off at a few times @xmath277 .
this is consistent with the observed cut - off at high mass of the stellar mass function at low redshift ( e.g. * ? ? ?
furthermore , a decreasing stellar mass fraction with increasing halo mass has been observed in groups and clusters ( e.g. * ? ? ?
? * ; * ? ? ?
* ; * ? ? ?
@xcite show that brightest cluster galaxies ( bcgs ) contribute a large fraction to the total light of their host cluster , but become progressively less important in the overall luminosity budget in the highest mass clusters .
this effect is explained by differential growth between the bcgs and their host clusters , due to clusters accreting nearby galaxies in lower mass groups , while the bcgs grow modestly by merging or cannibalism @xcite .
similarly we can imagine that the clusters will accrete more dark matter in a similar process while the central galaxy grows more slowly .
in section [ sec : ratio - lit - centr ] we focussed our analysis on the central galaxy of the dmh . indeed , as explained in section [ sec : dmm - mass ] , our clustering measurements are made in a way where we can assume there is only one massive galaxy per dmh , from which we measure the mass of the host halo .
however , an overabundance of satellite galaxies has been observed in massive dmhs ( e.g. * ? ? ? * ; * ? ? ?
it is therefore possible that a non - negligible fraction of the stellar mass in the most massive dmhs resides in satellite galaxies ( e.g. * ? ? ?
* ; * ? ? ?
to study this possibility in more detail we use the sdss group catalogue @xcite , as it provides stellar masses for each member of each group .
we limit our analysis of these groups to those at @xmath278 in order to be complete at stellar masses @xmath279 , and at @xmath280 for @xmath281 .
we plot the ratio of the total stellar mass in the sdss group dmhs to the total mass of the dmhs in figure [ fig : ratio - moda ] .
as shown in this figure , the limited completeness of the sample does not allow us to conclude if the mass in satellite galaxies can account for the totality of the `` missing '' stellar mass in the dmh . however , the stellar mass distribution of satellite galaxies in the sdss , hosted by groups with dmhs masses of @xmath282@xmath283 , is shown in figure [ fig : sat - distr ] .
this study is limited to @xmath278 , and shows a plateau before reaching the completeness limit of @xmath279 .
this flattening at the faint end is similar to the observations of the luminosity function of the local group ( e.g. * ? ? ?
furthermore , a decrease in the stellar - mass fraction is also observed in local clusters .
@xcite observed a deficiency of x - ray gas in galaxy groups compared to clusters .
@xcite explain these observations by the fact that the missing x - ray gas has cooled in less massive systems to form stars .
this implies that a larger fraction of the baryonic matter in clusters is in the form of hot gas in more massive systems ( e.g. * ? ? ?
however , while the total mass can be derived from the virial mass for these massive clusters , the total stellar mass is harder to estimate .
a non - negligible amount of the light emitted by a cluster resides in its intracluster light ( icl ) , as dynamical processes can strip stars from galaxies which orbit in the intracluster space ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) , making the estimate of the total stellar - mass in a cluster difficult to measure accurately . in our study of the stellar - mass fraction from the sdss group catalogue ,
given the likely incompleteness in establishing group membership , we can only conclude that there is a weak trend in which the stellar - mass fraction in more massive systems is lower than in less massive ones , as shown in figure [ fig : ratio - moda ] .
further study of the role of satellite galaxies in the total amount of stellar - mass in a dmh is required to fully understand this result .
figure [ fig : ratio - litb ] shows that , at fixed stellar mass , the stellar mass fraction increases at lower redshift . in other words , the stellar component apparently grows more rapidly than the dark component between @xmath276 for galaxies with @xmath284 .
however , as shown in table [ tab : results2 ] , the average stellar mass in each stellar mass bin does not change significantly with cosmic time , while the average dmh mass decreases by a factor @xmath285 within these stellar mass selected samples .
furthermore , the star - formation rate does not change dramatically in the redshift range @xmath276 , and the high - mass end of the stellar mass function does not evolve significantly below @xmath286 .
this suggests that , instead of an increase of the average stellar mass , a statistical process is responsible for an apparent decrease of the average mass of massive dmhs between @xmath269 and @xmath6 for galaxies with stellar masses @xmath284 .
we propose that these observations for very massive galaxies ( @xmath284 ) can be explained by a `` halo downsizing '' effect ( e.g. * ? ? ?
we assume that the most massive galaxies are already in place at @xmath8 in the most massive dmhs , in terms of their stellar material .
gradually the stellar mass bin we select is populated by galaxies that have gained stellar mass with time but that are hosted by less massive dmhs than the older systems we find at higher redshifts .
meanwhile the most massive galaxies do not gain in stellar - mass , nor in dmh mass , due to a quenching of their star formation and a weak merging rate @xcite . as a consequence the average mass of the dmhs for these galaxies decreases , and the stellar mass ratio increases , which is what we find .
such an idea is supported by the `` archaeological downsizing '' @xcite , and the observation that the most massive galaxies appear to have assembled their stellar mass earlier than younger galaxies .
galaxies with @xmath287assembled half of their stellar mass before @xmath288 , and more than 90% of their mass was already in place at @xmath289 @xcite .
such observations imply that the mechanism responsible for quenching star - formation in galaxies is strongly mass - dependent and that it occurs earlier in the most massive galaxies @xcite . using semi - analytical models ,
@xcite have shown that the downsizing effect observed for red galaxies , could naturally result from a shutdown in star - formation .
more massive central galaxies in today s universe formed earlier and over a shorter period of time than less massive galaxies . a critical mass of dmhs of @xmath290 , above which a shutdown in star - formation occurs , is also observed in these studies , corresponding roughly to the mass where we observe a decline in the stellar - to - dark - matter - mass - ratio in figure [ fig : ratio - lita ] . as mentioned in section [ sec : dmm - mass ] , the `` assembly bias '' effect could also mimic the observed increase of stellar - mass ratio with lower redshifts .
@xcite have shown that , at fixed halo mass , more massive and clustered central galaxies tend to occupy haloes that formed earlier , while less clustered central galaxies occupy haloes that formed later .
however , @xcite have demonstrated that this effect is unobservable for massive dmhs ( @xmath291 ) .
although we can not rule out that assembly bias could artificially increase the ratio of stellar to dark matter mass which we observe , this effect is likely not large enough to account for the observed evolution .
the halo downsizing " effect could also be used to explain the observation that galaxies in more massive haloes contain a lower stellar - mass fraction than those in less massive haloes , as seen in figure [ fig : ratio - lita ] .
a tentative explanation of the formation of massive galaxies could be that at higher redshift the most massive dmhs formed rapidly by accretion or mergers @xcite .
this would provide galaxies less time to form stars and become more subject to a shutdown of their star - formation , compared with less massive dmhs @xcite .
moreover , the gas accreted in the more massive dmhs is hot , which can not cool quickly in such massive potentials , suppressing further the star - formation in these massive galaxies , as observed for groups and clusters @xcite .
the main difficulties in testing directly the scenario described above are in measuring accurately the stellar and dark matter masses involved .
we assume within our clustering - derived masses , and through the other mass estimations , that masses of dmhs , dynamical masses , and total masses , are equivalent . however , it is not always easy to compare dynamical masses with the often used @xmath292 and @xmath293 masses , derived for a radius of a system corresponding to overdensities of 500 and 200 , relative to the critical density at a given redshift . furthermore , dynamical masses are extremely difficult to measure in low mass systems , particularly at high redshift . on the other hand ,
the total amount of stellar mass and gas mass in the most massive systems is also difficult to measure , and requires extensive observations .
for instance , a deeper study in the near - infrared bands of local groups and clusters is required to obtain a better measure of the stellar mass fractions in the satellites galaxies and in the icl .
in addition , a better estimate of the gas content of clusters and groups , through very detailed x - ray studies of a large sample , would be extremely valuable .
semi - analytic models are known to struggle in reproducing the observed formation of the most massive galaxies , pushing the formation of these systems to very late epochs ( e.g. * ? ? ?
* ; * ? ? ?
by introducing agn feedback ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) the most recent models now manage to reproduce a star formation `` downsizing '' effect , implying that the stars in more massive galaxies form early in the universe @xcite . in order to check
if these models are able to reproduce the behaviour we observe between stellar mass fraction and total mass , how this fraction evolves with redshift , and to better understand the underlying physical processes involved , we compare our results to those from one of the current leading models .
we use the millennium simulations @xcite in which were implemented the semi - analytic models developed by @xcite .
the millennium model predictions for the stellar mass fraction of the central galaxy of dmhs , at different redshifts ( @xmath294 , @xmath295 , @xmath39 ) , are shown in figure [ fig : ratio - moda ] , with the dashed area showing the rms for predicted dmhs at @xmath295 .
as expected , the peak in the mass ratio vs. halo mass occurs at @xmath296 , where star formation is known to be the most efficient . in less massive dmhs
the infall time of gas clouds is too long to form stars efficiently , while in more massive haloes it is the cooling time of the gas cloud that is too long @xcite . by construction ,
the mean value of this ratio in the models is similar to the observed stellar - mass fraction in the local universe .
the model from @xcite reproduces well the behaviour observed for the central galaxies in the most massive dmhs at low redshifts .
however , at higher redshifts the agreement is not as good , with the model showing lower stellar mass fractions compared to our observations .
these models seem to not form massive galaxies quickly enough to reproduce the observations ( see also * ? ? ?
* for this comparison done on stellar mass densities ) .
previously , @xcite used @xcite s models to study the high - redshift galaxy population .
they compared the observed mass functions from @xcite and with simulations , and found an overall overprediction by the simulations between @xmath297 in the mass range @xmath298@xmath299 . on the other hand
their prediction shows a deficiency in this redshift range for the most massive galaxies with @xmath300 .
@xcite use a different semi - analytic model with a different recipe for reproducing agn feedback , also based on the millennium simulation , but find results very similar to the @xcite model @xcite . by comparing the mass functions derived from their models with the measurements by @xcite ,
@xcite also find an underprediction of the number densities for the most massive galaxies at high redshifts .
using the sample of @xmath19-band selected galaxies at @xmath301 presented here , @xcite confirm this result by studying the mass growth of the most massive galaxies .
these underpredictions in the number of massive galaxies at higher redshift by the semi - analytic models is likely responsible for the disagreement seen in figure [ fig : ratio - moda ] . on the other hand ,
observed mass errors in massive systems can be partially responsible for the overestimation at the steep end of the mass function ( e.g. * ? ? ?
furthermore , @xcite introduced a simple model based on n - body simulations and abundance matching between galaxies and dmhs that can shed some light on this problem .
this model is tuned a posteriori to reproduce the observed evolution of the stellar - mass function and star - formation rate history within galaxies . with this simple model ,
@xcite reproduce the shape of the stellar mass fraction as a function of dmh mass as shown in figure [ fig : ratio - moda ] .
their model does not address explicitly , by construction , the growth history of galaxies more massive than @xmath302 .
however , their model infers a shift in the masses of dmh for which the stellar mass fraction peaks from higher mass at high redshift , to lower mass at lower redshift ( @xmath303at @xmath8 to @xmath304at @xmath305 ) , where the millennium simulation predicts the opposite .
this model favours a range of characteristic masses of dmhs for which a shutdown in star - formation occurs , and is not defined for dmh masses of @xmath306(e.g .
our current dataset can not address this issue , but larger and deeper surveys should be able to observe such a phenomenon . as mentioned in section [ sec : ratio - lit ] , a non - negligible amount of stellar mass within massive dark matter haloes could be contained in satellite galaxies .
we investigate this idea further to determine the actual stellar to halo mass ratio for our most massive systems .
we also use the available semi - analytic catalogue derived from @xcite , in which central and satellite galaxies are differentiated , to estimate the total stellar mass in each simulated dmh .
we do this by adding the stellar mass included in satellites to the mass of the central galaxies .
we find that in high mass dmhs the stellar mass fraction almost matches the value of the mean stellar fraction in the universe , as shown by the red long - dash - short - dashed line in figure [ fig : ratio - moda ] .
this prediction of a nearly constant stellar mass fraction is not surprising , as it is an effect of the construction of the semi - analytical models .
the most massive systems ( groups and clusters ) in the models are built from dmhs with a similar mass distribution , and over a similar time - scale .
they will therefore form a similar amount of stars , resulting in a similar stellar - mass fraction . in figure
[ fig : ratio - moda ] we show the predicted evolution of the stellar mass fraction with mass for dmhs derived by @xcite using the model of @xcite , with a nearly horizontal dotted line with a slope of @xmath307log@xmath308log@xmath309 .
the model from @xcite leads to similar conclusions as @xcite .
furthermore , the trend found within the sdss groups , with a lower mass completeness limit of @xmath310 , seems to converge to the slope determined by the models . on the other hand , @xcite derived an estimate of total stellar masses and dynamical masses from the sample of clusters from @xcite and @xcite , both taking into account intracluster light .
the slope derived from their conservative model , @xmath307log@xmath308log@xmath311 , does not match the stellar mass fraction from the sdss groups .
this could indicate an underestimate of the total stellar - mass in their study . in order to verify
if the stellar mass fraction derived from the sdss groups , and the predictions from the @xcite s model , are in good agreement , we show in figure [ fig : sat - distr ] the distribution in stellar mass of satellite galaxies selected in dmhs of masses @xmath282@xmath312from the sdss .
this figure shows that the satellite galaxies in the sdss groups reach a plateau before the completeness limit of @xmath313 , while the number of satellites in the simulations keeps rising .
such a flattening at the faint - end is observed in the luminosity function of the local group ( e.g. * ? ? ?
* ) , confirming that simulations fail to reproduce the luminosity / mass function of satellites galaxies in groups .
this is part of the well known missing satellite problem in cdm ( e.g. * ? ? ?
at the completeness limits of the sdss data there are twice as many satellites per dmh in the simulations than in the observations . within the most massive part of this distribution it appears that the simulations lack massive satellite galaxies when compared to the observed sdss groups .
this is more likely due to the limited - volume of the simulations , but could also be due to the lack of massive galaxies in the overall mass functions as discussed above .
however , the observations contain many uncertainties as well .
estimating the stellar - masses of faint objects is still tricky , and the group membership of satellite galaxies , especially at the faint end , is difficult to ascertain . to solve this issue , a deeper study of groups and clusters
is required to estimate , in a better manner , their total stellar and dark - matter masses .
finally , figure [ fig : ratio - modb ] shows the variation of the stellar - mass fraction with redshift predicted from the simulation compared to the data themselves .
this figure shows that no variation of the stellar mass fraction with redshift is predicted by the simulation for galaxies with stellar masses @xmath314 .
the n - body simulations present a very tight correlation between the mass of the dmhs and the number of galaxies hosted by each dmh at any epoch , and are in this sense very close to smooth particle hydrodynamic simulations and analytic halo occupation models ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
therefore , the stellar mass fraction in a given range of stellar mass is constant with epoch , and by construction equal to the mean value in the local universe , explaining their failure to reproduce the observed increase in stellar mass fraction with lower redshift .
moreover , as shown in figure [ fig : ratio - modb ] , the effect of the fraction of mass in satellites is negligible here as well .
however , because of the limited volume of the simulations , the number of galaxies with @xmath302does not allow for a direct comparisons with our observations .
simulations over larger volume are required , and observations of the stellar mass ratio of less massive galaxies at @xmath276 would be very useful as well .
we present in this paper an analysis of the clustering properties of a stellar mass selected sample of galaxies over 0.7 deg@xmath10 taken from the palomar observatory wide - field infrared survey ( powir ) . by utilising information from three separate fields ,
we measure the two - point correlation function , and based on this we derive the spatial correlation length and bias for our sample of galaxies within various stellar mass bins at @xmath0 .
our major results are : i. we find that the correlation length ( @xmath315 ) varies with both redshift and stellar mass .
we find that the largest correlation lengths ( most clustered systems ) are found for galaxies at the highest stellar masses with @xmath277@xmath316at @xmath8 .
the correlation length for galaxies at all stellar mass selections are found to decrease at lower redshift .
we derive the dark matter halo masses for these stellar mass selected systems utilising the correlation lengths and models from @xcite . due to a likely contamination between our samples , the correlation lengths and therefore the dark matter halo masses
are lower limit estimates
. however we take into account this effect in our error budgets .
we find that halo masses for our galaxies selected by @xmath317are a factor of 30 to 170 times the stellar masses of each galaxy .
this is in very good agreement with other methods for measuring halo masses , including lensing and kinematics .
we find a remarkable relation between dark matter halo mass ( @xmath212 ) , and the ratio of stellar - to - dark - matter - mass ( @xmath318 ) for the central galaxy in the halo , and find that this relation does not vary much with redshift .
this correlation is such that the more massive a dark matter halo mass @xmath212 is , the lower the ratio of stellar - to - halo - mass ( @xmath318 ) .
we further find that this correlation exists over at least two orders of magnitude from halo masses @xmath319 to @xmath320 .
this is true for all methods of determining halo masses , from lensing , kinematics , and clustering .
\iv . within our massive galaxy sample at @xmath286
we find that the ratio of stellar - to - halo - mass increases from high to low redshifts .
this correlation implies that at a given stellar mass selection , the average underlying mass of the host halo decreases at lower redshifts .
this may imply that the most stellar massive systems at the highest redshifts are hosted by very massive haloes , but at lower redshift they stop forming new stars , and therefore do not increase their stellar mass , while other systems in the lower mass dark matter haloes grow to reach similar stellar mass .
this is an example of `` halo downsizing '' .
the ` halo assembly bias ' implying that clustering depends not only on the mass of the dark matter halo , but also on the assembly history and environment of the galaxy , can also contribute partially to this observed increase in the stellar mass fraction with redshift .
v. we compare our results to the semi - analytic models built using the millennium simulation .
we find that there is roughly a good agreement between models and the data for the central galaxy .
the predicted value of the total - stellar - to - total - halo mass ratio decreases only slightly with increasing halo mass , also in rough agreement with our observational results .
we use the sdss group catalogue from @xcite to examine whether the missing stellar mass can be accounted for by observed satellite galaxies .
we find that , even though the overall agreement is good , the model marginally overpredicts the number of faint galaxies in comparison to the sdss catalogue .
moreover the millennium simulation fails to reproduce the `` halo downsizing '' effect we observe .
we argue that this missing mass , in the more massive haloes , is most probably in the form of baryons in a warm / hot phase .
we thank the anonymous referee for his detailed and constructive report .
the palomar and deep2 surveys would not have been completed without the active help of the staff at the palomar and keck observatories , and the deep2 and powir teams , particularly sandy faber and richard ellis .
funding to support this effort came from a national science foundation astronomy & astrophysics fellowship , and grants from the uk particle physics and astronomy research council ( pparc ) .
we also thank simone weinmann and xiaohu yang for making their group catalogue publicly available .
we are also grateful to the sdss collaboration for producing a wonderful data set and for making their products publicly available . the millennium simulation databases used in this paper and the web application providing online access to them were constructed as part of the activities of the german astrophysical virtual observatory . , m. , faber , s. m. , newman , j. , et al .
2003 , in discoveries and research prospects from 6- to 10-meter - class telescopes ii . edited by guhathakurta , puragra .
proceedings of the spie , volume 4834 , pp .
161 - 172 ( 2003 ) . , 161172 , s. m. , phillips , a. c. , kibrick , r. i. , et al
. 2003 , in presented at the society of photo - optical instrumentation engineers ( spie ) conference , vol . 4841 , society of photo - optical instrumentation engineers ( spie ) conference series , ed . m. iye & a. f. m. moorwood , 16571669
|
we present a study on the clustering of a stellar mass selected sample of 18,482 galaxies with stellar masses @xmath0at redshifts @xmath1 , taken from the palomar observatory wide - field infrared survey .
we examine the clustering properties of these stellar mass selected samples as a function of redshift and stellar mass , and discuss the implications of measured clustering strengths in terms of their likely halo masses .
we find that galaxies with high stellar masses have a progressively higher clustering strength , and amplitude , than galaxies with lower stellar masses .
we also find that galaxies within a fixed stellar mass range have a higher clustering strength at higher redshifts .
we furthermore use our measured clustering strengths , combined with models from @xcite , to determine the average total masses of the dark matter haloes hosting these galaxies .
we conclude that for all galaxies in our sample the stellar - mass - to - total - mass ratio is always lower than the universal baryonic mass fraction . using our results , and a compilation from the literature ,
we furthermore show that there is a strong correlation between stellar - mass - to - total - mass ratio and derived halo masses for central galaxies , such that more massive haloes contain a lower fraction of their mass in the form of stars over our entire redshift range . for central galaxies in haloes with masses @xmath2we
find that this ratio is @xmath3 , much lower than the universal baryonic mass fraction .
we show that the remaining baryonic mass is included partially in stars within satellite galaxies in these haloes , and as diffuse hot and warm gas .
we also find that , at a fixed stellar mass , the stellar - to - total - mass ratio increases at lower redshifts .
this suggests that galaxies at a fixed stellar mass form later in lower mass dark matter haloes , and earlier in massive haloes .
we interpret this as a `` halo downsizing '' effect , however some of this evolution could be attributed to halo assembly bias .
[ firstpage ] galaxies : evolution galaxies : high - redshift cosmology : observations large - scale structure of universe .
|
diffractive scintillations of small angular - diameter radio sources indicate that the interstellar electron density spectrum on lengthscales @xmath0 is nearly kolmogorov ; i.e. r.m.s .
density fluctuations across a lengthscale @xmath1 are nearly proportional to @xmath2 .
they also establish that there are large variations in the amplitude of the density spectrum along different lines of sight .
rickett ( 1977 , 1990 ) and @xcite review the observations of diffractive scintillation and their interpretation .
they also discuss refractive scintillations and dispersion measure fluctuations , which probe density fluctuations on scales larger than the diffractive scales .
non - diffractive measurements tend to indicate that the kolmogorov spectrum extends to much larger scales . however , we focus primarily on diffractive measurements because they are much more sensitive
. density fluctuations that obey the kolmogorov scaling occur in homogeneous subsonic _ hydrodynamic _ turbulence .
they are due to the entropy mode , a zero - frequency isobaric mode whose density fluctuations are offset by temperature fluctuations . since subsonic turbulence is nearly incompressible ,
the velocity fluctuations follow kolmogorov s scaling . to a good approximation ,
the entropy mode is passively mixed by the velocity field , so it also conforms to the kolmogorov spectrum .. ] density fluctuations in the earth s atmosphere , which cause stars to twinkle , obey the kolmogorov scaling .
they arise from the passive mixing of the entropy mode .
the electron density spectrum in the interstellar medium can not be explained by hydrodynamic turbulence . because the medium is ionized
, magnetic effects must be accounted for .
this is evident since the lengthscales probed by diffractive scintillations are smaller than the collisional mean free paths of both electrons and protons .
if the magnetic field were negligible , freely streaming plasma would wipe out density fluctuations at diffractive scales . in the presence of a magnetic field ,
electrons and protons are tied to fieldlines at the scale of their gyroradii . for typical interstellar field strengths ,
these gyroradii are smaller than the diffractive scales .
a magnetic field thus impedes the plasma from streaming across fieldlines , and allows the turbulent cascade and the associated density fluctuations to reach very small scales across the fieldlines before dissipating .
therefore , a theory for compressible turbulence in magnetized plasmas is required to explain the observed density spectra .
our objective is to develop this theory . until now , the only description of density fluctuations in interstellar plasmas was by higdon ( 1984 , 1986 ) .
these papers , while prescient , preceded a theory for _ incompressible _ mhd turbulence , and therefore did not account for the full dynamics of the cascade .
we compare higdon s theory with ours in [ sec : higdon ] .
our compressible theory extends the theory of incompressible mhd turbulence given by @xcite by including a slightly compressible slow mode and a passive entropy mode .
we also consider kinetic effects : on sufficiently short lengthscales , the mean free paths of the particles are significant , and the equations of compressible mhd must be modified .
this is especially important for damping . in a future paper ii
, we apply the theory developed here to estimate amplitudes for density fluctuations produced in supernova shocks , hii regions , stellar winds , and thermally unstable regions .
these are then compared to scattering measures observed along different lines of sight . before considering compressible turbulence ,
we discuss incompressible mhd turbulence , focusing on issues that are important for the compressible case .
goldreich & sridhar ( 1995 , 1997 ) propose a picture of the dynamics of incompressible strong mhd turbulence and describe the power spectra of alfvn waves , slow waves , and passive scalars .
we extend their picture to cover additional features such as the parallel cascades of both slow waves and passive scalars . throughout this paper ,
`` parallel '' and `` transverse '' refer to the orientation relative to the `` local mean magnetic field '' , which is the magnetic field averaged over the scale of interest .
our discussion of incompressible mhd turbulence , while somewhat lengthy , is important for understanding the extension to compressible turbulence that follows .
consider a uniform unperturbed plasma with an embedded magnetic field .
turbulence is excited at the mhd outer scale , @xmath3 , by random and statistically isotropic forcing , with r.m.s .
velocity fluctuations and r.m.s . magnetic field fluctuations
( in velocity units ) which are comparable to the alfvn speed , @xmath4 . , in which case @xmath3 would be defined as the lengthscale at which the fluctuations extrapolate to @xmath4 . ] as the turbulence cascades from the mhd outer scale to smaller scales , power concentrates in modes with increasingly transverse wave vectors .
the inertial range velocity spectrum applies to lengthscales below @xmath3 but above the dissipation scale .
it is anisotropic and is characterized by @xmath5 the inertial range magnetic field spectrum is identical . here
@xmath6 is the lengthscale transverse to the local mean magnetic field , @xmath7 is the r.m.s .
velocity fluctuation across @xmath6 , and @xmath8 is the lengthscale parallel to the local mean magnetic field across which the velocity fluctuation is @xmath7 .
we interpret @xmath8 as the elongation along the magnetic field of an `` eddy '' which has a size @xmath6 transverse to the magnetic field ; it is not an independent variable , but is a function of @xmath6 .
deep within the inertial range , where @xmath9 , eddies are highly elongated along the magnetic field : @xmath10 . in the following subsections , we explain the physics underlying the spectrum , and consider some of the implications .
arbitrary disturbances can be decomposed into alfvn waves and slow waves .
the appendix summarizes the properties of these waves in the more general case of compressible mhd . in incompressible mhd , alfvn waves and slow waves
are usually referred to as shear - alfvn waves and pseudo - alfvn waves , but the former designation is more convenient for making the connection with compressible mhd .
our understanding of the mhd turbulence is based on two facts : ( i ) mhd wave - packets propagate at the alfvn speed either parallel or antiparallel to the local mean magnetic field ; and ( ii ) nonlinear interactions are restricted to collisions between oppositely directed wave - packets .
these facts imply that in encounters between oppositely directed wave - packets , each wave - packet is distorted as it follows field lines perturbed by its collision partner .
a wave - packet cascades when the fieldlines that it is propagating along have spread by a distance comparable to its transverse size .
alfvn waves have quasi two - dimensional velocity and magnetic field fluctuations which are confined to planes perpendicular to the local mean magnetic field . as their more complete name shear - alfvn implies
, they dominate the shear of the mapping of planes transverse to the local mean magnetic field produced by field line wander .
thus , alfvn waves control the dynamics of mhd cascades ; slow waves may be ignored when considering the dynamics of alfvn waves . in strong mhd turbulence the cascade time of an alfvn
wave - packet is comparable to its travel time across the parallel length of a single oppositely directed alfvn wave - packet of similar size .
@xcite refer to this balance of timescales as `` critical balance '' .
it relates the parallel size of a wave - packet , @xmath8 , to its transverse size , @xmath6 .
wave - packets of transverse size @xmath6 cascade when the fieldlines they follow wander relative to each other by a transverse distance @xmath6 .
critical balance implies that this occurs over a parallel distance @xmath8 .
the alfvn wave spectrum is given by equations ( [ eq : spectrum1 ] ) and ( [ eq : spectrum2 ] ) , with @xmath7 referring to the velocity fluctuations of the alfvn waves .
it is deduced from two scaling arguments : ( i ) kolmogorov s argument that the cascade time @xmath11 leads to an energy cascade rate , @xmath12 , which is independent of lengthscale ; and ( ii ) the `` critical balance '' assertion that the linear wave period which characterizes the alfvn waves in a wave - packet is comparable to the nonlinear cascade time of that wave - packet , i.e. @xmath13 . before considering slow waves in mhd turbulence ,
we discuss two topics that are governed by the dynamics of alfvn waves only : eddies and passive scalars . because of their transverse polarization , alfvn waves are responsible for the wandering of magnetic fieldlines .
a snapshot of wandering fieldlines is shown in figure [ fig : fieldlines ] .
each of these fieldlines passes through a localized region of size @xmath6 in one plane transverse to the mean magnetic field .
away from this plane the bundle of fieldlines diverges due to the differential wandering of the individual lines . at a second plane ,
the bundle s cross sectional area has approximately doubled .
critical balance implies that the distance to this second plane is comparable to the parallel wavelength which characterizes the bundle , @xmath8 . as the bundle spreads , other fieldlines , not depicted , enter from its sides . in general , the neighboring fieldlines of any individual field line within a region of transverse size @xmath6 change substantially over a parallel distance of order @xmath8 .
it is natural to think of @xmath8 as the parallel size of an `` eddy '' that has transverse size @xmath6 .
two eddies with the same transverse lengthscale that are separated by a parallel distance greater than their @xmath8 incorporate different fieldlines , and hence are statistically independent .
eddies are distinct from wave - packets .
the former are rooted in the fluid whereas the latter propagate up and down magnetic fieldlines at the alfvn speed . aside from their anisotropy ,
eddies in mhd turbulence are similar to those in hydrodynamic turbulence . they are spatially localized structures with characteristic velocity fluctuations and lifetimes .
the r.m.s .
velocity difference between two points is determined by the smallest eddy that contains both .
different eddies of a given size are statistically independent . the three - dimensional spectrum for r.m.s . velocity fluctuations across transverse lengthscales @xmath6 and parallel lengthscales @xmath14 is @xmath15 there is negligible additional power within an eddy on parallel lengthscales smaller than @xmath8 , so for @xmath16 , @xmath17 .
for @xmath18 , the smallest eddy that contains both @xmath6 and @xmath14 has a transverse lengthscale @xmath19 which satisfies @xmath20 .
the velocity fluctuation of this eddy is obtained by solving this equation for @xmath19 ( eq . [
[ eq : spectrum2 ] ] ) , and inserting this @xmath19 in equation ( [ eq : spectrum1 ] ) .
contours of the three - dimensional spectrum are plotted in figure [ fig : contours ] .
each contour represents eddies of a characteristic size .
@xcite give the three - dimensional spectrum in fourier - space . since eddies that are separated by more than @xmath8 are statistically independent , the power spectrum at a fixed transverse wavenumber @xmath21 is independent of the parallel wavenumber @xmath22 in the corresponding region of fourier - space , i.e. where @xmath23 .
the turbulent cascade is generally viewed as `` proceeding '' from larger eddies to smaller eddies as this is the direction of energy transfer . however , smaller eddies cascade many times in the time that a large eddy cascades .
this is particularly important in turbulent mixing .
consider the evolution of two fluid elements whose initial separation is larger than the dissipation scale . on cascade timescale @xmath24
, their transverse separation will random walk a distance @xmath6 as the result of the cascade of smaller eddies .
therefore , on a timescale comparable to an eddy s cascade time , the transverse locations of its component fluid elements whose sizes may be considered to be comparable to the dissipation scale are completely randomized .
moreover , since mixing at the dissipation scale causes neighbouring fluid elements to be rapidly homogenized , transverse smoothing of the eddy occurs on the timescale that it cascades .
rapid transverse mixing in mhd turbulence is similar to the more familiar isotropic mixing in hydrodynamic turbulence
. a passive scalar , @xmath25 , satisfies the continuity equation , @xmath26 , and does not affect the fluid s evolution .
it could represent , for example , the concentration of a contaminant .
we consider the spectrum of a passive scalar mixed by the alfvn wave cascade .
these considerations are important for our subsequent investigation of compressible turbulence .
they are also helpful for understanding the slow wave spectrum .
we discuss the passive scalar spectrum both in the inertial range and also below the scale at which the alfvn wave spectrum is cutoff . as we show in this subsection , the transverse spectrum of the passive scalar in the inertial range is @xmath27 where @xmath6 is the lengthscale transverse to the local mean magnetic field and @xmath28 is the r.m.s .
fluctuation in the passive scalar across @xmath6 .
the passive scalar parallel spectrum is the same as the alfvn wave parallel spectrum given in equation ( [ eq : spectrum2 ] ) , where @xmath8 is now to be interpreted as the lengthscale parallel to the local mean magnetic field across which the passive scalar fluctuation is @xmath28 .
mixing of the passive scalar is due to alfvn waves .
slow wave mixing is negligible .
this is because transverse velocity gradients are much larger than parallel ones in mhd turbulence .
thus alfvn waves , whose velocity fluctuations are perpendicular to the magnetic field , are much more effective at mixing than slow waves , whose velocity fluctuations are nearly parallel to the magnetic field .
the transverse cascade arises from the shuffling of fieldlines as alfvn waves propagate through the fluid . the transverse spectrum ( eq . [
[ eq : cspectrum1 ] ] ) follows from the kolmogorov - like hypothesis that the cascade rate of the `` energy '' in the scalar field is independent of lengthscale ; i.e. @xmath29 is constant , where @xmath24 is the passive scalar cascade time , which is assumed to be proportional to the cascade time of alfvn waves .
comparing this with the constancy of the kinetic energy cascade rate , @xmath30 , we conclude that @xmath31 , which implies equation ( [ eq : cspectrum1 ] ) .
a similar argument holds for the cascade of a passive scalar in hydrodynamic turbulence ( e.g. , tennekes & lumley 1972 ) .
the parallel cascade of a passive scalar is more subtle .
it might appear that a passive scalar can not cascade along fieldlines since , neglecting dissipation , both the scalar and the magnetic field are frozen to the fluid , and thus the scalar must be frozen to fieldlines .
in that case there certainly could not be a parallel cascade .
if the scalar were injected on large scales , then fluctuations with smaller wavelengths along the magnetic field would not be generated .
however , dissipation can not be neglected .
it is an essential part of mhd turbulence , as it is of hydrodynamic turbulence .
for example , the description of turbulent mixing in [ subsec : eddies ] depends crucially upon small scale dissipation .
perhaps the best way to understand the parallel cascade is to consider mixing on the transverse lengthscale @xmath6 within two planes which are perpendicular to the local mean magnetic field , and which are separated by a parallel distance larger than @xmath8 .
velocity fluctuations within the two planes are statistically independent .
this is evident because a bundle of fieldlines can not be localized within a transverse distance @xmath6 over a parallel separation greater than @xmath8 .
even a pair of fluid elements , one in each plane , that are initially on the same fieldline are mixed into two regions with different values of passive scalar concentration .
it follows that the parallel cascade of the passive scalar also obeys equation ( [ eq : spectrum2 ] ) .
a passive scalar cascade may extend below the transverse scale at which the mhd cascade is cut - off . mixing on these scales is driven by fluid motions at @xmath32 which results in a scale - independent mixing time equal to the cascade time at @xmath32 .
this yields @xmath33 a similar argument applies in hydrodynamic turbulence .
@xcite call this regime in hydrodynamic turbulence the `` viscous - convective subrange '' .
the slow wave spectrum is the same as that of the alfvn waves .
it is given by equations ( [ eq : spectrum1 ] ) and ( [ eq : spectrum2 ] ) , with @xmath7 referring to the velocity fluctuations of the slow waves .
this is a consequence of the similar kinematics of slow waves and shear alfvn waves and the fact that both are cascaded by shear alfvn waves .
slow waves obey the same linear wave equation as alfvn waves , and to lowest nonlinear order they travel up and down the local mean magnetic field lines at the alfvn speed just as alfvn waves do . however , the dynamics of the mhd cascade is controlled by the alfvn waves @xcite because their velocity and magnetic field fluctuations are perpendicular to the local mean magnetic field , whereas those of the slow waves are nearly parallel to it . since perpendicular gradients are much larger than parallel ones in the mhd cascade , alfvn waves are much more effective at mixing than are slow waves . hence
, alfvn waves cascade both themselves and slow waves , whereas slow waves cascade neither .
the transverse mixing of the slow waves by alfvn waves is analogous to the mixing of a passive scalar . as discussed in
[ sec : passivescalar ] , a passive scalar assumes the same inertial range spectrum as that of the velocity field which is responsible for its mixing .
thus , equation ( [ eq : spectrum1 ] ) is also applicable to the velocity fluctuations of the slow waves .
similarly , the parallel cascade of slow waves is analogous to the parallel cascade of a passive scalar .
since alfvn waves cascade in the time they move a distance @xmath8 , slow waves separated by this distance are independently mixed .
thus , equation ( [ eq : spectrum2 ] ) also applies to slow waves .
there is , however , a conceptual difference between the parallel cascades of the passive scalar and of the slow mode . in the absence of dissipation
a passive scalar is frozen to fieldlines , whereas slow mode wave - packets travel along them at the alfvn speed .
a passive scalar has a parallel cascade because alfvnic fluctuations are statistically independent within two transverse planes _
frozen in the fluid _ and separated by @xmath8 .
the parallel cascade of slow waves occurs because two transverse planes separated by @xmath8 and _ travelling at the alfvn speed in the same direction _ experience uncorrelated sequences of distortions suffered as a result of interactions with oppositely directed alfvn waves .
nevertheless , these two requirements are both satisfied in the mhd cascade when the transverse planes are separated by a distance greater than @xmath8 , and so the passive scalar and the slow mode have the same parallel spectrum . whereas passive scalar mixing is due to eddies , slow wave mixing may be thought of as due to `` travelling eddies '' , i.e. eddies which travel up and down the magnetic field at the alfvn speed .
numerical simulations offer some support for the above description of incompressible mhd turbulence .
those by @xcite support both equations ( [ eq : spectrum1 ] ) and ( [ eq : spectrum2 ] ) , and those by @xcite support equation ( [ eq : spectrum1 ] ) .
however , although the simulations of @xcite support equation ( [ eq : spectrum2 ] ) , they yield @xmath34 instead of equation ( [ eq : spectrum1 ] ) .
because the simulations of @xcite are stirred highly anisotropically , whereas those of @xcite and @xcite are stirred isotropically , it is not clear whether these disparate results conflict .
@xcite speculate that the discrepancy between their spectrum and the scaling prediction of @xcite results from intermittency . in any case
, we expect that the physical picture of a critically balanced cascade , which underlies goldreich & sridhar s description of mhd turbulence , remains valid .
even if the spectrum is proportional to @xmath35 , we expect that the results of this investigation which assumes a spectrum proportional to @xmath36would not be significantly altered .
the simulations of @xcite confirm that a passive scalar has the same transverse spectrum as that of alfvn waves , although both are proportional to @xmath35 .
they also indicate that the parallel cascade of a passive scalar conforms to equation ( [ eq : spectrum2 ] ) .
@xcite present results from simulations of the interaction between oppositely directed slow and alfvn waves ; the slow waves cascade whereas the alfvn waves do not .
they also compute the spectrum of slow waves in a simulation of mhd turbulence and find that both its transverse and longitudinal behavior matches that of the alfvn wave spectrum .
scaling arguments do not yield values for the `` kolmogorov constants '' , the order - unity multiplicative constants of the spectrum .
however , they can be obtained from simulations .
we define them such that equation ( [ eq : spectrum1 ] ) remains valid , i.e. we take @xmath3 to be the separation at which the r.m.s . velocity difference is equal to , or extrapolates to , @xmath4 .
two kolmogorov constants , @xmath37 and @xmath38 , are needed in this paper : @xmath39 in these definitions , @xmath6 and @xmath8 are the inverses of the wavenumbers transverse and parallel to the local mean magnetic field , and @xmath24 is the cascade time of waves with transverse wavenumber @xmath40 . from the numerical simulations of @xcite , @xmath41 because these simulations yield a transverse spectrum which is proportional to @xmath35 instead of @xmath36 , the resulting `` kolmogorov constants '' are not truly constant .
our primary concern is interstellar scintillation , which is affected by electron density fluctuations on very small scales , typically @xmath42 cm for diffractive scintillation . in the remainder of this paper
we calculate the spectrum of density fluctuations which results from compressible turbulence in magnetized plasmas with parameters appropriate to the interstellar medium . throughout
, we consider plasmas that have more ions than neutrals , and that have 1 @xmath43 , where @xmath44 is the ratio of the thermal pressure to the mean magnetic pressure : @xmath45 . here , @xmath46 is the isothermal sound speed , @xmath4 is the alfvn speed , @xmath47 is the thermal pressure , @xmath48 is the magnetic field strength , and @xmath49 is the mass density . in [ sec : betalt1 ] . ]
the incompressible limit corresponds to @xmath50 . on the lengthscales that we consider , compressible mhd is a good approximation for the dynamics of the ionized interstellar medium .
kinetic effects , where important , may be accounted for by simple modifications to the mhd equations . therefore , we turn our attention to turbulence in compressible mhd .
the turbulent velocity spectrum in compressible mhd is approximately the same as the turbulent velocity spectrum in incompressible mhd , because the alfvn mode remains incompressible in a compressible medium , and the slow mode is only slightly compressible .
thus , the velocity spectrum for both of these modes is given by equations ( [ eq : spectrum1 ] ) and ( [ eq : spectrum2 ] ) .
the appendix summarizes the properties of the relevant modes in compressible mhd .
there are two additional modes in compressible mhd which are not present in incompressible mhd .
one of these is the fast mode .
however , as long as @xmath51 a few , the fast mode is essentially a sound wave .
its phase speed is larger than the phase speed of either the alfvn mode or the slow mode , and so the fast mode does not couple to them .
thus , we ignore the fast mode . an analogous approximation is often made in subsonic hydrodynamic turbulence , where sound waves have little influence on the turbulent cascade . the second mode that is present only in compressible mhd is the entropy mode .
this is a zero - frequency mode with unperturbed pressure , and with density perturbation offset by temperature perturbation ( see appendix ) .
the entropy mode exists when the turbulent motions are adiabatic ; the entropy mode is then the only source of entropy fluctuations . in hydrodynamics ,
simple scaling arguments show that the entropy mode does not affect the dynamics of the fluid
the fluid obeys the incompressible equations of motion provided that fluid motions are subsonic and that density fluctuations are smaller than the mean density ; these arguments carry over directly to mhd ( e.g. higdon 1986 , and references therein ) . furthermore , since the entropy of any fluid element is conserved in the inertial range of adiabatic turbulence ,
the mixing of the entropy mode is identical to the mixing of a passive scalar ( see eq . [ [ eq : cspectrum1 ] ] ) , yielding the transverse spectrum @xmath52 where @xmath53 is the r.m.s .
entropy fluctuation across @xmath54 .
the parallel spectrum is given by equation ( [ eq : spectrum2 ] ) .
analogously , in hydrodynamic turbulence the effect of the entropy mode on turbulent motions can often be neglected , and the entropy mode is mixed as a passive scalar @xcite .
based on the above discussion , there are two sources of density fluctuations on small scales : the slow mode and the entropy mode . since the slow mode density perturbation is proportional to its velocity perturbation , and
since the entropy mode density perturbation is proportional to its entropy perturbation , both yield a kolmogorov spectrum of density perturbations : @xmath55 in the remainder of this paper , we investigate the density spectrum in more detail .
we calculate the density spectrum on diffractive lengthscales for given values of the number density , @xmath56 , and outer scale , @xmath57 . typically , in regions of the interstellar medium which are relevant for scintillation , @xmath58 @xmath59 .
values of @xmath57 are more uncertain , though it is likely that @xmath57 is typically within a few orders of magnitude of 1 pc .
it is also plausible that the value of @xmath57 is related to that of @xmath56 . in paper ii , we discuss in much more detail the values of these parameters in turbulent regions of the ionized interstellar medium , e.g. in supernova shocks , hii regions , and stellar winds .
a naive guess for the resulting density fluctuation at the lengthscale @xmath6 is @xmath60 , which we refer to as the fiducial spectrum .
however , there are a number of physical effects which suppress the small - scale density spectrum in the interstellar medium relative to the fiducial spectrum . when considering scintillation observations , it is the transverse not parallel lengthscale which is relevant .
each line of sight crosses regions with different orientations of the local mean magnetic field .
the observational effects of the parallel lengthscale are washed out if the orientation varies by an angle greater than a tiny ratio : @xmath61 at the diffractive scale .
since the variation in angle due to eddies on scales larger than the diffrative scale is @xmath62 , it increases with scale , so these eddies render the parallel lengthscale unobservable . as a result , we frequently refer to the transverse lengthscale as , simply , the lengthscale . because of the multitude of special lengthscales involved ,
we organize the discussion by decreasing lengthscale .
we begin at the outer scale , and proceed down to the smallest scales relevant to interstellar scintillation , while considering the effects of the alfvn mode , the slow mode , and the entropy mode simultaneously . since most of the relevant lengthscales in the interstellar medium have a similar dependence on the background density
they decrease with increasing density the ordering of lengthscales is fairly universal . the lengthscales which we consider are the outer scale , the mhd scale , the cooling scale , the collisionless scale of the neutrals , the collisionless scale of the ions , and the proton gyro scale .
we conclude with a summary of the most important effects .
table [ tab : lengths ] lists the lengthscales that are used in this paper .
as a model for the excitation of the turbulence , we consider plasma which is stirred on an outer scale @xmath57 with velocity fluctuations that are of order the sound speed : @xmath63 .
the generalization from this case of mach 1 turbulence to subsonic turbulence with arbitrary mach number @xmath64 is trivial : @xmath57 would be interpreted as an effective outer scale at which velocity fluctuations extrapolate to @xmath65 .
however , we focus on the mach 1 case because it is probably the most relevant for interstellar scintillation . if , initially , the strength of the magnetic field is negligible , then random fieldline stretching amplifies the mean field @xcite .
it is uncertain both how quickly the magnetic field is amplified , and whether its energy density is amplified until it approaches equipartition with the turbulent kinetic energy density .
if it does reach equipartition within a few outer - scale crossing times , then @xmath66 would be appropriate for mach 1 turbulence . recall that @xmath44 is the ratio of thermal pressure to magnetic pressure , or equivalently , @xmath67 .
we generally assume that the mean magnetic field can approach equipartition with the gas pressure , so we think of @xmath44 as close to but a little larger than unity , although we leave its exact value unspecified .
provided @xmath44 is larger than unity and @xmath68 on scale @xmath57 , the turbulent kinetic energy dominates the mean magnetic energy on scales just below @xmath57 .
thus the cascade is hydrodynamic on these lengthscales , and the velocity fluctuations are given by kolmogorov s isotropic scaling : @xmath69 .
the kinetic energy density decreases towards smaller scales . at a critical scale which we denote @xmath3the kinetic energy density is sufficiently small that it is comparable to the mean magnetic energy density : @xmath70 , which implies that @xmath71 below this scale , the turbulent kinetic energy density is smaller than the magnetic energy density , and the theory of compressible mhd turbulence is applicable .
note that the effects of large - scale velocity fields can be neglected when considering fluctuations on smaller scales , because a large - scale velocity field can be eliminated by a galilean transformation .
however , since large - scale magnetic fields can not be transformed away , they affect the dynamics of small - scale eddies .
hydrodynamic turbulent motions on scales slightly larger than @xmath3 have speeds @xmath72 .
thus they couple to , and efficiently excite , alfvn waves and slow waves whose phase velocities @xmath73 are of comparable magnitude . provided @xmath74 and @xmath75 , fast waves , which have @xmath76 , are weakly excited at @xmath3 . in what follows , we neglect fast waves .
alfvn waves excited at @xmath3 cascade to smaller scales , and these cascading alfvn waves in turn cascade slow waves .
the spectra of both of these cascades are given by equations ( [ eq : spectrum1 ] ) and ( [ eq : spectrum2 ] ) , so that @xmath77 for both modes , even on scales smaller than @xmath3 .
we have not yet discussed density fluctuations on the scales which have been considered in this section .
it is more convenient to do so in the following section .
on the lengthscales in the interstellar medium that we have considered thus far , the radiative cooling time is shorter than the eddy turnover time .. ] consequently , in photoionized regions , turbulence on these lengthscales is expected to be isothermal ( sridhar 1998 ; goldreich 1998 ; higdon & conley 1998 ) .
this has only a marginal effect on the turbulent dynamics described in the previous section , because isothermal alfvn waves are identical to adiabatic alfvn waves , and isothermal slow waves are only slightly different from adiabatic slow waves .
however , there is no entropy mode in isothermal turbulence . as a consequence
, small - scale density fluctuations may be significantly suppressed .
there are two possible solutions to this `` cooling catastrophe '' : either ( i ) the outer scale is extremely small , small enough that the turbulence at the outer scale is nearly adiabatic ; or ( ii ) there are significant density fluctuations associated with the slow mode .
however , in the latter case , the mean magnetic field must be amplified almost to equipartition with the gas pressure , so that @xmath66 .
either of these two solutions would place stringent constraints on the nature of the turbulence which is responsible for observed density fluctuations . in the following , we consider the cooling scale in more detail .
there is a critical scale for the turbulence , which we call the `` cooling scale '' , @xmath78 . above this scale
the turbulence is isothermal , and below it the turbulence is adiabatic .
we assume that @xmath79 throughout this paper , except in [ sec : coolinglargebeta ] where we consider the case in which this inequality is reversed .
the cooling scale is where the eddy turnover time , @xmath80 , is equal to the cooling time , @xmath81 ; i.e. , with @xmath82 , @xmath83 as we will discuss in detail in paper ii , most plausible astrophysical sources of interstellar scintillation have @xmath84 , and thus @xmath85 . in general , the kinetic power per unit mass which is dissipated in mach 1 turbulence is @xmath86 , where @xmath87 is the eddy turnover time at the outer scale .
the thermal power per unit mass which is required to keep the gas hot is @xmath88 .
hence , to avoid the conclusion that the cooling scale is smaller than the outer scale , one must require that more energy go into turbulent motions than into heat . as we will see in paper ii , this requirement is difficult to satisfy .
nonetheless , in isothermal shocks , and in plasmas which are thermally unstable , the two powers are comparable , and the cooling scale is comparable to the outer scale . as an example , we consider an hii region .
photoionized plasma is thermally stable .
heating is primarily due to photoionizing photons , and cooling is primarily due to electron impact excitation of metal line transitions ( e.g. , spitzer 1978 ) .
a characteristic temperature for plasma photoionized by hot stars is @xmath89 8,000k , which implies that the speed of sound is @xmath90 km / sec .
the cooling time is @xmath91 years , where @xmath56 is the number density of electrons .
therefore , @xmath92 a plausible value for @xmath57 might be the radius an hii region ( the `` strmgren radius '' ) , which is @xmath93 pc @xcite .
therefore , with typical values of @xmath56 in hii regions ( @xmath58 @xmath59 ) , @xmath57 is significantly larger than @xmath94 , and so the cooling scale is significantly smaller than the outer scale .
the fact that the turbulence is isothermal on large scales has important implications .
had cooling been ignored , i.e. had it been implicitly assumed that the cooling time is infinite , then one would have calculated the density spectrum as follows : there should be density fluctuations of order unity on the outer scale , implying excitation of entropy modes on the outer scale which are passively mixed by the alfvnic turbulence to small scales .
this would yield the fiducial density spectrum , @xmath95 .
however , if the turbulence is isothermal on large scales , the above calculation is incorrect : the entropy mode does not exist under isothermal conditions . in the following , we consider separately two cases : first the low-@xmath44 case ( @xmath96 ) , and then the high-@xmath44 case ( @xmath97 ) . in each case , we show that the small - scale density fluctuations are substantially smaller than those predicted by the fiducial spectrum unless a relatively extreme condition holds : either @xmath66 or @xmath98 . with regards to interstellar scintillation
, the crucial difference between high-@xmath44 and low-@xmath44 turbulence lies in the compressibility of the slow mode .
since the sum of thermal pressure and magnetic pressure vanishes for the slow mode , the mode s density perturbation satisfies @xmath100 ( see appendix ) .
so , assuming that the energy in slow waves is comparable to that in alfvn waves , @xmath101 from the slow mode , both above and below the cooling scale .
thus , slow waves can produce density fluctuations which are not much smaller than the fiducial spectrum if @xmath44 is not much larger than unity . comparing with equation ( [ eq : bigbeta ] ) below
, the contribution of the slow mode to the density fluctuations at @xmath78 exceeds the contribution from reynolds stresses provided @xmath96 .
the compressibility of the slow mode also yields density fluctuations associated with the entropy mode .
the reason is as follows .
entropy fluctuations are associated with isothermal compressible waves .
those due to isothermal slow modes are passively mixed by alfvn waves . mixing due to alfvn waves with wavelengths smaller than the cooling scale takes place in less than the cooling time .
it produces a spectrum of entropy modes for @xmath102 that gives rise to a density spectrum similar to the one given in equation ( [ eq : slowdens ] ) . nonetheless , there is some damping of entropy fluctuations when they are mixed from scales larger than the cooling scale to scales smaller than the cooling scale .
we expect that the damping is of order unity . in this paper , however , we do not quantify the amount of damping more precisely , because it is a difficult calculation ; it is more difficult than the damping of the slow mode , which we calculate in the following section .
although there is negligible damping of the slow mode on lengthscales which are either much larger or much smaller than the cooling scale , there is some damping on lengthscales which are comparable to the cooling scale .
as we show in this section , the damping of the slow wave cascade is small , because the slow mode is not very compressible . in the limit that @xmath103 , the slow mode is incompressible , and the damping disappears . in the following , we calculate the damping to first order in @xmath104 . first , we calculate the damping rate of a slow wave of a fixed wavenumber . from the appendix ,
the dispersion relation of the slow mode , in the limit that @xmath105 , is @xmath106 the notation which we use in this section is the same as the notation which we define in the appendix , except that here we label the axis parallel to the mean magnetic field by @xmath107 instead of by @xmath108 .
note that we define @xmath109 , where primes denote eulerian perturbations .
the linearized equation of energy conservation is @xmath110 where @xmath111 is the perturbed entropy per particle .
combining this equation with the following monatomic ideal gas relations @xmath112 yields @xmath113 where @xmath114 is the square of the isothermal sound speed . on large scales ,
@xmath115 , and @xmath116 is the isothermal sound speed ; on small scales , @xmath117 , and @xmath116 is the adiabatic sound speed .
the damping rate is obtained by substituting the above @xmath116 ( eq . [ [ eq : tildec ] ] ) into the dispersion relation , which gives , to lowest order in @xmath104 , @xmath118 } \ ,
\label{eq : imcool}\ ] ] where @xmath119 is the imaginary part of @xmath120 . only on lengthscales where @xmath121 is this ratio non - negligible . to obtain the total decrement in the amplitude of the slow mode through the cooling scale
, we solve a kinetic equation obtained by balancing the slow mode @xmath122-space energy flux with the loss - rate of slow mode @xmath122-space energy density due to damping ; i.e. @xmath123 where @xmath124 is the transverse wavenumber , @xmath24 is the cascade time of a slow wave with this wavenumber , and @xmath7 is the velocity perturbation of the slow mode ; since @xmath125 , we drop the @xmath126 subscript on @xmath122 .
we rewrite the kinetic equation as follows : @xmath127 where @xmath128 is a kolmogorov constant ( see [ sec : kolmogorov ] ) . substituting equation ( [ eq : imcool ] ) into equation ( [ eq : coolcas ] ) and integrating , using @xmath129 ( eq . [
[ eq : spectrum2 ] ] ) and @xmath130 , yields the net damping through the cooling scale : @xmath131\vert_{{\lambda_\perp}\ll{{\rm l}_{\rm cool } } } \over [ v_{\lambda_\perp}/\lambda_\perp^{1/3}]\vert_{{\lambda_\perp}\gg{{\rm l}_{\rm cool } } } } & = & \exp\big ( { -{m_t\over \beta}\int_0^\infty { v_a k_\parallel t_{\rm cool } \over 1+(5v_a k_\parallel t_{\rm cool}/2)^2 } { dk\over k } } \big ) \nonumber \\ & = & \exp\big(-{3\pi m_t \over 10\beta}\big ) \nonumber \\ & \simeq & 1 - 1.3/\beta \ , \end{aligned}\ ] ] to lowest order in @xmath104 .
thus for @xmath44 slightly larger than unity , slow mode damping can be ignored . in this subsection only
, we assume that @xmath97 , which implies @xmath133 . in this case
, the isothermal hydrodynamic turbulent cascade extends from the outer scale down to the cooling scale . because the entropy mode does not exist under isothermal conditions and hydrodynamic turbulence is incompressible to order @xmath134 , to this order there are no density fluctuations on lengthscales larger than @xmath78
however , pressure fluctuations due to reynolds stresses , @xmath135 , create second order density and entropy fluctuations , @xmath136 these entropy fluctuations couple to the entropy mode at the cooling scale . on scales below the cooling scale ,
the entropy mode is mixed as a passive scalar yielding @xmath137 hydrodynamic turbulence mixes the entropy mode down to @xmath3 . at this scale ,
the hydrodynamic motions turn into alfvn and slow waves , so the entropy mode continues to be mixed below @xmath3 , and equation ( [ eq : bigbeta ] ) remains valid . comparing with equation ( [ eq : slowdens ] ) , the density fluctuations associated with the entropy mode
exceed those associated with the slow mode when @xmath97 .
the entropy mode s density spectrum ( eq . [ [ eq : bigbeta ] ] ) is smaller than the fiducial spectrum by the small factor @xmath138 .
as will be seen in paper ii , it yields density fluctuations that are too small to explain much of the observed interstellar scintillation for most plausible values of @xmath57 , such as those given by strmgren radii of hii regions .
there are two alternatives ; either @xmath57 is not much larger than @xmath139 , or @xmath96 .
in the remainder of this paper , we consider lengthscales which are comparable to , or smaller than , the mean free paths of the various particles ; thus , kinetic effects must be considered . in this section ,
we calculate the damping of the turbulence by neutrals .
we assume that the neutral density is smaller than the electron density , i.e. @xmath140 , and that most of the ions are protons .
the neutral particles which we consider are hydrogen atoms and helium atoms .
the relevant lengthscales are the hydrogen and helium mean free paths for collisions with protons : @xmath141 cm and @xmath142 cm @xcite .
the hydrogen mean free path is significantly smaller than the helium mean free path because its collisions with protons are due to resonant charge exchange .
there are two main results for this section . *
\(i ) only if the neutral fraction is sufficiently small ( eq . [ [ eq : neutralfrac ] ] , below ) can the cascade survive on lengthscales smaller than @xmath143 .
if the neutral fraction does not satisfy this condition , then all three modes alfvn , slow , and entropy are damped at the same lengthscale .
* \(ii ) regions where the cascade does not survive contribute to an excess of density fluctuations on large scales .
this might explain observations which detect an excess of power in large - scale density fluctuations .
the organization of the calculation is as follows : first , the change in wave frequency due to neutrals is calculated .
second , we consider the effect of the frequency change on the cascade .
third , we consider the cases in which the neutrals are hydrogen atoms and in which they are helium atoms .
although helium atoms have a larger mean free path , we consider them after neutral hydrogen atoms because they are of lesser importance .
finally , we consider regions where neutral damping terminates the cascade . a similar , though less detailed calculation is performed by @xcite for the case in which the neutrals are hydrogen atoms .
consider an incompressible wave , either an alfvn wave or a slow wave in the incompressible limit .
although the slow wave is , in fact , slightly compressible , this does not have a large effect on the final result .
we calculate the frequency change of an incompressible wave with fixed @xmath144 : @xmath22 along the mean magnetic field and @xmath21 transverse to the mean magnetic field , where @xmath145 .
we define @xmath146 to be the mean velocity of the protons averaged over the proton distribution function .
thus @xmath146 satisfies the equation of motion for the alfvn wave or for the slow wave derived in the appendix .
we then calculate the force that these protons exert on the neutrals . since this force is the same as the force of the neutrals on the protons ,
inserting it into the equation of motion for the protons yields the change in frequency .
we denote the perturbed distribution function for the neutrals , i.e. perturbed to first - order in @xmath147 , after being fourier - transformed in space and time , by @xmath148 , where @xmath147 is the velocity of the neutrals .
the evolution of the neutrals is determined by the boltzmann equation : @xmath149 where @xmath150 is the fourier - transformed , perturbed , collision integral of the neutrals with the protons .
neutral - neutral collisions are negligible relative to neutral - proton collisions
. the collision integral is simplified by making the approximation that , on the timescale that a neutral collides with protons , @xmath151 is driven towards a maxwellian with mean velocity @xmath146 : @xmath152 \ , \ ] ] where the first term in the square brackets is generated by expanding a maxwellian distribution function with mean velocity @xmath146 to linear order in @xmath146 and retaining only the perturbation , and @xmath153 is the frequency with which a given neutral atom collides with protons . with the above approximation to the collision integral
, the solution to the boltzmann equation reads @xmath154 next we verify that the perturbed neutral number density vanishes , @xmath155 \nonumber \\ & \times & \big [ \int d^2{\mbox{\boldmath$v$\unboldmath}}_{n\perp } { \exp({-m_nv_{n\perp}^2/2 t ) } \over \nu_{n , p}-i\omega+i{\mbox{\boldmath$k$\unboldmath}}\cdot{\mbox{\boldmath$v$\unboldmath}}_{n\perp } } \big ] \\ & = & 0 \ , \end{aligned}\ ] ] since the first square - bracketed integral vanishes by antisymmetry . for the purpose of evaluating the above integrals , we use @xmath126 and @xmath107 for the perpendicular and parallel projections of @xmath147 relative to @xmath146 , and not relative to the magnetic field as in the rest of this paper ; i.e. @xmath156 and @xmath157 .
we also use the incompressibility relation @xmath158 , which implies that @xmath159 . with the neutral distribution function given by equation ( [ eq : fn ] ) , the force per unit volume of the protons on the neutrals may be calculated as @xmath160 \nonumber \\ & \times & \big [ \int d^2{\mbox{\boldmath$v$\unboldmath}}_{n\perp } { ( -i\omega+i{\mbox{\boldmath$k$\unboldmath}}\cdot{\mbox{\boldmath$v$\unboldmath}}_{n\perp})\exp({-m_nv_{n\perp}^2/2 t ) } \over \nu_{n , p}-i\omega+i{\mbox{\boldmath$k$\unboldmath}}\cdot{\mbox{\boldmath$v$\unboldmath}}_{n\perp } } \big ] \\ & = & { m_nn_n{\mbox{\boldmath$v$\unboldmath}}\nu_{n , p}\over ( 2\pi t / m_n)^{1/2 } } \nonumber\\ & \times & \int_{-\infty}^\infty ds { ( -i\omega+iks)\exp({-m_ns^2/2 t ) } \over \nu_{n , p}-i\omega+iks } \\ & = & { m_n n_n { \mbox{\boldmath$v$\unboldmath}}\nu_{n , p } \over \pi^{1/2}\pi_1}\nonumber\\&\times & \int_{-\infty}^{\infty } { it \over { 1+it } } \exp \big[-(t/\pi_1+\pi_2/\pi_1)^2\big ] dt \ .\end{aligned}\ ] ] the second equality follows after replacing the overall multiplicative @xmath147 ( @xmath161 ) by @xmath162 , because the @xmath163 term vanishes by antisymmetry of the integral over @xmath164 . in the second equality ,
the first bracketed integral is equal to @xmath165 .
the second bracketed integral is a double integral ; the integral over the component of @xmath163 which is perpendicular to @xmath144 is equal to @xmath166 .
the remaining integral is over the component of @xmath163 which is parallel to @xmath144 ; we label this component @xmath167 in the third equality .
finally , the fourth equality follows from the change of variables @xmath168 , and from the following definitions @xmath169 where @xmath170 is the proton mass .
the dimensionless parameter @xmath171 is the number of wavelengths a neutral crosses before colliding , and the dimensionless parameter @xmath172 is the number of waveperiods a neutral travels before colliding . since @xmath173 and @xmath174 , @xmath175 in the inertial range of the mhd cascade . performing the integral for the real and imaginary parts of @xmath176 to lowest order in @xmath177 yields @xmath178 where @xmath179 since @xmath180 is equal to the force of the neutrals on the protons , we insert @xmath180 into the equation of momentum conservation for the protons .
more precisely , we add @xmath181 to the right - hand side of equation ( [ eq : pcons ] ) in the appendix .
we label the resulting change of frequency caused by the presence of neutrals as @xmath182 . assuming that @xmath183 , it follows that @xmath184 the physical interpretation of this frequency change is straightforward . recall that @xmath173 , so that the wavelength , @xmath185 , is nearly identical to the wavelength transverse to the magnetic field , @xmath186 .
it is convenient to define the neutral mean free path , @xmath187 , as follows : @xmath188 where @xmath189 is the isothermal sound speed .
the imaginary part of @xmath182 is the negative of the damping rate , and is given by @xmath190 the above expression may be explained as follows @xcite . for wavelengths larger than the neutral mean free path , @xmath191 , the neutrals are nearly locked to the protons .
they damp the mhd waves by transferring the protons momentum across a neutral mean free path . the damping rate increases with decreasing wavelength provided @xmath192 . for wavelengths smaller than the neutral mean free path , the neutral atoms are effectively freely streaming .
this leads to a damping rate which is independent of wavelength .
the real part of the frequency change , @xmath193 , is @xmath194 its physical significance is apparent . for small wavelengths
the neutral atoms are effectively freely streaming , the motion of the protons nearly decouples from the motion of the neutrals , and there is negligible real frequency change associated with the presence of the neutrals . for large wavelengths
the neutrals are locked to the protons , so the mass density of the fluid which participates in the waves is larger than that of the protons .
since the alfvn speed , and hence the wave frequency , is inversely proportional to the square root of the mass density , i.e. @xmath195 , this results in a decrease of the frequency by a fractional amount which is equal to one - half of the mass ratio of neutrals to protons .
the turbulent cascade is quenched if the damping rate exceeds the eddy cascade rate or , equivalently , if @xmath196 . from equation ( [ eq : omegai2 ] ) and @xmath129
, it is seen that for small wavenumbers @xmath197 , and for large wavenumbers @xmath198 .
therefore @xmath199 is a maximum for transverse wavelengths , @xmath200 , comparable to the neutral mean free path , and decreases for both larger and smaller wavelengths .
the requirement that the cascade survive damping by neutrals is then that @xmath201 for @xmath200 comparable to the neutral mean free path , or equivalently , @xmath202 at this scale ( eq . [ [ eq : omegai2 ] ] ) .
since @xmath203 is the rate at which a given proton collides with neutrals , the cascade survives to small scales if the wave frequency , and hence the cascade rate , at the scale of the neutral mean free path is faster than the rate at which a proton collides with neutrals . to obtain the total decrement in the amplitude of both the alfvn mode and the slow mode through the damping scale , we solve the kinetic equation for the cascade . for slow mode damping at the cooling scale .
the principal difference , aside from the fact that the imaginary part of the frequency is different , is that here we must account for the decrease in the cascade rate , @xmath24 , caused by damping .
this is not necessary when treating slow mode damping at the cooling scale because there the alfvn modes which control the cascade rate are undamped . ] as in
[ sec : slowdamp ] , the kinetic equation is obtained by balancing the @xmath122-space energy flux with the loss - rate of @xmath122-space energy density due to damping : @xmath123 where @xmath204 is the transverse wavenumber , and @xmath24 is the cascade time .
the kinetic equation can be rewritten as follows : @xmath205 where @xmath38 is a kolmogorov constant ( see [ sec : kolmogorov ] ) . before integrating this equation , we re -
express the damping frequency in terms of the relevant lengthscales , @xmath57 and @xmath187 : @xmath206 the first equality above follows from equation ( [ eq : omegai ] ) .
the second equality follows from the kolmogorov constant @xmath207 ( see [ sec : kolmogorov ] ) .
the third equality follows from the velocity spectrum , @xmath208 ( eq . [ [ eq : vspec ] ] ) , and from the definition of the neutral mean free path ( eq . [ [ eq : ln ] ] ) . for convenience ,
we define @xmath57 in this section such that @xmath209 when @xmath210 . integrating the kinetic equation ( eq . [ [ eq : kinetic ] ] ) , and using @xmath211 , yields the net decrement due to damping by collisions with neutrals : @xmath131\vert_{\lambda_\perp\ll { \rm l}_n } \over [ v_{\lambda_\perp}/\lambda_\perp^{1/3}]\vert_{\lambda_\perp\gg { \rm l}_n } } & = & \exp\big [ { 2m_t\over 3}\int_0^\infty { \omega_i\over v_a k_\parallel}{dk\over k } \big ] \nonumber\\ & = & \exp\big [ -0.2\big({m_n\over m_p}\big)^{1/2}{n_n\over n } \big({{{\rm l}_{\rm out}}\over{\rm l}_n}\big)^{1/3 } \big ] \ , \label{eq : neutraldamp}\end{aligned}\ ] ] after substituting from equation ( [ eq : om ] ) .
the numerical prefactor in the second equality , @xmath212 , follows after inserting the values of the kolmogorov constants ( eq . [
[ eq : kolmogorov ] ] ) and the value of the integral @xmath213 which was integrated numerically ; see equation ( [ eq : gdef ] ) for the definition of @xmath214 . for the cascade to continue to small scales
, the right - hand side of equation ( [ eq : neutraldamp ] ) can not be very small .
this can be viewed as an upper limit on the neutral fraction : @xmath215 .
recall that the condition for the cascade to reach small scales is that the cascade rate at @xmath187 be faster than the rate at which a proton collides with neutrals .
if the neutral fraction is too large , then so is the proton collision rate , and the cascade is quenched .
moreover , for a fixed @xmath187 , a large value for @xmath57 implies that the cascade time at @xmath187 is large , and hence that the cascade is more susceptible to damping by neutrals .
we postpone consideration of the damping due to hydrogen and helium atoms until after we evaluate the effect of @xmath216 on the cascade .
we can picture the cascade as proceeding from large scales to small scales .
as it crosses the scale of the neutral mean free path ( eq . [ [ eq : omegar ] ] ) , the effective alfvn speed , and hence the real part of wave frequency , increases by the fraction @xmath217 .
consequently , the cascade time decreases by the same amount .
to the extent that the flux of energy in alfvn waves from large scales to small scales is constant in the turbulent cascade , @xmath218 constant , where @xmath24 is the cascade time .
therefore , a fractional decrease of @xmath217 in the cascade time causes a decrement in the small - scale alfvnic velocity perturbation by the fractional amount @xmath219 .
this decrement is in addition to that due to damping .
moreover , it applies to the slow mode and the entropy mode as well as to the alfvn mode . because @xmath220 must be small for the cascade to pass through the scale of neutral damping , this decrement is also small , and we ignore it from here on . when the neutrals are hydrogen atoms we set @xmath221 , @xmath222 , @xmath223 , and @xmath224 .
collisions between hydrogen atoms and protons are due to resonant charge exchange . from equation ( [ eq : ln ] ) , with the value of @xmath225 taken from @xcite , @xmath226 at @xmath227k ; the temperature dependence of @xmath228 is very weak . for the cascade to survive damping by neutral hydrogen atoms
, the right - hand side of equation ( [ eq : neutraldamp ] ) can not be very small .
this sets an upper limit on the neutral fraction of @xmath229 the real part of @xmath230 is larger on scales below @xmath231 than it is on scales above @xmath143 by the fractional value @xmath232 .
although helium has a lower abundance than hydrogen , it has a higher ionization potential .
therefore , in some regions helium might comprise the majority of neutrals . when the neutrals are helium atoms we set @xmath233 , @xmath234 , @xmath235 , and @xmath236 .
from equation ( [ eq : ln ] ) , with the value of @xmath237 taken from @xcite , @xmath238 at @xmath227k , and @xmath239 . to place an upper bound on damping by helium atoms
, we assume that most of the helium is neutral , i.e. @xmath240 . then the decrement is given by the right - hand side of equation ( [ eq : neutraldamp ] ) : @xmath241\vert_{\lambda_\perp\ll { \rm l}_{{\rm he } } } \over [ v_{\lambda_\perp}/\lambda_\perp^{1/3}]\vert_{\lambda_\perp\gg { \rm l}_{{\rm he } } } } = \exp\big [ -\big ( { ( n/{\rm cm}^{-3}){{\rm l}_{\rm out}}\over 2\times 10^{19 } { \rm cm } } \big)^{1/3 } \big ] \ .\ ] ] only if @xmath242 could the cascade be terminated at the scale of the helium mean free path . if there are not many neutral helium atoms , or if the outer scale is not sufficiently large , then damping at @xmath243 may be neglected , and the cascade extends at least to the scale of the hydrogen mean free path .
collisions of neutral helium atoms with singly ionized helium ions might also be significant .
although there are fewer helium atoms than protons , he@xmath244-he@xmath245 collisions have a larger cross - section than he@xmath244-proton collisions because they are due to resonant charge exchange . nonetheless , given the cosmic abundance of helium , the mean free path of neutral helium due to he@xmath244-he@xmath245 collisions can not be significantly smaller than that due to he@xmath244-proton collisions regardless of the ionization fraction of helium . the real part of @xmath246 increases by @xmath247 below the scale @xmath243 , or 0.2 if most of the helium is neutral . in subsequent sections ,
we consider regions in the interstellar medium where there are too few neutrals to damp the cascade . however , there are almost certainly many regions where the neutrals do damp the cascade .
we discuss these regions here .
suppose both alfvn waves and slow waves are damped by neutrals .
what happens to the entropy waves ?
if undamped , they would be mixed down to smaller scales by alfvn waves at the neutral damping scale
. the resulting density spectrum would be @xmath248 constant ( see eq . [ [ eq : cspectrum3 ] ] ) . because it is flatter than @xmath249 , regions in which damping by neutrals truncated the alfvn cascade might be important contributors to small scale density fluctuations .
however , the fact that they would contribute a density spectrum different from that which is observed suggests that the entropy wave cascade is not more resistant than the alfvn wave cascade to damping by neutrals .
indeed that is the case .
recall that the condition for the truncation of alfvn and slow wave cascades is that each proton collide with at least one neutral atom during one wave period at the neutral damping scale . under this condition
, the neutrals would damp the entropy waves by conducting heat across them .
although regions in which the cascade is damped by neutrals do not contribute small scale density fluctuations , they may still be significant .
observational evidence indicates that there is more power in density fluctuations on large scales , @xmath250 cm , than would be predicted by extrapolating from small scales , @xmath251 cm , with the kolmogorov scaling .
see , for example , @xcite for a review of the observations .
perhaps this excess arises in regions where the cascade is damped by neutrals .
we complete this section by briefly considering and then rejecting the possibility that a turbulent cascade truncated at the neutral damping scale might be regenerated on a much smaller scale due to stirring by eddies at the damping scale .
although the ratio of the damping rate to the wave frequency decreases below the scale of the neutral mean free path , the absolute damping rate approaches a constant value .
provided the cascade is truncated by neutral damping , this rate is larger than the stirring rate and the cascade can not be regenerated .
if the cascade survives the neutral collisionless scales , then , proceeding to smaller scales , the next scale of importance is the ion collisionless scale .
this scale is set by the mean free path of protons to collide with other protons : @xmath252 at a temperature of 8,000k @xcite . since the electron and proton gyroradii
( see eq . [ [ eq : protongyro ] ] for protons ) are very small compared to the proton mean free path , the electrons and protons are tied to magnetic fieldlines . therefore , when considering the collisionless effects of electrons and protons , the relevant lengthscales are those parallel to the magnetic field
. since turbulent eddies are highly elongated along the magnetic field , their transverse lengthscales are much smaller than their parallel ones ( eq . [ [ eq : spectrum2 ] ] ) .
it is the transverse lengthscale which is relevant when considering observations of the density spectrum , because each line of sight averages over regions with different magnetic field orientations . in this section ,
we show that both the slow mode and the entropy mode are cut off at the lengthscale where the parallel eddy size is comparable to the proton mean free path .
the lengthscale where the density spectrum is _ observed _ to cut off , i.e. the transverse lengthscale , is therefore significantly smaller than the mean free path . before discussing the damping of the slow mode and the entropy mode , we consider two larger lengthscales , which are set by the effects of the electrons .
although the behaviour of the cascade at these lengthscales is interesting , it is shown to be unimportant for our purposes . throughout this section ,
we neglect numerical factors of order unity , such as the factors which are associated with kinetic corrections to the fluid equations ( given in braginskii 1965 ) and the kolmogorov constants .
however , we retain the dependences on @xmath44 , which is assumed to satisfy @xmath74 , and on the ratio of the proton mass to the electron mass : @xmath253 .
the electrons have the same mean free path as the protons , but they are faster than the protons by the square root of the mass ratio : @xmath254 where @xmath255 is the electron thermal speed and @xmath65 is the sound speed , which is comparable to the proton thermal speed .
because of charge neutrality , the electrons have the same density , both perturbed and unperturbed , as the protons .
viscous damping caused by the electrons may be neglected : since the dynamic viscosity of the electrons is smaller than that of the protons by the square root of their mass ratio , electron viscous damping is always subdominant .
as we now show , electrons are important for conducting heat on parallel lengthscales slightly larger than the proton mean free path . electrons diffuse parallel to the magnetic field across a distance @xmath8 in the time @xmath256 .
this is equal to the cascade time of an eddy with parallel lengthscale @xmath8 , i.e. it is equal to @xmath257 , at the `` electron diffusion scale '' , given by @xmath258 , where @xmath259 in eddies with parallel lengths smaller than this , the electrons diffuse across the parallel lengths of the eddies , thereby conducting heat , and the electrons are isothermal ; in eddies with parallel lengths larger than this , conduction is unimportant .
we must also consider the effect of electron conduction on the protons .
electrons and proton temperatures approach a common value on the timescale that there are @xmath260 collisions per particle .
since electrons are faster , the collision time is set by the electron speed , and the time for electron and proton temperatures to equilibrate is @xmath261 this time is smaller , by a factor @xmath44 , than the cascade time of eddies with parallel size @xmath262 .
thus in eddies of this parallel size , and in those which are slightly smaller , protons are at the same temperature as the electrons . and , since the electrons are isothermal , so are the protons . as a result of the above considerations ,
when the parallel cascade crosses the electron diffusion scale , the cascade becomes isothermal .
this is similar to the crossing of the cooling scale , discussed in
[ sec : coolingsmallbeta ] , though backwards , and similar considerations apply .
in particular , the alfvn mode is incompressible , and hence unaffected .
the slow mode is nearly incompressible , and is only slightly affected : it suffers some damping to first order in @xmath104 .
the entropy mode , however , does not exist under isothermal conditions ; as the electron diffusion scale is crossed , entropy waves are converted into slow waves , and the density fluctuations which had been associated with entropy waves now become associated with slow waves .
continuing to slightly smaller scales , the isothermal cascade reaches the `` equilibration scale '' , where the cascade time , @xmath257 , is comparable to the electron - proton equilibration time ( eq . [ [ eq : equil ] ] ) . at this scale , @xmath263 , where @xmath264 on larger scales , the protons are thermally coupled to the electrons , and hence they are isothermal on slightly larger scales ; on smaller scales , the protons are thermally independent of the electrons , and hence adiabatic .
the transition through the equilibration scale is nearly identical to the transition through the cooling scale ( [ sec : coolingsmallbeta ] ) : the alfvn waves are unaffected , and they mix larger - scale isothermal slow waves into smaller - scale adiabatic slow waves and into smaller - scale entropy waves , although with some damping . therefore , below the equilibration scale , the entropy mode reappears .
the result of the calculations in both this subsection and the previous subsection is that , in crossing the electron diffusion scale and the equilibration scale , density fluctuations which were associated with the entropy mode on large scales are transferred from the entropy mode to the slow mode and then back to the entropy mode .
density fluctuations which were associated with the slow mode on large scales are unaffected .
therefore , for our purposes , these two lengthscales have little net effect on the density spectrum .
although there is some damping , the amount of damping is comparable to the amount at the cooling scale , and hence is not very significant .
entropy waves and slow waves with parallel wavelengths smaller than the proton mean free path both damp on the timescale that protons stream across their parallel wavelengths @xcite .
however , because we consider @xmath74 , there is a scale slightly larger than the proton mean free path at which these waves damp in the turbulent cascade .
this is the proton diffusion scale , which is the scale at which protons can diffuse across an eddy within a cascade time . below the proton diffusion scale ,
proton viscosity kills the slow waves and proton heat conduction kills the entropy waves .
alfvn waves are unaffected by either of these effects .
the density spectrum therefore cuts off below the proton diffusion scale . to evaluate the proton diffusion scale , we equate the time for protons to diffuse across an eddy of parallel lengthscale @xmath8 , i.e. @xmath265 , with the cascade time , @xmath257 .
this gives @xmath266 , where @xmath267 however , from an observational point of view , it is the transverse size of a damped eddy , @xmath6 , which would be observed .
this is related to the parallel size through equation ( [ eq : spectrum2 ] ) .
alternatively , we use the kolmogorov constant @xmath268 ( see [ sec : kolmogorov ] ) .
then , the transverse scale of a damped eddy , i.e. an eddy with parallel size @xmath266 , is @xmath269 which we solve for the cut - off lengthscale : @xmath270 , where @xmath271 below this scale , the density spectrum is cut off . for plausible values of @xmath44 , @xmath57 , and @xmath56 ,
this lengthscale is significantly smaller than @xmath272 .
it is also larger than the proton gyroradius ( eq . [ [ eq : protongyro ] ] , below ) .
@xcite summarize the observations of the density spectrum cut - off .
there is considerable evidence that the cut - off scale is smaller than about @xmath273 cm along many directions .
there is weaker evidence , from refractive scintillation , that along some lines of sight the cut - off scale is larger than around @xmath274 cm .
our theory might have implications for these observations .
we conclude this section with two remarks .
first , we re - emphasize the importance of the fact that the cascade is anisotropic .
it is this fact which allows the density spectrum to reach lengthscales which are significantly smaller than the proton mean free path .
second , we note an important consequence of the parallel cascades of the slow mode and of the passive scalar ; these parallel cascades were explained in great detail when considering incompressible mhd turbulence ( [ sec : incomp ] ) .
had there been no parallel cascade , then neither the slow mode nor the entropy mode would damp at the proton diffusion scale .
the wavelength along the magnetic field would be effectively infinite , and so the ions could not diffuse across wavelengths .
rather , the slow mode and the entropy mode would be mixed down to the proton gyro scale , where the alfvnic cascade is cut off . on scales smaller than @xmath275 ,
density fluctuations are wiped out . since on these scales protons
diffuse across the lengths of many eddies before the eddies cascade , the density within neighbouring eddies is homogenized .
homogenization occurs not only in the direction parallel to the local mean magnetic field as might have been expected since protons are tied to fieldlines but also in the transverse direction .
this is because eddies that are adjacent in the parallel direction incorporate substantially different fieldlines , so proton diffusion also wipes out density differences amongst eddies with transverse separations .
the result is that the density spectrum on scales smaller than @xmath275 is determined by density fluctuations at @xmath275 ; equivalently , @xmath276
the alfvn mode is undamped in a collisionless medium @xcite .
thus , the alfvn wave cascade survives below the proton diffusion scale , without the accompaniment of the slow waves and entropy waves .
the alfvnic cascade is damped at the scale of the proton gyroradius : @xmath277 at this scale , alfvn waves are converted into whistlers .
the whistlers cascade to smaller scales , where they are damped by the collisionless effects of the electrons ( quataert 1998 , viz .
the curve in his fig .
1b that corresponds to equal electron and proton temperatures ) .
the primary goal of this paper has been to calculate the small - scale density spectrum in turbulent interstellar plasmas .
our theory of compressible mhd turbulence is based upon the incompressible theory of @xcite , for which there is growing support from numerical simulations .
we hypothesize that the compressible theory is similar to the incompressible theory of mhd turbulence , but with two main modifications : a compressible slow mode and an entropy mode which is passively advected .
while we believe that this hypothesis is plausible , it can , and should , be tested with numerical simulations of compressible mhd .
because of the multitude of special lengthscales encountered when discussing the turbulent cascade in the interstellar medium , a recapitulation might be useful . in the following
, we list the most important lengthscales , and summarize their significance to the turbulent cascade . _ the outer scale _ ( @xmath57 ) : this is the lengthscale at which the turbulent motions are stirred . on slightly smaller scales , there is isothermal hydrodynamic turbulence for most plausible astrophysical sources . _ the mhd scale _ ( @xmath3 ) : at this lengthscale there is a transition from hydrodynamic to mhd turbulence .
larger - scale hydrodynamic motions couple to smaller - scale alfvn waves and slow waves . _ the cooling scale _ ( @xmath78 ) : at this lengthscale there is a transition from isothermal to adiabatic turbulence . in high-@xmath44 turbulence , where small scale density fluctuations are due to the entropy mode , entropy and hence density fluctuations are suppressed by cooling . in low-@xmath44 turbulence ,
i.e. @xmath278 , density fluctuations due to the slow mode are important , and cooling has a negligible effect on the small - scale density spectrum . _ the collisionless scale of the neutrals _ ( @xmath228 ; @xmath279 ) : neutrals decouple from ions across these scales . if the neutral fraction is not sufficiently small , then all three modes alfvn , slow , and entropy are damped .
_ the collisionless scale of the ions _ ( @xmath280 ; @xmath281 ; @xmath282 ) : across these lengthscales there is a gradual transition from fluid behaviour to collisionless plasma behaviour .
the first two of these lengthscales , the electron diffusion scale and the equilibration scale , are set by the electrons .
these lengthscales have only a small net effect on the cascade .
the proton diffusion scale , however , is critically important for density fluctuations : both the slow mode and the entropy mode are cut off at this scale , and hence there are no density fluctuations below @xmath275 . _ the proton gyro scale _ ( @xmath283 ) : alfvn waves are cut off at this scale . however
, this scale has little importance for the density spectrum , because there are no density fluctuations below the proton diffusion scale .
in his 1984 paper , higdon attributes small - scale density fluctuations to the passive mixing of the entropy mode . in his 1986 paper
, higdon notes that if the entropy mode varies along the magnetic field , it is damped in a collisionless medium .
he then attributes small - scale density fluctuations to the passive mixing of tangential pressure balances .
tangential pressure balances are structures which are parallel to the mean magnetic field .
they are composed of both entropy waves and slow waves that have purely transverse wave vectors . considering that higdon s papers preceded even a theory of incompressible mhd turbulence , they are a remarkable accomplishment . however
, higdon does not account for the parallel cascade .
entropy and slow waves with purely transverse wave vectors contain negligible power .
they are cascaded along the magnetic field by alfvn waves .
consequently , they are damped when protons can diffuse across eddies in a cascade time , i.e. below the proton diffusion scale .
in this section only , we consider compressible turbulence in plasmas that have @xmath284 , e.g. in isothermal shocks and in the solar wind . since alfvn waves are unaffected by the value of @xmath44 , and since nearly transverse slow waves are only slightly affected ( see appendix ) , the dynamics of the cascade is nearly independent of @xmath44 .
while the dispersion relation of the slow mode is changed from @xmath285 to @xmath286 , the slow waves are still passively mixed by the alfvn waves .
therefore , the alfvn , slow , and entropy spectra in @xmath284 mhd turbulence are the same as when @xmath74 .
however , the damping of the slow and entropy waves is significantly changed . when @xmath74 , the proton thermal speed is faster than the alfvn speed .
therefore protons can stream across small eddies before they cascade .
conversely , when @xmath284 , the time for protons to cross an eddy is always shorter than the cascade time . therefore slow and entropy waves can not be damped by protons which cross eddies , and the density spectrum extends to smaller scales .
we have ignored the effects of the electrons . however , in the following we show that the electrons behaviour may be ignored for our purposes .
our discussion closely parallels that when @xmath74 in [ sec : collion ] , and uses similar notation . when @xmath284 , the equilibration scale ( eq . [ [ eq : lequ ] ] ) , is larger than the electron diffusion scale ( eq . [ [ eq : led ] ] ) .
thus the largest scale at which kinetic effects are significant is the scale at which the cascade time is comparable to the time for electrons and protons to equilibrate their temperatures . below this scale ,
electrons and protons are thermally decoupled . nonetheless , this thermal decoupling has no effect on the alfvn , slow , or entropy waves on scales larger than the electron diffusion scale . at the electron diffusion scale ,
the cascade time is comparable to the electron diffusion time .
below this scale , the electrons are isothermal . however , since the protons are thermally decoupled from the electrons , the slow waves and entropy waves are cascaded to smaller scales . continuing to smaller scales ,
the next scale of importance is that at which the parallel size of an eddy is comparable to the proton mean free path . below this scale ,
the cascade is collisionless .
the alfvn waves are undamped by collisionless effects .
the entropy waves are undamped since protons can not cross eddies within a cascade time .
the slow waves are also undamped within a cascade time : although they damp within a waveperiod by barnes damping , their waveperiod is longer than the cascade time .
consequently , the kolmogorov density spectrum extends down to the proton gyro scale .
in a future paper , we will examine in detail the density spectrum in the solar wind . in another paper ( paper ii )
, we will relate the theory of compressible plasma turbulence developed here to observations primarily those of diffractive scintillation .
we will demonstrate that the observed amplitude of small - scale density fluctuations is surprisingly large , especially along certain lines of sight , such as the one toward the galactic center . then we will attempt to determine which astrophysical sources contribute the bulk of the density fluctuations and why they do so .
we thank jason maron for analyzing some of his simulations for us , for showing us how to use his numerical code , and for informative discussions .
research reported in this paper was supported by nsf grant 94 - 14232 .
in this appendix we derive the properties of the alfvn mode , the slow mode , and the entropy mode . the fast mode is not relevant to interstellar scintillation for reasons discussed in the body of the paper .
the fourier - transformed , linearized equations of ideal mhd , with @xmath287 and @xmath288 , read @xmath289-{\mbox{\boldmath$b$\unboldmath}}v_a k_z \ , , \qquad\mbox{(momentum conservation ) } \label{eq : pcons } \\ \omega{\mbox{\boldmath$b$\unboldmath}}&=&{\mbox{\boldmath$\hat{z}$\unboldmath}}v_a({\mbox{\boldmath$k$\unboldmath}}\cdot{\mbox{\boldmath$v$\unboldmath } } ) -{\mbox{\boldmath$v$\unboldmath}}v_a k_z \ , \
, \qquad\mbox{(faraday 's law)}\end{aligned}\ ] ] where @xmath56 is number density , @xmath49 is mass density , @xmath48 is the background magnetic field intensity , and @xmath290 is the alfvn speed .
the @xmath108-axis is chosen to lie parallel to the background magnetic field .
some other variables are the perturbed number density , @xmath291 , the fluid velocity , @xmath146 , and the perturbed magnetic intensity , @xmath292 , or in velocity units , @xmath293 .
the sound speed , @xmath116 , is defined less conventionally here to be the square root of the ratio of perturbed pressure to perturbed mass density , @xmath294 . for an isothermal gas , @xmath295 , where @xmath296 is the isothermal sound speed ; @xmath297 . for an adiabatic ideal gas , @xmath298 , where @xmath65 is the adiabatic sound speed ; @xmath299 for a monatomic plasma .
more generally , however , the equation of energy conservation must be specified before @xmath116 can be determined .
while in general @xmath116 depends on @xmath120 , we ignore the @xmath120-dependence in this appendix , and treat @xmath116 as a constant .
this treatment is valid for the purposes of this paper .
we can express the equations of motion in terms of @xmath146 as follows : @xmath300{\mbox{\boldmath$k$\unboldmath } } -v_a^2k_z({\mbox{\boldmath$k$\unboldmath}}\cdot{\mbox{\boldmath$v$\unboldmath}}){\mbox{\boldmath$\hat{z}$\unboldmath } } \ .
\label{eq : eomexpand}\ ] ] the alfvn mode is incompressible ; @xmath158 .
thus the term involving the sound speed in the momentum equation vanishes , and the properties of the alfvn mode are independent of the equation of energy conservation .
we obtain the dispersion relation @xmath301 by forming the cross product of equation with @xmath144 .
the eigenfunction satisfies @xmath302 note that both @xmath146 and @xmath303 are perpendicular to @xmath304 : the alfvn wave is polarized transverse to the unperturbed magnetic field .
we summarize the properties of the slow mode to lowest order in @xmath305 , the limit appropriate to the mhd cascade . to obtain the dispersion relation we assume , subject to verification , that the perturbation in total pressure i.e .
thermal plus magnetic pressure vanishes to second order in @xmath306 .
note that the perturbation in total pressure is proportional to the terms in square brackets in both equation ( [ eq : pcons ] ) and equation ( [ eq : eomexpand ] ) . from the vanishing of this pressure term in equation ( [ eq : eomexpand ] ) , we then have @xmath307k_zv_z\ .
\label{eq : kdotv}\ ] ] the z - component of equation ( [ eq : eomexpand ] ) , with @xmath308 given by equation ( [ eq : kdotv ] ) , then yields the dispersion relation : @xmath309v_a \vert k_z \vert\ .\ ] ] next , we solve for the eigenfunction to lowest order in @xmath306 : @xmath310 where the x - axis has been chosen to lie in the plane containing @xmath144 and @xmath304 .
we now see that our assumption of negligible perturbed pressure is self - consistent to lowest order in @xmath306 .
while we have not used the x - component of equation ( [ eq : eomexpand ] ) , both the left - hand side and the right - hand side of this equation are of the same order in @xmath306 . note that both @xmath146 and @xmath303 are nearly parallel to @xmath304 : the slow wave polarization is nearly aligned with the unperturbed magnetic field . moreover , in the limit that @xmath311 , the slow mode is nearly incompressible . for adiabatic fluid motions ,
the linearized equation of energy conservation reads @xmath312 where @xmath167 is the entropy per particle .
thus , there exists a zero frequency , @xmath313 , mode with @xmath314 , which has @xmath315 with @xmath316 denoting temperature .
note that there is no corresponding mode in an isothermal fluid .
armstrong , j. w. , rickett , b. j. , & spangler , s. r. 1995 , apj , 443 , 209 banks , p. m. 1966 , planet .
space sci .
, 14 , 1105 barnes , a. 1966 , phys .
fluids , 9 , 1483 barnes , a. 1979 , in solar system plasma physics vol 1 , ed .
e. n. parker , c. f. kennel , & l. j. lanzerotti ( new york : north - holland ) , 249 braginskii , s. i. 1965 , rev .
plasma phys .
, 1 , 205 cho , j. , & vishniac , e. t. 2000a , apj , 538 , 217 cho , j. , & vishniac , e. t. 2000b , apj , 539 , 273 goldreich , p. 1998 , aas , 192 , 4603 goldreich , p. , & sridhar , s. 1995 , apj , 438 , 763 goldreich , p. , & sridhar , s. 1997 , apj , 485 , 680 higdon , j. c. 1984 , apj , 285 , 109 higdon , j. c. 1986 , apj , 309 , 1342 higdon , j. c. , & conley , a. 1998 , aas , 192 , 4609 lambert , h. c. , & rickett , b. j. 2000 , apj , 531 , 883 maron , j. , & goldreich , p. 2001 , apj , 554 , in press monin , a. s. , & yaglom , a. m. 1971 , statistical fluid mechanics ( cambridge : mit press ) mller , w. & biskamp , d. 2000 , phys .
, 84 , 475 quataert , e. 1998 , apj , 500 , 978 rickett , b. j. 1977 , ara&a , 15 , 479 rickett , b. j. 1990 , ara&a , 28 , 561 spitzer , l. , jr .
1978 , physical processes in the interstellar medium ( new york : john wiley & sons ) sridhar , s. 1998 , talk at conference on _ interstellar turbulence _ ,
puebla , mexico , 1998 tennekes , h. , & lumley , j. l. 1972 , a first course in turbulence ( cambridge : mit press )
|
radio - wave scintillation observations reveal a nearly kolmogorov spectrum of density fluctuations in the ionized interstellar medium . although this density spectrum is suggestive of turbulence , no theory relevant to its interpretation exists .
we calculate the density spectrum in turbulent magnetized plasmas by extending the theory of incompressible magnetohydrodynamic ( mhd ) turbulence given by @xcite to include the effects of compressibility and particle transport .
our most important results are as follows .
\(1 ) density fluctuations are due to the slow mode and the entropy mode .
both modes are passively mixed by the cascade of shear alfvn waves . since the shear alfvn waves have a kolmogorov spectrum , so do the density fluctuations .
\(2 ) observed density fluctuation amplitudes constrain the nature of mhd turbulence in the interstellar medium .
slow mode density fluctuations are suppressed when the magnetic pressure is less than the gas pressure .
entropy mode density fluctuations are suppressed by cooling when the cascade timescale is longer than the cooling timescale .
these constraints imply either that the magnetic and gas pressures are comparable , or that the outer scale of the turbulence is very small .
\(3 ) a high degree of ionization is required for the cascade to survive damping by neutrals and thereby to extend to small lengthscales .
regions that are insufficiently ionized produce density fluctuations only on lengthscales larger than the neutral damping scale .
these regions may account for the excess of power that is found on large scales .
\(4 ) provided that the thermal pressure exceeds the magnetic pressure , both the entropy mode and the slow mode are damped on lengthscales below that at which protons can diffuse across an eddy during the eddy s turnover time .
consequently , eddies whose extents _ along the magnetic field _ are smaller than the proton collisional mean free path do not contribute to the density spectrum .
however , in mhd turbulence eddies are highly elongated along the magnetic field . from an observational perspective
, the relevant lengthscale is that _ transverse _ to the magnetic field .
thus the cut - off lengthscale for density fluctuations is significantly smaller than the proton mean free path .
\(5 ) the alfvn mode is critically damped at the transverse lengthscale of the proton gyroradius , and thus cascades to smaller lengthscales than either the slow mode or the entropy mode .
# 1([#1 ] )
|
the work of morris and thorne@xcite has led to a great deal of interest in the formation and properties of three - dimensional wormholes ( topological connections between separated regions of space - time ) that are solutions of the einstein s equations of general relativity . subsequently visser@xcite suggested a wormhole configuration , a flat - space wormhole that is framed by `` struts '' of an exotic material , a variant of the cosmic string solutions of einstein s equations@xcite . to satisfy the einstein field equations the cosmic string framing visser wormholes
must have a _ negative _ string tension@xcite of @xmath0 and therefore a negative mass density .
however , for the total mass of the wormhole system , the negative mass density of the struts should be combined with the effective positive mass density of the wormhole s gravitational field .
the overall object could , depending on the details of the model , have positive , zero , or negative net external mass .
note that in hypothesizing the existence of such a wormhole , one has to abandon the averaged null energy condition@xcite .
therefore , the hypotheses underlying the positive mass theorem no longer apply and there is nothing , in principle , to prevent the occurrence of negative total mass@xcite .
some of the visser wormhole configurations have the shape of a cube or other geometrical solid , but one particularly simple configuration is a flat - space wormhole mouth framed by a single continuous loop of exotic cosmic string .
it has been suggested@xcite that the inflationary phase of the early universe might produce closed loops of cosmic string .
it is therefore at least plausible that a similar mechanism might produce negative - mass string loops framing stable visser wormholes .
if a particle with positive electric charge passes through such a wormhole , its lines of force , threading through the wormhole aperture , give the entrance mouth an effective positive charge ( flux lines radiating outward ) and give the exit mouth an effective negative charge ( flux lines converging inward . ) similarly , when a massive object passes through the wormhole , the same back - reaction mechanism might cause the entrance mouth to gain mass and the exit mouth to lose mass@xcite
. now let us consider a stable visser wormhole with near - zero mass residing in the mass - energy rich environment of the early universe .
the expected density fluctuations of the early universe suggest that the separated wormhole mouths will reside in regions of differing mass density , leading to a mass flow between the regions they connect .
as this mass passes through the wormhole , the entrance mouth will gain mass while the exit mouth will lose mass by the same amount .
soon , if the mass flow continues , the exit wormhole mouth will acquire a net negative mass .
this will lead to a gravitational instability , since the positive mass mouth will attract more mass through its aperture while the negative mass mouth will gravitationally repel nearby mass@xcite .
thus the positive - negative imbalance will be fed by gravity and will continue to grow .
if this process proceeds without interruption , the exit wormhole mouth might develop a stellar - scale negative mass .
visser wormholes , therefore , provide at least one motivation for seriously considering the possible existence of naturally occurring astronomical objects of sizable negative mass .
we have tried to sketch only one scenario for the possible existence of negative mass objects .
we must , however , point out that the non - existence of such objects would not rule out the existence of natural wormholes , since their properties are highly model - dependent .
also , observations providing evidence of negative mass objects would not require the existence of natural wormholes , since other negative - mass objects could conceivably exist , but would indicate a direct violation of the averaged null energy condition mentioned above .
negative - mass objects , while repelling all nearby mass ( positive or negative ) , would themselves be attracted@xcite by the mass of a nearby galaxy and might form part of a galactic halo .
they would have unusual gravitational properties that could produce detectable gravitational lensing effects .
these lensing effects , for the same absolute mass , are of the same magnitude as those recently detected for massive cosmic halo objects ( machos)@xcite , but , as we will show below , are qualitatively different in shape .
we here examine in detail the lensing effects of gravitationally negative anomalous compact halo objects ( gnachos ) .
naively , if a gravitationally attractive positive mass acts as a converging lens that brightens a background star , one might expect a gravitationally repulsive negative mass to act as a diverging lens that causes a background star to briefly grow dimmer .
actually the lensing of a negative mass is not analogous to a diverging lens . in certain circumstances
it can produce more light enhancement than does the lensing of an equivalent positive mass .
[ fig:1 ] shows the geometry of starlight that is gravitationally lensed by an object of negative mass .
starlight is radiated from the stellar source @xmath1 and is detected by the observer at @xmath2 . a gravitationally negative object
@xmath3 lies between source and observer at an impact - parameter distance @xmath4 from the source - detector axis @xmath5 .
an off - axis ray of light passes near the negative mass object @xmath3 , coming within a distance @xmath6 of it , and is deflected@xcite by an angle @xmath7 = @xmath8 is the absolute value of the deflecting mass , and @xmath9 is the velocity of light .
the angle @xmath7 is an external angle of the triangle formed by the direct and deflected rays .
this triangle has interior angles @xmath10 and @xmath11 , so @xmath7 = @xmath12 .
if the detector - source distance is @xmath13 and the detector - mass distance is @xmath14 , then we have the following equations : @xmath15 these lead to the dimensionless quadratic equation : @xmath16 where @xmath17 is the dimensionless distance of closest approach of the deflected ray , @xmath18 is the dimensionless impact parameter distance of the deflecting mass , and @xmath19 is the characteristic gravitational length scale of the problem . for a positive lensing mass , @xmath20 would be the radius of the einstein ring produced when the mass is positioned at zero impact parameter ( @xmath4=0 ) .
to give some feeling for this length scale , if the stellar source @xmath1 is in the large magellanic cloud and a negative lensing mass @xmath3 of one solar mass is in the galactic halo , then @xmath13 = 2 @xmath21 10@xmath22 m , @xmath14 = 5 @xmath21 10@xmath23 m , and @xmath20 = 1.5 @xmath21 10@xmath24 m or about 10 au .
solving quadratic ( [ eq:4 ] ) for @xmath25 gives two solutions : @xmath26.\ ] ] when @xmath27 there are two real solutions of the quadratic , corresponding to two rays that are deflected to the observer .
when @xmath28 there are no real solutions , indicating that the deflection is blocking all rays from reaching the observer . at @xmath29
= 2 , @xmath30 = @xmath31 = @xmath32 and , as will be discussed below , the rainbow - like caustic occurs , allowing many rays to reach the observer and producing a dramatic brightening of the background star . fig .
[ fig:2 ] shows this schematically .
the negative mass deflects rays in inverse proportion to their distance of closest approach , creating a shadowed umbra region where light from the source is extinguished . at the edges of the umbra the light rays
accumulate to form the caustic and give a very large increase in light intensity .
the intensity falls slowly to normal at larger transverse distances .
if the unmodified intensity of the background star is @xmath33 and the altered intensity of the background star in the presence of the negative lensing mass for each solution is @xmath34 , then the partial amplification factors @xmath35 are given by@xcite : @xmath36 the overall relative intensity @xmath37 is the modulation in brightness of the background star as detected by the observer , and is given by : @xmath38 it is interesting to compare this with the similar expression for relative intensity @xmath39 of a background source that is lensed by an object of _ positive _ mass : @xmath40 for the same dimensionless impact parameter @xmath29 with @xmath27 , it is always true that @xmath41 , so for large impact parameters a negative mass actually provides _ more _ light enhancement through lensing than an equivalent positive mass . when @xmath42 , the overall intensity @xmath43 , a condition indicating a caustic at which light rays from many trajectories are deflected toward the observer .
it provides a distinctive and unusual signature for negative - mass lensing .
the intensity modulation that is actually observed occurs when the lensing mass , which is assumed to be moving with transverse velocity @xmath44 , crosses near the source - detector axis @xmath5 with a minimum impact parameter @xmath45 and a minimum dimensionless impact parameter @xmath46 = @xmath47 .
the time - dependent impact parameter is therefore @xmath48 , and @xmath49 where @xmath50 = @xmath51 is the transit time across the distance of the minimum impact parameter and is the characteristic time scale of the problem .
[ eq:8 ] and [ eq:10 ] are used to calculate the light enhancement profile of the process , taking @xmath52 = 0 when @xmath53 , and there are no real solutions to eqn .
[ fig:3 ] shows these light enhancement profiles , plotting @xmath52 vs. @xmath54 for a range of minimum dimensionless impact parameters ranging from @xmath46 = 0.50 to 2.20 .
[ fig:4 ] shows similar light enhancement profiles for a positive mass of the same size and range of @xmath46 values .
as can be seen , the light enhancement profiles are qualitatively different for positive and negative lensing masses of the same magnitude and geometry . in particular
, the negative mass curves are much sharper , show stronger but briefer light enhancements , and for @xmath55 show a precipitous drop to zero intensity , i. e. , extinction of the light from the source when the lensing mass deflects all light from @xmath1 away from the observer .
such a signature might possibly be confused with that of occultation by a dark foreground object . however , an occultation would not be preceded by a dramatic rise in intensity and might , if the object had a significant atmosphere , have different light profiles at different wavelengths .
therefore , the negative gravitational lensing presented here , if observed , would provide distinctive and unambiguous evidence for the existence of a foreground object of negative mass .
the calculations presented above show that objects of negative gravitational mass , if they exist , can provide a very distinctive light enhancement profile .
since three groups are presently conducting searches for the gravitational lensing of more normal positive mass objects , we suggest that these searches be slightly broadened so that the signatures of the objects discussed above are not overlooked by over - specific data selection criteria and software cuts .
we recommend that macho search data be analyzed for evidence of gnachos .
this work was initiated at the nasa jet propulsion laboratory ( jpl ) workshop on advanced quantum / relativity theory propulsion concepts organized by dr .
robert h. frisbee of jpl and sponsored by dr .
gary l. bennett of the nasa office of advanced concepts and technology .
the authors wish to thank the other workshop participants for stimulating discussions and useful suggestions .
the work of jgc was supported in part by the division of nuclear sciences of the u. s. department of energy under grant de - fg06 - 90er40537 .
the work of rlf was supported in part by nasa through jpl .
the work of mv was supported in part by the u. s. department of energy .
|
visser has suggested traversable 3-dimensional wormholes that could plausibly form naturally during big bang inflation . a wormhole mouth embedded in high mass density might accrete mass , giving the other mouth a net _ negative _ mass of unusual gravitational properties .
the lensing of such a gravitationally negative anomalous compact halo object ( gnacho ) will enhance background stars with a time profile that is observable and qualitatively different from that recently observed for massive compact halo objects ( machos ) of positive mass .
we recommend that macho search data be analyzed for gnachos .
|
canonical analysis la dirac is a straightforward , if sometimes labyrinthine , approach to counting a system s physical degrees of freedom ( dof ) in the presence of gauge symmetries and non - linear interactions .
however , this approach can uncover unexpected subtleties , as already exemplified by some toy models in @xcite . in this paper , we show that more physically motivated theories can also contain similar subtleties , with qualitatively important consequences .
our focus will be on a specific self - interacting spin-2 gravity model , but similar effects may well arise in other interacting theories with higher - spin gauge symmetries , at least if some ( listed ) necessary conditions are met .
the theory we shall study in detail is the truncation of @xmath0 `` nmg '' @xcite to its pure quadratic curvature , yet ghost - free , part @xcite , @xmath1= \frac{1}{16}\int { \operatorname{d}\!}^3x\sqrt{-g}\ , \left[g^{\mu\nu}g_{\mu\nu}-\frac12\,g^2 \right ] = \frac{1}{16}\int { \operatorname{d}\!}^3x\sqrt{-g}\ , g_{\mu\nu}s^{\mu\nu } , \quad s_{\mu\nu } : = r_{\mu\nu}-\frac14 g_{\mu\nu}r\ , ; \label{eq : action2}\ ] ] here @xmath2 is the einstein tensor , @xmath3 is its trace and @xmath4 is the @xmath0 schouten tensor .
we have set an overall dimensional constant to unity and used mostly plus signature .
this model provides a remarkable example of symmetry - breaking through the clash between its two local , conformal and coordinate , invariances .
their co - existence at linearized level underlies this fourth derivative metric system s `` miraculous '' transmutation into single ghost - free vector excitation , or equivalently to a propagating torsion with non - propagating metric .
nonlinearly , however , conformal- is necessarily sacrificed to coordinate - invariance .
this paper is organized as follows : in section [ se:2 ] we perform the hamiltonian analysis . in section [ se:3 ]
we exhibit the bifurcation mechanism and count the number of physical degrees of freedom . in section [ se:4 ] ,
we first transmute the free field into a geometric model with propagating torsion , relating it to an ( improved ) version of einstein weyl gravity , as well as to its other , maxwell vector , avatar .
we then exhibit the nonlinear obstructions and their effects , in particular , introduction of propagator - less variables . in section [ se:5 ]
we comment on possible applications to other interacting gauge theories .
in keeping with our hamiltonian approach , we use a first - order formulation of the action , @xmath5= \int{\operatorname{d}\!}^3x \left[\frac12 { \epsilon}^{\mu\nu\rho}f^i{}_{\mu}r_{i\nu\rho}+\frac12 { \epsilon}^{\mu\nu\rho } { \lambda}^i{}_{\mu } t_{i\nu\rho}-\frac{e}{4}\left(f_{ik}f^{ik}-f^2\right)\right]\ , .
\label{eq : lag1}\ ] ] roman / greek indices are local / world . in form notation , the cartan , dreibein @xmath6 ( with determinant @xmath7 ) and ( dualized ) spin - connection @xmath8 , variables define the torsion and ( dualized ) curvature @xmath9 , @xmath10 .
the lagrange - multiplier @xmath11 ensures the on - shell torsion constraint @xmath12 , while the auxiliary @xmath13 is essentially the same as @xmath14 in @xcite and in the appendix a of the second reference @xcite .
the first - order action is just an ostrogradsky auxiliary variable form of the original fourth - order action , so the two are classically equivalent and share the same excitation content .
we now proceed to analyze canonically , following the earlier methods of @xcite , to which we refer for a more extensive discussion . the lagrangian s variables ( @xmath15 have conjugate momenta @xmath16 @xmath17 @xmath18 @xmath19 , leading to the primary constraints ( @xmath20 means weakly equal , i.e. , equal on the constraint surface ) : @xmath21 the constraints @xmath22 are second class
. it will be useful to define the linear combination @xmath23 eliminating the momenta @xmath24 leads to a partly reduced phase space , in which the nontrivial dirac brackets are given by @xmath25 the remaining dirac brackets are the same as the corresponding poisson brackets .
the canonical hamiltonian @xmath26 can be written as a sum over secondary constraints @xmath27 up to boundary terms . here ,
@xmath28 and @xmath29 the consistency conditions of the secondary constraints lead to the ternary constraints @xmath30 that establish the symmetry of the auxiliary fields .
similarly , the consistency conditions of the ternary constraints yield quaternary constraints @xmath31 at this stage the constraint procedure fortunately ends : no further constraints are generated through consistency conditions .
having found all constraints in the previous section , we establish their first / second class properties in order to count the number of physical dof .
we summarize the main result in tables [ tab:1 ] and [ tab:2 ] .
the fact that we have two different tables is a consequence of the advertised subtlety we called `` bifurcation '' and now explain .
the consistency condition of the quaternary constraint @xmath32 leads to the following dirac bracket @xmath33 where @xmath34 is the total hamiltonian @xmath35 and ( @xmath36 , @xmath37 , @xmath38 , @xmath39 ) are lagrange - multipliers of primary constraints .
all components @xmath40 and @xmath41 are determined at this stage ( and hence the corresponding constraints are second class ) , except for the lagrange multiplier @xmath42 that multiplies the primary constraint @xmath43 .
the consistency conditions allows us to determine the remaining lagrange multiplier @xmath42 , unless the condition @xmath44 holds .
the resulting bifurcation consists in the following : if condition does / not hold , the quaternary constraint @xmath32 and the primary constraint @xmath43 are both first / second class .
1.8pt ' '' '' & first class & second class + ' '' '' primary & @xmath45 & @xmath46 + ' '' '' secondary & @xmath47 & @xmath48 + ' '' '' ternary & & @xmath49 + ' '' '' quaternary & & @xmath50 + 1.8pt ' '' '' & first class & second class + ' '' '' primary & @xmath51 , @xmath43 & @xmath52 , remaining @xmath53 + ' '' '' secondary & @xmath47 & @xmath48 + ' '' '' ternary & & @xmath49 + ' '' '' quaternary & @xmath32 & @xmath54 + according to table [ tab:1 ] ,
i.e. , when is absent , we have a 48-dimensional phase space with 12 first class and 20 second class constraints .
consequently , the theory in general exhibits two local physical dofs , namely the massive bulk gravitons .
however , if the condition holds , then table [ tab:2 ] states that there are 14 first class constraints and 18 second class constraints . in that case
the number of local physical dofs is reduced to one .
translating condition into metric form , by using @xmath55 and @xmath56 , one finds that it holds automatically if the metric is a solution of the einstein equations @xmath57 this observation concurs with the fact that the linearized theory around einstein solutions has one additional gauge symmetry due to partial masslessness @xcite .
again , this symmetry enhancement results from the fact that the constraint @xmath43 becomes first class , leading to the additional gauge symmetry .
however , the latter is an artifact of linearization , being broken in the full nonlinear theory , as we have shown above .
in this section we provide an alternative derivation of the loss of gauge - invariance beyond linearization in a straightforward perturbative approach .
this sheds additional light on the bifurcation mechanism and exhibits it as a clash between diffeomorphism and conformal invariance .
moreover , the perturbative analysis can be useful particularly for possible higher - rank / higher - spin generalizations , where a full canonical analysis is often less accessible than a perturbative one .
we work with the second order , `` ostrogradski '' action [ equivalent to ] , using auxiliary , symmetric tensor density , variables @xmath58 : @xmath59= \frac{1}{4}\int { \operatorname{d}\!}^3x\ ; \big\{g_{\mu\nu}(g)\ , f^{\mu\nu } - \frac12 \big[f_{\mu\nu}^2 -(\textrm{tr}\ , f)^2\big]/ \sqrt{-g}\big\ } \label{eq : saction2}\ ] ] where we have omitted the contracting metrics .
[ completing squares and integrating out @xmath60 recovers ( [ eq : action2 ] ) . ]
we will study ( [ eq : saction2 ] ) , initially at linear , then full non - linear , metric levels .
the linearization of ( [ eq : action2 ] ) is manifestly ( linear ) diffeo - invariant , while use of the bianchi identity easily confirms its conformal invariance , under @xmath61 with our convention @xmath62
. both invariances also hold in ( [ eq : saction2 ] ) , of course , with @xmath60 transforming as a ( linear ) diffeo tensor and conformally like the schouten tensor : @xmath63 .
now we count dof : varying @xmath64 yields @xmath65 , where @xmath3 is the usual linear einstein operator . in @xmath0
, there are no einstein excitations ( riemann and ricci being equivalent ) , so @xmath66 is a pure gauge `` metric '' .
varying @xmath60 , we learn that @xmath64 obeys the einstein equation with linear source , so the general solution is:@xmath67 inserting legally ( [ eq : sol4 ] ) into ( [ eq : saction2 ] ) , we note first that its @xmath68 term vanishes by the bianchi identities upon part integration , leaving the quadratic @xmath60-terms : these precisely combine into the promised reduced one - dof maxwell action ( second reference in @xcite ) : @xmath69 \rightarrow \ ; -\frac14 \int { \operatorname{d}\!}^3x\ ; f_{\mu\nu}^2 , \quad f_{\mu\nu}:=(\partial_{\mu } a_{\nu } - \partial_{\nu } a_{\mu})\ , ; \label{eq : eq5}\ ] ] it is invariant under @xmath70 . that @xmath71 is indeed our conformal transformation parameter @xmath72
is then verified by ( [ eq : eq3],[eq : sol4 ] ) .
the above set of field equations is consistent with both underlying invariances ; for example , upon taking the divergence of einstein equation in ( [ eq : sol4 ] ) , its left side vanishes by the bianchi identity , while the divergence / conservation of its ( symmetric ) right side matter source also does , being proportional to the latters s field , i.e. , maxwell s , equations . in this connection ,
note that there is no `` spin - loss '' paradox in the above tensor - to - vector transmutation because all massless fields in @xmath0 are necessarily spinless @xcite , at least in flat space ; it would be useful to learn if this persists in ( a)ds . the above , conformal - to - gauge transmutation
should not be confused with a separate , surprising @xcite , @xmath0 conformal invariance enjoyed by maxwell , by virtue of its further transmutability into a scalar .
we provide a concise derivation , emphasizing the nonlocal ( as usual with such transmutations ) price involved : the first order maxwell action is @xmath73 = - \frac12 \int { \operatorname{d}\!}^3x \big[f^{\mu\nu } ( \partial_{\mu } a_{\nu } - \partial_{\nu } a_{\mu } ) - \frac12 f_{\mu\nu}^2\big]\ , , \label{eq : a}\ ] ] where @xmath14 and @xmath74 are independent variables .
varying @xmath74 gives @xmath75 , whose general solution is @xmath76 , the scalar @xmath77 having dimension of @xmath78 ; @xmath79 s dual vector @xmath80 is just @xmath81 , a relation whose highly non - local inverse we also note , @xmath82 replacing @xmath83 by @xmath84 in immediately yields the free scalar action , @xmath85 \rightarrow i[s ] = - \frac12 \int { \operatorname{d}\!}^3x ( \partial s)^2 $ ] . to cancel the scalar stress - tensor s trace , add to its minimal @xmath86
the usual identically conserved ( hence not affecting the poincare generators ) improvement @xcite term , @xmath87 .
its on - shell @xmath88 cancels that of @xmath89 .
the above procedure is nonlocal in @xmath90 , inherited from @xmath91 in .
but one can not improve directly at the vector level either .
any such attempt is doomed from the start : it is easily verified that there is no identically conserved @xmath92 as required by dimension that is even on - shell gauge - invariant , so one can not cancel the maxwell stress - tensor s trace , @xmath93 . nor is any other local choice available , since it must , just by dimensions , depend on @xmath94 rather than on the field strengths @xmath95 ; the scalar itself is the only gauge - invariant , dimensionally correct , but nonlocal , option .
[ we have not attempted to find a direct descent from to the scalar action . ] in canonical terms , the associated dof - reducing maxwell gauge constraint is the standard ( @xmath96 ) , leaving a @xmath97 photon with just one transverse dof , whose sign is fixed by that chosen for ( [ eq : action2 ] ) or ( [ eq : saction2 ] ) .
more explicitly , the maxwell action s kinetic term is ( @xmath98 ) , in terms of the spatial transverse - longitudinal orthogonal vector decomposition @xmath99 , which commutes with time - derivatives .
the longitudinal excitation is removed by the gauss constraint @xmath100 that enforces @xmath101 .
instead , the cubic correction ( [ eq : eq6 ] ) contains terms quadratic in @xmath102 , thereby replacing this constraint by an irrelevant ( because it integrates out ) perfect square plus ( cubic ) terms that depend on @xmath103 , so the longitudinal dof are reinstated .
( we have checked that cubic terms @xmath104 are absent though . )
not having studied the hamiltonian in detail , we can not assert that it is no longer bounded below , but that seems likely for any cubic : one would have to include quartic corrections for a meaningful conclusion , though that is not very relevant any more . this paradoxical second order form of a fourth order action is explained by the original metric propagator indeed being @xcite @xmath105 , thus agreeing with @xmath60 having dimension of curvature , @xmath106 , hence @xmath107 . [ actually , our method also applies , with appropriate dimensional numerical differences , to the @xmath108 weyl action , which can also be written ( modulo coefficient differences in schouten ) in the form .
the @xmath0 argument is unchanged , in particular that @xmath109 and @xmath79 each obey the einstein equation , so each has 2 , rather than @xmath0 s 0 , dof
. the @xmath74 , coordinate vector gauge , part of @xmath60 still represents a ( now 2 dof ) photon @xcite ( see also @xcite ) .
again , there is no spin - transmutation paradox ( despite the presence of spin @xmath108 ) , because the propagator of a fourth derivative order theory can ( and usually does ) include lower - spin poles . ]
our linearized system has a second , more general geometrical interpretation : it represents a propagating vector torsion and contortion , but a non - propagating ( because @xmath0 ) metric .
indeed , this is already suggested by the einstein equation in .
we recall that one may extend riemannian geometries by extending its metric affinity @xmath110 to include an antisymmetric torsion , @xmath111}$ ] , and a contortion , whose symmetric part we denote by @xmath112 [ we define ( anti-)symmetrization with a factor 1/2 ] .
for our system , the choices are @xmath113}\ , , \qquad k^{\alpha}{}_{(\mu\nu ) } = c\ ; { \delta}^{\alpha}_{(\mu } a_{\nu ) } + d \;\eta_{\mu\nu } a^{\alpha}\ , .
\label{eq : geomint}\ ] ] the parameter choice @xmath114 kills the ricci tensor s antisymmetric part .
the further choices @xmath115 and @xmath116 make the vanishing of the einstein tensor , but now of the full connection , @xmath117 , @xmath118 , coincide with the source - ful einstein equation of .
the divergence of the right hand side of that equation is proportional to the maxwell operator , whose vanishing is then assured by consistency with the metric bianchi identity .
note that this total connection @xmath119 is metric compatible in the einstein
weyl sense .
@xmath120 in this interpretation the maxwell field is recognized as the weyl potential , up to a factor and the presence of torsion , see e.g. @xcite .
we stress that the inclusion of torsion leads to an improvement of the usual torsionless einstein
weyl connection , since our ricci tensor is symmetric , while the einstein
weyl ricci - tensor in general is not .
more explicitly , for einstein weyl we have @xmath121 and @xmath122 ; note that metric - compatibility in the einstein
weyl sense is guaranteed if @xmath123 .
interestingly , the tracefree part of the right eq . coincides precisely with the einstein
weyl equations ( up to terms quadratic in the weyl potential ) , see @xcite and refs . therein .
this provides yet - another - way to see why conformal invariance is broken cubically : the einstein
weyl equations are weyl - invariant , but they coincide with our equations only at the linearized level , not non - linearly .
we now show explicitly how enforcing diffeo- destroys conformal- invariance beyond linear order , as it must since the full action ( [ eq : action2 ] ) involves the factor @xmath124 , rather than @xmath125 .
we need only consider the first , cubic , deviation , where the effect will be manifested as loss of maxwell gauge invariance .
returning to ( [ eq : saction2 ] ) and inserting the linearized values , we find that the cubic action reduces , schematically , to the `` bare @xmath74-form '' @xmath126 \sim \frac14\int { \operatorname{d}\!}^3x\,\big[2 g_2^{\mu\nu}(h)\partial_\mu a_\nu + h_{\mu\nu } q^{\mu\nu}(a)\big ] \ , , \label{eq : eq6}\ ] ] where @xmath127 ( and also its integral ) is an ( irreducibly ) gauge _ variant _ quantity bi - linear in the gauge potential @xmath78 .. ] this manifest loss of maxwell gauge invariance means of course , loss of conformal invariance . a concrete , `` no - calculation '' realization of invariance loss is now easy .
consider first the vacuum state in the gauge @xmath128 , @xmath129 : all linearized and cubic terms manifestly vanish .
now gauge - vary this `` null '' configuration , with local parameter @xmath130 , where @xmath131 are the minkowski coordinates .
then @xmath132 : this means @xmath74 is a conformal killing vector of the flat background .
the ( gauge invariant ) quadratic action still vanishes of course . instead , @xmath133 . this
, manifestly constant trace , @xmath134 , is easily found to be non - zero ; hence the cubic action is a non - invariant , proportional ( in this gauge ) to the vector `` mass '' term @xmath135 , already at vacuum .
we conclude that the simple one - dof content of the linearized level is indeed lost here , with the implied consequent presence of additional , ghost , modes that always plagues generic quadratic actions , albeit without propagators of their own .
we have studied a novel , higher - order , clash between two local invariances that characterize dynamical quadratic curvature schouten schouten - ghost - free @xmath0 gravity , at linear order . after performing a canonical analysis and exhibiting a bifurcation in the constraint analysis that leads to different counts of dofs , we considered the theory perturbatively , providing a transparent derivation of the free theory s one - dof , second order , character : our geometrical , torsionful , representation complementing an earlier vectorial transmutation .
we then traced the unavoidable breaking , already at cubic level , of the model s conformal symmetry by its nonlinear diffeo - invariant , but conformal - gauge dependent , completion .
the culprits were the dynamical ( rather than minkowski ) metrics that contract indices beyond lowest order .
the above symmetry - breaking raises one , also novel , field - theoretic puzzle posed by this otherwise consistent model : its propagator depends on fewer variables than do its vertices . how does one calculate ( at least perturbatively ) either classically or at quantum loop level ?
the new variables , having no lines of their own , can only lie on open , but presumably not on closed loop lines , yet they are not external fields either .
we have also checked the consistency of our perturbative analysis with a non - perturbative canonical analysis along the lines of @xcite , with perfect agreement .
a ( vaguely ) similar situation occurs in topologically massive gravity , whose metric s propagator is not uniformly of either second or third derivative order , since its third derivative cotton contribution is independent of the metric s conformal factor , one that is present in its einstein , second order term .
that problem is not one of principle , however all components have proper propagators , just ones of different momentum order and can only affect topologically massive gravity s uv behavior @xcite . rather , the nearest analog is perhaps massive @xmath108 einstein gravity with a `` wrong '' explicit mass term a la @xcite ; however that symmetry - breaking pathology is inserted by hand , rather than , as in our model , from enforcing a greater invariance ! clearly , some intriguing unsolved directions remain .
the general non - perturbative conditions of the canonical analysis we encountered here are a non - perturbative extension and here agree with the perturbative results .
moreover , a similar effect was observed in generalized massive gravity , see second reference in @xcite .
this suggests that the bifurcation effect we described here is not just confined to our specific model , but can be present in more general interacting higher - rank / higher - spin gauge theories as well as , perhaps , in vector models , see @xcite and refs . therein .
we conclude with a list of ingredients necessary , but not always sufficient , for the emergence of bifurcation . 1 .
higher ( at least spin 2 ) gauge symmetries 2 .
non - linear ( at least cubic ) interactions 3 .
the linearized theory must have additional ( linearized ) gauge symmetries such as conformal / weyl symmetry 4 .
the linearized gauge symmetry must be broken in the interacting theory it would be interesting to construct explicit examples of bifurcation in @xmath136 theories with spins higher than 2 , and/or towers thereof .
we thank h. afshar , j. franklin , o. hohm , r. jackiw and p. townsend for useful discussions .
dg thanks caltech and kitp for hospitality during the kitp program ` bits , branes , black holes ' while part of this work was begun .
sd was supported in part by nsf phy-1064302 and doe de - fg02 - 164 92er40701 grants .
se and dg are supported by the start project y435-n16 of the austrian science fund ( fwf ) and by the fwf project p21927-n16 .
this research was supported in part by the national science foundation under grant no .
nsf phy11 - 25915 .
[ [ note - added ] ] note added : + + + + + + + + + + + as this work was completed , and simultaneous with our submission of the perturbative analysis ( essentially section [ se:4 ] ) , the closely related posting @xcite appeared .
it discusses the chern - simons and other extensions of the present model .
where they overlap , our conclusions agree ; indeed the models of @xcite provide further evidence for our conjecture that the bifurcation we encountered here is a more generic feature of interacting higher - rank / higher - spin theories .
10 natexlab#1#1bibnamefont # 1#1bibfnamefont # 1#1citenamefont # 1#1url # 1`#1`urlprefix[2]#2 [ 2][]#2 e. a. bergshoeff , o. hohm , and p. k. townsend , `` massive gravity in three dimensions , '' _ phys .
* 102 * ( 2009 ) 201301 , http://www.arxiv.org/abs/0901.1766[arxiv:0901.1766 ] . * * , ( ) , 124042 , http://www.arxiv.org/abs/0905.1259[arxiv:0905.1259 ] .
s. deser , `` ghost - free , finite , fourth order d=3 ( alas ) gravity , '' _ phys .
lett . _ * 103 * ( 2009 ) 101302 , http://www.arxiv.org/abs/0904.4473[0904.4473 ] .
h. afshar , b. cvetkovic , s. ertl , d. grumiller , n. johansson , `` holograms of conformal chern - simons gravity '' * * , , http://arxiv.org/abs/arxiv:1106.6299[arxiv:1106.6299 ] .
`` conformal chern - simons holography - lock , stock and barrel . '' * * , , http://arxiv.org/abs/arxiv:1110.5644[arxiv:1110.5644 ] .
s. deser and a. waldron , `` partial masslessness of higher spins in ( a)ds , '' _ nucl .
phys . _ * b607 * ( 2001 ) 577604 , http://www.arxiv.org/abs/hep-th/0103198[hep-th/0103198 ] .
b. binegar , `` relativistic field theories in three - dimensions , '' _ j.math.phys . _ * 23 * ( 1982 ) 1511 .
s. deser and r. jackiw , `` statistics without spin : massless d = 3 systems , '' _ phys.lett .
_ * b263 * ( 1991 ) 431436 .
r. jackiw and s .- y .
pi , `` tutorial on scale and conformal symmetries in diverse dimensions '' http://arxiv.org/abs/1101.4886[arxiv:1101.4886 ] .
, , , , , * * : 223001 , ( ) .
d. grumiller and r. jackiw , `` einstein - weyl from kaluza - klein , '' _ phys .
lett . _ * a372 * , 2547 ( 2008 ) , http://www.arxiv.org/abs/arxiv:0711.0181 [ math - ph][arxiv:0711.0181 [ math - ph ] ] . s. deser and z. yang , `` is topologically massive gravity renormalizable ?
_ class.quant.grav .
_ * 7 * ( 1990 ) 16031612 .
d. boulware and s. deser , `` can gravitation have a finite range ?
_ phys.rev .
_ * d6 * ( 1972 ) 33683382 .
r. jackiw and s .- y .
pi , `` seeking an even parity mass term for 3-d gauge theory , '' _ phys.lett .
_ * b403 * ( 1997 ) 297303 , http://www.arxiv.org/abs/hep-th/9703226[hep-th/9703226 ] . , * * , ( ) , http://www.arxiv.org/abs/1104.0164[1104.0164 ] .
, , * * , ( ) , http://www.arxiv.org/abs/0704.1512[0704.1512 ] .
|
we present a non - perturbative canonical analysis of the @xmath0 quadratic - curvature , yet ghost - free , model to exemplify a novel , `` constraint bifurcation '' , effect .
consequences include a jump in excitation count : a linearized level gauge variable is promoted to a dynamical one in the full theory .
we illustrate these results with their concrete perturbative counterparts .
they are of course mutually consistent , as are perturbative findings in related models .
a geometrical interpretation in terms of propagating torsion reveals the model s relation to an ( improved ) version of einstein weyl gravity at the linearized level
. finally , we list some necessary conditions for triggering the bifurcation phenomenon in general interacting gauge systems .
|
stimulated raman adiabatic passage ( stirap ) is one of the most popular tools , used for manipulation of quantum structures @xcite .
this method transfers population adiabatically between two states @xmath0 and @xmath1 , in a three - level quantum system , without populating the intermediate state @xmath2 .
the applications of stirap cover a huge part of contemporary physics : coherent atomic excitation @xcite , control of chemical reactions @xcite , quantum information processing @xcite , coherent quantum state transfer and spatial adiabatic passage @xcite , waveguide optics @xcite , to name a few .
the technique of stirap is based on the existence of a dark state , which is an eigenstate of the hamiltonian and is a time - dependent superposition of the initial and target states .
because stirap is an adiabatic technique , it achieves high fidelity only in the limit of adiabatic evolution , which requires large temporal pulse areas .
if the adiabatic condition is not perfectly realized , the nonadiabatic coupling between the eigenstates causes the efficiency of the process to drop down .
several methods to achieve higher fidelity of stirap have been proposed .
one scenario is to use an additional field to cancel the nonadiabatic coupling @xcite .
another approach is to minimize the nonadiabatic losses by applying lewis - riesenfeld inverse engineering @xcite , which has been shown in @xcite to be potentially equivalent to berry s transitionless quantum driving @xcite .
one may also use the dykhne - davis - pechukas formula which leads to the so called parallel adiabatic passage @xcite .
these techniques imply a strict time dependence of the pulse shapes and detuning .
finally , one can use composite pulses to improve dramatically the fidelity of adiabatic passage @xcite , by choosing appropriately the relative phases of the pulses in the sequence .
other theoretical techniques proposed to engineer the shortcuts , with a detailed discussion of the experimental results and prospects , can be found in the recent review @xcite . in the recent years
there has been a growing interest in the use of non - hermitian ( nh ) hamiltonians @xcite , especially in the context of @xmath3-symmetric systems @xcite .
it was demonstrated , for instance , that a @xmath3-symmetric hamiltonian can produce a faster than hermitian evolution in a two - state quantum system , while keeping the eigenenergy difference fixed @xcite .
nh extensions have been done on the classical landau - zener model @xcite and some schemes for the realisation of @xmath3-symmetry have been proposed @xcite .
recently , an approximation of the adiabatic condition for nh systems was also derived @xcite . in this paper
, we propose a technique to increase the speed of stirap by using a nh term in the hamiltonian to cancel the nonadiabatic coupling ; hence we call this method a nh shortcut to stirap .
the proposed method is similar to the nh shortcut for two - level adiabatic processes , studied in @xcite .
the paper is organized as follows .
first , we shortly review the theory of stirap in the case of resonant excitation .
then we study how additional nh terms in the hamiltonian affect the evolution and we show how nh terms can be used to cancel the nonadiabatic coupling .
we then consider two concrete examples and examine the limitations of our method .
interestingly , for a stirap using delayed gaussian - shaped pulses in the counter - intuitive scheme the imaginary energy terms of the hamiltonian turn out to be time independent . finally , a possible physical implementation of nh shortcut of stirap in waveguide optics is briefly discussed .
, @xmath2 and @xmath1 form a lambda configuration .
the pump field @xmath4 couples states @xmath0 and @xmath2 and the stokes field couples states @xmath1 and @xmath2 .
( b ) tunneling - coupled optical waveguides realizing nh stirap in the counter - intuitive scheme for gaussian pulses .
populations of atomic states @xmath0 , @xmath2 and @xmath1 in ( a ) are mimicked by light power propagating in waveguides @xmath5 , @xmath6 and @xmath7 , respectively .
waveguide @xmath5 is lossy with loss rate @xmath8 , whereas waveguide @xmath7 provides gain @xmath9 .
the pump ( stokes ) rabi frequency @xmath10 corresponds to the tunneling rate @xmath11 of outer waveguides with the center waveguide.,width=321 ] we start by briefly reviewing the theory of stirap @xcite .
let us consider a three - level quantum system in a lambda configuration [ fig .
[ system](a ) ] , coherently driven by two external fields , the pump field @xmath4 , and the stokes field @xmath12 .
the goal is to transfer all the population from the initial state @xmath0 to the final state @xmath1 , without populating the intermediate state @xmath2 .
the evolution of the system is described by the schrdinger equation [ schr ] i _ t ( t ) = ( t)(t ) , where the vector @xmath13^t$ ] contains the three probability amplitudes of states @xmath0 , @xmath2 and @xmath1 . the hamiltonian in the rotating - wave approximation and assuming exact resonance reads @xcite [ h ] ( t ) = 2 , where the rabi frequencies @xmath4 and @xmath12 of the pump and stokes pulses are assumed to be real .
the mechanism of the population transfer in stirap and the limitation of adiabaticity can be easily understood if we introduce the so - called adiabatic basis .
this is the basis , formed of the instantaneous eigenstates @xmath14 , @xmath15 , @xmath16 of the time - varying hamiltonian @xmath17 .
these _ adiabatic states _ are connected to the original states @xmath0 , @xmath2 and @xmath1 by [ adbasis ] @xmath18 where the time - dependent mixing angle @xmath19 is defined as ( t ) = _ p(t)/_s(t ) .
the probability amplitudes of the adiabatic states @xmath20^t$ ] are connected to the original ones by using the transformation ( t)=(t)(t ) , where the transformation matrix @xmath21 is given by [ r ] ( t ) = .
the schrdinger equation in the adiabatic basis can be written as [ schra ] i _ t ( t ) = _ a(t)(t ) , with @xmath22 , or explicitly [ ha ] _
a(t ) = , where @xmath23 .
now , if we assume an adiabatic ( slow ) evolution , then @xmath24 which means that if the system is initially in one of the adiabatic states , then it stays in it during the evolution .
it is this type of adiabatic following , which is in the essence of stirap .
let us now take a closer look at the adiabatic state @xmath25 .
we see that it contains only states @xmath0 and @xmath1 and has no component on state @xmath2 [ see eq . ] .
furthermore , for a _ counterintuitive _ order of the pump and stokes pulses , we have the relations @xmath26 and @xmath27 .
hence , as time evolves , the mixing angle @xmath28 rises from @xmath29 to @xmath30 and we get = , = - .
this means , that if the evolution is perfectly adiabatic , and the system is initially prepared in state @xmath0 , at the end of the process we will achieve a complete population transfer to state @xmath1 , with negligible excitation of the intermediate state @xmath31 .
however , the evolution is never perfectly adiabatic and some nonadiabatic coupling @xmath32 is always present , which limits the efficiency of stirap . in the next section we show how this problem can be overcome by adding nh terms in the hamiltonian .
we consider again the stirap hamiltonian , but this time we add two time - dependent nh terms @xmath33 in the diagonal @xcite , [ hg ] ^(t ) = 2 . which correspond to time - varying complex energies of the bare states @xmath34 and @xmath35 . in the adiabatic basis , this hamiltonian has the form @xmath36.\end{aligned}\ ] ] our goal is to choose @xmath37 in such a way that the terms in the hamiltonian , which cause the evolution of the system to deviate from state @xmath25 , vanish .
this can be achieved by setting [ gamma ] ( t)= , which nullifies the first and the third element from the second column of @xmath38 . in this way
the evolution of the amplitudes of the adiabatic state obey the equations @xmath39 , \label{a0}\\ & a_-(t_f)=0 , \end{aligned}\ ] ] where we have assumed that initially the state of the system coincides with the adiabatic state @xmath40 , hence @xmath41 .
it can be shown that if @xmath4 and @xmath12 are reflections of each other , @xmath42 [ e.g. , if @xmath4 and @xmath12 are identical symmetric functions of time ] , where @xmath43 is the pulse delay , then @xmath37 is an even function of time and @xmath44 is an odd function of time . hence
, if we assume symmetric time interval , @xmath45 , the integral in is equal to zero .
this means that despite the fact that the hamiltonian is nh , the norm of the state vector at the end of the evolution will be equal to unity .
there is , however , an essential distinction between the hermitian and non - hermitian method in the transient population during the evolution . while in the hermitian case the norm of the state vector remains unity during the entire evolution , in the nh stirap , as seen from eq .
, the norm of the state vector will change in time .
as a function of time for gaussian pulse shapes with @xmath46 and @xmath47 . the black dashed line illustrates the standard stirap technique , the solid blue line shows the nh stirap , and the vertical dotted lines mark the initial and final moments of time @xmath48 and @xmath49 , used in the integration of the schrdinger equation .
( bottom ) shapes of the pump and stokes pulses and of the gain / loss function @xmath8(t).,width=566 ] in order to better illustrate the method of the nh shortcut of stirap , we consider two special cases , and we calculate the nh ( gain / loss ) functions @xmath37 for both of them . as a first example , let us assume that the pump and stokes pulses have gaussian shape , [ gaussians ] @xmath50 where @xmath51 is the peak rabi frequency , @xmath52 is the pulse duration and @xmath43 is the delay between the pulses . for this pulse shape
, it can be easily shown that the gain / loss function of eq .
is constant , [ eq15 ] ( t)=2 .
this time independence of @xmath8 allows one a relatively easy implementation of the nh hamiltonian in waveguide optics , where @xmath8 describes optical amplification / attenuation in the outer waveguides of a three - waveguide structure @xcite , as discussed in more detail in the next section . in fig .
[ populationgaus ] we plot the evolution in time of the population of the target state @xmath1 , for the regular stirap ( @xmath53 ) and for the nh extension ( @xmath54 ) .
we can see from the figure , that the nh stirap strongly outperforms the original hermitian stirap , by achieving complete population transfer , while for the standard stirap we have only @xmath55 for the same values of @xmath51 and @xmath43 . as a second example we assume hyperbolic secant pulse shapes , [ sech ] @xmath56 in this case
, the gamma term is given by ( t)=. in fig .
[ populationsech ] we show the evolution of the population of the target state , as well as the pulse shapes and @xmath8 .
again , we find that by using the nh stirap , we obtain much higher fidelity , even for parameter values where the ordinary stirap fails .
we want to note here , that both in the cases of gaussian and sech pulse shapes , the function @xmath37 does not depend on the particular value of the peak rabi frequency .
this is an interesting feature , which allows our technique to be used even for small values of @xmath51 , where the usual stirap technique would not work .
but for hyperbolic secant pulse shapes with @xmath57 and @xmath58.,width=566 ]
as anticipated above , our suggested method of nh shortcut to adiabaticity could be physically realized in waveguide optics @xcite . in this case , the system is composed of three optical waveguides @xmath5 , @xmath6 and @xmath7 in the geometrical setting schematically shown in fig . [ system](b ) .
in photonic stirap , the role of rabi frequencies @xmath59 and @xmath60 is played by the tunneling rates @xmath61 and @xmath62 between the center waveguide @xmath6 and the waveguides @xmath5 and @xmath7 respectively . therefore , to implement the desired pulse shape , the separation distance @xmath63 and @xmath64 between outer and center waveguides is slowly varied along the propagation direction @xmath65 . since the tunneling rate scales exponentially with the waveguide separation according to the law @xmath66\}$ ] ( with @xmath51 and @xmath67 two constant parameters to be determined from the fabrication process ) , in order to implement a gaussian pulse shape a quadratic law for @xmath63 and @xmath64 is required ( see ref . for further details ) . in our nh generalization of optical stirap , for a gaussian pulse shape , a longitudinally - independent optical loss @xmath9 and an equal rate of net optical amplification @xmath8 should be applied to the outer waveguides @xmath5 and @xmath7 , respectively .
a possible implementation of such an active / passive system can be achieved in ti in - diffused fe : linbo@xmath68 optical waveguides , which have been recently exploited to experimentally demonstrate @xmath3-symmetric directional couplers in the visible @xcite .
actually , the geometry of fig . [ system](b ) can be easily obtained in ti in - diffused linbo@xmath68 waveguides by using standard photolithography to shape the ti stripes before in - diffusion . moreover
, loss rate @xmath8 can be controlled by the amount of fe@xmath69 dopants , giving rise to optical excitation of electrons from fe@xmath69 centers to the conduction band of linbo@xmath68 . finally , optical gain
is provided under laser pumping through two - wave mixing process thanks to the photorefractive nonlinearity induced by fe - doping : a gain coefficient @xmath70 of few @xmath71 can be easily achieved ( @xmath72 @xmath71 in ref . ) .
note that , as in ref .
, the pump laser beam is incident from the top of the sample and a suitable amplitude mask ought to be employed to selectively control the amount of pump illumination in the three waveguides so to achieve : @xmath73 , i.e. no pumping , to have net attenuation with rate @xmath9 in @xmath5 ; @xmath74 , i.e. transparency , in @xmath6 ; and @xmath75 , so to have net optical amplification @xmath76 , in @xmath7 . to provide more evidence of feasibility of our scheme ,
we numerically studied beam propagation in a three - waveguide optical structure in the geometrical setting schematically shown in fig .
[ system](b ) . in the scalar and paraxial approximations , the optical schrdinger equation for the electric field envelope @xmath77 that describes
beam propagation in the optical structure reads ( see , for instance , @xcite ) @xmath78 where @xmath79 , @xmath80 is the wavelength of light wave propagating in the dielectric medium , @xmath81 is the substrate refractive index at wavelength @xmath80 , @xmath65 and @xmath82 are the longitudinal and transverse spatial coordinates , respectively , and @xmath83 is the optical potential , which is related to the refractive index change @xmath84 of the guiding structure from the substrate region by the simple relation @xmath85 .
an error function shape @xmath86-{\rm erf } [ ( x - w_1)/d_x ) ] \ } / [ 2 \ ; { \rm erf } ( w_1/d_x)]$ ] has been assumed for the refractive index profile of each of the three waveguides in the structure [ see the inset of fig .
[ fig4](a ) ] , with a channel width @xmath87 m , a diffusion length @xmath88 m and a maximum index change @xmath89 , which are typical to ti in - diffused linbo@xmath68 optical waveguides . to simulate loss and gain in the outer waveguides @xmath90 and @xmath91 , a small imaginary part @xmath92 of opposite signs , leading to a complex refractive index , has been added to @xmath93 for the waveguides @xmath90 and @xmath91 .
the optical structure is probed at @xmath94 nm ( ar - ion laser light ) , corresponding to a bulk refractive index @xmath95 . the outer waveguides @xmath90 and @xmath91 are parabolically - curved along the propagation direction @xmath65 ; the detailed behavior of the spacings @xmath63 and @xmath64 of waveguides @xmath90 and @xmath91 from the straight central waveguide @xmath96 are depicted in fig . [ fig4](a ) .
the minimum separation distance is @xmath97 m , whereas the longitudinal shift is @xmath98 mm [ see fig .
[ system](a ) ] .
the bent profiles of waveguides @xmath90 and @xmath91 yield an effective gaussian variation with @xmath65 of the coupling constants @xmath61 and @xmath62 [ see eqs . ] , corresponding to @xmath99 and @xmath100 mm , i.e. @xmath101 as in fig .
[ populationgaus](a ) .
the full sample length is @xmath102 mm .
figure [ fig4](c ) shows , in a pseudo - color map , the numerically - computed light intensity evolution along the sample for @xmath103 , i.e. in the absence of gain and loss in the outer waveguides . at the input plane , @xmath104
, the waveguide @xmath90 is excited by a gaussian beam which is well overlapped with the fundamental mode of the waveguide .
the corresponding behavior of the optical light power trapped in waveguide @xmath91 , normalized to the input optical power value in waveguide @xmath90 , is shown by the dashed curve of fig .
[ fig4](b ) , which reproduces very well the behavior predicted by coupled - mode equations ( the dashed curve in the upper panel of fig .
[ populationgaus ] ) . to cancel the non - adiabatic terms ,
according to eq .
a power loss / gain coefficient @xmath105 is required .
such a gain / loss value , which can be realized by fe - doping , is obtained by assuming an imaginary part @xmath106 of the refractive index change in waveguides @xmath90 and @xmath91 . in fig . [
fig4](d ) we show the numerically - computed light intensity evolution along the sample in the presence of loss and gain in waveguides @xmath90 and @xmath91 , respectively .
the corresponding behavior of the normalized optical light power trapped in waveguide @xmath91 is shown by the solid curve in fig .
[ fig4](b ) , which reproduces very well the curve predicted by the coupled - mode equation model ( the solid curve in the upper panel of fig.[populationgaus ] ) .
( b ) that realizes the nh hamiltonian .
( a ) behavior of the waveguide separation distances @xmath63 and @xmath64 of waveguides @xmath90 and @xmath91 from the central waveguide @xmath96 . the refractive index profile @xmath107 of the single waveguide channel used in the numerical simulations is depicted in the inset .
( b ) evolution of the optical beam power trapped in waveguide @xmath91 , normalized the optical power that excites waveguide @xmath90 at the input plane .
the dashed curve refers to the hermitian model ( i.e. absence of loss and gain in waveguides @xmath90 and @xmath91 ) , whereas the solid curve corresponds to the nh model with a uniform loss / gain rate defined by eq . .
panels ( c ) and ( d ) show the evolution of beam intensity along the waveguide system , as obtained by solving the optical schrdinger equation using a beam propagation method , in the hermitian [ panel ( c ) ] and non - hermitian [ panel ( d ) ] case .
parameter values are given in the text.,width=332 ]
in this paper we have studied theoretically a nh generalization of stirap , which allows one to cancel the nonadiabatic coupling and increase the speed and fidelity of the process .
we have examined two special cases , which allow to understand how the technique works for a particular shape of the pump and stokes pulses . remarkably , for a stirap using delayed gaussian - shaped pulses in the counter - intuitive scheme , the required imaginary terms of the hamiltonian turn out to be time independent .
this allowed us to propose a feasible implementation of nh stirap in optical waveguides with uniform gain / loss .
we want to note here , that our method can not be applied for arbitrary pulse shapes , but has some restrictions . in order to find the domain of applicability of our technique
, we can express the gain / loss function @xmath8 in terms of the pump and stokes rabi frequencies .
using eq .
, after some simple algebraic calculations , we obtain ( t)=- = hence , in order to be able to apply this method , the logarithm of the ratio of the two pulse functions needs to have well defined first derivative .
for instance , if we try to use @xmath108 pulse shapes , one can easily check that the @xmath8 term is divergent , which obviously limits the applicability of our method .
another point , which is important to stress is that the technique of nh shortcut is very sensitive to the initial conditions .
the initial state of the system must be very closely prepared to the adiabatic state @xmath40 .
otherwise , the norm of the state vector will not be conserved and some extra gain or loss will be introduced . finally , we point out that the scheme , which we propose , only works for the case of resonant stirap .
if the one - photon detuning is not equal to zero , the nonadiabatic couplings will depend on two mixing angles ; hence it is not straightforward to nullify the losses and further investigation may be necessary .
one possible direction for further investigation , in order to overcome this issue , might be to use the non - hermitian generalization of the lewis - resenfield theory @xcite to find a shortcut for such hamiltonian .
k. bergmann , h. theuer , and b. w. shore , rev .
phys . * 70 * , 1003 ( 1998 ) ; n. v. vitanov , m. fleischhauer , b. w. shore , and k. bergmann , adv . at . ,
, opt . phys . * 46 * , 55 ( 2001 ) ; n. v. vitanov , t. halfmann , b. w. shore , and k. bergmann , annu . rev .
phys . chem . *
52 * , 763 ( 2001 ) .
m. hennrich , t. legero , a. kuhn , and g. rempe , phys .
lett . * 85 * , 4872 ( 2000 ) ; a. kuhn , m. hennrich , and g. rempe , ibid . * 89 * , 067901 ( 2002 ) ; t. wilk , s. c. webster , h. p. specht , g. rempe , and a. kuhn , ibid . * 98 * , 063601 ( 2007 ) ; j. l. srensen , d. mller , t. iversen , j.b .
thomsen , f. jensen , p. staanum , d. voigt , and m. drewsen , new j. phys . * 8 * , 261 ( 2006 ) .
k. eckert , m. lewenstein , r. corbalan , g. birkl , w. ertmer , and j. mompart , phys .
a * 70 * , 023606 ( 2004 ) ; a. d. greentree , j. h. cole , a. r. hamilton , and l. c. l. hollenberg , phys .
b * 70 * , 235317 ( 2004 ) ; l. c. l. hollenberg , a. d. greentree , a. g. fowler , and c. j. wellard , phys . rev .
b * 74 * , 045311 ( 2006 ) ; k. eckert , j. mompart , r. corbalan , m. lewenstein , and g. birkl , opt . commun . *
264 * , 264 ( 2006 ) .
e. paspalakis , opt . commun . * 258 * , 30 ( 2006 ) ; s. longhi , phys . rev .
e * 73 * , 026607 ( 2006 ) ; s. longhi , g. della valle , m. ornigotti , and p. laporta , phys . rev .
b * 76 * , 201101 ( 2007 ) ; y. lahini , f. pozzi , m. sorel , r. morandotti , d. n. christodoulides , and y. silberberg , phys .
* 101 * , 193901 ( 2008 ) ; s .- y . tseng and m .- c .
wu , ieee photon .
* 22 * , 1211 ( 2010 ) ; t .- y .
hsiao , y .- w .
jhang , c. hu , and s .- y .
tseng , opt .
express * 20 * , 24085 ( 2012 ) ; a. p. hope , t. g. nguyen , a. d. greentree , and a. mitchell , opt .
express * 21 * , 22705 ( 2013 ) .
r. g. unanyan , l. p. yatsenko , b. w. shore , k. bergmann , opt . commun . * 139 * , 48 ( 1997 ) ; xi chen , i. lizuain , a. ruschhaupt , d. gury - odelin , and j. g. muga , phys . rev
. lett . * 105 * , 123003 ( 2010 ) ; l. giannelli and e. arimondo , phys .
rev . a * 89 * , 033419 ( 2014 )
. e. m. graefe , h. j. korsch , czech . j. phys . * 56 * , 1007 ( 2006 ) ; s. a. reyes , f. a. olivares and l. morales - molina , j. phys .
a : math . theor . * 45 * , 444027 ( 2012 ) ; r. uzdin and n. moiseyev , j. phys .
a : math . theor .
* 45 * , 444033 ( 2012 ) .
note that , at each time instant @xmath65 , the nh matrix @xmath109 turns out to be @xmath3-symmetric if @xmath110 [ see : c. m. bender , p. n. meisinger , and q. wang , j. phys .
a * 36 * , 6791 ( 2003 ) ] . to realize nh stirap ,
a counter - intuitive sequence of rabi frequencies is needed , so that in our case the nh matrix @xmath109 in not @xmath3 invariant .
|
we propose a non - hermitian generalization of stimulated raman adiabatic passage ( stirap ) , which allows one to increase speed and fidelity of the adiabatic passage .
this is done by adding balanced imaginary ( gain / loss ) terms in the diagonal ( bare energy ) terms of the hamiltonian and choosing them such that they cancel exactly the nonadiabatic couplings , providing in this way an effective shortcut to adiabaticity . remarkably , for a stirap using delayed gaussian - shaped pulses in the counter - intuitive scheme the imaginary terms of the hamiltonian turn out to be time independent .
a possible physical realization of non - hermitian stirap , based on light transfer in three evanescently - coupled optical waveguides , is proposed .
|
first argument of the pbh formation was presented by carr and hawking @xcite , in which the key concept of the pbh formation is that the three different length scales of the particle horizon , the schwarzschild radius , and the jeans length of any overdense region at radiation - dominated epoch are of the same order of magnitude .
once an initially superhorizon - size density fluctuation crosses the particle horizon , competition between self gravity and centrifugal pressure force determines whether such a fluctuation collapses into black hole or not@xcite . if the self gravity overcomes the pressure force , then the fluctuation will form a black hole with the order of horizon mass .
jedamzik @xcite found that , during cosmic qcd phase transition , there is almost no pressure response on adiabatic compression of fluctuation in the mixed phase of quark - gluon plasma and hadron gas .
this could occur because the enhanced energy density due to the compression of the fluctuations will turn into the high energy quark - gluon phase , instead of increasing centrifugal pressure force . as a result the overdense region from initially existing fluctuations tends to collapse efficiently to form the pbh during the qcd epoch when horizon mass scale is around 1@xmath1 .
pbh mass spectrum is most likely dominated by the horizon mass at the phase transition epoch @xcite .
green and liddle @xcite verified that this speculation turns out to be a good approximation in their detailed studies of the pbh mass spectrum .
these theoretical speculations have been studied more quantitatively in 1d general - relativistic hydrodynamic calculations @xcite , where the required critical overdensity parameter , @xmath2 , was found to be @xmath3 for the pbh formation .
similar sophisticated calculations @xcite have also shown numerically that @xmath4 is even as low as @xmath5 by taking account of the dynamics of the qcd phase transition in supercooling phase by the use of eos in the mit bag model for the mixed phase .
recently an interesting cosmological proposal has been made that our universe is occupied fractionally by a new class of extra - energy component in addition to the ordinary matter , known as `` quintessence '' .
the quintessential scalar field causes the accelerating universal expansion at present @xcite and it also has a possibility of making significant contribution to the total energy density at radiation dominated epoch @xcite . such an extra - energy density is interpreted as to increase relativistic degrees of freedom , @xmath0 , and thereby causes more rapid expansion rate at the same temperature .
similar effect is expected in the neutrino - degenerate universe models @xcite , and the universal lepton asymmetry was critically studied in the context of constraining the primordial nucleosynthesis and cosmic microwave background anisotropies . on the other hand ,
another attempt has been made quite recently to understand the einstein gravity in the light of the unified theory @xcite .
five dimensional brane world cosmology leads to an effective theory of gravity , which is motivated by the superstring theory or m - theory , which is a strong candidate for the unified theory . in this scenario
the `` standard '' friedmann equation should be slightly modified by adding several new terms which arise from the extra dimensions .
several authors @xcite showed that there is a new class of effective energy among them which diminishes in proportion to @xmath6 called `` dark radiation '' , where @xmath7 is the cosmic scale factor .
the interesting point is that this dark radiation term could be even negative because it originates purely from fifth dimensional geometry .
if this is the case , negative dark radiation causes the cosmic expansion slower .
we can therefore interpret this effect as a decrease in effective degrees of freedom , @xmath0 .
the purpose of the present letter is to study how strongly the change of the relativistic degrees of freedom @xmath0 affects the pbh formation during qcd phase transition .
the energy density of the universe gets larger for larger @xmath0 at the same temperature , but the duration of qcd phase transition becomes shorter .
these two effects operate in opposite direction to form pbhs . therefore
, it is not a trivial subject since many physical quantities may be affected in a complicated manner by this parameter @xmath0 . in order to clarify its effects we use a simple model for the numerical analysis instead of performing complicated hydrodynamic calculations .
we then try to find the parameter dependence of the typical mass scale and the amount of pbhs which are formed during the qcd phase transition .
we first briefly review important quantities involved in the pbh formation during the qcd phase transition .
horizon mass at the qcd epoch is written as @xmath8 where @xmath9 is the number of relativistic degrees of freedom , and t is the cosmic temperature .
this equation shows that the horizon mass at the qcd phase transition is compatible to typical mass scale of machos @xcite .
recent progress in lattice qcd calculations and accelerator experiments have provided with rich information of the qcd phase transition .
we employ the mit bag model to express the quark - gluon plasma phase @xcite . in this model
the energy densities of quark - gluon plasma ( @xmath10 ) and hadron gas ( @xmath11 ) and the pressures ( @xmath12 and @xmath13 ) respectively are given by @xmath14 @xmath15 where @xmath16 and @xmath17 are the numbers of degrees of freedom of the quark - gluon plasma and hadron gas phases , and b is the bag constant .
the condition of pressure balance between the two phases , @xmath18 , for a first order phase transition yields @xmath19 @xmath20 where @xmath21 is the latent heat of the phase transition .
this implies that the intensity of the first order qcd phase transition depends on the difference between the two degrees of freedom before ( @xmath16 ) and after ( @xmath17 ) the phase transition .
duration of the phase transition is expressed by the change of the cosmic scale factor during the phase transition . from the conservation law of entropy
we estimate the rate @xmath22 where @xmath23 and @xmath24 are the scale factors at the beginning and the end of the qcd phase transition , respectively .
we can learn from this that the duration of the phase transition is comparable to the hubble time at the qcd epoch , and it is enough for fluctuations to collapse gravitationally .
for the numerical analysis of the pbh formation we mainly follow the method developed by cardall and fuller @xcite .
the condition for the gravitational collapse of pbh is that the self gravity is stronger than the centrifugal pressure force , which is expressed as @xmath25 where @xmath26 is energy density , p is pressure , and @xmath27 is the size of an overdense region at the turn around .
this condition becomes @xmath28 where @xmath29 in the radiation dominated era and @xmath30 is the jeans length .
equation ( 8) is a well known condition for the gravitational collapse .
this condition is also expressed , using the density contrast @xmath31 at horizon crossing @xcite , as @xmath32 this means that fluctuations having density contrast above @xmath33 at the horizon crossing can collapse to form pbh at radiation dominated era .
we need modification of eq.(9 ) , taking account of the effects of the qcd phase transition . in order to proceed quantitative discussions
, we introduce the quantity @xmath34 and modify the above condition by @xmath35 @xmath36 where @xmath37 and @xmath38 are the sizes of initially expanding fluctuation at the beginning and the end of the phase transition , and @xmath39 is the size at turn around @xcite .
this quantity @xmath34 works for softening the equation of state @xmath40 and takes the value , @xmath41 .
when @xmath34 is the minimum value , @xmath42 , the fluctuation is not affected by the qcd phase transition at all , and eq.(10 ) turns out to be the simplest case of eq.(9 ) .
when @xmath34 is the maximum value , @xmath43 , the fluctuation has been affected the most by the phase transition because the fluctuation starts collapsing early enough so that the overdense region includes the mixed phase of quark - gluon plasma and hadron gas . in the latter case
the system does not feel any effective pressure response . @xmath37 and @xmath38 are calculated from the energy density of the background and the energy contrast of the fluctuation when the fluctuation crosses the particle horizon .
we solve the flat friedmann equation for the background @xmath44 and the closed friedmann equation for the fluctuation @xmath45 in these equations , we assume that the fluctuation has the top hat profile @xmath46 , where @xmath47 means the density contrast at horizon crossing .
we normalize the density evolution and fix the gauge by setting @xmath48 , @xmath49 , and @xmath50 when the fluctuation crosses the horizon .
we show the quantity @xmath51 as the function of @xmath47 in fig.1 , from which we can show the effect of qcd phase transition on the pbh formation . here ,
@xmath52 , @xmath53 is the mean energy density when the fluctuation crosses the horizon , and @xmath54 is the mean energy density above which the universe is in pure quark - gluon plasma phase .
this variable @xmath55 indicates when the fluctuation under consideration crosses the particle horizon .
modification of collapsing condition for @xmath56 .
@xmath57 corresponds to a fluctuation with mass scale @xmath58 @xcite .
@xmath51 , for @xmath59 , and @xmath47 are shown as indicated .
collapse region @xmath60satisfies the collapsing condition of eq.(10).,scaledwidth=100.0% ] in order to connect the physics of the qcd phase transition with the relativistic number of degrees of freedom , we introduce effective background degrees of freedom,@xmath61 , where @xmath62 ( @xmath63 , @xmath64 , @xmath65 , three kinds of neutrinos in the standard case ) , and @xmath66 is for the extra - component at the qcd phase transition epoch as to be discussed in the next section .
the numbers of degrees of freedom of the hadron phase and quark - gluon phase are respectively given by , @xmath67 ( pions ) , and @xmath68 ( 21 for quarks and 16 for gluons ) .
we thus obtain the relation between the parameter @xmath55 and the number of effective degrees of freedom @xmath0 for horizon mass scale @xmath69 , @xmath70 we consider the significance of the pbhs formed in the qcd phase transition in this sub - section .
present cosmological density parameter of pbhs formed in the qcd phase transition is represented by @xmath71 where @xmath72 is the hubble parameter in units of 100 km / s / mpc , @xmath73 is the relativistic degrees of freedom , and @xmath74 is the fraction of the radiation energy density which is converted into black holes at temperature t. @xmath74 is defined by @xmath75 where @xmath76 is the probability function of a horizon volume fluctuation to have overdensity parameter @xmath47 @xcite @xmath77 @xmath78 in eq .
( [ sigma ] ) @xmath79 is the spectral index of primordial density fluctuations .
we consider the fluctuation which has overdensity larger than @xmath80 [ see eq . ( [ epsilon ] ) ] and can collapse into black holes , so that it satisfies eq.(10 ) for a given @xmath55 ( see fig.1 ) .
it is to be noted that even a small fraction @xmath81 can result in the significant amount of cosmological energy density at present . from eq.(15 ) we can estimate the mass of the pbh and the variance @xmath82 for a given @xmath55 and n. we can also calculate the efficiency @xmath81 and @xmath83 using fig.1 and eqs.(15)-(19 ) .
if the universe is modeled to have a new class of extra - energy component , it leads to increase relativistic degrees of freedom , @xmath73 . in the neutrino - degenerate universe models
@xcite , additional degrees of freedom are written as @xmath84,\ ] ] where @xmath85 s correspond to @xmath86-,@xmath87- and @xmath88-neutrino species , and @xmath89 is the neutrino chemical potential divided by the neutrino temperature , @xmath90 .
we note that slightly non - zero chemical potential is favored in recent bbn analysis with better goodness of fit to the recent data of cmb anisotropies @xcite , which constrains @xmath91 .
another class of extra - energy component appears in quintessential inflationary model @xcite @xmath92 where @xmath93 is the background photon energy density , @xmath94 , and @xmath95 is the energy density of quintessence field,@xmath96 . @xmath97 and @xmath98 are the quintessence scalar field value and its time derivative at the qcd phase transition .
@xmath99 is constrained to be @xmath100 so as to satisfy the bbn and cmb analyses @xcite . in a recent theory of brane world cosmology @xcite , completely a new class of cosmic energy , called dark radiation ,
emerges from extended friedman equation .
its extra - component is expressed by @xmath101 where @xmath102 is the dark - radiation term , @xmath103 , and @xmath87 is a constant which comes from the electric coulomb part of the five - dimensional weyl tensor @xcite .
we note that the dark radiation can be negative depending on the sign of @xmath87 .
this is in remarkable contrast to eqs.(20 ) and ( 21 ) which are always positive .
the bbn and cmb constraints @xcite on @xmath99 in the brane world cosmology is @xmath104 ( at the @xmath105 c.l . ) .
note that only the bbn constraint allows @xmath106 .
we show the calculated results for @xmath107 and @xmath108 in the figs.2 and 3 .
@xmath109 corresponds to @xmath110 just after the qcd phase transition , which is the lowest value deduced from only the bbn constraint in brane world cosmology @xcite .
@xmath111 is for the standard case , and @xmath112 corresponds to @xmath113 , which is a likely case in the quintessential inflationary scenario @xcite . refer to ref .
@xcite for more critical discussions of the observational constraints on @xmath114 in quintessence scalar fields .
the key results are summarized as follows .
fine tuned gaussian blue spectra fluctuation is needed with spectral index @xmath115 , @xmath116 , @xmath117 in the cases of @xmath109 , @xmath118 , and @xmath108 respectively , in order to have significant pbhs energy density @xmath83 at present , i.e. @xmath119 .
typical mass scales of pbh are @xmath120 , @xmath121 , and @xmath122 for @xmath123 and @xmath108 , respectively , where three mass scales refer to maximum @xmath124 s .
( eq.(16 ) and fig.2 ) .
the spectral indices as shown in fig.3 are compatible with those inferred from the data of cobe observations , but larger than the best fit analysis of the combined different cosmological data sets @xcite .
these indices are slightly larger than the values constrained from the evaporation of pbhs by hawking radiation in much smaller mass scale @xcite . increasing the effective number of degrees of freedom @xmath0
causes the shorter duration of the qcd phase transition because the universal expansion obeys @xmath125 instead of eq.(6 ) .
accordingly , the duration of the mixed phase also becomes shorter , making less efficient pbh formation ( fig.3 ) .
the typical mass scale of the pbhs , however , increases with increasing @xmath0 ( fig.2 ) .
this is because the fluctuations that are most effectively influenced by the qcd phase transition enter the horizon at later time with increasing number of degrees of freedom .
we studied the formation mechanism of pbhs during the qcd phase transition , being motivated by the cosmological interest in machos as a dark matter and also new classes of cosmic extra - energy components for the degenerate neutrinos , dark energy , and dark radiation . in the present paper
we take account of the effects of these extra - components by means of parameterizing their effect as an increase or decrease in effective relativistic degrees of freedom .
we conclude that the pbh formation during qcd phase transition becomes more efficient with an negative extra - component because of the increase of the duration of the qcd phase transition .
we also found that in such a case the typical mass scale of pbh becomes smaller .
these results suggest larger probability of finding more pbhs that have typically sub - solar mass @xmath126 if the dark radiation exists as allowed in the brane world cosmology .
|
it has recently been suggested that the formation of horizon - size primordial black hole ( pbh ) from pre - existing density fluctuations is effective during the cosmic qcd phase transition . in this letter
we discuss the dependence of pbh formation on effective relativistic degrees of freedom , @xmath0 during the cosmic qcd phase transition .
our finding is important in the light of recent cosmological arguments of several new classes of cosmic energy that appear from universal neutrino degeneracy , quintessential inflation , and dark radiation in brane world cosmology .
extra - energy component from the standard value in these new cosmological theories is represented as an effective radiation in terms of @xmath0 .
we conclude that the pbh formation during qcd phase transition becomes more efficient if negative extra - component of the cosmic energy is allowed because of the increase of the duration of the qcd phase transition , which leads to smaller mass scale of pbhs .
this suggests larger probability of finding more pbhs if the dark radiation exists as allowed in the brane world cosmology .
cosmology , quark - gluon plasma , black holes 98.80 , 12.38.m , 97.60.l
|
leptogenesis @xcite is an extremely successful theory in explaining the baryon asymmetry of the universe by adding three heavy right - handed neutrinos @xmath2 to the standard model , @xmath3 with masses @xmath4 at the scale of grand unified theories ( guts ) and yukawa couplings @xmath5 similar to the other fermions .
this also solves the problem of the light neutrino masses via the see - saw mechanism without fine - tuning @xcite . the heavy neutrinos decay into lepton and higgs boson after inflation , the decay is out of equilibrium since there are no gauge couplings to the standard model .
if the cp asymmetry in the yukawa couplings is large enough , a lepton asymmetry is created by the decays which is then partially converted into a baryon asymmetry by sphaleron processes . as temperatures are high , interaction rates and the cp asymmetry need to be calculated using thermal field theory @xcite rather than vacuum quantum field theory .
however , in the conventional approach @xcite , thermal masses have been put in by hand without investigating the validity of this approach in detail .
we have adressed this issue in @xcite and found that corrections arise through the occurence of two lepton dispersion relations in the thermal bath . in this paper
, we calculate the decay rate of the heavy neutrino in a consistent way ( chapter [ decay ] ) which automatically includes the effect of leptonic quasiparticles , compare it to the conventional approach ( chapter [ leptogenesis ] ) and give an outlook of what needs to be done to arrive at consistent descriptions of leptogenesis .
our calculation is general enough to be applied to all decays of a yukawa fermion at finite temperature , which has interesting implications for other early universe dynamics ( chapter [ conclusions ] ) .
if a particle reaction like scattering or decay takes place in the background of a heat bath , e.g. in the hot state of the early universe , thermal field theory has to be employed to describe this process .
there are two different approaches for considering finite temperatures within quantum field theory , the imaginary and real time formalism @xcite , both yielding the same results . in this work
, we will use the imaginary time formalism .
going from zero to finite temperature , ensemble - weighted expectation values of operators have to be used rather than vacuum expectation values .
for an operator @xmath6 , this reads @xmath7 where @xmath8 is the density operator describing the ensemble . in this way
it can be shown that the propagator at finite temperature @xmath9 is given by its usual vacuum expression where the zero component of the momentum is replaced by imaginary discrete matsubara frequencies @xmath10 in the case of bosons or @xmath11 in the case of fermions with integers @xmath12 ( see e.g. @xcite ) .
perturbation theory at finite temperature then follows from using these propagators and summing over the matsubara frequencies in loop diagrams .
however , using these bare thermal propagators can lead to inconsistent results , which are not complete to leading order , infrared divergent , and gauge dependent in the case of gauge theories .
a famous example is the damping rate of a plasma wave in the quark gluon plasma , which is different in different gauges . in order to cure this ,
the hard thermal loop ( htl ) resummation has been invented @xcite .
for this purpose one has to distinguish between hard momenta of the order @xmath9 or larger and soft momenta of the order @xmath13 or smaller , where @xmath14 is the coupling constant , which is strictly possible only in the weak coupling limit @xmath15 .
after all , the htl improved perturbation theory has been successfully applied to thermal qcd for the description of the quark - gluon plasma ( see e.g. @xcite ) .
the basic idea is that the bare propagators are replaced by resummed propagators , if the external momentum is soft @xmath16 . for a scalar field ( fig .
[ resummed ] ) , this resummation follows from the dyson - schwinger equation as resummed propagator ] @xmath17 the thermal self - energy @xmath18 of the scalar field then acts as a thermal mass @xmath19 and gives a correction to the zero - temperature mass @xmath20 . since @xmath18 is of the order @xmath21 , the resummation will only affect the propagator when @xmath22 , which is reflected in the prescription to resum only soft momenta .
the resummed fermion propagator has a more complicated structure and will be explained in the next section . in general , the self - energy is momentum dependent , e.g. the photon self - energy in qed . in this case , the leading order gauge independent self - energy follows from integrating only over hard momenta in the loop diagram defining the self - energy .
this htl contribution in the resummed propagator leads to a correction of the order @xmath13 which can not be neglected if the momentum of the propagator is soft .
the poles of the htl resummed propagators then describe the dispersion relations in the medium , e.g. plasma waves following from the resummed photon propagator .
in addition to propagators also htl effective vertices related to the propagators by ward identities might have to be used .
in the neutrino decay we want to calculate , the higgs boson and the lepton acquire thermal masses of the order @xmath23 via their interactions with other standard model particles . in the regime where the temperature is of the order of the neutrino mass @xmath24 , one of the momenta of the decay products can be soft and
has to be resummed . in the regime
where @xmath25 , both higgs boson and lepton momentum will be soft and need to be resummed .
we are interested in both regimes , therefore we will resum both higgs boson and lepton propagator .
the case of resumming only one propagator is included in this approach , since resumming a hard propagator gives only a negligible correction to the bare propagator .
the htl resummation has been invented for the weak coupling limit @xmath15 .
this limit does not apply in our case , our more phenomenological approach is rather motivated by the desire to capture effects beyond perturbation theory and justified a posteriori by the sizeable corrections it reveals , similar to the treatment of meson correlation functions in @xcite .
we consider a leptogenesis - inspired model with a massive majorana fermion _
n _ coupling to a massless dirac fermion @xmath26 and a massless scalar @xmath27 .
the interaction and mass part of the lagrangian then reads @xmath28 the htl resummation technique has been considered in @xcite for the case of a dirac fermion with yukawa coupling , from which the htl resummed propagators for the lagrangian in eq .
follow directly .
we like to calculate the interaction rate @xmath29 of @xmath30 .
we cut the @xmath31 self energy and use the htl resummation for the fermion and scalar propagators ( fig .
[ cut ] ) .
@xmath31 decay via the optical theorem with dressed propagators denoted by a blob ] according to finite - temperature cutting rules @xcite , the interaction rate reads @xmath32.\ ] ] at finite temperature , the self - energy reads @xmath33 where @xmath34 and @xmath35 are the projection operators on left- and right - handed states , @xmath36 and we have summed over majorana and dirac spins . the htl - resummed scalar propagator is @xmath37 where @xmath38 is the thermal mass of the scalar , created by the interaction with fermions . due to the reduced majorana degrees of freedom , @xmath39 differs from the dirac - dirac case by a factor 1/2 @xcite .
the effective fermion propagator in the helicity - eigenstate representation is given by @xcite @xmath40 where @xmath41^{-1}\ ] ] and @xmath42 this again differs from the dirac case by a factor 1/2 @xcite .
the trace can be evaluated as @xmath43= \delta_+ ( p_0-p \eta)+\delta_- ( p_0 + p \eta),\ ] ] where @xmath44 is the angle between * p * and * k*. we evaluate the sum over matsubara frequencies by using the saclay method @xcite . for the scalar propagator , the saclay representation reads @xmath45
e^{-\omega_q \tau } + n_b(\omega_q ) e^{\omega_q \tau}\},\ ] ] where @xmath46 , @xmath47 is the bose - einstein distribution and @xmath48 . for the fermion propagator
it is convenient to use the spectral representation @xcite @xmath49 e^{-\omega \tau'},\ ] ] where @xmath50 is the fermi - dirac distribution and @xmath51 the spectral density @xcite .
the fermion propagator in eq
. has two different poles for @xmath52 , which correspond to two leptonic quasiparticles with a positive ( @xmath53 ) or negative ( @xmath54 ) ratio of helicity over chirality @xcite . the spectral density @xmath51 has two contributions , one from the poles and one discontinuous part .
since the quasi - particles are our final states , we will set @xmath55 such that @xmath56 .
thus we are only interested in the pole contribution @xmath57 where @xmath58 are the dispersion relations for the two quasiparticles , i.e. the solutions for @xmath59 such that @xmath60 , shown in fig .
[ modes ] .
there exists an analytical solution for @xmath58 making use of the lambert w function which has not yet been reported in the literature .
the analytical solution is explained in detail in the appendix .
one can assign a momentum - dependent thermal mass to the two modes as shown in fig .
[ masses ] and for very large momenta the heavy mode @xmath61 approaches @xmath62 , while the light mode becomes massless .
the two leptonic dispersion relations compared with the standard dispersion relation @xmath63 in blue are shown . ]
the momentum - dependent quasiparticle masses @xmath64 are shown . ] in order to execute the sum over matsubara frequencies , we write @xmath65 with @xmath66 and remember that when evaluating frequency sums , also @xmath67 can be written as a matsubara frequency and later on be continued analytically to real values of @xmath68 @xcite .
in particular @xmath69 .
we can write @xmath70 then @xmath71 since @xmath72 .
after evaluating the sum over @xmath59 and carrying out the integrations over @xmath73 and @xmath74 , we get @xmath75 .
\end{split}\ ] ] integrating @xmath76 over the pole part of @xmath51 in eq . , we get @xmath77 \right . \\ & \label{frequencysum } \left
. + \frac{\omega_\mp^2-k^2}{2 m_\ell^2 } \left [ \frac{n_b+n_f}{p_0+\omega_\mp-\omega_q}+ \frac{1+n_b - n_f}{p_0+\omega_\mp+\omega_q } \right ] \right \ } \end{split}\ ] ] where @xmath78 and @xmath79 or @xmath80 , respectively .
the four terms in eq .
correspond to the processes with the energy relations indicated in the denominator , i.e. the decay @xmath81 , the production @xmath82 , the production @xmath83 and the production of @xmath84 from the vacuum , as well as the four inverse reactions @xcite .
we are only interested in the process @xmath85 , where the decay and inverse decay are illustrated by the statistical factors @xmath86 our term thus reads @xmath87 for carrying out the integration over the angle @xmath88 , we use @xmath89 where @xmath90\ ] ] denotes the angle for which the energy conservation @xmath91 holds .
the integration over @xmath88 then yields @xmath92,\ ] ] where @xmath93 .
it follows that @xmath94 \\ = & \frac{1}{2 p_0 } \ ; { \rm i m } \left \ { g^2 t \sum_{k_0 } \int \frac{{\rm d^3 } k}{(2 \pi)^3 } \ ; { \rm tr } [ ( \slashed{p } + m ) p_l s^ * p_r ] d^ * \right \ } \\ = & \frac{g^2}{8 \pi^2 p_0 } \ ; { \rm i m } \left \ { t \sum_{k_0 } \int { \rm d } k \ , { \rm d } \eta \ : k^2 d^ * [ \delta_+ ( p_0-p \eta ) + \delta_- ( p_0 + p \eta ) ] \right \ } \\ = & \frac{g^2}{32 \pi p_0 p } \sum_\pm \int_{-1 \leq \eta_\pm \leq 1 } { \rm d } k \ : \frac{\omega_\pm^2-k^2}{2 m_\ell^2 } [ 1+n_b(\omega_{q \pm } ) -n_f(\omega_\pm ) ] \\ & \hspace{3.4 cm } \times [ 2 p_0 ( k \mp \omega_\pm ) \pm m^2 \pm ( \omega_\pm^2-k^2 ) \mp m_\phi^2 ] , \end{split}\ ] ] where we only integrate over regions with @xmath95 .
using finite temperature cutting rules , one can also write the interaction rates for the two modes in a way which resembles the zero - temperature case @xcite @xmath96 , \end{split}\ ] ] where @xmath97 and @xmath98 analogously and the matrix elements are @xmath99 now that we have arrived at an expression for the full htl decay rate of a yukawa fermion , we would like to compare it to the conventional approximation adopted by @xcite . to this end , we do the same calculation for an approximated fermion propagator @xmath100 this yields the following approximated interaction rate : @xmath101 [ m^2 + m_\ell^2 -m_\phi^2 ] \\ = & \frac{1}{2 p_0}\int \ , { \rm d } \tilde{k } \ : { \rm d } \tilde{q } \ : ( 2 \pi)^4 \delta^4(p - k - q ) |\mathcal{m}|^2 \\ & \hspace{1.6 cm } \times [ 1+n_b(\omega_q)-n_f(\omega ) ] , \end{split}\ ] ] where @xmath63 , @xmath102 and the integration boundaries @xmath103\ ] ] ensure @xmath95 , where @xmath104.\ ] ] we see that the matrix element is @xmath105 this result resembles the zero temperature result @xmath106\ ] ] with zero temperature masses @xmath107 , @xmath39 . the missing factor @xmath108 accounts for the statistical distribution of the initial or final particles .
as pointed out in more detail in @xcite , we have shown that the approach to treat thermal masses like zero temperature masses in the final state @xcite is justified since it equals the htl treatment with an approximate fermion propagator .
however this approach does not equal the full htl result . concluding this calculation , a caveat has to be added : the external majorana fermion will also acquire a thermal mass of order @xmath13 .
thus , if its zero temperature mass is smaller than that , the external fermion also needs to be described by leptonic quasiparticles to be consistent .
however , in our leptogenesis application , the yukawa coupling giving rise to the majorana neutrino decay is much smaller than the couplings giving rise to the thermal masses of the higgs boson ( scalar ) and the lepton ( dirac fermion ) and thus the thermal mass of the heavy neutrino can be neglected .
we have calculated the decay rate assuming a majorana particle , but the result can be very easily generalized to the case of two dirac fermions by inserting the appropriate factors of two in the decay rate and the thermal masses .
when turning to leptogenesis with @xmath109 we sum over the two components of the doublets , particles and antiparticles and the three lepton flavors .
thus we need to replace @xmath110 .
integrating over all neutrino momenta , the decay density in equilibrium is @xmath111 where @xmath112 , @xmath113^{-1}$ ] is the equilibrium distribution of the neutrinos and @xmath114 \ , \gamma$ ] . since @xmath115 , the thermal masses are @xmath116 and @xmath117 the couplings denote the su(2 ) coupling @xmath118 , the u(1 ) coupling @xmath119 , the top yukawa coupling @xmath120 and the higgs self coupling @xmath121 , where we assume a higgs mass of @xmath122 gev .
the other yukawa couplings can be neglected since they are much smaller than unity and the remaining couplings are renormalized at the first matsubara mode @xmath123 as explained in @xcite . in fig .
[ comp ] , we compare our consistent htl calculation to the approximation adopted by @xcite , while we add quantum statistical distribution functions to their calculation which then equals the approach of using an approximated lepton propagator @xmath124 as in eq . @xcite .
we assume the heavy neutrino masses to be hierarchical and evaluate the decay rate for the typical value @xmath125 gev , which is inspired by putting @xmath126 to the gut scale ( @xmath127 gev ) and assuming @xmath128 analogous to the quark sector .
the combination of yukawa couplings @xmath129 which governs the decay rate is often parametrized by the so called effective neutrino mass @xmath130 , where is the vacuum expectation value of the higgs field .
we take , inspired by the mass scale of the atmospheric mass splitting .
however , our results can be generalized to all regions of parameter space .
the neutrino decay density with the one lepton mode approach @xmath131 and the two - mode treatment @xmath132 for @xmath133 gev and @xmath134 ev . the thresholds for the two modes ( + ) , ( - ) and one mode ( 0 ) are indicated . ] in the one - mode approach , the decay is forbidden when the thermal masses of higgs boson and lepton become larger than the neutrino mass @xmath135 .
considering two modes , the kinematics exhibit a more interesting behavior .
for the positive mode , the phase space is reduced due to the larger quasi - mass and at @xmath136 , the decay is only possible into leptons with small momenta , thus the rate drops dramatically .
the decay into the negative , quasi - massless mode is suppressed since its residue is much smaller than the one of the positive mode . however , the decay is possible up to @xmath137 . due to the various effects , the two mode rate differs from the one mode approach by more than one order of magnitude in the interesting temperature regime of @xmath138 .
it is extremely tempting to put this result in a boltzmann - solver and obtain an effect for the produced baryon asymmetry .
however , in the quest for consistent treatments which capture effects of the same origin and size , other effects need to be included as well . at higher temperatures , when @xmath139 , the higgs can decay into neutrino and lepton modes and this process acts as production mechanism for neutrinos @xcite .
moreover , the cp asymmetry needs to be calculated taking into account the two lepton modes in order to have a consistent treatment .
as discussed in detail in @xcite , we have , by employing htl resummation and finite temperature cutting rules , confirmed that treating thermal masses as kinematic masses as in @xcite is a reasonable approximation .
however , quantum statistical functions need to be included as they always appear in thermal field theory .
moreover , the full htl lepton propagator shows a non - trivial two - mode behavior which is not accounted for by the conventional approach .
we have calculated the effect of the two modes in a general way which is applicable to any decay and inverse decay rates involving fermions at high temperature .
thus , this calculation is a valuable tool for other particle processes in the early universe , as other leptogenesis processes , the thermal production of gravitinos or the like .
the behavior of the decay density of the two lepton modes can be explained by considering the dispersion relations @xmath58 of the modes and assigning momentum - dependent quasi - masses to them .
the thresholds for neutrino decay reported in @xcite are shifted and the decay density shows deviations of more than an order of magnitude in the interesting temperature regime @xmath140 .
thus we expect these effects to have a sizeable impact on the final baryon asymmetry
. however , in order to arrive at a minimal consistent treatment , also the decay @xmath141 at high temperatures needs to be included as well as a cp asymmetry that is corrected for lepton modes . in a further step , it will be interesting to include the effect of thermal widths in the calculations .
as for all effects arising from thermal field theory , the effects are only important in the weak washout regime , where leptogenesis takes place at high temperatures .
we are aware of the progress that is currently being made in approaching the effects of quantum statistics @xcite , quantum transport equations @xcite or other collective phenomena as e.g. the landau - pomeranchuk effect @xcite .
these efforts contribute to getting an idea of the size and impact of various thermal effects by approaching the extremely complex situation from different angles .
+ * acknowledgements * we would like to thank georg raffelt , florian hahn - wrnle , steve blanchet , matthias garny , marco drewes , wilfried buchmller and annika whner for fruitful and inspiring discussions .
the dispersion relations of the two lepton modes are given by the poles of the corresponding propagator .
hence , we seek the zeros of @xmath142^{-1}\ ] ] the equations @xmath143 can be transformed by the substitutions @xmath144 this yields @xmath145 further introducing @xmath146 leads to @xmath147.\ ] ] since the prefactor does not have poles for the values of @xmath55 we are looking at , solving @xmath143 amounts to solving @xmath148 which in turn means @xmath149
this is the defining equation of the lambert w function @xcite , thus the solution reads @xmath150 according to the definition in eq .
@xmath151 thus the two real branches of the lambert function , @xmath152 and @xmath153 , correspond to the two solutions we seek . in the range given by eq .
@xmath154 and @xmath155 .
for @xmath156 we have @xmath157 and @xmath158 .
hence , the physical solutions for @xmath159 read @xmath160 a. d. sakharov , `` violation of cp invariance , c asymmetry , and baryon asymmetry of the universe , '' pisma zh .
( 1967 ) 32 . m. fukugita and t. yanagida , `` baryogenesis without grand unification , '' phys .
b * 174 * ( 1986 ) 45 .
p. minkowski , `` mu @xmath163 e gamma at a rate of one out of 1-billion muon decays ?
, '' phys .
b * 67 * ( 1977 ) 421 .
t. yanagida , `` horizontal gauge symmetry and masses of neutrinos , '' proceedings of the workshop on unified thoeries and baryon number in the universe , tsukuba , japan 1979 ( ed.s .
o. sawada and a. sugamoto , kek report no .
79 - 18 , tsukuba ) .
m. gell - mann , p. ramond and r. slansky , `` complex spinors and unified theories , '' proceedings of the supergravity stony brook workshop , new york , 1979 , ed.s .
p. van nieuwenhuizen and d. freedman ( north - holland , amsterdam ) .
l. covi , n. rius , e. roulet and f. vissani , `` finite temperature effects on cp violating asymmetries , '' phys .
d * 57 * ( 1998 ) 93 [ arxiv : hep - ph/9704366 ] .
g. f. giudice , a. notari , m. raidal , a. riotto and a. strumia , `` towards a complete theory of thermal leptogenesis in the sm and mssm , '' nucl .
b * 685 * ( 2004 ) 89 [ arxiv : hep - ph/0310123 ] .
d. besak and d. bodeker , `` hard thermal loops for soft and collinear external momenta , '' arxiv:1002.0022 [ hep - ph ] .
c. p. kieig and m. plmacher , `` thermal masses in leptogenesis , '' [ arxiv : hep - ph/0910.4872 ] .
m. le bellac , _ thermal field theory _ , cambridge university press , cambridge , uk , 1996 .
e. braaten and r. d. pisarski , `` soft amplitudes in hot gauge theories : a general analysis , '' nucl .
b * 337 * ( 1990 ) 569 .
e. braaten and r. d. pisarski , `` deducing hard thermal loops from ward identities , '' nucl .
b * 339 * ( 1990 ) 310 .
m. h. thoma , `` applications of high - temperature field theory to heavy - ion collisions , '' in : _ quark - gluon plasma 2 _ , ed .
hwa ( world scientific , singapore 1995 ) , p.51134 [ arxiv : hep - ph/9503400 ] .
f. karsch , m. g. mustafa and m. h. thoma , `` finite temperature meson correlation functions in htl approximation , '' phys .
b * 497 * ( 2001 ) 249 [ arxiv : hep - ph/0007093 ] .
m. h. thoma , `` damping of a yukawa fermion at finite temperature , '' z. phys .
c * 66 * ( 1995 ) 491 [ arxiv : hep - ph/9406242 ] .
h. a. weldon , `` simple rules for discontinuities in finite temperature field theory , '' phys . rev .
d * 28 * ( 1983 ) 2007 .
r. l. kobes and g. w. semenoff , `` discontinuities of green functions in field theory at finite temperature and density . 2 , '' nucl .
b * 272 * ( 1986 ) 329 .
e. braaten , r. d. pisarski and t. c. yuan , `` production of soft dileptons in the quark - gluon plasma , '' phys
. rev .
* 64 * ( 1990 ) 2242 .
j. i. kapusta , p. lichard and d. seibert , `` high - energy photons from quark - gluon plasma versus hot hadronic gas , '' phys .
d * 44 * ( 1991 ) 2774 [ erratum - ibid .
d * 47 * ( 1993 ) 4171 ] .
e. braaten and r. d. pisarski , `` calculation of the quark damping rate in hot qcd , '' phys .
d * 46 * ( 1992 ) 1829 .
r. d. pisarski , `` computing finite temperature loops with ease , '' nucl .
b * 309 * ( 1988 ) 476 .
r. d. pisarski , `` renormalized gauge propagator in hot gauge theories , '' physica a * 158 * ( 1989 ) 146 .
v. v. klimov , `` spectrum of elementary fermi excitations in quark gluon plasma .
( in russian ) , '' sov .
j. nucl .
phys . *
33 * ( 1981 ) 934 [ yad
. fiz .
* 33 * ( 1981 ) 1734 ] .
h. a. weldon , `` covariant calculations at finite temperature : the relativistic plasma , '' phys .
d * 26 * ( 1982 ) 1394 .
h. a. weldon , `` effective fermion masses of order gt in high temperature gauge theories with exact chiral invariance , '' phys .
d * 26 * ( 1982 ) 2789 .
h. a. weldon , `` dynamical holes in the quark - gluon plasma , '' phys .
d * 40 * ( 1989 ) 2410 .
g. baym and n. d. mermin , `` determination of thermodynamic green s functions , '' j. math .
* 2 * ( 1961 ) , 232 l. dolan and r. jackiw , `` symmetry behavior at finite temperature , '' phys .
d * 9 * ( 1974 ) 3320 .
a. basboll and s. hannestad , `` decay of heavy majorana neutrinos using the full boltzmann equation including its implications for leptogenesis , '' jcap * 0701 * ( 2007 ) 003 [ arxiv : hep - ph/0609025 ] .
j. garayoa , s. pastor , t. pinto , n. rius and o. vives , `` on the full boltzmann equations for leptogenesis , '' jcap * 0909 * ( 2009 ) 035 [ arxiv:0905.4834 [ hep - ph ] ] .
f. hahn - woernle , m. plumacher and y. y. y. wong , `` full boltzmann equations for leptogenesis including scattering , '' jcap * 0908 * ( 2009 ) 028 [ arxiv:0907.0205 [ hep - ph ] ] .
f. hahn - woernle , `` wash - out in @xmath164-dominated leptogenesis , '' arxiv:0912.1787 [ hep - ph ] .
a. anisimov , w. buchmller , m. drewes and s. mendizabal , `` nonequilibrium dynamics of scalar fields in a thermal bath , '' annals phys .
* 324 * ( 2009 ) 1234 [ arxiv : hep - th/0812.1934 ] .
m. garny , a. hohenegger , a. kartavtsev and m. lindner , `` systematic approach to leptogenesis in nonequilibrium qft : vertex contribution to the cp - violating parameter , '' phys . rev .
d * 80 * ( 2009 ) 125027 [ arxiv : hep - ph/0909.1559 ] .
m. garny , a. hohenegger , a. kartavtsev and m. lindner , `` systematic approach to leptogenesis in nonequilibrium qft : self - energy contribution to the cp - violating parameter , '' [ arxiv : hep - ph/0911.4122 ] .
a. anisimov , w. buchmuller , m. drewes and s. mendizabal , `` quantum interference in a thermal bath , '' [ arxiv : hep - ph/1001.3856 ] .
m. garny , a. hohenegger and a. kartavtsev , `` medium corrections to the cp - violating parameter in leptogenesis , '' [ arxiv : hep - ph/1002.0331 ] .
m. beneke , b. garbrecht , m. herranen and p. schwaller , `` finite number density corrections to leptogenesis , '' arxiv:1002.1326 [ hep - ph ] .
r. m. corless , g. h. gonnet , d. e. g. hare , d. j. jeffrey and d. e. knuth , `` on the lambert w function , '' adv .
comput . math . * 5 * ( 1996 ) 329 . f. chapeau - blondeau and a. monir , `` numerical evaluation of the lambert w function and application to generation of generalized gaussian noise with exponent 1/2 , '' ieee trans .
signal processing , vol .
50 ( 2002 ) 2160
|
we calculate the decay rate of a yukawa fermion in a thermal bath using finite temperature cutting rules and effective green s functions according to the hard thermal loop resummation technique .
we apply this result to the decay of a heavy majorana neutrino in leptogenesis .
compared to the usual approach where thermal masses are inserted into the kinematics of final states , we find that deviations arise through two different leptonic dispersion relations .
the decay rate differs from the usual approach by more than one order of magnitude in the temperature range which is interesting for the weak washout regime .
we discuss how to arrive at consistent finite temperature treatments of leptogenesis .
mpp-2010 - 31 + * decay of a yukawa fermion at finite temperature and applications to leptogenesis * + clemens p. kieig@xmath0 , michael plmacher@xmath0 , markus h. thoma@xmath1 , @xmath0 _ max - planck - institut fr physik ( werner - heisenberg - institut ) , + fhringer ring 6 , d-80805 mnchen , germany + @xmath1 _ max - planck - institut fr extraterrestrische physik , + giessenbachstrae , d-85748 garching , germany _ _
|
several mechanisms have been suggested so far for the formation and evolution of elliptical galaxies .
one scenario is based on an early monolithic collapse of a gas cloud or early merging of lumps of gas where dissipation plays a fundamental role ( larson 1974 ; arimoto & yoshii 1987 ; matteucci & tornamb 1987 ) . in this scenario
the star formation stops soon after a galactic wind develops and the galaxy evolves passively since then .
bursts of star formation in merging subsystems made of gas had been also suggested ( tinsley & larson 1979 ) ; in this picture star formation stops after the last burst and gas is lost via stripping or wind .
the alternative and more recent scenario is the so - called hierarchical clustering scenario , where merging of early formed stellar systems in a wide redshift range and preferentially at late epochs , is expected ( kauffmann et al .
the main difference between the monolithic collapse and the hierarchical merging relies in the time of galaxy formation , occurring quite early in the former scenario ( at reshifts @xmath0 ) and continuously in the latter scenario .
there are arguments either in favour of the monolithic or the hierarchical scenario , but the former one gives a more likely picture since it can reproduce the majority of the properties of stellar populations in ellipticals , in particular some fundamental facts such as that the ellipticals are dominated by old stars ( k - giants ) and that the [ @xmath1/fe]@xmath2 in the dominant stellar population ( worthey et al .
1992 ; weiss et al . 1995 ; kuntschner et al .
this high [ @xmath1/fe ] ratio is the clear signature of the pollution from massive stars .
the same occurs in the most metal poor stars in our galaxy and is due to the fact that sne ii are the main producers of @xmath1-elements ( o , ne , mg , si . s and ca ) whereas sne ia , which explode with a delay relative to type ii sne , are thought to be responsible for the production of fe .
therefore , the high [ @xmath1/fe ] ratio in ellipticals argues strongly in favor of a short period of star formation during which type ia sne did not have time to substantially pollute the insterstellar medium ( ism ) . in this paper we will discuss a monolithic model ( pipino et al .
2002 ) for the formation and evolution of ellipticals , where these objects suffer a short ( @xmath3 1gyr ) but intense star formation period halted by a sn driven galactic wind . after the onset of the wind , which devoids the galaxy of all the gas present
, star formation is assumed to stop .
this is because the galaxy , after the wind , contains hot and rarified gas , a situation which is unfavorable to star formation .
the time for the occurrence of the wind , @xmath4 , is therefore crucial in determining the evolution of the galaxy and the intracluster / intergalactic medium .
therefore , the assumptions about the energy transferred from sne into the ism are very important . unfortunately , very little is known about this feed - back and one has to choose the assumptions which produce a realistic model for ellipticals . in section 2
we will discuss the condition for the occurrence of a galactic wind and the chemical evolution model . in section 3 we will present the results and draw a few conclusions .
the model we are adopting here is described in detail in pipino et al .
the main feature of the model is that it assumes a strong star formation rate for ellipticals ( roughly 20 times stronger than adopted in the solar vicinity ) and takes into account in detail the contributions from sn ii and ia .
it is very important to compute detailed sn rates , taking into account the stellar lifetimes ( @xmath5 ) , in order to study the different roles played by snii and sn ia in galaxy evolution .
type ii sne originates from the explosion of massive stars ( @xmath6 ) , the snii rate is : @xmath7 with @xmath8 and @xmath9 and @xmath10 being the imf .
type ia sne originate from the thermonuclear explosion of a co - white dwarf ( wd ) in a binary system .
the binary system can be made of a co wd plus a red giant star or by two co wds .
the type ia sn rate in the single degenerate case can be written as : @xmath11dm\nonumber \\ \end{aligned}\ ] ] where @xmath12 ( with @xmath13 mass of the secondary star ) and @xmath14 , the total mass of the binary system , is defined in the range 3 - 16 @xmath15 ( see matteucci & greggio , 1986 ) .
type ii sne produce mainly @xmath1-elements ( o , ne , mg , si , s , ca ) and part of fe .
the adopted yields are from woosley & weaver ( 1995 ) .
type ia sne produce mainly fe - peak elements ( @xmath16 of fe ) .
the adopted yields are from thielemann et al .
( 1993 ) the yields from low and intermediate mass stars ( @xmath17 ) are from renzini & voli ( 1981 )
the condition for the occurrence of a wind , where for wind we intend an outflow which carries the gas out of the potential well of the galaxy , is : @xmath18 where @xmath19 is the thermal energy of the gas and @xmath20 is the potential energy of the gas .
the thermal energy of gas due to sn and stellar wind heating is : @xmath21 with : @xmath22 and @xmath23 for the contribution from sne and stellar winds , respectively .
the efficiencies of energy transfer from sne into the ism is @xmath24 with @xmath25erg ( typical sn energy ) and that typical of stellar wind is : @xmath26 with @xmath27erg ( typical energy injected by a @xmath28 star ) .
the simplest approach is to consider the @xmath29 and @xmath30 efficiencies as constant .
bradamante et al .
( 1998 ) estimated such efficiencies by computing the ratio between the energy in the shell of the supernova remnant and in the initial blast wave , at the time of the merging of the shell with the ism , by taking into account results from hydrodynamical calculations .
they found that under typical conditions , namely @xmath31 erg , @xmath32 ( density of the ism ) and @xmath33 ( sound speed ) , @xmath29=0.007 - 0.13 and @xmath34 .
therefore , the majority of the initial blast wave energy of sne and stellar winds is radiated away . however , this hypothesis is strictly valid for an isolated object and multiple sn explosions can radically change the situation .
in addition , the contributions from different sn types can be different .
for example , sne ia exploding after type ii should provide more energy into the ism since they explode in a hot cavity . in the present model
we adopt a more complex formulation which assumes that the @xmath35 is varying in time .
in particular , we assume the formulation of cox ( 1972 ) for the efficiency of energy injection from sne : @xmath36 for @xmath37 , where @xmath38 is the cooling time of a supernova remnant , @xmath25 erg is the explosion energy and @xmath39 is the time elapsed from the sn explosion
. for @xmath40 holds : @xmath41 for the cooling time we adopt the formulation of cioffi et al .
( 1988 ) as a function of metallicity : @xmath42 it is worth noting that also with these prescriptions @xmath43 .
recchi et al . (
2001 ) assumed that @xmath44 and @xmath45 in successful chemodynamical models of dwarf irregular galaxies , in order to account for the fact that type ia sne occur with a delay and in an already hot cavity produced by type ii sne .
pipino et al .
( 2002 ) tried several assumptions for sn feedback in ellipticals including this one and concluded that no more than 35 - 40% of the initial blast wave energy of sne ii+ ia should heat the ism in order to have realistic models for ellipticals and the intracluster medium ( icm ) .
the total mass of the galaxy is expressed as @xmath46 with @xmath47 being the luminous mass .
the binding energy of the gas is computed as in matteucci ( 1992 and references therein ) . for all galaxies here we assumed @xmath48 and @xmath49 , being the ratio between the effective radius and the radius of the dark matter core .
the galactic winds in the models for ellipticals in the mass range @xmath50 occur on a timescale less than 1 gyr ( pipino et al . 2002 ) .
therefore , due to the short star formation period assumed for ellipticals , the abundance ratios in their stars show the signature of type ii sn nucleosynthesis , namely [ @xmath1/fe]@xmath510 , as it is evident in figure 1 where we show the predictions of the model for the chemical composition of the gas in a galaxy with initial luminous mass @xmath52 , salpeter ( 1955 ) imf and feedback prescriptions as in recchi et al .
( 2001 ) . in the figure
is marked the age for the occurrence of a galactic wind ( @xmath53 gyr ) .
after the wind no more star formation is taking place , so it is evident from the figure that the majority of the stars in this galaxy will show [ @xmath1/fe ] @xmath51 0 , whereas the gas that goes into the icm will show [ @xmath1/fe]@xmath54 .
this is due to the occurrence of type ia sne which have a maximum at 0.3 - 0.5 gyr ( matteucci & recchi , 2001 ) , namely after star formation has stopped .
the same is true for the energetics of type ii and ia sne if we assume @xmath55 . on the other hand ,
if sne ia are assumed to inject all of their initial blast wave energy , the situation is reverted , as shown in figure 1 ( right side ) ; sne ia dominate the energetics even before the onset of the wind.this also implies that both the energetics and the chemistry of the icm are dominated by type ia sne ( see pipino et al .
finally , in figure 2 we show the predicted [ x / fe ] vs. [ fe / h ] relations for ellipticals of different initial luminous mass .
overimposed are the data for a lyman - break galaxy ms 1512-cb58 at redshift @xmath56 obtained by pettini et al .
( 2002 ) .
the good agreement between the predictions for ellipticals and the observations strongly suggest that lyman - break galaxies at high redshift could be ellipticals in formation , thus supporting the monolithic scenario .
arimoto , n. , yoshii , y. 1987 , bradamante , f. , matteucci , f. , dercole , a. 1998 , 337 , 338 cioffi , d.f . ,
mckee , c.f . , bertschinger , e. 1998 , 334 , 252 cox , d.p .
1972 , 178 , 159 kauffmann , g. , white , s.d.m . ,
guiderdoni , b. , 1993 , 264 , 201 kuntschner , h. , lucey , j.r . ,
smith , r.j . ,
hudson , m.j . ,
davies , r.l . , 2001 323 , 615 larson , b.r .
1974 , 166 , 585 matteucci , f. 1992 , 397 , 32 matteucci , f. pipino , a. 2002 , 569 , l69 matteucci , f. recchi , s. 2001 , 558 , 351 matteucci & tornamb 1987 , 185 , 51 pettini , m. , rix , s.a . ,
steidel , c.c . ,
adelberger , k.l . ,
hunt , m.p . ,
shapley , a.e . , 2002 , 569 , 742 pipino , a. , matteucci , f. , borgani , s. , biviano , a. , 2002 , 7 , 227 recchi , s. , matteucci , f. , dercole , a. 2001 , 322 , 800 renzini , a. , voli , m. 1981 , 94 , 175 salpeter , e.e .
1955 , 121 , 161 thielemann , f.k.,nomoto , k. , hashimoto , m. a. 1993 , in `` origin and evolution of the elements '' , eds .
n. prantzos et al . , cambridge univ .
press p. 297 tinsley , b.m , larson , r.b .
1979 , 186 , 503 weiss , a. , peletier , r.f .
, matteucci , f. , 1995 296 , 73w woosley , s.e . , weaver , t.a . , 1995 , 101 , 181 worthey , g , faber , s.m . , gonzalez , j.j .
, 1992 , 398 , 69
|
models of sn driven galactic winds for ellipticals are presented .
we assume that ellipticals formed at high redshift and suffered an intense burst of star formation .
the role of supernovae of type ii and ia in the chemical enrichment and in triggering galactic winds is studied . in particular , several recipes for sn feed - back together with detailed nucleosynthesis prescriptions
are considered .
it is shown that sne of type ii have a dominant role in enriching the interstellar medium of elliptical galaxies whereas type ia sne dominate the enrichment and the energetics of the intracluster medium .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
the effective potential constitutes a fundamental tool in quantum field theory , widely used to study a multitude of physical phenomena , such as spontaneous symmetry breaking , tunnelling rates due to vacuum instability , and thermal phase transitions . with the ever increasing experimental precision on the determination of the standard model ( sm ) higgs boson mass at the cern large hadron collider ( lhc )
, more accurate computations of the sm effective potential can now be performed up to next - to - next - to - leading order ( nnlo ) .
assuming no new physics and ignoring quantum - gravity effects , such nnlo studies @xcite imply that the sm vacuum may well be metastable .
most remarkably , the actual profile of the effective potential at large higgs - field values was found to be extremely sensitive to very small variations of the input parameters at the electroweak scale @xcite , thereby rendering such precision computations rather subtle .
interestingly enough , this unusual feature seems to have sparked in the last years renewed interest in studying several field - theoretical aspects of the effective potential , which include the development of new semi - analytical techniques beyond ordinary perturbation theory @xcite , the treatment of infrared ( ir ) divergences due to the masslessness of the sm would - be goldstone bosons @xcite , more precise field - theoretical approaches to estimating tunnelling rates @xcite and assessing their gauge dependence @xcite , as well as evaluations of sm metastability rates in the presence of planck - scale suppressed operators @xcite .
the two - particle - irreducible ( 2pi ) effective action , as introduced by cornwall , jackiw and tomboulis @xcite , provides a powerful approach to resumming infinite series of higher - order diagrams of a given loop - graph topology in a systematic and self - consistent way . nevertheless , simple restrictions of the 2pi effective action to considering only a finite number of loop - graph topologies usually cause residual violations of possible underlying symmetries of the theory , which in turn introduce inconsistencies to the equations of motion .
most notably , for theories realizing spontaneous breakdown of global symmetries , naive truncations of the 2pi effective action lead to massive goldstone bosons ( see , e.g. @xcite ) . a potentially interesting theoretical framework for addressing this problem is the so - called symmetry improved two - particle - irreducible ( si2pi ) formalism @xcite , in which symmetry - consistent restrictions to the 2pi truncated equations of motion were implemented .
the resulting si2pi effective action has a number of satisfactory field - theoretical properties @xcite , including the masslessness of the goldstone bosons and the proper prediction of a second order thermal phase transition for @xmath2 scalar theories .
further applications and developments of the sip2pi formalism may be found in @xcite . in particular , in @xcite ( see also @xcite ) , the effect of fermions has been included in a semi - perturbative way and the si2pi effective potential has been used to address from _ first principles _ the ir divergence problem of the perturbative sm effective potential due to the masslessness of the would - be goldstone bosons .
hence , the results of this si2pi approach have proved to be very useful , as they enable one to assess the regime of validity of other simplified and partially _ ad hoc _
approaches @xcite .
in this paper we study the renormalization - group ( rg ) properties of the si2pi formalism , and prove for the first time a number of remarkable field - theoretical results .
first , after briefly reviewing the formalism in sec . [
sec : si2pi ] , we show in sec . [
sec : finite ] that the runnings of all proper 2pi quartic couplings are ultra - violet ( uv ) finite , without ignoring higher - order terms , in the hartree fock and sunset approximations of the 2pi effective action .
in fact , as this result is highly non - trivial , one may even conjecture that the uv finiteness of all proper 2pi quartic couplings will hold true to any arbitrarily high loop order of truncation of the si2pi effective action .
an equally important result of our study is presented in sec .
[ sec : rg_inv ] , where we show how the si2pi effective potential is _ exactly _ rg invariant at a given loop order of truncation , in stark contrast to what happens in the usual 1pi formalism , where the perturbative effective potential is rg invariant only _ up to higher order terms_. this is a remarkable result , as it helps us to clarify the relation between the 2pi resummation and the frequently considered rg - improved approach to the 1pi effective potential .
specifically , the former takes into account precisely the higher - order contributions needed to restore exact rg invariance when the effective potential is evaluated at a given loop level , whereas the latter attempts to minimize these higher - order terms by plausibly guessing the value of the rg scale @xmath3 , so as to avoid large logarithmic contributions . at this point , it is important to clarify that the approach we follow here differs conceptually from the one presented in @xcite , where functional methods were employed to reinforce rg invariance in the 2pi effective action . in sec .
[ sec : rgsi2pi ] we modify the si2pi formalism by appropriately improving the rg running of the quartic couplings .
this enables us to reach a high level of accuracy , and compare our results with the 1pi nnlo potential , which includes two - loop threshold corrections matched with 3-loop rg running .
we show that , as a consequence of the exact rg - invariance shown in sec .
[ sec : rg_inv ] , the predictions for the profile of the effective potential in the si2pi framework can reach a higher level of accuracy , even up to one order of magnitude , than the corresponding one obtained in the 1pi formalism .
this is indeed a significantly new result obtained for a simple @xmath0 theory that certainly encourages further studies within the context of more realistic theories . in a similar vein , in sec .
[ sec : conclusions ] we draw our conclusions and present possible future directions to this field .
in this section we briefly review the 2pi formalism by applying it to the scalar @xmath0 model and present the equations of motion used to analyze the rg properties derived from this formalism .
a more detailed discussion may be found in @xcite .
our starting point is the lagrangian of the @xmath0 model , @xmath4 where @xmath5 , @xmath6 is the @xmath0 scalar multiplet , which can be expanded in terms of a higgs and a goldstone mode as @xmath7 the 2pi effective action is obtained by introducing , in addition to a local source @xmath8 as in the usual 1pi effective action , a bi - local source @xmath9 , both with implicit @xmath0 group structure . by legendre - transforming the connected generating functional with respect to these sources , one obtains the 2pi effective action @xmath10 $ ] , depending on the background field @xmath11 and the _ dressed _
propagators @xmath12 . by expanding @xmath10 $ ] up to two - loop graph topologies
, one finds @xcite @xmath13 \ = \ & \int \
! \bigg[\frac{z_0}{2 }
( \partial_\mu \phi)^2 \:+\:\frac{m^2 + \delta m_0 ^ 2}{2 } \,\phi^2 \ : -\ : \frac{\lambda + \delta \lambda_0}{4 } \,\phi^4 \bigg ] \notag \\[3pt ] & -\ : \frac{i \hbar}{2 } \operatorname{tr}\big(\ln\delta_h\big ) \ : -\ : \frac{i \hbar}{2 } \operatorname{tr}\big(\ln \delta_g \big ) \displaybreak[0]\notag\\[3pt ] & - \ : \frac{i \hbar}{2 } \operatorname{tr}\big\{\big [ z_1 \ , \partial^2\ : + \ : \big ( 3 \lambda + \delta \lambda_1^a + 2 \delta \lambda_1^b\big)\,\phi^2 - \big ( m^2 + \delta m^2_1\big ) \big]\,\delta_h \big\ } \notag\\[3pt ] & - \ : \frac{i \hbar}{2 } \operatorname{tr}\big\{\big [ z_1 \ , \partial^2\ : + \ : \big(\lambda + \delta \lambda_1^a\big ) \ , \phi^2 - \big(m^2 + \delta m^2_1\big ) \big]\,\delta_g \big\ } \displaybreak[0 ] \notag\\[3pt ] & - \ , \frac{i \hbar^2}{4 } \ , \bigg\ { -i(3 \lambda + \delta \lambda_2^a + 2 \delta \lambda_2^b ) \int \ !
i\delta_h i\delta_h \;-\ ; 2i ( \lambda + \delta \lambda_2^a ) \int \ !
i\delta_h i\delta_g\notag\\ & -\ ; i ( 3\lambda + \delta \lambda_2^a + 2 \delta \lambda_2^b ) \int \ !
i\delta_g i\delta_g \bigg\ } \notag\\ & - \ , i \hbar^2 \ , \bigg\ { \ ; \parbox{1.7cm}{\includegraphics[width=1.7cm]{hhh } } \ ; + \ ; \parbox{1.7cm}{\includegraphics[width=1.7cm]{hgg } } \ ; \bigg\ } \;,\end{aligned}\ ] ] where the integrals are meant to be evaluated in position space over the common spacetime variable of the relevant fields and propagators . in the last line of , thick lines denote the dressed green s functions @xmath12 , whereas in the following thin lines are reserved to represent tree - level propagators .
the additional counter - terms ( cts ) in with respect to the 1pi formalism are related to the appearance of several operators with mass - dimensions 2 and 4 in the 2pi effective action @xcite .
in addition , there is a perturbative 1pi - like ct @xmath14 for the quartic coupling @xmath15 that appears at two- and higher loop orders @xcite . at the order of truncation considered here , however , this last ct @xmath16
is not needed to be included in the vertices of the diagrams in the last line of , as such a ct would be necessary to cancel subdivergences of higher - order diagrams , in close analogy to the usual 1pi perturbation theory . for the _ full _ effective action @xmath10 $ ] ,
the equation of motions ( eoms ) are given by @xmath17 in practice , however , the effective action can be calculated by considering only a finite number of 2pi topologies , for instance up to two - loop graph topologies as in , so that we can write @xmath18 , where @xmath19 is unknown .
a common strategy is to set @xmath20 , based on our ignorance of it , so that the eoms are obtained by replacing @xmath21 in .
however , in the presence of symmetries this leads to a number of artefacts , most importantly to violations of the usual ward identities , thus giving rise to massive goldstone bosons ( see @xcite and references therein ) in theories with spontaneous breaking of global symmetries .
this long - standing issue was addressed in @xcite ( see also @xcite ) by considering the so - called symmetry improved two - particle irreducible ( si2pi ) formalism . in this approach , rather than setting @xmath20 , one exploits the symmetries of the system to improve upon the expected form of @xmath19 , implying that the eoms needed to be solved are @xmath22 notice that , although we have written the above equations using the two - loop order truncation of the 2pi effective action in , these will still be valid at any order of truncation .
it has been shown in @xcite that this approach has a number of satisfactory field - theoretical properties , besides the masslessness of the goldstone bosons in the spontaneously - symmetry - broken phase of the theory .
moreover , the si2pi effective potential @xmath23 may be uniquely determined as a solution to the first - order differential equation @xmath24 subject to the initial condition @xmath25 , where @xmath26 is the vev of the scalar field @xmath27 . as discussed above , the eoms for the dressed propagators @xmath28
are obtained by differentiating @xmath29 $ ] in with respect to these propagators .
after wick - rotating to the euclidean momentum space , we obtain @xmath30 & \ + \ : ( 3 \lambda + \delta\lambda_2^a + 2 \delta \lambda_2^b ) \
, \hbar \bm{\mathcal{t}}_{\!\!h } \ : + \ : ( \lambda + \delta \lambda_2^a ) \ , \hbar \bm{\mathcal{t}}_{\!\!g } \notag\\[3pt ] & \ - \ : 18 \lambda^2 \phi^2 \ , \hbar \bm{\mathcal{i}}_{hh}(k ) \:-\ : 2 \lambda^2 \phi^2 \ , \hbar \bm{\mathcal{i}}_{gg}(k ) \ ; , \label{eq : eombarehscal } \displaybreak[0]\\[9pt ] \delta^{-1}_g(k)\ = & \ \ ( 1 + \delta z_1 ) \ , k^2 \:+\ : ( \lambda + \delta \lambda_1^a ) \ , \phi^2 \:-\ : ( m^2 + \delta m_1 ^ 2 ) \notag\\[3pt ] & \ + \ : ( \lambda + \delta \lambda_2^a ) \ ,
\hbar\bm{\mathcal{t}}_{\!\!h } \:+\ : ( 3 \lambda + \delta\lambda_2^a + 2 \delta \lambda_2^b)\,\hbar \bm{\mathcal{t}}_{\!\!g } \ : - \ : 4 \lambda^2 \phi^2 \ , \hbar \bm{\mathcal{i}}_{hg}(k ) \ ; , \label{eq : eombaregscal}\end{aligned}\ ] ] where @xmath31 . here
, we have introduced the tadpole and sunset integrals : @xmath32 where @xmath33 and @xmath34 , and @xmath3 is the so - called @xmath35 renormalization mass scale . at the two - loop level of the 2pi
effective action , there is no wavefunction renormalization for the radial field @xmath11 , and so we can set its ct @xmath36 to zero .
the eoms and are renormalized by cancelling separately the subdivergences proportional to the renormalized tadpole integrals and the overall divergences proportional to the field powers @xmath37 and @xmath38 , as described in detail in @xcite . out of @xmath39 relations ,
only 5 of them are found to be independent , which uniquely fixes the value of the 5 cts appearing in and . following this procedure ,
the renormalized eoms are found to be @xcite @xmath40 \delta^{-1}_g(k ) \
k^2 \:+\ : \lambda_1^a \phi^2 \,-\ , m^2 \:+\ : \lambda_2^a \ , \hbar\mathcal{t}_h^{\rm fin } \:+\ : ( \lambda_2^a + 2 \lambda_2^b ) \,\hbar \mathcal{t}_g^{\rm fin } \
: 4\lambda^2\phi^2 \ , \hbar \mathcal{i}_{hg}^{\rm fin}(k ) \notag\\ & + \ : \hbar^2\pi^{\mathrm{2pi } , ( 2)}_g \ ; , \label{eq : eomren_g}\end{aligned}\ ] ] where the analytic expressions for the renormalized , uv - finite integrals @xmath41 and @xmath42 are given in [ app : sunset ] .
keeping only the tadpole integrals @xmath41 in and is known as the hartree
fock ( hf ) approximation .
including also the remaining one - loop integrals @xmath42 is commonly referred to as the sunset approximation . in order to obtain a level of accuracy higher than the two - loop order in perturbation theory , we have included also the renormalized two - loop 2pi self - energies @xmath43 .
however , as discussed in @xcite , the latter contributions result from a three - loop order truncation of the 2pi effective action , and so we approximate them by their usual 1pi form evaluated in the zero - momentum limit @xmath44 .
their analytic expressions are given in [ app : pert_2loop ] .
we will include them only in sec .
[ sec : rgsi2pi ] when calculating explicitly the si2pi effective potential . in
and we have generalized the analysis in @xcite by taking the various 2pi quartic couplings different from each other , even at the renormalized level ( not only the corresponding cts ) .
this is done so in anticipation of the results that will be derived in the next section , where we will show that the rg running of the various 2pi quartic couplings will be , in general , different and so they can be taken to be equal only at some fixed rg scale @xmath45 . ,
scaledwidth=90.0% ] in the eoms and .
notice that the propagators belonging to two - loop 2pi topologies do not get dressed , since these are approximated by their usual 1pi forms.[fig : resum2loop],scaledwidth=80.0% ] as shown in figs .
[ fig : resum1loop ] and [ fig : resum2loop ] , the eoms and represent selective resummations of infinite set of diagrams of certain loop - graph topologies , which are obtained by dressing the propagators with one- and two - loop self - energies by an arbitrary number of times . here
, we must note that the one - loop self - energies can also be dressed , in a recursive way by an arbitrary number of times in fair analogy to considerations from fractal geometry , whereas the two - loop self - energies can not , since these have been approximated by their usual 1pi forms . before concluding this section , we briefly comment on the potential issue regarding the ir sensitivity of the sunset approximation in the si2pi formalism , which was recently raised in @xcite , concerning the ( ir - divergent ) higgs propagator in the limit @xmath46 .
we have checked explicitly that for values of scalar quartic couplings under study , i.e. @xmath47 , the alleged ir - sensitive effects on ir - safe observables of interest here and in @xcite , such as the pole mass of the unstable higgs particle and the 2pi effective potential , are absent at the level of accuracy we have been considering .
in fact , we have numerically varied the ir cutoff by a few orders of magnitude and carefully verified that our results for the effective potential remain stable .
this check for the effective potential away from its minimum complements analogous checks that were performed already in @xcite at the minimum of the potential .
in this section we calculate the rg running of the proper 2pi couplings in the hf and sunset approximations , and prove that they are uv finite without ignoring higher order terms , as usually done in the 1pi perturbation theory .
this is the first important result of this analysis , which holds true for both the 2pi and si2pi formalisms .
in fact , the uv finiteness of the proper 2pi couplings is a highly non - trivial rg property . as we illustrate in sec .
[ sec:1pi_finite ] , it requires delicate cancellations between @xmath48 poles with different @xmath49-powers and at different orders in the loop ( @xmath50 ) expansion . while in 1pi perturbation theory this result is guaranteed by `` @xmath50-expanding '' the full 1pi effective action , no analogous argument holds for the respective 2pi action , since no series - expansion parameter exists in this case for the solutions to the eoms .
following @xcite , the bare quartic couplings @xmath51 assume in the 1pi @xmath35 scheme the general form @xmath52 in @xmath53 dimensions , where @xmath54 is a dimensional - regularization ( dr ) parameter .
upon imposition of @xmath3-independence on the bare quartic couplings in , @xmath55 one finds that the rg running of the renormalized quartic couplings @xmath56 depends only on the coefficients of the single poles @xmath57
, i.e. @xmath58 with @xmath59 instead , the coefficients @xmath60 of the higher - order @xmath48 poles , with @xmath61 , have to conspire , in such a way that the rhs of vanishes identically .
this can happen , _ iff _ the recursive relations , @xmath62 are satisfied for all @xmath63 .
we note that these recursive relations are a consequence of subtle cancellations required in order to have finite @xmath64-functions @xmath65 .
most remarkably , they involve coefficients @xmath60 of poles with different powers of @xmath54 , and when and are expanded in series of loops , these cancellations occur among different loop levels .
all the relations presented in this subsection are valid to all orders in perturbation theory . in the 1pi formalism , they can be expanded in a series of the loop parameter @xmath50 , and are therefore valid at a given loop order as well .
this result shows the uv finiteness of the rg running at fixed loop order in the 1pi formalism .
we first consider the rg running of all 2pi couplings in the hf approximation , which will help us to gain valuable insight into the structure of the cts and their rg properties .
then , we extend our considerations to the less trivial case of the sunset approximation . in order to renormalize the eoms in the hf approximation ( cf .
and with the sunset integrals @xmath66 neglected ) as sketched in sec .
[ sec : si2pi ] and detailed in @xcite , the following cts are needed : @xmath67 \delta \lambda_1^a \ = \
\delta \lambda_2^a \ & = \
\frac{l \lambda^a}{\epsilon } \ ; \frac{\displaystyle \lambda^a \bigg(2 - \frac{4 l \lambda^b}{\epsilon } \bigg ) \,+\ , 4 \lambda^b \bigg ( 1 - \frac{l \lambda^b}{\epsilon}\bigg)}{\displaystyle 1 - \frac{2 l ( \lambda^a + 2 \lambda^b)}{\epsilon } + \frac{4 l^2 \lambda^b(\lambda^a+\lambda^b)}{\epsilon^2 } } \;,\label{eq : dl_a } \\ \delta \lambda_1^b\ = \ \delta
\lambda_2^b \ & = \
\frac{l \lambda^b}{\epsilon } \ ; \frac{2 \lambda^b}{\displaystyle 1- \frac{2 l \lambda^b}{\epsilon } } \;,\label{eq : dl_b}\\ \delta m^2 \ & = \
\frac{l m^2}{\epsilon } \ ; \frac{2(\lambda^a + \lambda^b)}{\displaystyle 1 - \frac{2 l ( \lambda^a + \lambda^b)}{\epsilon}}\;,\label{eq : dm}\end{aligned}\ ] ] with @xmath68 and @xmath69 . here , we have generalized the results reported in @xcite , by taking @xmath70 , with @xmath71 and @xmath72 .
as we will see below , we have done so , since @xmath73 and @xmath74 run differently .
notice that the cts stated in contain contributions from all orders in @xmath75 or @xmath50 , including infinite inverse powers of @xmath54 .
hence , it is far from obvious that these cts will lead to a finite running for the renormalized couplings @xmath73 and @xmath74 , according to our discussion in sec .
[ sec:1pi_finite ] . however , as we will now show , the rg running of the proper 2pi couplings @xmath73 and @xmath74 is _ exactly _ uv finite , within the hf approximation _
itself_. to this end , let us first consider the expansion of in powers of @xmath76 , @xmath77 employing the notation of sec .
[ sec:1pi_finite ] , we may identify @xmath78 and @xmath79 .
hence , yields @xmath80 and the @xmath76 pole in the running is cancelled , provided , for @xmath81 , is satisfied , i.e. @xmath82 it is not difficult to verify that this is indeed the case . to prove the cancellation of all @xmath48 ,
it is more convenient to follow a different strategy and use directly with the all - orders expressions . for instance , by plugging into , we must have @xmath83\ & = \ \mu \frac{d}{d\mu } \bigg [ \ , \overline{\mu}^{2 \epsilon } \bigg ( \lambda^b + \frac{2 l ( \lambda^b)^2}{\epsilon } \ , \frac{1 } { 1- 2 l \lambda^b/\epsilon } \bigg ) \bigg ] \notag \\ & = \
\overline{\mu}^{2 \epsilon } \frac{2 \epsilon \lambda^b } { 1- 2 l \lambda^b/\epsilon } \;+\ ; \overline{\mu}^{2 \epsilon } \frac{1}{(1- 2 l \lambda^b/\epsilon)^2}\ , \mu \frac{d \lambda^b}{d \mu}\ = \
0 \;.\end{aligned}\ ] ] this last constraint implies the finite rg running for @xmath74 , @xmath84 within the hf approximation only . by analogy , the rg running of the remaining parameters in the hf approximation are found to be @xmath85 from the finiteness of the above rg equations , we see that the rg running is fully determined by the single @xmath76 poles of the cts , whereas all higher order @xmath86 poles occurring in the resummed 2pi cts in cancel precisely against each other .
evidently , we observe the extraordinary rg property that _ higher - order poles in the resummed 2pi cts appear in such a way that make no contributions to the rg running , in complete analogy with what happens in the 1pi formalism to all orders . _ at @xmath87{gev}$ ] . for comparison , we also show the one - loop running of the coupling @xmath88 in the 1pi formalism .
[ fig : running],scaledwidth=40.0% ] we now turn our attention to the rg running of the 2pi kinematic parameters in the sunset approximation . proceeding as above and using the formulae for the cts reported in [ app : sunset ]
, the following rg equations may be derived : @xmath89\\ \displaystyle \mu \frac{d}{d \mu } \lambda_1^b \ & = \ - 2 \epsilon \
, \lambda_1^b \,+\ , 4 l \ , ( \lambda_1^b \lambda_2^b + 4\lambda^2 ) \,-\ , 32 l^2 \lambda^2 \lambda_2^b \ ; , \label{eq : running_sunset_l1b}\displaybreak[1]\\ \displaystyle \mu \frac{d}{d \mu } \lambda_2^a \ & = \ - 2 \epsilon \ , \lambda_2^a \,+\ , 4 l \ , \lambda_2^a ( \lambda_2^a + 2\lambda_2^b ) \ ; , \label{eq : running_sunset_l2a}\displaybreak[1]\\ \displaystyle \mu \frac{d}{d \mu } \lambda_2^b \ & = \ - 2 \epsilon \ , \lambda_2^b \,+\ , 4 l \ , ( \lambda_2^b)^2\;,\label{eq : running_sunset_l2b } \displaybreak[1]\\ \displaystyle \mu \frac{d}{d \mu } m^2 \ & = \ \ , 4 l \
, m^2 \ , ( \lambda_2^a + \lambda_2^b)\;. \label{eq : running_sunset_m}\end{aligned}\ ] ] notice that the rg equations and for the proper 2pi couplings @xmath90 and @xmath91 , respectively , contain contributions up to two - loop order . even though the sunset eoms are formally expressed in terms of 2pi one - loop integrals only , the resulting @xmath92 terms in the rg runnings of @xmath90 and @xmath91 are a consequence of non - trivial resummation effects . consequently ,
in the sunset approximation all the quartic couplings exhibit different rg runnings .
this is shown in fig .
[ fig : running ] , where the rg running of all quartic couplings is displayed , using the common value @xmath93 at @xmath87{gev}$ ] as initial renormalization condition .
this choice of values for @xmath94 and @xmath45 was motivated by the expected strength of the sm higgs quartic coupling at scales close to the mass of the observed higgs boson at the lhc .
although the rg runnings of all different quartic couplings will coincide when the full 2pi effective action is considered , we observe that the spread between them is still significant in the sunset approximation .
in this section our aim is to prove that the si2pi effective potential is _ exactly _ rg invariant in the hf approximation . in [ app : rg_invariance ] , we extend our proof of exact rg invariance to the full one - loop truncated si2pi effective potential which includes the sunset approximation .
this explicit demonstration of exact rg invariance of the si2pi effective potential is one of the central results of this work .
we should remind ourselves that no exact rg invariance can be achieved in the perturbative 1pi effective potential .
in fact , the perturbative 1pi effective potential is approximately rg invariant , as the residual rg - violating terms are formally of higher order in the @xmath50 expansion .
to minimize the significance of these higher order terms , one makes educated guesses for the value of the rg scale @xmath3 , usually taken to be close to the field value @xmath11 , in order to avoid potentially sizeable uncertainties from large logarithmic contributions to the effective potential . this approach forms the basis of the well - known procedure of rg improvement of the 1pi effective potential .
instead , this approach of rg improvement is no longer necessary within the si2pi framework . as we explicitly show below
, the si2pi effective potential is _ exactly _ rg invariant by itself , and no _ ad hoc _ choices of the value of the rg scale @xmath3 need to be made . in particular , for a fixed value of @xmath11 , the si2pi effective potential does not depend on @xmath3 , so any choice of @xmath3 can be made , including @xmath95 , for example . proving the exactness of rg invariance in the hf approximation becomes a straightforward exercise , because the integrals involved can be performed analytically . in particular , we have @xmath96 in the euclidean space and @xcite @xmath97 with @xmath98 .
let us now analyze the responses of all relevant parameters under an infinitesimal shift of the renormalization scale : @xmath99 , with @xmath100 . obviously , the field @xmath11 remains constant under this shift , i.e. @xmath101 , since the wavefunction renormalization vanishes at this order @xcite .
the other parameters of the 2pi effective action then vary as follows : @xmath102 where @xmath103 and the explicit form of the @xmath64- and @xmath104-functions can be read off from for @xmath105 .
likewise , the variation of the tadpole integral in is given by @xmath106 under the same infinitesimal rg shift : @xmath107 , the variation of the eom for the field @xmath108 in the hf approximation may be calculated by virtue of and to be @xmath109 \notag \\ & + \ , \beta_a \hbar \mathcal{t}_g^{\rm fin } \delta t \,+ \ , \lambda^a\bigg [ l \ln \bigg(\frac{m^2_g}{\mu^2}\bigg)\delta m_g^2 \ , - \ , 2 l m_g^2 \delta t \bigg ] \;.\end{aligned}\ ] ] making now use of the explicit forms for the @xmath64- and @xmath104-functions , as given by and , we finally arrive at @xmath110 \delta m_h^2 \ , -\ , \bigg[\lambda^a l \ln \bigg(\frac{m^2_g}{\mu^2}\bigg ) \bigg ] \delta m_g^2 \ = \ 0\times \delta t\ = \ 0\;.\ ] ] it is important to remark here that all terms proportional to @xmath111 vanish identically .
we may now proceed in a similar fashion by considering the variation of the eom for the goldstone field @xmath112 in the hf approximation . in this case
, we obtain another homogeneous equation linearly independent of for @xmath113 and @xmath114 , @xmath115 \delta m_h^2\ , -\ , \bigg[1 - ( \lambda^a+2 \lambda^b ) l \ln \bigg(\frac{m^2_g}{\mu^2}\bigg ) \bigg ] \delta
m_g^2\ = \ 0\times \delta t\ = \ 0\;,\ ] ] which implies @xmath116 this means that the solutions to the @xmath108- and @xmath112-boson eoms in the hf approximation are both independent of @xmath3 , provided the 2pi rg equations
are used . since the si2pi effective potential is obtained precisely from the solution to the eom of the goldstone propagator [ cf . ]
, we conclude that _ the si2pi effective potential is exactly rg invariant _ , at least in the hf approximation .
the same statement holds true for the full one - loop truncated si2pi potential , which includes the sunset approximation , but the proof is more laborious and is therefore presented in [ app : rg_invariance ] . as these findings are highly non - trivial , we feel tempted to conjecture that this basic property of exact rg invariance of the si2pi potential will persist for any arbitrary high loop - order truncation of the 2pi effective action differs from the one that would be obtained in the si2pi formalism . ] . .
for numerical purposes , we have worked in the massless approximation @xmath117 , since mass effects for field values much larger than the electroweak scale are expected to be negligible .
[ fig : effpot_exactrg],scaledwidth=40.0% ] in fig .
[ fig : effpot_exactrg ] we plot the si2pi effective potential in the sunset approximation , along with the 1pi unimproved ( @xmath118 ) and rg - improved ( @xmath95 ) one - loop effective potentials . here and in the following
, we neglect the electroweak - scale mass @xmath119 , since we are interested in large field values @xmath120 .
as expected , we see that the difference between the unimproved and improved 1pi effective potentials becomes significant for high field values @xmath11 . instead , the si2pi effective potential is @xmath3-independent , yielding the same value for any choice of @xmath3 .
nevertheless , we see that the results for the si2pi effective potential differ significantly from the corresponding ones obtained in the rg - improved 1pi framework . in the next section
we identify the origin of this difference , which we address . as a consequence
, this will enable us to develop a novel method to calculate the effective potential semi - perturbatively , thereby increasing considerably its precision in comparison to the standard rg - improved 1pi approach .
in the previous section , we have shown that the complete one - loop truncated si2pi effective potential is exactly rg invariant , provided all the 2pi parameters run as given by in the sunset approximation .
nevertheless , the predictions obtained in this si2pi framework for the effective potential differ from those found for the rg - improved 1pi effective potential . in this section
, we first clarify the origin of this difference and then present a new computational method of higher precision for the effective potential by unifying the two theoretical approaches of rg improvement and symmetry improvement in the context of the 2pi effective action .
we call the effective potential computed with this new method the renormalization group and symmetry improved 2pi ( rg - si2pi ) effective potential . to start with ,
let us first compare the rg equations of the 2pi quartic couplings with the one for the 1pi quartic coupling @xmath88 at the one - loop level , @xmath121 we observe that , if we take all 2pi quartic couplings equal in , some @xmath122 contributions are still missing .
in fact , the omission of these contributions is exactly the origin of the limited accuracy of the predictions for the si2pi effective potential through this order .
the remaining @xmath122 terms are generated when one includes the higher - order 2pi self - energy diagrams as shown in fig .
[ fig : missing_diagrams ] .
in particular , their @xmath122 contribution to the rg running of the quartic couplings is described by the extra terms @xmath123 when these terms are added to the rhs of the rg equations , and , the 1pi @xmath122 result is recovered in the limit where all the quartic couplings are taken to be equal . in other words , although the contribution of the diagrams in fig .
[ fig : missing_diagrams ] to the eoms is subleading in the sunset approximation , their contribution to the running of the couplings is still at the same leading order .
therefore , without going through the strenuous procedure of 2pi renormalization at higher orders , one practical way to obtain accurate results will be to appropriately amend the @xmath64-functions as given by ordinary perturbation theory through the desired loop order for all the 2pi couplings and calculate the si2pi effective potential using these improved rg equations . we will adopt this novel method below in order to compute a rg and symmetry improved 2pi effective potential for a @xmath0 theory , beyond the nnlo . in detail , the rg equations that we will be utilizing with this method are @xmath124 where @xmath125 collectively denotes all running quartic couplings and @xmath126 is the running field , which enters both the rg - si2pi effective potential and the corresponding nnlo - improved 1pi effective potential . in and , we use the perturbative three - loop @xmath64- and @xmath104-functions for the quartic coupling @xmath15 and the running field @xmath127 ( see e.g. @xcite ) . by construction , within the adopted method , the so - derived rg - si2pi effective potential contains all the two - loop graphs present in the respective perturbative 1pi effective potential .
in addition , the rg - si2pi effective potential contains selective fractal - type resummations of higher - order diagrams as sketched in fig .
[ fig : resum1loop ] and [ fig : resum2loop ] , whereas the running of the field @xmath127 and of the quartic couplings is the same as that used for evaluating the corresponding 1pi effective potential . running . as can be seen by comparison with and ,
the first two diagrams contribute to the running of @xmath15 , and the third diagram to the running of @xmath128 .
[ fig : missing_diagrams],title="fig : " ] running . as can be seen by comparison with and ,
the first two diagrams contribute to the running of @xmath15 , and the third diagram to the running of @xmath128 .
[ fig : missing_diagrams],title="fig : " ] running . as can be seen by comparison with and ,
the first two diagrams contribute to the running of @xmath15 , and the third diagram to the running of @xmath128 .
[ fig : missing_diagrams],title="fig:",height=68 ] within the approximative method that we follow here , the rg - si2pi effective potential is no longer exactly rg invariant , because _ not all _ restoring corrections of rg invariance as dictated by the 2pi formalism were introduced to the running of the kinematic parameters .
thus , in order to minimize the impact of the residual large logarithms , we may choose a sliding renormalization scale @xmath129 .
this is standard procedure for rg improvement of the 1pi effective potential , but it introduces some subtleties in the si2pi framework . for a @xmath11-dependent rg scale @xmath130 , the 1pi ward identity , used to obtain the si2pi effective potential , gives now only the _ partial _ derivative of the effective potential : @xmath131 where the running field @xmath132 becomes now an implicit function of @xmath11 : @xmath133 .
however , in order to compute the si2pi effective potential , we are interested in the _ total _ derivative @xmath134 hence , this last relation can be rewritten in the equivalent integral form as @xmath135 while the first term on the rhs of is obtained by virtue of , the remaining terms need to be approximated by their 1pi expressions through the two - loop order considered here . these additional terms can be computed , given the explicit two - loop order expression for the 1pi effective potential given in appendix c [ cf . ] . in this way
, one can integrate with respect to @xmath132 and obtain @xmath136 .
notice that this additional step is introduced by the procedure of rg - improving the effective potential , within our approximative method .
instead , it is not necessary for the exactly rg - invariant si2pi effective potential calculated in sec . [
sec : rg_inv ] . to make this explicit , note that the latter satisfies @xmath137 so that the total and partial derivatives coincide , which implies that the second and third term on the rhs of cancel exactly . .
we have normalized the results to the 1pi @xmath138 case , for convenience of presentation .
the area between the different blue and red lines provides an estimate for the accuracy of the results obtained in the two methods .
the input values at the scale @xmath87{gev}$ ] are given in table [ tab : lambda].[fig : rgsi2pi],scaledwidth=40.0% ] following the approximative method outlined above , we can now solve numerically the 2pi eoms and .
we remind the reader that as described in sec .
[ sec : si2pi ] , these include contributions from one- and two - loop 2pi self - energies ; the latter were approximated by their perturbative 1pi forms .
we have adapted the numerical code in @xcite by including these two - loop 2pi contributions and a @xmath11-dependent renormalization scale @xmath139 ( see also discussion below ) .
we first solve numerically the coupled system of eoms for the @xmath108- and @xmath112-propagators for different values of @xmath132 , and then utilize the solution for the @xmath112-propagator in to find the si2pi effective potential by means of . to maximize the numerical accuracy
, we have rescaled the eoms by re - expressing all mass - dimensional terms in units of @xmath132 , so that no large mass hierarchies are present during the numerical computation .
also , in the large - field @xmath11 regime of interest here , we have safely set @xmath140 .
we have used grids of 1001 points in momentum space , with a uv cutoff @xmath141 , tolerating a relative threshold error of @xmath142 for the convergence of the numerical algorithm .
the ir cutoff is chosen to be @xmath143 , and we have checked that all our predictions for the rg - si2pi effective potential remain stable against variations of either the uv or the ir cutoff .
these checks provide firm support of our regularization treatment in numerically solving the relevant 2pi eoms in the sunset approximation , whilst the potential ir - sensitive effects claimed in @xcite have no noticeable impact on the physical quantities that we have been studying . in fig .
[ fig : rgsi2pi ] we present numerical estimates of the rg - si2pi effective potential and the corresponding rg - improved 1pi effective potential through nnlo , as functions of the field @xmath11 .
the predicted values are displayed for several choices of the rg scale @xmath3 .
the first obvious choice is @xmath138 .
in addition , we have considered other relevant mass - scales of the theory , such as @xmath144 and @xmath145 .
following @xcite , we have taken @xmath3 to be equal to either of them , i.e. @xmath146 and @xmath147 .
for all the different cases , we have chosen the value of @xmath15 at the scale @xmath87{gev}$ ] , such that the value of the effective potentials calculated at different rg scales coincides at @xmath148 .
the precise values of @xmath149 required to achieve to this are given in table [ tab : lambda ] .
c|l method & @xmath149 + ' '' '' si2pi @xmath147 & 0.13 + ' '' '' si2pi @xmath150 & 0.13000235 + ' '' ''
si2pi @xmath138 & 0.13000469 + ' '' '' 1pi @xmath147 & 0.13000003 + ' '' ''
1pi @xmath150 & 0.13000287 + ' '' ''
1pi @xmath138 & 0.13000985 + the bands formed by the different lines in fig .
[ fig : rgsi2pi ] give a kind of a measure for the @xmath3-dependence of the theoretical predictions , and as such , they should be viewed as theoretical uncertainties inherent to these nnlo calculations .
we observe that the bands predicted for the rg - si2pi potential all lie within the respective ones for the 1pi effective potential , which clearly shows that the predictions for the former potential suffer less from @xmath3-dependent theoretical uncertainties .
most impressively , the uncertainty of the si2pi effective potential is significantly smaller than the 1pi one , which may be reduced by up to one order of magnitude when compared with the 1pi results , e.g. for the bands restricted by the lines @xmath151 and @xmath150 .
these findings should not be too surprising , as the rg - si2pi effective potential , albeit not exactly rg invariant , carries a sort of a _ memory _ effect of exact rg invariance that the si2pi formalism seems to be endowed with .
we have studied the rg properties of the symmetry improved two - particle - irreducible effective action truncated to two - loop order , within the context of a simple @xmath0 scalar theory .
we have been able to prove a number of remarkable field - theoretical results , which to the best of our knowledge have not been discussed adequately in the existing literature . unlike the perturbative 1pi effective action
, the loop - truncated 2pi effective action of a simple @xmath152-theory requires different renormalizations for all mass and coupling parameters , in order to obtain uv finite results for all possible operators consisting of the renormalized field @xmath11 and its dressed propagator @xmath12 . working in this loop - truncated 2pi framework ,
we have shown that the rg equations for all relevant 2pi masses and couplings of a simple @xmath0 model are _ exactly _ uv finite in both the hf and sunset approximations , as all @xmath48 poles of order @xmath153 and higher vanish identically in dimensional regularization .
this is the first important result of our study , which is highly non - trivial .
in fact , in ordinary 1pi perturbation theory , such cancellations of the higher order @xmath76 poles occur at different loop orders , and they become only complete to all orders . as shown in fig .
[ fig : running ] , the different 2pi couplings exhibit different rg runnings , even though they are taken to be equal at some given scale @xmath118 .
note that this variance in the running of the quartic couplings is expected to decrease , as the order of loop truncation of the 2pi effective action increases .
thus , we expect to obtain a universal rg running for all 2pi quartic couplings to all loop orders .
another important result of our study was to prove that the one - loop truncated si2pi effective potential is _ exactly _ rg invariant in the hf and the sunset approximations .
this is an extraordinary property and should be contrasted with what happens in the usual 1pi formalism , where the effective potential becomes rg invariant , only upon ignoring _
unknown _ higher order terms . therefore ,
whilst one needs to do educated guesses for the rg scale @xmath3 so as to rg - improve the effective potential and minimize the impact of these unknown terms in the 1pi formalism , there is no need to do so in the 2pi framework , as the si2pi effective potential is _ exactly _ @xmath3-independent and as such , its value will be equal for both fixed @xmath118 and sliding @xmath154 , e.g. @xmath95 .
however , in the sunset approximation that we have been considering here , not all rg equations of the 2pi couplings have the required accuracy , as not all of them match to the one - loop @xmath64-function , @xmath155 , known from the standard 1pi perturbation theory .
we have identified the origin of the missing terms in the one - loop 2pi @xmath64-functions , which stem from higher order loop - graph topologies in the 2pi effective action that first appear at the three - loop order of truncation . to improve the accuracy of the rg running , we have adopted an approximative method , where the three - loop order 1pi @xmath64- and @xmath104-functions were utilized for all 2pi quartic couplings and the running field @xmath11 . in this approximative method ,
the exact rg invariance of the si2pi effective potential is lost , so we have resorted to a rg - improvement approach similar to the one performed for the usual 1pi effective potential .
the so - derived rg - si2pi effective potential contains all two - loop diagrams present in the 1pi effective potential , including the three - loop rg - running of the quartic coupling and the radial scalar field @xmath11 .
in addition , the rg - si2pi effective potential goes beyond the nnlo , as it contains selective resummations of infinite series of graphs as illustrated in figs .
[ fig : resum1loop ] and [ fig : resum2loop ] , which are inherent in the 2pi formalism . as shown in fig .
[ fig : rgsi2pi ] , the rg - si2pi effective potential gives rise to predictions that suffer less from @xmath3-dependent theoretical uncertainties .
specifically , we have found that the @xmath3-dependent uncertainties of the si2pi effective potential are considerably smaller than the one obtained in the 1pi formalism . in particular , the theoretical uncertainties may be lowered by up to one order of magnitude when compared to those obtained in the 1pi framework , as can be seen from the bands confined by the lines @xmath156 and @xmath150 in fig .
[ fig : rgsi2pi ] .
this last finding of our study underlies the remarkable rg properties of the rg - si2pi effective potential . on the other hand , this should not be considered too surprising , as the rg - si2pi effective potential seems to have inherited a good degree of exact rg invariance from the original si2pi formalism .
the present study opens up new exciting vistas that deserve further detailed and rigorous exploration .
in particular , it would be interesting to extend the present computation of the si2pi effective potential by including dressed chiral fermion propagators , and check its invariance under rg running in both the hf and sunset approximations
. this would be an important milestone in this line of investigations , enabling for the first time higher order precision computations for the sm effective potential beyond the nnlo with considerably reduced theoretical uncertainties .
an equally interesting avenue that one could follow would be to go beyond the sunset approximation , and investigate whether the exactness of rg invariance persists for the si2pi effective potential .
the ultimate goal of these studies would be to obtain an _ exactly _ rg - invariant si2pi effective potential beyond the nnlo , which in turn will help us to obtain more accurate determinations of several physical processes , such as vacuum - to - vacuum tunnelling transitions , beyond their state of the art .
the work of a.p . is supported by the lancaster manchester
sheffield consortium for fundamental physics , under stfc research grant : st / l000520/1 .
the work of d.t . is funded by a ulb postdoctoral fellowship .
the renormalization in the sunset approximation is described in detail in @xcite . here
we present useful formulae for the renormalized tadpole and sunset integrals , @xmath157 and @xmath42 , and for the 2pi quartic coupling cts @xmath158 and @xmath159 in dr , which are relevant to our study . here
we assume that all the renormalized proper 2pi couplings , @xmath160 and @xmath161 , are , in principle , different from each other .
in this way , we generalize the results of @xcite , beyond the hf approximation . working in the euclidean momentum space
, we may conveniently express the renormalized sunset integral as @xmath162 where @xmath33 , and @xmath163 is an auxiliary propagator introduced here with the purpose to remove uv infinities in dr . in order to calculate the renormalized tadpole integral , we parametrize the renormalized propagators as @xmath164 where @xmath165 stands for all momentum - independent contributions that originate from the tree - level terms and the tadpole integrals in the renormalized eoms and , whereas @xmath166 is the contribution of the sunset integrals , i.e. @xmath167 the renormalized tadpole integral is then found to be @xmath168 \notag\\ & \quad - \ : \frac{\mu^2}{16 \pi^2 } \ : + \ : \hbar \ , \frac{\nu_a \ , \lambda^2 \phi^2}{(16 \pi^2)^2 } \,\frac{\eta}{2 } \ ; , \label{eq : ren_t } \end{aligned}\ ] ] where @xmath169 , @xmath170 and @xmath171 requiring now that the unrenormalized eoms and be uv finite , we may derive the 2pi cts @xmath172 and @xmath173 in the sunset approximation , @xmath174 \notag\\ & - \ ; 4 \ , l^2 \bigg [ \frac{1}{\epsilon } ( 3 \lambda_2^a \lambda^2 + \lambda_2^b \lambda^2 ) + \frac{1}{\epsilon^2 } ( \lambda_1^a \lambda_2^a \lambda_2^b + \lambda_1^a { \lambda_2^b}^2 - \lambda_2^a \lambda^2 + 3 \lambda_2^b \lambda^2 ) \bigg ] \notag\\ & + 8 \ , l^3 \bigg [ \frac{1}{\epsilon^2 } ( \lambda_2^a \lambda_2^b \lambda^2 + { \lambda_2^b}^2 \lambda^2 ) + \frac{1}{\epsilon^3 } ( \lambda_2^a \lambda_2^b \lambda^2 + { \lambda_2^b}^2 \lambda^2 ) \bigg ] \bigg\ } \ ; , \displaybreak[1]\\[6pt ] \delta \lambda_1^b \ & = \ \frac{1}{\displaystyle 1 - 2 \frac{l}{\epsilon } \lambda_2^b } \bigg\ { 2 \frac{l}{\epsilon } ( \lambda_1^b \lambda_2^b + 4 \lambda^2 ) \,-\ , 8 \ , l^2 \bigg ( \frac{1}{\epsilon } + \frac{1}{\epsilon^2}\bigg ) \lambda_2^b \lambda^2 \bigg\ } \;.\end{aligned}\ ] ] instead , the 2pi cts @xmath175 , @xmath176 and @xmath177 may be obtained by , and , upon replacing @xmath178 with @xmath179 .
in this appendix we prove that the 2pi eoms and are _ exactly _ rg invariant in the sunset approximation , in close analogy to the proof presented in sec .
[ sec : rg_inv ] in the hf approximation . to start with , let us first point out that in the sunset approximation , it is important to also include the rg runnings @xmath180 for the field @xmath11 and the quartic coupling @xmath15 that are induced at the tree level in dr .
these tree - level effects @xmath181 give non - trivial finite contributions when multiplied with uv poles @xmath182 .
thus , under an infinitesimal rg shift : @xmath183 , with @xmath184 , the 2pi kinematic parameters transform as follows : @xmath185 where @xmath186 and @xmath187 , and the @xmath64- and @xmath104-functions can be read off from . our next step is to calculate the variation of the renormalized sunset and tadpole integrals stated in and , respectively , under the same infinitesimal shift : @xmath183 .
to do so , we first consider the integral @xmath188 in . in this
integral , the term depending on the auxiliary propagator @xmath189 can be integrated out analytically in dr .
the latter allows us to write the unrenormalized sunset integral @xmath190 as follows : @xmath191 hence , the variation of @xmath188 is given by @xmath192 \
= \ 2 l \delta t \,+\ , 2 \epsilon \ , \hbar \mathcal i_{ab}^{\rm fin } \,+\ , { \cal o}(\delta \delta ) \notag\\ & = \ 2 l \delta t \,+\ , { \cal o}(\epsilon ) \,+\ , { \cal o}(\delta \delta)\;,\end{aligned}\ ] ] where only the leading terms in a series expansion in powers of @xmath54 and @xmath193 are kept .
proceeding analogously for the tadpole integral , we get @xmath194 given that @xmath195 , we have @xmath196 as can be seen from the last expression , one key ingredient is the variation @xmath197 .
this can be found by observing that implies @xmath198 , with @xmath98 , and by varying and . in this way
, we have @xmath199 notice that we do _ not _ neglect relevant terms @xmath180 , as they can potentially give finite contributions when multiplied by @xmath76 poles . taking such terms into account , we end up with @xmath200 thus , after rg - varying the eom pertinent to the inverse higgs propagator @xmath201 , we obtain in the limit @xmath202 , @xmath203 \ , + \ , { \cal o}(\delta \delta ) \;.\end{aligned}\ ] ] equipped now with all key equations for the variations of @xmath204 and @xmath205 in , the eoms and , and the @xmath64- and @xmath104-functions as deduced from , and after some lengthy algebra , we finally obtain @xmath206 we note that an analogous result is obtained for @xmath207 as well . employing the identity of variations : @xmath208 on the rhs of for @xmath209 and its corresponding counterpart for @xmath207
, we get a linear system of two equations : @xmath210 where @xmath211 and @xmath212 are lengthy expression depending on @xmath213 , and all other kinematic parameters of the theory , such as the field @xmath11 , the mass parameter @xmath214 and the 2pi quartic couplings .
it can be shown that the two equations in and are linearly independent .
this implies that barring fine - tunings , @xmath215 is a non - trivial solution to and , for arbitrary values of @xmath216 and @xmath11 .
hence , the solutions to the 2pi eoms for the @xmath108- and @xmath112- propagators are exactly rg invariant .
therefore , we conclude that the si2pi effective potential , obtained from @xmath217 , is _ exactly _ rg invariant also in the sunset approximation .
in this appendix , we present analytical expressions for the two - loop 1pi effective potential @xmath218 and the renormalized two - loop self - energies @xmath219 appearing in the eoms and , all in the limit of interest @xmath220 .
these are calculated in perturbation theory in the so - called @xmath35 scheme by means of standard techniques @xcite , and approximated by their zero - momentum value .
we adopt the compact notation of @xcite , and introduce the abbreviations : @xmath221 and @xmath222 , where @xmath216 is the euclidean momentum .
moreover , we define the one - loop function @xmath223 and the two - loop function @xmath224 evaluated at zero momentum , whose analytical expression may be found in @xcite .
the 1pi effective potential for the scalar @xmath0 model , up to two - loop order , is given by @xmath225 \notag\\ & + \
l^2 \ , \frac{\lambda}{4 } \ ,
\big [ 3 a(h)^2 + 2 a(h ) a(g ) + 3 a(g)^2 - 12 \lambda \phi^2 i(h , h , h ) - 4 \lambda \phi^2 i(h , g , g)\big]\;,\end{aligned}\ ] ] with @xmath226 and @xmath227 .
the 2pi two - loop topologies contributing to @xmath219 are shown in fig .
6 of @xcite , after neglecting in that work the fermion and charged goldstone - boson contributions . for the @xmath0 model at hand , we find @xmath228 \notag \\ & - \ 6 \lambda^2 i(h , h , h ) \;-\ ; 2 \lambda^2 i(h , g , g ) \notag\\ & - \ 8 \lambda^3 \phi^2 \big [ 27 \ , i(h',h , h ) \ ; + \ ; 3\ , i(h',g , g ) \ ; + \ ; 2 \ , i(g',g , h)\big ] \notag \\ & - 8 \lambda^4 \phi^4 \big [ 81 \ , i(h',h',h ) \ ; + \ ; 6 \ , i(h',g',g ) \;+\ ; i(g',g',h)\big ] \;,\\[6pt ] \pi^{\mathrm{2pi } , ( 2)}_g / l^2 \ & = \ 8 \lambda^3 \phi^2 \big[a(h ) - a(g ) \big]^2/(h - g)^2 \notag \\ & - \ 2 \lambda^2 i(g , h , h ) \ ; - \ ; 6 \lambda^2 i(g , g , g ) \notag \\ & + \ 8 \lambda^3 \phi^2 \big [ 3
\,i(g , g , g ) \;-\ ; 3 \,i(h , h , h ) \;+\ ; i(g , h , h ) \;-\ ;
i(g , g , h)\big]/(h - g ) \notag\\ & + \ 16 \lambda^4 \phi^4 \big [ - i(g , g , g ) \,-\ , 3 \ , i(h , h , h ) \ , + \ , 5 \ , i(g , g , h ) \,-\,i(g , g , h ) \big]/(h - g)^2 \;.\end{aligned}\ ] ] where now @xmath229 , @xmath230 and the prime ( @xmath231 ) denotes differentiation of the function @xmath224 with respect to the primed argument , e.g. @xmath232 , @xmath233 etc .
|
the symmetry improved two - particle - irreducible ( si2pi ) formalism is a powerful tool to calculate the effective potential beyond perturbation theory , whereby infinite sets of selective loop - graph topologies can be resummed in a systematic and consistent manner . in this paper
we study the renormalization - group ( rg ) properties of this formalism , by proving for the first time a number of new field - theoretic results .
first , the rg runnings of all proper 2pi couplings are found to be uv finite , in the hartree fock and sunset approximations of the 2pi effective action .
second , the si2pi effective potential is _ exactly _ rg invariant , in contrast to what happens in the ordinary one - particle - irreducible ( 1pi ) perturbation theory , where the effective potential is rg invariant only up to higher orders .
finally , we show how the effective potential of an @xmath0 theory evaluated in the si2pi framework , appropriately rg improved , can reach a higher level of accuracy , even up to one order of magnitude , with respect to the corresponding one obtained in the 1pi formalism .
@xmath1 man / hep/2017/02 , ulb - th/17 - 03 + march 2017 2pi effective action ; renormalization group ; effective potential . makefntext[1]=0em1emmakefnmark#1
|
x - ray novae ( or soft x - ray transients ) are compact binaries in which a neutron star or black hole ( bh ) primary accretes from a donor star via roche - lobe over - flow .
most of these systems are usually in a quiescent state with an x - ray luminosity of 10@xmath6 to 10@xmath7 erg s@xmath3 .
however , they undergo episodic outbursts that last for months with x - ray luminosities that can sometime reach or exceed the eddington limit ( @xmath8 10@xmath9 erg s@xmath3 for a 10 m@xmath10 bh ) . despite the residual activity of these quiescent objects ,
very little is known about their emission properties at very low accretion rates ( mcclintock & remillard 2005 ) . with the sensitivity of current x - ray missions ( in particular _
chandra _ and _ xmm - newton _ ) , it is now possible to study in more detail the physical processes that take place in this accretion regime
. was discovered by _
rxte_/asm on 1998 september 7 ( smith 1998 ) .
a brief and intense radio / x - ray flare , associated with a massive plasma ejection , was observed two weeks later ( hannikainen et al .
subsequent radio and x - ray observations revealed the formation of large scale jets moving away from the black hole over the course of several years ( corbel et al . 2002 ; tomsick et al . 2003a ; kaaret et al . 2003 ) .
after its discovery outburst in 19981999 , had a second strong outburst in 2000 , and fainter and shorter outbursts were detected in 2001 , 2002 and 2003 ( figure 1 ) .
optical observations indicate that the compact object in is likely a black hole of 10.5 @xmath5 1.0 m@xmath10 at a distance of about 5.3 kpc ( orosz et al .
2002 ) .
h was discovered with _
ariel v _ in august 1977 ( kaluzienski & holt 1977 ) and was precisely localized by _
heao-1 _ a few weeks later ( doxsey et al .
1977 ) . based on its x - ray properties ,
h has been classified as a black hole candidate ( white & marshall 1983 ) . in march 2003 , _ integral _ detected new activity from igr j17464@xmath113213 ( revnivtsev et al .
2003 ) that was later found to correspond to h. during outburst , h went through several x - ray states with properties typical of bhc .
the 2003 outburst ended around late november 2003 .
h was observed again in outburst from july 2004 ( swank 2004 ) to november 2004 . a bright radio flare ( likely associated with a massive ejection event )
was observed in 2003 by rupen et al .
the ejected plasma was later found to interact with the interstellar medium causing in - situ particle acceleration and the formation of two large - scale , synchrotron - emitting radio and x - ray jets ( corbel et al .
2005 ) as in the case . in this paper ,
we present the results of ten _ chandra _ observations of the two black holes and h carried out during their quiescent state .
these two sources have been detected at their faintest level ever observed . the high quality
_ chandra _ spectra for allow us to study and monitor the emission properties of this black hole at very low accretion rate .
we then investigate the quiescent emission of black hole systems .
our _ chandra _ program of monitoring x - ray jets from and h has also allowed us to study in great detail the black holes in these two systems .
was observed by _
chandra _ on five occasions : 2002 march 11 ( mjd 52344.8 ) , 2002 june 19 ( mjd 52444.5 ) , 2002 september 24 ( mjd 52542.0 ) , 2003 january 27 ( mjd 52667.3 ) and 2003 october 23 ( mjd 52935.6 ) . for completeness
, we also include the results of two published _
chandra _ observations on 2000 august 21 ( mjd 51777.4 ) and 2000 september 11 ( mjd 51798.3 ) that were performed during the decay of the 2000 outburst of ( tomsick et al .
_ observations of h were carried out on three occasions : 2004 february 12 ( mjd 53048 ) , march 24 ( mjd 53089 ) and march 27 ( mjd 53092 ) .
figure 1 shows the _
rxte_/asm 1.512 kev light - curves of and h , and the arrows indicate the dates of the _ chandra _ observations .
this illustrates that all our observations were conducted a few months ( or even up to a year for ) after or before a period of significant x - ray activity ( outburst ) .
this allows us to monitor the emission properties of and h during their quiescent phases . for all
_ chandra _ observations , we used the advanced ccd imaging spectrometer spectroscopic array ( acis - s ; bautz et al .
1998 ) in imaging mode , with the target placed on one of the back - illuminated acis chips ( s3 ) . for the first observation of in 2002 and the first observation of h , only the s3 chip was read out and a 1/2 sub - array mode was used to limit pile - up . for the later observations ,
the sources were known to be at lower flux and the full acis - s imaging mode array was used .
we produced 0.38 kev acis images using the `` level 2 '' event lists from the standard data processing using the _ chandra _ interactive analysis of observations ( ciao ) software package .
we constructed light curves with all valid events on the s3 chips to search for times of high background .
periods with background flares were removed using the standard ciao script _
analyse_ltcrv.sl _ .
we searched for x - ray sources in each 0.38 kev image using _ wavdetect _ ( freeman et al .
2002 ) , the wavelet - based source detection routine in ciao .
for all _ chandra _ observations , an x - ray source is found at the location of the black hole . with an absolute astrometric precision of 0.6 ( 90% confidence level ) , the _ chandra _ locations are consistent with the positions reported at other wavelengths ( see also corbel et al .
2005 for h ) .
as the focus of this paper is the spectra of the two black holes in quiescence , we refer the reader to the studies of the x - ray jets presented in corbel et al .
( 2002 ; 2005 ) , tomsick et al .
( 2003a ) , kaaret et al .
( 2003 ) as well as the tomsick et al .
( 2001 ) black hole study for details of the _ chandra _ data reduction .
we extracted energy spectra in the 0.38 kev energy range for the black holes in all _ chandra _ observations using ciao tools , and we fitted these spectra using xspec .
we used a circular source extraction region with radii of 1.4and 4 , for h and , respectively .
we extracted background spectra from annuli with inner radii of 9and 6and outer radii of 19and 16 for h and , respectively .
these regions were centered on the black hole positions as given by _
wavdetect_. the source aperture size is adpated to enclose most of the source counts .
background region are adapted to cover a sufficiently large empty region .
due to the low numbers of source counts for h ( 52 counts or less ) , we used the w statistic for fitting ( wachter et al .
1979 ; arnaud , in prep . ) the un - binned spectra .
this is adapted from the cash statistic ( cash 1979 ) and is valid for background subtracted spectra . for , which is brighter than h
, we re - binned the spectra in order to have enough counts in each bin to be able to use the @xmath12 statistic ( with the exception of the _ chandra _ observations on mjd 51777 , 51798 and 52444 , for which we used the w statistic due to the low number of counts ) .
these spectra are adequately fitted with a power - law model including interstellar absorption .
we used this model because this spectral shape is typically seen for black holes in quiescence ( garcia et al .
2001 ; kong et al .
2002 ) , but we emphasize that we can not statistically rule out other spectral shapes .
however , bremsstrahlung models provide un - physical temperatures as already mentioned in kong et al .
( 2002 ) . for h
, we fixed the equivalent hydrogen absorption column density , n@xmath13 to the constant value ( 2.3 @xmath0 10@xmath14 @xmath15 ) measured in _
observations of h during its 2003 outburst ( miller et al .
2005 ) . for
, we left n@xmath13 as a free parameter in our fit .
as the values obtained ( table 1 ) are consistent with results from previous x - ray outbursts ( 9 @xmath0 10@xmath16 @xmath15 ; tomsick et al .
2001 ) , we also fixed the column density of to this value .
this column density is inconsistent with the value inferred from the interstellar absorption lines ( snchez - fernndez et al .
1999 ) , but agrees with the hydrogen column density deduced from x - ray spectra of or its associated x - ray jets ( tomsick et al .
2001 , 2003a ; kaaret et al .
2003 ) , and also with the galactic column density along the line of sight ( jain et al .
the difference when compared to the optical measurement of snchez - fernndez et al .
( 1999 ) may be explained in various ways .
one possibility is that the x - ray absorption column density may be intrinsically higher than the optically - inferred value due to additional local material covering the x - ray source .
another possibility is that the absorption column density deduced from the interstellar absorption lines are artificially lower ( due to saturation for example , hynes et al .
we believe that the latter is more likely as and its x - ray jets ( which are separated by up to 30 ) have similar column densities .
in addition , it is interesting to point out the absence of variation of the hydrogen column density between outburst and quiescence . for h ,
the best - fit photon indices are 1.3@xmath17 , 1.6@xmath18 and 2.2 @xmath5 0.6 for the observations on mjd 53048 , 53089 and 53091 , respectively .
refitting these three data sets simultaneously ( allowing the normalization to vary ) leads to a photon index of 1.96 @xmath5 0.46 ( with 90% confidence errors ) . as there are more frequent observations of ,
its x - ray spectrum is better constrained .
the fit results of the individual observations are reported in table 1 . to provide the best constraint on the spectral parameters , we simultaneously fit all seven datasets simultaneously ( again allowing the normalization to vary ) . leaving the hydrogen column density as a free parameter leads to n@xmath13 = ( 8.8 @xmath5 1.0 )
@xmath19 10@xmath16 @xmath15 , which is consistent with previous estimates . with n@xmath13 frozen to 9 @xmath0 10@xmath16 @xmath15 ( tomsick et al . 2001 ) , we obtain the best constraint on the photon index for in quiescence , @xmath20 = 2.25 @xmath5 0.08 ( with 90% confidence errors ) .
figure 2 shows the 68% , 90% and 99% error contours allowing two parameters ( n@xmath13 and @xmath20 ) to vary .
the results indicate that these parameters are very well constrained by our observations . the photon indices in all _ chandra _ observations of ( figure 3 ) are consistent with the value obtained by fitting the spectrum from the observation with the brightest flux ( on mjd 52345 ) alone .
this indicates that the x - ray spectra of are consistent with the same spectral shape for all the observations , even though the flux varies by a factor of at least 16 .
the fact that the x - ray flux varies between the observations indicates that accretion onto the black hole has not stopped in quiescence .
we use the _ chandra _ observations of on mjd 52344 to construct a light - curve of with a time resolution of 1 ks ( figure 4 ) .
a fit with a constant level leads to a @xmath21 of 74 for 27 degrees of freedom , clearly indicating that there is also significant x - ray variability on short ( @xmath8 ks ) time - scale .
the un - absorbed 0.510 kev fluxes for h and are reported in table 1 . for h ,
we fixed the power - law photon index to a value of 2.0 , whereas for , the photon index was left at its fitted value .
for both sources , the column densities were fixed as indicated previously .
the quoted errors are based on the numbers of source and background counts and poisson statistics . [ cols="^,^,^,<,^,^,^,^ " , ] most of the x - ray novae have not been observed / detected in x - rays in their quiescent state .
however , we note that the softening of their x - ray spectra has been seen at flux level well above quiescence . in the decay phase of their recent outburst , ( corbel et al 2004 ) , as well as xte j1908@xmath22094 ( jonker et al .
2004 ) , are first observed to harden significantly in the hard state ( we note that this is also the case for at intermediate luminosity , possibly related to compton reflection , nowak et al .
however , a clear softening ( @xmath20 = 1.96 @xmath5 0.09 ) , like in , is then observed in in the decay at a luminosity of 2 @xmath0 10@xmath23 erg s@xmath3 ( tomsick et al .
no information is available on the quiescent spectra of or xte j1908@xmath22094 , but it would be interesting to see if they also soften ( as ) in quiescence . similarly , 4u
1543@xmath1147 gradually softened at intermediate luminosity ( @xmath20 = 2.22 @xmath5 0.12 ) , whereas the hardest spectra in the hard state had @xmath20 = 1.64 ( kalemci et al . 2005 ) .
_ xmm - newton _ confirmed this softening with a photon index of @xmath20 = 1.94 @xmath5 0.04 at a luminosity level of 4 @xmath0 10@xmath23 erg s@xmath3 ( 10@xmath24 l@xmath25 ) ( la palombara & mereghetti ( 2005 ) .
4u 1543@xmath1147 was not in its true x - ray quiescence phase ( at the time of the _ xmm - newton _ observation ) which is believed to occur below 3 @xmath0 10@xmath4 erg s@xmath3 ( garcia et al .
however , the optical / infra - red monitoring ( buxton & bailyn 2004 ) indicates that 4u 1543@xmath1147 reached its quiescent optical level only ten days after the _ xmm - newton _ observations . it is unlikely that the x - ray flux could have decayed by three orders of magnitude on timescale of ten days .
this may possibly point out that the optical emission settled down to its quiescent level much before the x - rays .
other examples of similar softening at intermediate luminosity during the decay phase include xte j1748@xmath11288 ( kotani et al .
2000 ) and gs 1124@xmath1168 ( ebisawa et al .
1994 ) .
one of the main results from our study is that a significant fraction of bhs have an x - ray spectrum that softens at lower luminosity .
the emission of quiescent bhs is usually consistent with an extension of the hard state properties to lower luminosities ( corbel et al . 2000 , 2003 ; kong et al .
2000 ; tomsick et al .
2004 ) , and hard state models have been applied to explain the emission in the quiescent state . among these models ,
mcclintock et al . ( 2003 ) explained the power - law photon index of @xmath20 = 2.02 @xmath5 0.16 of ( one of the best studied black holes in quiescence ) with an advection dominated accretion flow ( adaf ) located inside a truncated standard thin disk .
however , a pure adaf model was insufficient to reproduce the x - ray spectrum of and they had to reduce the mass accretion rate close to the black hole .
specifically , they assumed that the mass accretion rate varies as a function of radius as @xmath26 and obtained adequate fits only for @xmath27 .
the fate of the missing mass " is not described in mcclintock et al .
( 2003 ) but could possibly takes the form of an outflow ( e.g. quataert & narayan 1999 ) , which is then like the adios model ( blandford & begelman 1999 ) .
this is also quite similar to yuan et al .
( 2004 ) , who added a jet contribution to the standard adaf model in order to fit the spectral energy distribution of in the hard state . within this context
, an x - ray power - law photon index of @xmath8 2 is predicted for bh in quiescence ( yuan & cui 2005 ) .
however , at low luminosity the accretion flow is known to be convectively unstable ( igumenshchev & abramowicz 1999 ; quataert & gruzinov 2000 ) , and this convection - dominated accretion flow ( cdaf ) has a very different structure than the standard adaf , as convection significantly reduces the mass accretion rate ( equivalent to setting adaf p parameter to 1 in eq . 1 of mcclintock et al .
2003 ) . in this cdaf framework
, the x - ray spectra are expected to soften significantly at very low luminosity ( ball et al .
2001 ) as observed in our study .
so , it is likely that adaf models would be able to reproduce the soft and hard spectra described above , by tuning the amount of outflow or convection with the p parameter . to reproduce the harder spectra of longer orbital period systems , such as v404 cyg , larger values of the @xmath28 parameter , corresponding to stronger outflow or stronger convection , would be required .
alternatively , in standard sphere@xmath22disk model , a softening of the x - ray spectra would also be expected due to a decrease of the coronal optical depth as the mass accretion rate decreases ( see , for example , the discussion in tomsick et al .
2004 ) . in the magnetic corona model ,
active regions above the disk are responsible for producing the hard x - ray emission and constitute the base of the outflow ( merloni & fabian 2002 ) . in that model
, the x - ray spectrum is expected to soften at lower accretion rate as the accretion power is carried away into the jets rather that being used to heat the electrons ( merloni & fabian 2002 ) .
such a model could easily explain the soft spectra of our sources in quiescence .
however , in this model , the inner boundary of the accretion disk does not vary with mass accretion rate .
the characteristic frequencies of qpos and of broad - band timing noise are known to vary at state transitions and in the hard state ( tomsick et al .
2004 ; kalemci et al .
2005 ) , with lower frequencies generally occurring at lower luminosities .
this is easy to understand in models where the inner disk radius increases at low mass accretion rates , but is difficult to explain in the models where the inner disk radius is constant .
alternatively , a new possibility for the origin of x - ray emission has emerged in recent years . indeed , a strong correlation is observed between the radio and x - ray emission of and v404 cyg in the hard state , and this correlation seems to extend down to the quiescent level for these two sources ( corbel et al .
2003 ; gallo et al . 2003 ) .
the radio emission originates from a powerful self - absorbed compact jet ( corbel et al .
2000 ) , and it also seems possible that this jet may contribute to the high energy emission ( markoff et al .
2001 , 2003 ) of bh in the standard hard state .
furthermore , radio observations show that the compact jet ( at least for v404 cyg ) is maintained in quiescence ( gallo et al .
2004 ) . even if the debate is still open regarding the contribution of the compact jet ( markoff & nowak 2004 , homan et al .
2005 , markoff et al .
2005 ) at high energy , we can check if the jet model could explain the softening of x - ray spectra for bh in quiescence . to explain the x - ray spectra in the hard state
, the electron energy distribution should have an energy index of the order of 2.22.4 ( markoff et al .
if the electrons that could be responsible for the x - ray emission of bh in quiescence are above the cooling break due to spectral ageing , then the energy index of the electron distribution ( above the break ) would be 3.23.4 , therefore increasing ( by 0.5 ) the photon index to @xmath29 2.1 as observed in a large part of the sources from our sample . in this picture , the reason for the change would be that the particles are not sufficiently re - accelerated at low accretion rate , which is indeed quite reasonable .
alternatively , the compact jet contribution may vanish at low luminosity giving the possibility that thermal emission from the nozzle dominates the x - ray band and therefore gives a different spectral shape .
the hard spectra ( @xmath29 1.6 ) would easily be explained if the compact jet properties are similar to the standard hard state .
this is consistent with the fact that these kind of spectra are observed for the sources ( v404 cyg , and v4641 sgr ) that have a large accretion disk , and which may be a necessary condition to sustain a powerful compact jet .
similarly , it is of some interest to note that acceleration processes are also believed to be inefficient in the jets of the low luminosity ( 10@xmath30 l@xmath25 ) supermassive black hole , sgr a@xmath31 , located in the center of our galaxy ( melia & falcke 2001 ) .
we have observed and detected and h during their quiescent states at their faintest level of x - ray emission with a 0.510 kev luminosity of 2 @xmath0 10@xmath1 ( d/5 kpc)@xmath2 erg s@xmath3 for and 9 @xmath0 10@xmath4 ( d/8
kpc)@xmath2 erg s@xmath3 for h. such levels of x - ray emission are in the upper range compared with the levels observed in other black hole systems .
the _ chandra _ observations also provide the best constraint on the power - law index of with an index of 2.25 @xmath5 0.08 .
we focused our analysis on the power - law model as this spectral shape is typically seen for black holes in quiescence and can represent an approximation for most theoretical emission mechanisms , but we emphasize that we can not statistically rule out other spectral shapes .
all _ chandra _ spectra of are consistent with a soft power - law ; therefore , indicating that the x - ray spectrum softens at lower luminosity .
we bring to light that all systems with short orbital period ( @xmath8 @xmath32 60 hours ) are consistent with a softening of their x - ray spectra in quiescence
. however , the long orbital period systems may be consistent with a hardening of their x - ray spectra , but confirmation of this trend is required
. a possible and realistic test would be to obtain _ xmm - newton _ or _ chandra _ observations of long orbital period systems , like v4641 sgr or with a long exposure during quiescence .
in addition , , with an orbital period of 833 hours , may be a very good target if it returns to quiescence .
simultaneously , further observations of short orbital period systems ( like , or 4u 1543@xmath1147 at lower luminosity ) should be performed in order to test the softening of their spectra .
we found that various classes of models ( adaf corona + jet , cdaf , sphere@xmath22disk , magnetic corona or jet models ) are able to reproduce the softening of the spectra in quiescence , but we note that most of them need the presence of powerful outflow or significant convection in order to reproduce these soft x - ray spectra .
this may increase the likelihood that outflows are present in the most frequent phase ( quiescence ) of black hole binary s activity and have significant influence on the physics of these systems and neighbooring environment ( fender et al .
2003 , 2005 ) . s.c . would like to thank h. falcke , r. fender , a. king , e. kording , and s. markoff for discussions on various aspects of bh emission models .
rxte_/asm results are provided by _
rxte_/asm team at mit .
we thank the anonymous referee for constructive comments .
pk acknowledges partial support from nasa chandra grants go4 - 5038 and go4 - 5039 and from a university of iowa faculty scholar award .
jat acknowledges partial support from nasa chandra grant go3 - 4041x .
|
we report the results of observations of the black hole binaries and h in their quiescent state using the _ chandra x - ray observatory_. both sources are detected at their faintest level of x - ray emission ever observed with a 0.510 kev unabsorbed luminosity of 2 @xmath0 10@xmath1 ( d/5 kpc)@xmath2 erg s@xmath3 for and 9 @xmath0 10@xmath4 ( d/8
kpc)@xmath2 erg s@xmath3 for h. these luminosities are in the upper range compared to the faintest levels observed in other black hole systems , possibly related to residual accretion for these sources with frequent outbursts . for
, the _ chandra _ observations also constrain the x - ray spectrum as a fit with an absorbed power - law model yields a photon index of 2.25 @xmath5 0.08 , clearly indicating a softening of the x - ray spectrum at lower luminosities compared to the standard hard state .
similar softening at low luminosity is seen for several black hole transients with orbital periods less than 60 hours .
most of the current models of accreting black holes are able to reproduce such softening in quiescence .
in contrast , we find that systems with orbital periods longer than 60 hours appear to have hard spectra in quiescence and their behaviour may be consistent with hardening in quiescence .
|
let s arrange @xmath4 base pairs ( _ bps _ ) on a circle with radius @xmath5 such that , @xmath6 , as depicted in fig .
[ fig:1 ] . when all _ bps _ centers of mass lie on the circumference , which represents the molecule backbone , the system is in the ground state .
say @xmath7 , the inter - strand fluctuation for the _
i_-base pair ( @xmath8 ) with respect to the ground state .
we define the vector @xmath9 : @xmath10 the ground state is recovered once all _ bps_-fluctuations vanish hence , @xmath11 .
the polar angle , @xmath12 , measures the @xmath13 _ bp _ twisting around the molecule backbone , with @xmath14 being the number of _ bps _ per helix turn and @xmath15 is the twist number accounting for the coiling of the individual strands around the helical axis @xcite .
the azimuthal angle , @xmath16 , defines the bending between adjacent _ bps _ along the stack . as the polynucleotide chain has a direction due to the chemistry of the intra - strand bonds , a distribution of values for the twist ( @xmath17 ) and the bending ( @xmath18 ) of the first _ bp _ in the sequence is weighed in the computation .
the fluctuational orbits defined by @xmath19 and @xmath20 overlap consistently with the closure condition holding for the dna ring .
my previous path integral analysis of dna @xcite were based on the ansatz that the _ bps _ displacements could be treated as one dimensional paths @xmath21 , @xmath22 , with the imaginary time @xmath23 $ ] and @xmath24 being the inverse the temperature . here
i introduce a more general ( albeit more cpu time consuming ) space - time mapping technique , which does not pin a base pair to a specific @xmath25 thus avoiding the somewhat arbitrary partition of the @xmath24 length in @xmath4 intervals : @xmath26 the paths @xmath27 are expanded in fourier series , the constraint @xmath28 for a dna loop is implemented and , for the model in fig .
[ fig:1 ] , the _ bps _ fluctuations in the path integral formalism are given by : @xmath29^{1/2 } \ , \nonumber \\ & & f(\theta_i,\phi_i)=\,\sin\theta_i \cos^2 \phi_i + \sin^2 \phi_i \ , .
\ , \label{eq:006}\end{aligned}\ ] ]
this new mapping technique lets @xmath30 as an integration variable and permits to set a realistic rise distance between adjacent nucleotides along the stack . written in terms of the @xmath31 ,
the hamiltonian contains : _
i ) _ a morse potential , @xmath32 $ ] , describing the effective hydrogen bond interaction between _ bps _ mates , _ ii ) _ a solvent term , @xmath33 $ ] , accounting for hydrogen bond recombination with the counterions dissolved in water , _ iii ) _ a ( two particles ) stacking potential , @xmath34 $ ] , between adjacent bases along the strand .
the potential parameters are widely discussed in refs .
@xcite . accordingly , the classical action @xmath35 $ ] is a @xmath30-integral of the base pair _
kinetic _ plus _ potential _ energies @xcite and the partition function @xmath36 for a sequence with @xmath4 _ bps _ reads : @xmath37 \bigr\}\,\ , \nonumber \\ & & a[\eta_{i } ] = \ , \sum_{i=1}^{n } \int_{0}^{\beta } d\tau \biggl [ \frac{\mu}{2 } \dot{\eta } _
i^2(\tau ) + v_m[\eta_i(\tau ) ] + \,v_{sol}[\eta_i(\tau ) ] + v_{s}\bigl [ \eta _
i(\tau ) , \eta _ { i-1}(\tau ) \bigr ] \biggr ] \ , , \ , \label{eq:7}\end{aligned}\ ] ] where @xmath38 is the _ bp _ reduced mass .
the measure @xmath39 is a multiple integral over the path fourier coefficients while @xmath40 indicates that the paths @xmath27 are closed trajectories .
the integration over the two particles potential greatly enhances the computational times with respect to the previous method @xcite but it offers a more realistic model for the base stacking in sequences of any length .
there has been a widespread interest in the properties of small sequences of dna following measurements of high cyclization probabilities which have pointed to an intrinsic flexibility for fragments of @xmath41 _ bps _ or less @xcite .
it has been suggested that the formation of local disruptions in the helix of dna loops may be the mechanism which permits to release the torsional stress and energetically favors the stability of bent molecules .
certainly , the breaking of some _ bps _ and the opening of fluctuational bubbles change the average pitch of the helix , that is the number of _ bps _ per helix turn . in the theoretical framework synthesized by eq .
( [ eq:7 ] ) , i compute the free energy ( @xmath42 ) of three heterogeneous loops with different length , @xmath43 , but similar content of at-_bps _ and gc-_bps _ , @xmath44 each .
the rise distance is pinned to the experimental value .
the @xmath45 and @xmath46 _ bps _ loops have been prepared as described in ref.@xcite whereas the @xmath47 _ bps _ is a toy sequence here introduced for comparison . for each loop
, we simulate a broad set of values for the twist @xmath15 and , for any @xmath15 ( i.e. , @xmath48 ) , @xmath36 is calculated by summing over an ensemble of fluctuational paths representing a large number of molecule configurations , about @xmath49 paths for each base pair in eq .
( [ eq:7 ] ) .
hence , the free energy is obtained as a function of the helical repeat @xmath48 .
the computational time for a simulation , e.g. for the @xmath50 sequence , is about eight days on a workstation ( intel xeon e5 - 1620 v2 , 3.7ghz processor ) .
the room temperature results for @xmath51 are plotted in fig .
[ fig:2 ] .
while the shortest loop shows a free energy minimum also at @xmath52 , the minima for all loops are remarkably found for @xmath48 in the range @xmath53 in line with the well known values of helical pitch in dna sequences @xcite .
even more interestingly , the @xmath45 and @xmath46 minicircles also show free energy minima at larger @xmath48 due to a spontaneous unwinding of the complementary strands .
this @xmath15 reduction is consistent with the observations @xcite of _ bps _ disruptions occuring in small loops as a consequence of the strong rotational deformations of the helix .
the bending stress decreases by increasing the radius of the loop .
then , for an ensemble of molecules with @xmath4 _ bps _ , the free energy minimization evaluates the thermodynamically stablest values of helical repeat , that is an average property of the molecule ensemble .
furthermore , the path integral method can also determine the probabilities for the formation of fluctuational bubbles and select those base pairs along the dna sequence for which hydrogen bond breaking is more likely to occur .
by tuning the system temperature and the input parameters which control the counterions concentration in the solvent , we thus obtain a general and reliable computational scheme for the modeling of heterogeneous dna loops in various ambient conditions .
|
the thermodynamical stability of dna minicircles is investigated by means of path integral techniques .
hydrogen bonds between base pairs on complementary strands can be broken by thermal fluctuations and temporary fluctuational openings along the double helix are essential to biological functions such as transcription and replication of the genetic information .
helix unwinding and bubble formation patterns are computed in circular sequences with variable radius in order to analyze the interplay between molecule size and appearance of helical disruptions .
the latter are found in minicircles with @xmath0 base pairs and appear as a strategy to soften the stress due to the bending and torsion of the helix .
it is known that the helicoidal conformation of dna is essentially determined by the hydrophobicity of purine and pyrimidine bases and by the bond angles in the flexible sugar - phosphate backbone while sequence of the bases and environmental conditions due to the solvent also contribute to the molecule shape @xcite . as each strand bears a negative charge ( @xmath1 ) for each phosphate group and the rise distance between adjacent nucleotides
is @xmath2 , the bare double helix has a high linear charge density of @xmath3 .
although the effective charge density is reduced by the counterions in the solvent , the electrostatic strands repulsion is key to the stability of the helix and also affects the inter - helical chiral interactions in those condensed phases of dna assemblies ( such as liquid crystals ) which underlie the impressive growth of dna - based structures recently witnessed in materials science .
base pairing and base stacking are the fundamental interactions which control the synthesis of dna and determine the thermodynamic stability both of single helices and of helix aggregates . however
even stable duplexes at room temperature show local openings , temporary bubbles , which are intrinsic to the biological functioning as they permit the transcription and replication of the genetic code .
such bubbles are due to the strong fluctuational effects on the hydrogen bonds between complementary strands and cause the local unwinding of the double helix which ultimately leads to a state of negative supercoiled dna for almost all living beings .
while these processes are qualitatively understood , quantitative predictions of ( energetically ) optimal helical configurations for specific systems are scarce .
we contribute to fill this gap by introducing a new path integral computational method which readily applies to loops of dna as those found in bacterial plasmids , viral genomes and also mammalian cells .
|
the last decade has seen the emergence of the study of large - scale complex networks such as the web graph consisting of web pages and the links between them , the friendship network on facebook , and networks of interacting proteins in a cell .
several new random graph models were proposed for such networks , and existing models were modified in order to fit the data gathered from these real - life networks .
see the books @xcite for surveys of such models . a recent trend in stochastic graph modelling is the study of _ geometric graph models_. in geometric graph models , vertices are embedded in a metric space , and the formation of edges is influenced by the relative position of the vertices in this space .
geometric graph models have found applications in modelling wireless networks ( see @xcite ) , and in modelling the web graph and other complex networks ( @xcite ) . in real - world networks ,
the underlying metric space is a representation of the hidden reality that leads to the formation of edges .
thus , for the world wide web , web pages are embedded in a high dimensional _
topic space _ , where pages that are positioned close together in the space contain similar content .
the graph model we study is a variation on the _ random geometric graph _ , where vertices are chosen at random according to a given probability distribution from a given metric space , and two vertices are adjacent if the distance between the two vertices is no larger than some fixed real number .
random geometric graphs have been studied extensively in their own right ( see , for example , @xcite and the book @xcite ) .
analysis of stochastic graph models usually focusses on asymptotic results , which hold for cases where the number of vertices is sufficiently large .
an alternative approach is to study the
_ infinite limit _ ; that is , the infinite graph that results when the number of vertices reaches infinity . studying
the infinite limit is a well - known tool for studying scientific models , and it can help to recognize large - scale structure and long - term behaviour ( see @xcite ) . in this paper , we study the infinite limit of a geometric graph model that is a geometric extension of the classic random graph model @xmath8 .
one of the most studied examples of an infinite limit graph arising from a stochastic model is the infinite random graph .
the probability space @xmath8 consists of graphs with vertices @xmath9 so that each distinct pair of integers is adjacent independently with a fixed probability @xmath10 erds and rnyi @xcite discovered that with probability @xmath11 all @xmath12 are isomorphic .
a graph @xmath13 is _ existentially closed _ ( or _ e.c .
_ ) if for all finite disjoint sets of vertices @xmath14 and @xmath15 ( one of which may be empty ) , there is a vertex @xmath16 adjacent to all vertices of @xmath14 and to no vertex of @xmath17 we say that @xmath18 is _ correctly joined _ to @xmath14 and @xmath17 the unique isomorphism type of countably infinite e.c .
graph is named the _ infinite random graph _ , or the _ rado _ graph , and is written @xmath19 see the surveys @xcite for additional background on @xmath19 we now introduce the geometric graph model on which this paper is based . consider a metric space @xmath20 with distance function @xmath21 @xmath22 , a countable subset @xmath0 of @xmath20 , and @xmath1 .
the _ local area random graph @xmath23 _ has vertices @xmath24 and for each pair of vertices @xmath25 and @xmath26 with @xmath27 , an edge is added independently with probability @xmath28 .
note that @xmath0 may be either finite or infinite .
the larg model generalizes well - known classes of random graphs .
for example , special cases of the @xmath29 model include the random geometric graphs ( where @xmath30 ) , and the binomial random graph @xmath31 ( where @xmath20 has finite diameter @xmath32 , and @xmath33 ) . in the case
@xmath0 is infinite , our goals are to investigate what adjacency properties are satisfied by graphs generated by the larg model , and determine when the model generates a unique isomorphism type of countable graph .
we prove that with probability @xmath4 , graphs in @xmath23 satisfy a certain adjacency property which is a metric analogue of the e.c .
property ; see theorem [ random ] . the so - called geometric e.c .
property requires that vertices @xmath18 correctly joined to @xmath14 and @xmath15 may be found as close as we like to points in @xmath34 explicit examples of graphs with the geometric e.c .
property are given in theorem [ explicit ] .
for metric spaces such as @xmath2 with the @xmath3-metric , the geometric e.c .
property gives rise to a unique isomorphism type of graph ; see theorems [ thm:1d ] , [ thm:1disometry ] , and [ thm : nd ] . the main tool here is a generalization of isometry called a step - isometry .
our results are sensitive to the metric used .
non - isomorphism results for other metrics are given in section [ none ] .
in particular , we show in theorem [ non2 ] that in @xmath7 with the euclidean metric , there exist non - isomorphic geometric e.c
. graphs .
all graphs considered are simple , undirected , and countable unless otherwise stated .
if @xmath20 is a set of vertices in @xmath13 , then we use the notation @xmath35 $ ] for the subgraph of @xmath13 induced by @xmath36 we use the notation @xmath37 if @xmath13 is an induced subgraph of @xmath38 we refer to an isomorphism type as _ isotype _ , and denote isomorphic graphs by @xmath39 given a metric space @xmath20 with distance function @xmath32 , define the ( open ) _ ball of radius @xmath40 around @xmath41 _ by @xmath42 we will sometimes just refer to @xmath43 as a _ ball_. a subset @xmath0 is _ dense _ in @xmath20 if for every point @xmath44 , every ball around @xmath41 contains at least one point from @xmath0 .
we refer to @xmath45 as points or vertices , depending on the context . throughout ,
let @xmath46 denote the non - negative natural numbers , and @xmath47 denote the positive natural numbers .
for a reference on graph theory the reader is directed to @xcite , while @xcite is a reference on metric spaces .
as noted in the introduction , the unique isotype of the infinite random graph @xmath48 is characterized by the e.c . property . in this section
we define a geometric analogue of this property .
as will be demonstrated in section 3 , this property characterizes the unique isotype of graphs obtained from countable dense sets in @xmath49 , provided we consider a particular metric .
let @xmath50 be a graph whose vertices are points in the metric space @xmath20 with metric @xmath51 the graph @xmath13 is _ geometrically e.c_. _ at level @xmath40 _ ( or @xmath40-_g.e.c .
_ ) if for all @xmath52 so that @xmath53 , for all @xmath54 , and for all disjoint finite sets @xmath14 and @xmath15 so that @xmath55 , there exists a vertex @xmath56 so that * @xmath18 is correctly joined to @xmath14 and @xmath15 , * for all @xmath57 , @xmath58 , and * @xmath59 .
this definition implies that @xmath0 is dense in itself . also , if @xmath13 is @xmath40-e.c .
, then @xmath13 is @xmath52-e.c . for any @xmath60 . the geometrically e.c .
property bears clear similarities with the e.c . property .
the important differences are that a correctly joined vertex must exist only for sets @xmath14 and @xmath15 which are contained in an open ball with radius @xmath40 and centre @xmath41 , and it must be possible to choose the vertex @xmath18 correctly joined to @xmath14 and @xmath15 arbitrarily close to @xmath41
. see figure 1 .
[ random ] let @xmath61 be a metric space and @xmath0 a countable subset of @xmath20 which is dense in itself .
if @xmath62 and @xmath63 then with probability @xmath11 @xmath23 is @xmath40-e.c .
fix @xmath54 , disjoint finite subsets @xmath14 and @xmath15 in @xmath64 , and @xmath53 .
let @xmath65 then @xmath66 . let @xmath67 . consider the set @xmath68 note that @xmath69 is chosen so that for any @xmath70 , @xmath71 , and for all @xmath72 , @xmath73 for any graph @xmath13 in @xmath23 , the probability that any vertex @xmath70 is correctly joined to @xmath14 and @xmath15 equals @xmath74 .
the probability that no vertex in @xmath75 is correctly joined to @xmath14 and @xmath15 equals @xmath76 since @xmath0 is dense in itself , @xmath75 contains infinitely many points ; hence , @xmath77 as there are only countably many choices for @xmath41 , @xmath14 , and @xmath15 , and a countable union of measure @xmath78 sets is measure @xmath78 , the proof follows . a graph @xmath50 whose vertices are points in the metric space @xmath61 has _ threshold _
@xmath40 if for all edges @xmath79 , @xmath27 .
a graph that is geometrically g.e.c
. at level @xmath40 and has threshold @xmath40 is called a _
geometric @xmath40-graph . _ by definition , a graph @xmath13 generated by @xmath23 has threshold @xmath40 , and , if @xmath0 is countable and dense in itself , @xmath13 is a geometric @xmath40-graph .
thus , this random graph model generates geometric @xmath40-graphs .
we construct geometric @xmath40-graphs deterministically as follows .
given @xmath62 , a countable set @xmath0 which is dense in itself , and a linear ordering @xmath80 of @xmath0 , define @xmath81 as the limit of a chain of finite graphs @xmath82 , where @xmath83 for any @xmath84 , and @xmath85 .
let @xmath86 be the trivial graph with vertex set @xmath87 .
assume that @xmath82 is defined and @xmath88 we now define @xmath89 enumerate all pairs @xmath90 so that @xmath91 and @xmath92 so that @xmath93 , via a lexicographic ordering based on @xmath94 .
for each pair @xmath90 , in order , choose @xmath95 to be the least index point in @xmath0 ( according to @xmath94 ) such that @xmath18 has not been chosen for any previous pairs @xmath90 , @xmath96 and @xmath97 note that such a vertex exists as @xmath0 is dense and @xmath82 is finite .
join @xmath18 to all vertices in @xmath14 and to no other vertices of @xmath82 .
if necessary , add @xmath98 as an isolated vertex to form the graph @xmath89 observe that by ( [ ca ] ) , @xmath81 is a @xmath40-threshold graph .
[ explicit ] the graph @xmath81 is @xmath40-g.e.c . to show that @xmath99 is @xmath40-g.e.c .
, choose @xmath53 , a vertex @xmath100 and disjoint sets @xmath101 so that @xmath102 .
let @xmath103 be chosen so that @xmath104 and @xmath105 .
let @xmath95 be the vertex in @xmath106 added to @xmath82 to extend @xmath90 .
then @xmath18 is correctly joined to @xmath14 and @xmath15 and @xmath59 .
it should be emphasized that we do not claim that the graphs @xmath81 are the unique isotypes of @xmath40-g.e.c .
graphs with vertex set @xmath34 the theme of when two @xmath40-g.e.c .
graphs are isomorphic will be explored in the next two sections .
balls with radius @xmath40 in @xmath40-g.e.c .
graphs contain copies of @xmath48 , and hence , contain isomorphic copies of all countable graphs .
[ subset ] let @xmath107 be so that @xmath108 for some @xmath109 then a @xmath40-g.e.c . graph with vertex set @xmath110 is e.c . , and
so is isomorphic to @xmath48 .
let @xmath13 be a graph with vertex set @xmath110 , where @xmath110 is as stated .
assume that @xmath13 is @xmath40-g.e.c .
let @xmath14 and @xmath15 be any pair of disjoint , finite subsets of @xmath111 , and let @xmath53 . then by the @xmath40-g.e.c .
condition there exists a vertex @xmath112 so that @xmath18 is correctly joined to @xmath14 and @xmath15 .
the converse of theorem [ subset ] is false , in general .
for example , consider the metric space @xmath113 , where @xmath32 is the euclidean metric , and let @xmath114 .
fix @xmath110 an infinite clique in @xmath48 , and let @xmath115 .
embed the vertices of @xmath110 in @xmath116 so that they form a set that is dense in @xmath117 .
embed the vertices of @xmath118 so that they form a set that is dense in @xmath119 .
now choose @xmath120 so that @xmath121 , and let @xmath122 , and @xmath123 , where @xmath124 .
let @xmath125 .
note that @xmath126 .
the embedding of the vertices of @xmath48 is such that all vertices in @xmath127 are in @xmath110 , so they are all adjacent to @xmath128 .
thus , @xmath127 does not contain any vertex correctly joined to @xmath14 and @xmath15 , and thus , this embedding of @xmath48 is not @xmath40-g.e.c .
we finish with the following theorem which shows that there exists a close relationship between graph distance and metric distance in any graph that is @xmath40-g.e.c .
we denote the closure of set @xmath0 in @xmath20 by @xmath129 .
the set @xmath130 is _ convex _ if for every pair of points @xmath41 and @xmath131 in @xmath132 there exists a point @xmath18 such that @xmath133 [ mot ] let @xmath50 be geometric @xmath40-graph , and let @xmath129 be convex .
let @xmath134 so that @xmath135
. then the graph distance between @xmath25 and @xmath26 in @xmath13 equals @xmath136 .
theorem [ mot ] directly leads to the following corollary , which supplies motivation for proofs of the isomorphism results in the next section , and will be used to prove the non - isomorphism results of the final section .
[ useful ] if @xmath129 and @xmath137 are convex , and there is a @xmath40-g.e.c .
graph with vertices @xmath0 and a @xmath138-g.e.c . graph with vertex set @xmath130
which are isomorphic via @xmath139 , then for every pair of vertices @xmath134 , @xmath140 we supply a generalization of isometry , motivated by theorem [ mot ] and corollary [ useful ] .
given metric spaces @xmath141 and @xmath142 , sets @xmath143 and @xmath144 , and positive real numbers @xmath40 and @xmath138 , a _ step - isometry at level @xmath145 _ from @xmath0 to @xmath130 is a surjective map @xmath146 with the property that for every pair of vertices @xmath147 @xmath148 every isometry is a step - isometry , but the converse is false , in general . for example , consider @xmath116 with the euclidean metric , let @xmath149 , and let @xmath150 and @xmath151 .
then @xmath152 given by @xmath153 is a step - isometry , but is not an isometry .
let @xmath134 .
let @xmath154 . by assumption , @xmath155 .
note that the choice of @xmath156 supplies that @xmath157 let @xmath158 be the graph distance of @xmath25 and @xmath159 and note that @xmath160 since @xmath13 is a @xmath40-threshold graph . to show that @xmath161 , let @xmath162 , where @xmath163 , @xmath164 , be a shortest path in @xmath13 from @xmath25 to @xmath26 . since @xmath13 has threshold @xmath40 , @xmath165 for @xmath166 .
therefore , @xmath167 and so @xmath168 .
next , we show how to construct a path of length @xmath156 from @xmath25 to @xmath26 in @xmath13 , which will prove that @xmath169 .
let @xmath170 , so @xmath171 .
the set @xmath129 is convex , so for every pair of vertices @xmath172 , there exists a point @xmath173 so that @xmath174 . using this property
, we can obtain a sequence of points in @xmath129 between @xmath25 and @xmath26 whose successive distances add up to @xmath175 , and which are at most @xmath176 apart . we can then choose vertices @xmath177 from this sequence so that @xmath178 for @xmath179 , where @xmath180 and @xmath181 . for @xmath182
, we may then find @xmath183 so that @xmath184 .
letting @xmath185 and @xmath186 , we have that for @xmath179 , @xmath187 let @xmath188 .
now we successively apply the @xmath40-e.c .
property to choose @xmath189 so that * @xmath190 , and * @xmath191 is adjacent to @xmath192 in @xmath13 .
it then follows that @xmath193 is the desired path of length @xmath194 more precisely , fix @xmath195 , @xmath196 , and assume @xmath192 exists so that item @xmath197 holds .
note that ( @xmath195 ) and ( [ cha ] ) implies that @xmath198 therefore , @xmath199 , and so we can find a vertex @xmath200 in @xmath201 which is adjacent to @xmath192 . to choose the vertex @xmath202 ,
let @xmath203 . by the same argument as before , @xmath204 .
since @xmath205 , @xmath206 .
so @xmath207 .
therefore , there exist a vertex @xmath202 which is adjacent to @xmath208 and @xmath209 .
in this section we consider metric spaces where the geometric e.c .
property gives a unique isotype of graph .
we work in the space @xmath116 with the usual metric defined by @xmath210 .
( we will not mention this explicitly unless there is room for confusion . )
the first result of the section which serves as the template for more general results is the following .
[ thm:1d ] let @xmath0 and @xmath130 be two countable dense subsets of @xmath116 , and let @xmath211 .
if @xmath13 is a geometric @xmath40-graph with vertex set @xmath0 and @xmath212 is a geometric @xmath138-graph with vertex set @xmath132 then @xmath39 the proof of the theorem ( and others analogous to it in this section ) build up an isomorphism as a step - isometry . in the proofs we use the following alternative characterization of step - isometries .
fix @xmath62 and @xmath213 .
each @xmath214 may be uniquely represented as @xmath215 where @xmath216 and @xmath217 . in this representation
, we will refer to @xmath40 as the _ offset _ , and to @xmath218 as the _ anchor_. we will omit to state the anchor and offset explicitly wherever it is clear from the context .
the term @xmath219 is called the _ representative _ of @xmath26 and @xmath220 the _
quotient_. [ lemma : isometry_conditions ] let @xmath0 and @xmath130 be subsets of @xmath116 , and let @xmath40 and @xmath138 be two non - negative real numbers .
a surjective function @xmath221 is a step - isometry at level @xmath145 if and only if the following two conditions hold . 1 . for every @xmath134 ,
@xmath222 if and only if @xmath223 2 . for every @xmath224 , @xmath225 , where the representation of elements of @xmath0 has offset @xmath40 and that of @xmath130 has offset @xmath138 , and the anchor of the representation of @xmath130 is the image of the anchor of the representation of @xmath0 under @xmath139 .
assume first that items ( 1 ) and ( 2 ) hold , and fix @xmath134 .
let @xmath226 and @xmath227 .
assume without loss of generality that @xmath228 ; by hypothesis , this implies that @xmath229 .
then @xmath230 , and so @xmath231 hence , @xmath232 where @xmath233 if @xmath234 , and @xmath235 otherwise . similarly , @xmath236 where @xmath237 if @xmath238 , and @xmath239 otherwise . by hypothesis
, we have that @xmath240 and @xmath241 .
it follows that @xmath139 is a step - isometry at level @xmath145 .
now assume that @xmath139 is a step - isometry at level @xmath145 .
let @xmath242 and @xmath243 be so that @xmath244 and consider the representations of elements of @xmath0 and @xmath130 with offsets @xmath40 and @xmath138 , and anchors @xmath218 and @xmath245 , respectively .
condition ( 2 ) follows immediately from the definition of step - isometry . for the proof of ( 1 ) ,
fix any @xmath134 , and let @xmath226 and @xmath227 .
then @xmath246 where @xmath247 ; similarly , @xmath248 , where @xmath249 .
since @xmath139 is a step - isometry , @xmath240 . if @xmath250 , then both @xmath251 and @xmath252 ; if @xmath253 , then @xmath254 and @xmath255 .
thus item ( 1 ) holds .
the proof follows using a variant of the back - and - forth method ( used to show that @xmath48 is the unique isotype of e.c .
let @xmath256 and @xmath257 . for @xmath258 , we inductively construct a sequence of pairs of sets @xmath259 and isomorphisms @xmath260\rightarrow h[w_i]$ ] , so that for all @xmath261 , @xmath262 , @xmath263 , @xmath264 and @xmath265 , and @xmath266 extends @xmath267 .
it follows that @xmath268 is an isomorphism . as an additional induction hypothesis we require that @xmath267 is a step - isometry from @xmath269 to @xmath270 at level @xmath145 . specifically , we maintain conditions in items ( 1 ) and ( 2 ) from lemma [ lemma : isometry_conditions ] , where the representation of elements of @xmath0 has offset @xmath40 and anchor @xmath218 , and the representation of elements of @xmath130 has offset @xmath138 and anchor @xmath245 .
let @xmath271 , @xmath272 and define @xmath273 by @xmath274 then @xmath275 and @xmath276 , so the base case of the induction follows .
for the induction step , fix @xmath277 . to construct @xmath266 from @xmath278 we first _ go forth _ by finding an image of @xmath279 . in the following
, @xmath139 refers to @xmath278 and @xmath280 . define @xmath281
we claim that @xmath282 .
namely , let @xmath283 and @xmath284 be the elements in @xmath269 for which the maximum and minimum that define @xmath285 and @xmath128 are attained , respectively .
thus , @xmath286 and @xmath287 . by definition ,
@xmath288 . by the induction hypothesis ( specifically , item ( 1 ) from lemma [ lemma : isometry_conditions ] ) , this implies that @xmath289 . in order to maintain the induction hypothesis ,
@xmath290 must lie in @xmath291 , and @xmath292 must equal to @xmath220 .
let @xmath293 , and consider the interval @xmath294 any vertex in @xmath295 will qualify as a candidate for @xmath296 so that @xmath266 is a step - isometry at level @xmath297 we must then find a vertex in @xmath295 that will also guarantee that @xmath139 is an isomorphism , by making sure it has the correct neighbours .
for this , we apply the @xmath138-g.e.c .
condition of @xmath212 . in order to apply the @xmath138-g.e.c .
condition , we need to ensure that the images of all neighbours of @xmath26 in @xmath269 lie in a @xmath138-ball .
since @xmath13 has threshold @xmath40 , we consider all vertices of @xmath269 that lie in a @xmath40-ball around @xmath26 .
let @xmath298 , and fix @xmath299 .
such a vertex @xmath41 exists since @xmath130 is dense in @xmath116 . by definition of @xmath295 ,
we claim that @xmath301 to prove this , let @xmath302 since @xmath303 and @xmath27 , it follows that @xmath304 .
hence , @xmath305 is one of @xmath306 @xmath307 or @xmath308 if @xmath309 then @xmath310 by induction hypothesis , so @xmath311 . if @xmath312 , then @xmath313 , so @xmath314 by definition of @xmath128 .
hence , @xmath315 the final case is when @xmath316 then @xmath317 , so we have that @xmath318 in all cases , @xmath319 and ( [ moo ] ) follows . since @xmath13 has threshold @xmath40 , @xmath320 .
now let @xmath321 and @xmath322 .
then @xmath323 . let @xmath324 be chosen such that @xmath325 .
we now use the @xmath138-e.c .
property of @xmath212 to find a point @xmath326 which is adjacent to all vertices in @xmath14 and no other vertices of ( the finite set ) @xmath270 .
thus , we can add @xmath18 to @xmath270 to form @xmath327 and add @xmath26 to @xmath269 to form @xmath328 , and set @xmath329 .
observe that @xmath266 is an isomorphism .
to finish the induction step , if @xmath330 then we may _ go back _ , by finding an image @xmath331 in an analogous fashion .
we then add @xmath18 to @xmath328 , and maintain that @xmath266 is an isomorphism .
the proof of the following corollary is now immediate .
[ cor ] for all countable dense subsets @xmath0 of @xmath116 , @xmath332 , and @xmath333 with probability @xmath4 , there is a unique isotype of graph , written @xmath334 in @xmath23 .
the isomorphism type of @xmath335 does not depend on the choices of @xmath0 , @xmath40 or @xmath28 ; moreover , the same result holds for any 1-dimensional normed vector space with the metric derived from the norm . for this reason
, we name @xmath335 the _ infinite random geometric graph of dimension 1_. note that @xmath335 has infinite diameter ( unlike @xmath48 , which has diameter @xmath336 ) .
note that , for any countable set @xmath337 , any ordering @xmath94 of @xmath26 , and any real @xmath62 , the deterministic construction process @xmath338 described in the previous section gives explicit representations of @xmath335 .
we may extend theorem [ thm:1d ] to sets that are not necessarily dense in all @xmath116 .
the only additional condition required is that there exists a step - isometry between the two sets . for example , consider the rational intervals @xmath339 and @xmath340 , where @xmath341 .
consider the bijective map @xmath146 defined by @xmath342 where @xmath343 , @xmath344 and @xmath345 refer to the representation of elements @xmath0 with offset @xmath40 and anchor @xmath285 , and @xmath346 refers to the representation of elements @xmath130 with offset @xmath138 and anchor @xmath347 .
in other words , @xmath139 is a convex mapping of the intervals @xmath348 , @xmath349 , to the intervals @xmath350 , respectively , and of the intervals @xmath351 to the intervals @xmath352 .
it is straightforward to verify that @xmath139 is a step - isometry at level @xmath145 , so any geometric @xmath40-graph and geometric @xmath138 graph with vertex sets @xmath0 and @xmath130 , respectively , are isomorphic .
another setting we consider is where @xmath0 and @xmath130 are disjoint unions of rational intervals for which there exists a step - isometry between the endpoints of the intervals of @xmath0 to the endpoints of the intervals of @xmath130 .
[ thm:1disometry ] let @xmath0 and @xmath130 be two countable subsets of @xmath116 , and let @xmath211 .
let @xmath353 be a bijective step - isometry from @xmath0 to @xmath130 at level @xmath145 .
if @xmath13 is a geometric @xmath40-graph with vertex set @xmath0 and @xmath212 is a geometric @xmath138-graph with vertex set @xmath132 then @xmath39 let @xmath256 and @xmath257 , where @xmath354 .
as in the proof of theorem [ thm:1d ] , we inductively construct a sequence of pairs of sets @xmath259 ( @xmath277 ) and isomorphisms @xmath260\rightarrow h[w_i]$ ] , so that for all @xmath261 , @xmath262 , @xmath263 , @xmath264 and @xmath265 , and @xmath266 extends @xmath267 . as an additional part of the induction hypothesis
, we require that @xmath267 satisfies the following three conditions . 1 . for every @xmath134 , @xmath222 if and only if @xmath223 2 . for every @xmath134 ,
@xmath222 if and only if @xmath355 3 . for every @xmath224 , @xmath225 .
the first two conditions are those stated in lemma [ lemma : isometry_conditions ] , so this implies that @xmath267 is a step - isometry at level @xmath297 we can also conclude from this lemma that for all
@xmath134 , @xmath222 if and only if @xmath356 .
let @xmath357 and @xmath358 , and set @xmath359 .
conditions ( 1 ) and ( 3 ) follow as in the proof of theorem [ thm:1d ] .
condition ( 2 ) follows from the fact that @xmath360 . for the induction step , fix @xmath277 .
we construct @xmath266 from @xmath278 by first finding an image of @xmath279 . in the following
, @xmath139 refers to @xmath278 , and @xmath280 .
let @xmath361 and @xmath362 we have that @xmath282 , since the order of the representatives of vertices in @xmath269 is preserved under @xmath139 and under @xmath353 .
( see the similar argument in the proof of theorem [ thm:1d ] . ) in order to maintain conditions ( 1 ) and ( 2 ) of the induction hypothesis , @xmath290 should lie in @xmath291 , and because of condition ( 3 ) , @xmath292 must equal @xmath220 .
let @xmath293 , and consider the interval @xmath363 . from the definition of @xmath285 and @xmath128 it follows that @xmath364 .
the remainder of the proof is now analogous to the proof of theorem [ thm:1d ] and so is only sketched here .
let @xmath365 .
we can show that @xmath366 .
we can then invoke the @xmath138-g.e.c .
condition of @xmath212 to find a vertex @xmath18 in @xmath295 which is correctly joined to the vertices in @xmath270 so that an isomorphism is maintained if we set @xmath367 . finally , we finish the induction step by going back and finding a suitable image @xmath368 .
theorem [ thm:1d ] extends to @xmath2 with @xmath369 provided we use the _ product metric _ ; that is , the metric derived from the @xmath370 norm , defined by : @xmath371 where @xmath372 denotes the @xmath195-th component of @xmath25
. hence , we obtain unique isotypes of infinite random geometric graphs in all finite dimensions .
for the remainder of the section , @xmath373 is a positive integer , and @xmath32 is assumed to be the metric defined above .
[ thm : nd ] consider the metric space @xmath374 , where @xmath32 is the product metric defined above .
let @xmath0 and @xmath130 be two countable sets dense in @xmath49 , and let @xmath211 . if @xmath13 is a geometric @xmath40-graph with vertex set @xmath0 and @xmath212 is a geometric @xmath138-graph with vertex set @xmath130 , then @xmath375 . in particular , for all choices of @xmath0 and @xmath40 , there is unique isomorphism type of geometric @xmath40-graphs in @xmath374 , written @xmath376 .
theorem [ thm : nd ] is sensitive to the choice of metric . in section [ none ]
, we will show that the conclusion of theorem [ thm : nd ] for @xmath32 the euclidean metric fails even for @xmath377 the following provides the key tool for our proof of theorem [ thm : nd ] . as the proof is straightforward generalization of lemma [ lemma : isometry_conditions ] , it is omitted .
[ lemma : isometry_conditions_rn ] let @xmath0 and @xmath130 be subsets of @xmath49 with the @xmath370-metric , let @xmath242 and @xmath243 , and let @xmath40 and @xmath138 be two non - negative real numbers .
then a surjective function @xmath146 is a step - isometry at level @xmath145 if the following two conditions hold for all @xmath134 and for all @xmath195 , @xmath378 : 1 .
@xmath379 if and only if @xmath380 2 .
@xmath381 , where the representation of the @xmath195-th coordinate of elements of @xmath0 has offset @xmath382 and anchor @xmath383 the representation of the @xmath195-th coordinate of elements of @xmath130 has offset @xmath384 and anchor @xmath385 , let @xmath256 and @xmath257 .
we inductively construct a sequence of pairs of sets @xmath259 ( @xmath277 ) and isomorphisms @xmath260\rightarrow h[w_i]$ ] , so that for all @xmath261 , @xmath262 , @xmath263 , @xmath264 and @xmath265 , and @xmath266 extends @xmath267 .
as an additional induction hypothesis we require that @xmath267 satisfies conditions ( 1 ) and ( 2 ) from lemma [ lemma : isometry_conditions_rn ] . as in the proof of theorem [ thm:1d ] , for the base case we take @xmath357 , @xmath358 and @xmath359 . for the induction step , fix @xmath277 .
we construct @xmath266 from @xmath278 by first finding an image of @xmath279 . in the following
, @xmath139 refers to @xmath278 , and @xmath280 . for all @xmath386 , @xmath387 ,
define @xmath388 in order to maintain the induction hypothesis , for all @xmath386 , @xmath389 should lie in interval @xmath390 , and @xmath391 should be equal to @xmath392 .
let @xmath393 , and consider the product set @xmath394 any vertex in @xmath295 will qualify as a candidate for @xmath395 so that @xmath139 satisfies conditions ( 1 ) and ( 2 ) from lemma [ lemma : isometry_conditions_rn ] .
the remainder of the proof is analogous to that of theorem [ thm:1d ] , and so is omitted .
a _ step - isometric isomorphism _ is an isomorphism of graphs that is a step - isometry . in the base step in the proof of theorem [ thm : nd ] ,
if we are given induced subgraphs @xmath396 and @xmath397 such that @xmath398 is a step - isometric isomorphism , then the rest of the proof follows as before .
hence , we have the following corollary , which shows that the graphs @xmath376 act transitively on step - isometric isomorphic induced subgraphs .
let @xmath13 and @xmath212 be finite induced subgraphs of @xmath376 for some positive integer @xmath373 .
a step - isometric isomorphism @xmath399 extends to an automorphism of @xmath6 deleting a point from a dense set @xmath0 in @xmath49 gives another dense set .
hence , we have the following _ inexhaustibility _ property . for all @xmath400 and vertices
@xmath41 in @xmath376 , @xmath401 we can combine theorems [ thm:1disometry ] and [ thm : nd ] to obtain a result about isomorphisms between graphs with vertex sets in @xmath49 if there exist a special type of map between the sets .
given a set @xmath402 , denote the @xmath195-th component set of @xmath0 as : @xmath403 [ thm : ndisometry ] consider the metric space @xmath374 , where @xmath32 is the product metric defined above .
let @xmath0 and @xmath130 be two countable sets in @xmath49 , and let @xmath211 .
assume that for all @xmath378 , there exists a step - isometry at level @xmath145 from @xmath269 to @xmath270 .
if @xmath13 is a geometric @xmath40-graph with vertex set @xmath0 and @xmath212 is a a geometric @xmath138 graph with vertex set @xmath132 then @xmath375 .
the proof is a straightforward generalization of the proofs of theorems [ thm:1disometry ] and [ thm : nd ] , and is therefore omitted .
the choice of metric plays an important role in our isomorphism results in section 3 .
we demonstrate that there are non - isomorphic geometrically e.c .
graphs in the plane with the usual euclidean metric ( denoted by @xmath32 ) .
[ noni ] let @xmath0 be a countable set dense in @xmath408 equipped with the euclidean metric , and @xmath332 fixed .
then there exist infinitely many pair - wise non - isomorphic @xmath40-g.e.c .
graphs with vertex set @xmath34 assume for a contradiction that there is a step - isometry @xmath146 at level @xmath145 that is not an isometry .
without loss of generality , we assume that @xmath409 . for each @xmath224 , let @xmath410 . since @xmath139 is not an isometry , there must exist points @xmath411 and @xmath412 so that @xmath413 . since @xmath139 is a bijection , we may assume , without loss of generality , that @xmath414 .
the proof follows by the following two claims .
given @xmath415 , define _ the discrepancy of _ @xmath172 as @xmath416 the discrepancy is a measure of the error in the distance between pairs of points and their images under @xmath139 .
since @xmath139 is a step - isometry , we have that @xmath417 for all @xmath418 to see how the lemma follows from the claims , note that by hypothesis , there are two points of @xmath0 with discrepancy @xmath419 . by claim [ claim2 ] , there are two points of @xmath0 with discrepancy at least @xmath430 and with distance at least @xmath431 apart . by claim [ claim1 ] there are points with discrepancy at least @xmath432 apart . by induction , we obtain a sequence of pairs of points @xmath433 of @xmath0 whose discrepancy equals @xmath434 which tends to infinity in @xmath195 . in particular , there are points @xmath435 of @xmath0 so that @xmath436 , which contradicts the fact that @xmath139 is a step - isometry .
we now prove claim [ claim1 ] .
we first define some constants that will be useful in the proof .
let @xmath437 choose @xmath438 since @xmath0 is dense in @xmath408 , we can find points @xmath439 and @xmath440 in @xmath0 so that @xmath441 so @xmath442 are the vertices of a quadrilateral whose sides have length between @xmath443 and @xmath156 .
see figure [ onee ] .
let @xmath444 .
the distance between @xmath439 and @xmath440 is smallest when all sides of the quadrilateral equal @xmath443 , so using the pythagorean theorem we have that @xmath445 where the last step follows from the choice of @xmath446 .
( note that the calculation above is only valid if we use the euclidean metric . ) since @xmath139 is a step - isometry , @xmath453 now @xmath454 is largest when the quadrilateral has all sides equal to @xmath156 .
it follows that @xmath455 where the third inequality follows from ( [ eqn : r ] ) , and the last inequality follows from ( [ eqn : half ] ) .
in particular , @xmath456 as @xmath457 the proof of claim [ claim1 ] follows . we now prove claim [ claim2 ] .
let @xmath458 and assume @xmath459 .
let @xmath460 .
choose @xmath461 so that @xmath462 further , choose points @xmath463 so that @xmath464 @xmath465 and @xmath466 the choice of such points is possible since @xmath0 is dense .
see figure 4 .
subject to the given constraints , the points @xmath439 and @xmath440 are furthest apart when the distances achieve the upper bound of ( [ upper ] ) and the lower bound from ( [ lower ] ) , and when @xmath467 .
assume this to be the case . then @xmath468 form a rectangle , and the line segments @xmath469 , @xmath470 and @xmath471 are parallel .
hence , @xmath478 and @xmath479 and so @xmath480 the expression above is based on the case where @xmath439 and @xmath440 are furthest apart , and thus , it gives an upper bound for the general case .
combining this upper bound with the condition on @xmath481 given by ( [ cc ] ) , and with the assumption ( [ difference ] ) we have for @xmath482 equals @xmath483 and @xmath484 that @xmath485 where the first and last steps follow from ( [ cc ] ) .
we next consider the images of these points , and assume without loss of generality that @xmath486 .
since @xmath139 is a step - isometry , @xmath487 for @xmath482 equals @xmath488 or @xmath489 or @xmath490 or @xmath491 , and @xmath492 for @xmath482 equals @xmath493 or @xmath494 .
moreover , by assumption @xmath495 . now @xmath496 and @xmath451 are closest together when @xmath497 and @xmath498 for all @xmath482 for which @xmath499 is even .
as in the previous case , the line segments @xmath500 , @xmath501 , and @xmath502 are parallel , and @xmath503 is a rectangle . under these assumptions
we can compute @xmath454 similarly to the computation for @xmath504 , and obtain that @xmath505 since our assumptions hold for the case where @xmath496 and @xmath451 are closest together , we have in general , for @xmath482 equal to @xmath483 and @xmath484 , that @xmath506 a direct consequence of this lemma is the existence of many non - isomorphic g.e.c
. graphs with vertex sets dense in @xmath408 , equipped with the euclidean metric . a set @xmath0 in @xmath408 is @xmath40-_free _ if no pair of points in @xmath0 are distance @xmath40 apart .
for instance , one may consider @xmath507 , @xmath0 to be the set of all rational points in @xmath408 which are @xmath40-free , and @xmath508 ( which is also @xmath40-free ) .
it is straightforward to see there is no isometry from @xmath0 onto @xmath130 .
hence , a @xmath4-g.e.c .
graph on @xmath0 can not be isomorphic to a @xmath4-g.e.c .
graph on @xmath509 an enumeration @xmath512 of @xmath0 is _ good _ if @xmath513 for all @xmath514 and @xmath515 are not collinear .
we claim that a countable set @xmath0 dense in @xmath116 has a good enumeration . for a positive integer @xmath373 , we call @xmath516 a _ partial good enumeration _ of @xmath34
we prove the claim by constructing a chain of partial good enumerations by induction . using the density of @xmath24
choose three points @xmath515 that are not collinear , so that each are within @xmath40 of each other .
let @xmath517 enumerate @xmath518 as @xmath519 starting from @xmath520 , we inductively construct a chain of partial good enumerations @xmath521 , @xmath522 , so that for @xmath523 , @xmath521 contains @xmath524 we now want to form @xmath525 by adding @xmath526 . if @xmath527 , then let @xmath528 .
assume without loss of generality that @xmath529 let @xmath530 .
if @xmath531 then let @xmath532 and add it to @xmath521 to form @xmath533 otherwise , by the density of @xmath24 choose a shortest finite path @xmath534 of points of @xmath535 starting at @xmath536 and ending at @xmath537 so that two consecutive points in the path are distance at most @xmath40 . then add the vertices of @xmath538 to @xmath521 to form @xmath525 and enumerate them so that @xmath539 for @xmath540 .
taking the limit of this chain , @xmath541 is a good enumeration of @xmath0 , which proves the claim .
let @xmath542 be a good enumeration of @xmath24 and for any @xmath373 , let @xmath543 let @xmath13 and @xmath212 be as stated .
we say that two pairs @xmath544 and @xmath545 of vertices are _ compatible _ if @xmath544 are adjacent in @xmath13 and @xmath546 are adjacent in @xmath212 or @xmath544 are non - adjacent in @xmath13 and @xmath545 are non - adjacent in @xmath212 . for two pairs @xmath544 and @xmath545 such
that @xmath547 , the probability that they are compatible equals @xmath548 by corollary [ useful ] and lemma [ lemnoni ] , any isomorphism between subgraphs of @xmath13 and @xmath212 must be an isometry .
the images of three points in @xmath408 that are not collinear determine the isometry .
let @xmath549 be the event that there exists a partial isomorphism @xmath139 from @xmath550 $ ] into @xmath212 so that @xmath551 , and let @xmath552 note that @xmath553 for all @xmath373 .
next , we estimate the probability of @xmath554 .
note first that @xmath555 for all @xmath556 . for any tuple @xmath557 of three distinct vertices in @xmath521 , let @xmath558 be the event that there exists a partial isomorphism @xmath139 from @xmath550 $ ] to @xmath212 so that @xmath559 for @xmath560 . since the images of three points that are not collinear determine the isometry @xmath139 ,
if @xmath561 happens then all pairs @xmath562 and @xmath563 must be compatible , for @xmath564 .
thus , @xmath565 now @xmath566 so for @xmath567 we have that @xmath568 , and @xmath569 if @xmath15 is the event that @xmath13 and @xmath212 are isomorphic , then @xmath570 since the union of countably many sets of measure zero has measure zero , we conclude that @xmath571 , and thus , with probability @xmath11 @xmath407
|
we introduce a new class of countably infinite random geometric graphs , whose vertices @xmath0 are points in a metric space , and vertices are adjacent independently with probability @xmath1 if the metric distance between the vertices is below a given threshold .
if @xmath0 is a countable dense set in @xmath2 equipped with the metric derived from the @xmath3-norm , then it is shown that with probability @xmath4 such infinite random geometric graphs have a unique isomorphism type .
the isomorphism type , which we call @xmath5 is characterized by a geometric analogue of the existentially closed adjacency property , and we give a deterministic construction of @xmath6 in contrast , we show that infinite random geometric graphs in @xmath7 with the euclidean metric are not necessarily isomorphic .
|
let us consider the problem of determining the surface gravity waves generated by a body moving in a two - dimensional potential fluid . at the free surface
, @xmath0 , bernoulli s equation requires that @xmath1 where @xmath2 is the fluid speed and @xmath3 is the gravitational parameter .
the nonlinearity of forms the primary difficulty of analysis ; in order to make any sort of progress , the equation must usually be linearized . as explained by tuck @xcite , this linearization will typically involve making one of two possible assumptions . in the first
, the relevant flow quantities are expressed as a series expansion in powers of a geometric parameter such as @xmath4 .
for instance , one might assume that at leading order , the object is asymptotically thin or streamline in one or several of its dimensions .
as particular examples of note , a ship can be approximated by a single point source ( as studied by lord kelvin @xcite ) or a series of point sources along its centre plane ( as studied by michell @xcite ) .
alternatively , but also consistent with the limit @xmath5 , is the assumption that the body is submerged at a large depth compared to its own dimensions . then , for such an approach , the leading - order solution describes flow past the body in an infinite fluid without a free surface .
however , approximation schemes dependent on taking the limit of some geometric parameter can be difficult to justify formally as boundary conditions are applied at approximate locations rather than exact locations @xcite . thus for instance , flow past a circular cylinder can be approximated by flow past a dipole , but a finite - order approximation ( in powers of the cylinder radius ) can never yield a closed streamline around the body . even beyond the inability to satisfy the exact body conditions ,
another limitation of such approaches is that they say very little about the situation of flow around nonlinear or bluff bodies ; this is a crucial limitation for our purposes .
the second linearization of is more suitable for developing a theory of thick or bluff bodies . in this approach
, the relevant small parameter is typically related to the froude number , written as @xmath6 . as @xmath7 ,
the body is assumed to move at low speeds , and at leading - order , the free surface is entirely flat .
the great advantage of this approach is that there is no geometrical restriction placed on the body ; thus the associated boundary conditions are satisfied exactly to all orders .
the challenge , however , is that the limit @xmath7 is singular .
the free - surface waves are exponentially small and their amplitudes scale as @xmath8 .
thus they are said to be _ beyond - all - orders _ of any regular expansion in powers of @xmath9 .
the inability to detect the surface waves using standardized methods in perturbation theory is known as ogilvie s _ low - froude paradox _ , based on francis ogilvie s seminal observations in 1968 @xcite . in the fifty years that followed ogilvie s original work ,
the low - froude paradox was eventually resolved through the combined efforts of many researchers .
we mention , for example , works by dagan & tulin @xcite , vanden - breock & tuck @xcite , and ogilvie & chen @xcite , but a more comprehensive review can be found in @xcite .
chapman & vanden - broeck @xcite were the first to leverage modern techniques of exponential asymptotics to the study of the low - froude flows in prior work , without such techniques , the derived surface quantities were often asymptotically inconsistent .
the difficulty , though , in applying such techniques is their relative complexity ; these methods requires notions of estimating asymptotic divergence , optimal truncation , inner - and - outer boundary layer matching , and the stokes phenomenon .
( see _ e.g. _ works by boyd @xcite , costin @xcite , olde daalhuis _
et al . _
@xcite , berry and howls @xcite ) .
we ask : _ is there a more intuitive approach to the study of such low - froude flows ? _ in this paper , we shall demonstrate that , for the particular case of gravity - driven flow , there exists a powerful and visual methodology for the study of wave - body interactions .
this approach is made possible through a reduction of the water wave equations to a single first - order complex - valued differential equation that can be integrated .
in fact , there is a rich history that underlies the development of the methodology , as introduced in the companion paper @xcite , and originating from the works of tulin @xcite and tuck @xcite . before embarking on the present paper ,
we highly recommend a reading of the historical significance of the problem , as presented in @xcite .
we first provide a brief review of the main results of the companion paper .
. the surface profile , @xmath10 , in ( a ) corresponds to the numerical solution of the full nonlinear equations at @xmath11 .
the profiles in ( b ) correspond to the surface speed @xmath12 , as a function of the potential @xmath13 ; as seen , there is close agreement with the simplified models . the riemann surface ( c )
is explained in fig .
[ fig : surfship ] .
[ fig : outline ] ] for the moment , let us consider potential free - surface flow past the stern of a surface - piercing ship , modeled as the two - dimensional semi - infinite geometry shown in fig .
[ fig : outline](a ) .
the physical coordinates are denoted by @xmath14 and the complex potential by @xmath15 , for velocity potential @xmath13 and streamfunction @xmath16 .
then , our task is to solve for the speed , @xmath17 , and streamline angle , @xmath18 , on the free surface , where @xmath19 and @xmath20 .
the governing water - wave equations are reviewed in appendix [ sec : gov ] , and are given in non - dimensional form by @xmath21 here , we have introduced an operator for a principal - value integral known as the hilbert transform .
it is defined by @xmath22 = \frac{1}{\pi } \dashint_{0}^\infty \frac{\theta(\varphi)}{\varphi - \phi } \ , { \operatorname{d\!}{}}{\varphi}.\ ] ] the first equation in is the differentiated form of bernoulli s condition [ _ c.f .
_ ] , with @xmath23 being the square of the froude number for upstream speed @xmath24 , gravity @xmath3 , and geometrical length scale @xmath25 .
the second is a boundary - integral equation , written so as to split the integral on the solid body @xmath26 where @xmath27 is assumed to be known , with the hilbert transform of @xmath27 over the unknown free surface @xmath28 . for the surface - piercing configuration
, we also specify that the free surface attaches to the body at a stagnation point , @xmath29 . for a semi - infinite ship with the single stern - face of angle @xmath30 shown in fig .
[ fig : outline ] , the only length scale is the distance from the stagnation point to the corner .
thus , we may non - dimensionalize the problem so that the corner lies at @xmath31 , and the body can be specified as @xmath32 the above body shape is then used in , and the two equations can be numerically solved using the finite - difference schemes discussed in _ e.g. _ @xcite . in fig .
[ fig : outline](a ) , we have plotted the physical free surface height , @xmath10 , obtained by integrating @xmath33 once @xmath2 and @xmath27 are known . a typical numerical solution for the case of a rectangular stern , @xmath34 , is shown in fig .
[ fig : outline](a ) at @xmath35 .
although the two - dimensional semi - infinite geometry of may seem rather idealized , it presents a canonical model of wave - structure interaction with a variety of numerical and analytical challenges .
these two - dimensional hull shapes have been considered in _
e.g. _ the works of @xcite , but similar considerations can be extended to ships with multiple corners @xcite , or ships with smooth or bulbous profiles @xcite .
now the major difficulty in solving is that the hilbert transform is a global operator , and thus it requires the very values of @xmath27 that we seek .
earlier , tuck @xcite had proposed the idea that toy models of certain wave - structure problems could be derived by ignoring or replacing the hilbert transform , and changing to a local problem of initial - value type . in the companion paper @xcite ,
such a reduction was shown to be formally possible in the limit @xmath7 .
the argument is as follows .
let us expand the solution of as a truncated regular series expansion and an error term , @xmath36 + { \bar{q}}\quad \text{and } \quad \theta = \bigl[\theta_0 + { \epsilon}q_1 + \ldots + { \epsilon}^{n-1 } \theta_{n-1 } \bigr ] + { \bar{\theta}}.\ ] ] it follows from setting @xmath37 in that at leading order , @xmath38 , and the free - surface is entirely flat .
effectively , the surface has been replaced by a rigid plane , and the solution is known as the _
double - body flow_. in this case , the leading - order speed is given by @xmath39 as seen in fig .
[ fig : outline ] , the leading - order solution fails to capture the wave phenomena .
in fact , the situation is much worse , and it can be argued through the study of higher - order terms in , that none of the terms , @xmath40 and @xmath41 , will contain an oscillatory component at any order .
indeed , this is ogilvie s low - speed paradox @xcite .
let us take @xmath2 and @xmath27 , defined on the free surface , @xmath19 , and analytically continue the solutions to complex values of @xmath13 .
relabeling @xmath42 , the system can be reduced to the following complex - valued differential equation : @xmath43{\bar{q}}\sim r(w ; { \hat{{\mathscr{h}}}}[{\bar{\theta } } ] ) \quad \text{where } \quad { \hat{{\mathscr{h}}}}{\bar{\theta}}(w ) = { \hat{{\mathscr{h}}}}[{\bar{\theta } } ] = \int_0^\infty \frac{{\bar{\theta}}(\varphi)}{\varphi - w } \ , { \operatorname{d\!}{}}{\varphi}.\ ] ] the derivation of
is reviewed in appendix [ sec : reduce ] .
the idea of extending the governing system to complex values may seem odd , but the rationale can be seen _ a posteriori_. above , we have assumed that @xmath44 lies in the upper half - plane ( an analogous derivation can be done for the lower half - plane ) .
we have also introduced @xmath45 for the complexified hilbert transform , which is non - singular for @xmath44 off the real axis .
the two functions , @xmath46 and @xmath47 , are known , typically in closed form , while @xmath48 may be approximated ; we will specify these quantities later . integrating the equation , and taking the limit whereby we return to the physical free surface , @xmath49
, we obtain the result of @xmath50 \left [ \int_{\phi_\text{start}}^\phi r(\varphi ; { \hat{{\mathscr{h}}}}[{\bar{\theta } } ] ) { \mathrm{e}}^{p_1(\varphi ) } \ , { \mathrm{e}}^{\chi(\varphi)/{\epsilon}}\ , { \operatorname{d\!}{}}{\varphi } \right ] { \mathrm{e}}^{-\chi(\phi)/{\epsilon}}.\ ] ] in the example of flow past a surface - piercing ship , we take @xmath51 in order to account for condition of a stagnation point , @xmath52 , at the edge of the hull .
there are many subtleties to the reduction process that takes the full system to the first - order problem , and then again to the integral .
indeed , a rigorous explanation of why this reduction is valid forms the basis of the companion work @xcite . for now , let us take for granted , and seek an approximation of the integral . in the limit @xmath7 , the initial contour of @xmath53
$ ] can be defomred along paths of steepest descent .
once done , an asymptotic expansion of the integral is derived by consideration of dominant contributions from the endpoints , as well as any relevant saddle or critical points of the integrand . in @xcite , it was shown that the endpoint contributions would lead to a re - expansion of higher - order algebraic terms from , while the water waves were caused due to contributions _ away _ from the positive real axis .
therefore @xmath54 $ ] , which only depends on the free surface values of @xmath55 , can be ignored at leading order .
this subtle reduction process explains why reduced wave models ( such as the ones reviewed in @xcite ) remove or substitute alternatives for the hilbert transform .
let @xmath56 be the exponentially waves that arise from the integral according to this process .
the free surface will then be approximated by @xmath57 + 2 { \operatorname{re}}({q_\mathrm{exp}}),\ ] ] which includes both the wave - free flow ( the square - bracketed terms ) and the surface waves .
note that we are required to add both @xmath56 from and also its complex conjugate , due to an analogous argument for the analytic continuation of into the lower half - plane , and subsequent application of the schwartz reflection principle . combining these two contributions
then leads to @xmath58 .
let us be more concrete and provide a numerical example .
suppose we return to the differential equation and , having been sufficiently convinced by the above argument , we ignore the @xmath54 $ ] contributions to @xmath48 .
using the initial condition @xmath52 , the values and for @xmath46 and @xmath47 , and the approximation for @xmath48 , the differential equation is solved as an intial - value problem .
the result , , is shown in fig .
[ fig : outline](b ) .
the agreement with the full nonlinear solution is superb , even at the moderate value of @xmath35 . in @xcite , a careful check of numerical vs. asymptotic results as @xmath7 verifies these claims properly .
now we arrive at the current aims of this paper .
previously in @xcite , the asymptotic approximations of the integral were made based on the general assumption that the initial contour of integration @xmath59 $ ] , could be deformed for any given body geometry , and that such a deformation would include the key critical points generating the water waves .
in actuality , the topology of the steepest descent paths may be rather complicated .
different body geometries yield different contributing singularities ; or perhaps for some bodies , there may be branch cuts that complicate the deformation procedure . thus , our main goal is to demonstrate , through the use of several representative examples , the correspondence between the physical flow and the topology of steepest descent paths .
the power of this approach is two - fold .
first , it provides an analytical methodology for determining a wide range of wave - structure interaction problems .
second , it provides an intuitive approach to visualizing the effects of changing particular geometrical properties .
we will see , for example , how the complex - valued topology of a surface - piercing ship differs from that of channel flow , or how flow past a shallow angled step differs from flow past a rectangular step .
this correspondence process is illustrated in fig .
[ fig : outline](c ) .
let us now explain how the riemann surfaces connected to the steepest descent paths are computed . a general reference for the method of steepest descents can be found in _
* chap . 7 ) .
we first re - write the integral as @xmath60 ) { \mathrm{e}}^{p_1(\varphi)},\ ] ] and @xmath46 , a quantity known as the _ singulant _ , is given by , or repeated here as @xmath61 in the definition of @xmath46 , we have introduced the sign constant , @xmath62 .
the constant takes the value @xmath63 for surface - piercing flows , and @xmath64 for channel flows .
this distinction is necessary due to the difference in surface flow direction relative to the solid body , as explained in .
the function @xmath65 is the double - body solution discussed in
[ sec : mathintro ] , and encodes the information for the geometry of the moving body .
any choice for the initial point of integration , @xmath66 , is valid so long as the integral is defined ; a selection will be compensated by the numerical pre - factors associated with @xmath67 .
we will discuss specifics of the @xmath46 function later . returning to , the contour is initially along the real axis , @xmath68 $ ] , but once it is deformed along the paths of steepest descent , this would yield the general decomposition of @xmath69 thus the integral is approximated by the endpoint contributions , @xmath70 , from @xmath71 , and further contributions , @xmath72 , from critical points ( saddle points , poles , or branch points ) , presumed to correspond to the surface waves .
the paths of steepest descent are given by level sets of @xmath73 , and thus the criterion for determining whether @xmath44 lies on the steepest descent path from @xmath66 is @xmath74 = { \operatorname{im } } [ \chi(w_0 ) ] \quad \text{and } \quad { \operatorname{re}}[\chi(w ) ] \leq { \operatorname{re } } [ \chi(w_0)].\ ] ] the first condition ensures that @xmath44 is on a path of steepest descent or ascent ; the second ensures that it is , in fact , a path of steepest descent . in practice , however , calculating @xmath75 for different values of @xmath44 is not necessarily straightforward , as depending on the body function , @xmath65 , @xmath46 may possess a complicated branch structure .
a similar issue was encountered in the development of asymptotic solutions to the problem of thin - film rupture @xcite . in order to explain the terminology for the branch structure ,
let us return to the example ship flow of [ sec : mathintro ] . substituting @xmath65 from into , and choosing the initial point of integration , @xmath76 , to correspond to the corner of the ship , we have @xmath77 to define a single - valued function for the above integral
, we must specify the branches and branch cuts associated with the two critical points at @xmath78 .
each single - valued specification of @xmath46 is called a _ branch _ ( the output or range ) , and each branch is defined on a _ riemann sheet _ , @xmath79 of the complex plane ( the input or domain ) . however , as @xmath44 varies in the complex plane , the contour of integration may cross a branch cut ; here , the single - valued specification of @xmath46 encounters a discontinuity . for our purposes , we wish to continue the trajectory of the steepest descent contours of beyond such discontinuities . in other words ,
when a branch cut is traversed , we will move onto the adjacent riemann sheet , and take the values of @xmath46 from the next branch .
the collection of riemann sheets is known as the _ riemann surface _ for @xmath46 .
it is this notion of a riemann surface , represented by copies of the complex plane attached to one - another , that allows us to unambiguously define a multi - valued function for @xmath46 .
we refer readers to ( * ? ? ?
2 ) or ( * ? ? ?
* chap . 7 ) for more background on such complex - variable terminology .
continuing with our ship example in , let us choose the branch cuts so as to lie along the rays of constant angle from each branch point , @xmath80 .
typically , we will choose them so as to lie parallel to one of the coordinate axes .
let @xmath81 be the winding number around the point @xmath82 .
that is , @xmath83 marks the number of rotations associated with a given analytic continuation around the point @xmath82 , with positive and negative integers for counterclockwise and clockwise rotations , respectively .
notationally , we will denote each distinct branch of @xmath46 by the pair @xmath84 , indicating the number of revolutions required in order to reach the current branch .
for example , consider the surface - piercing ship with branch points at @xmath85 ( the corner ) and @xmath86 ( the stagnation point ) .
three illustrative branches are labeled as @xmath87 the first , @xmath88 , is the principal branch obtained by integrating from @xmath89 to @xmath90 without crossing a branch cut .
the second , @xmath91 is found by integrating from @xmath89 , then circling the stagnation point , @xmath92 , in the positive sense , and arriving at @xmath90 . the third , @xmath93 , begins from the previous branch , and then performs an additional rotation about the corner , @xmath94 .
when it may be ambiguous , we use the arrow notation to clarify the order in which branch cuts are crossed .
a three - dimensional cross - section of the riemann surface can be visualized by plotting @xmath95 using @xmath88 and its adjacent sheets . in general , these surfaces will self - intersect within the graphical representation but they are otherwise free of discontinuities
. for similar visualizations of riemann surfaces , see @xcite or ( * ? ? ?
the riemann surfaces that appear in this paper , such as the one in fig .
[ fig : outline ] , are graphed from the numerical integration of @xmath46 using the trapezoid rule .
in such computations , we have carefully defined the locations of the branch cuts , and implemented a numerical scheme allowing the integration path to smoothly switch between the individual branch sheets .
both interactive plots of the riemann surfaces , and also the data for the contour plots appearing in this paper can be found within the electronic supplementary material .
we are now ready to study the steepest descent paths associated with the integral .
we have chosen to focus on three representative geometries : the previous rectangular semi - infinite ship in infinite depth [ in [ sec : steep][sec : ship ] ] , a rectangular step in a channel [ in [ sec : steep][sec : step ] ] , and an angled step in a channel [ in [ sec : steep][sec : stepangle ] ] .
the methods we present can be extended to cover many other classes of moving bodies , notably those for which a boundary integral formulation exists .
this includes , for instance , models of an air jet stream blowing on stationary water @xcite , flow over a submerged source or sink @xcite , and flows past sluice gates or surf boards @xcite .
let us return to the case of flow past a semi - infinite rectangular stern .
the geometry is specified by @xmath96 in with @xmath34 , and the leading - order speed is given by .
our task is to approximate the integral , which involves the singulant function .
the critical points in the @xmath44 or @xmath97-plane include the corner @xmath98 and the stagnation point @xmath99 , and the branch cuts from both are taken along the positive real axis . in the integral
, we would like to take the initial point of integration to be the stagnation point , @xmath100 , so as to make use of the boundary condition @xmath52 .
however , because the origin is a degenerate point , it is easier to first consider the descents paths for the integral @xmath101 where @xmath102 is a small and positive number . also , it can be verified , through examination of the pre - factor @xmath67 , that the steepest descent contours will only depend on @xmath46 .
below , we will use @xmath44 interchangeably with @xmath97 ( the dummy variable ) . using the methodology outlined in [ sec : riemann ] , we generate the riemann surface associated with @xmath46 , and the result is shown in fig .
[ fig : surfship ] .
the visualized surface is composed of the three listed branches in .
however , as it turns out , in the rectangular geometry , the adjacent riemann sheets due to the corner point remain inactive , and are untouched by the relevant steepest descent paths
. this will not always be the case for all geometries and through all values of the angle @xmath103 . in
both the examples of a rectangular and angled step in [ sec : steep][sec : step ] and [ sec : steep][sec : stepangle ] , the steepest descent contours will indeed traverse onto adjacent riemann sheets .
we continue to examine fig .
[ fig : surfship ] , but now turn our attention to the planar projection , shown in grey , where several constant phase curves , @xmath104 , are projected down from the riemann surface .
the dashed lines are constant phase lines emerging from the corner point , @xmath98 , with one set of lines , on the right , forming a homoclinic , and the other , on the left , forming an open triangular region . setting @xmath105
, we find that the local emergence angles of such paths are @xmath106 where @xmath107 .
thus the homoclinic is formed by the two local angles , @xmath108 , while the triangular region is formed by the two local angles , @xmath109 . the most important quantity to note
is that along the dashed line with @xmath110 , we have @xmath111 , which indicates that the trajectory along this curve lies uphill of the point @xmath98 ( note that @xmath46 in was chosen so that @xmath112 at the corner ) .
we have also plotted two steepest descent trajectories , shown thick . within the interior of the homoclinic
, the steepest descent trajectory forms an arc that tends towards @xmath99 in a direction tangential to the negative real axis .
exterior of the homoclinic , and to the right of the plane , the trajectory moves to @xmath113 along the dashed line with local angle @xmath114 .
the dashed line , called the _ stokes line _ , marks a critical curve across which the steepest descent paths change in dramatic fashion ; it is of key importance in what follows . for the rectangular ship geometry with @xmath34 , shown in the @xmath95 plane .
the two thick contours correspond to steepest descent trajectories for an upstream ( near the stagnation point ) and downstream point on the free surface .
the lower plane shows the projection of curves onto the @xmath115-plane .
the surface is generated from numerical integration of @xmath46 .
note that @xmath97 is the dummy variable for integration in the @xmath44-plane.[fig : surfship],scaledwidth=100.0% ] let us now turn to fig .
[ fig : shipstokes](a ) , and consider the integral between two points , @xmath102 and @xmath116 , on the upstream side and sufficiently close to the stagnation point at the hull .
consequently , both such points lie within the dashed homoclinic .
we see that the contour from @xmath102 is deformed in a counterclockwise manner and tends to @xmath117 where @xmath118 . within this valley
, the contour then unwinds clockwise , with @xmath46 increasing until arriving at @xmath119 .
thus , when the integral , @xmath120 , is evaluated at @xmath121 , where @xmath122 marks the intersection of the dashed line with the positive real axis , the result of this process yields two dominant endpoint contributions , written loosely as @xmath123 the terms of which can be derived by integrating by parts . as was shown in @xcite , this process will yield the higher - order algebraic corrections in .
now that we have understood the steepest descent paths , we may freely take @xmath124 , that is , take the initial point of integration in to the stagnation point ; then , the proper steepest descent deformation for the contour @xmath125 $ ] involves only the single contour in the upper half - plane connected to @xmath116 . _
we conclude through this process that when sufficiently close to the body of the ship , there are no free - surface waves in the limit @xmath7 ( to this order of exponential accuracy ) . _ the dashed line in fig .
[ fig : shipstokes](a ) is known as the _ stokes line _ , and it marks a critical line across which the analytic continuation of the integral , @xmath120 , switches - on a subdominant contribution .
the intersection of the stokes line with the original contour along @xmath126 $ ] , is the location across which the deformation process makes a dramatic topological change , and this occurrence is known as the _ stokes phenomenon _ @xcite .
let us now consider the method of steepest descents applied to an initial contour in beginning at @xmath102 and ending at @xmath127 , which lies outside of the homoclinic region .
thus @xmath128 has sufficiently far downstream so as to pass the stokes line .
this is shown in fig .
[ fig : shipstokes](b ) .
conceptually , it is easier to split the integration range so that @xmath129 where @xmath130 and @xmath131 are the limiting points just on the interior and exterior of the homoclinic orbit .
the deformed contour is shown in the figure .
the steepest descent contours from @xmath102 and @xmath130 are the same as before , with both curves unraveling within the dashed homoclinic and then connected by the valley at @xmath117 where @xmath132 .
however , for the second integral in , the contours from @xmath133 and @xmath128 tend to the valley in the upper left , as @xmath134 .
since the two integrals that run alongside the top of the homoclinic orbit will sum to zero , this leaves the final deformed contour in fig .
[ fig : shipstokes](c ) .
thus , in addition to the endpoint contributions , we have @xmath135 the evaluation of the @xmath136 integral along the steepest descent path depends only on the local properties of the integrand near @xmath89 .
this is why we may think of the surface waves as being _ generated _ by the ship s corner .
the exponentially small contribution , corresponding to surface waves , that results from the integral approximation is given in @xcite .
_ we conclude that when sufficiently far from the body of the hull , the free - surface waves arise from the switching - on of a saddle - point contribution due to the corner of the ship ( the stokes phenomenon)_. ) -plane ( for dummy integration variable @xmath97 ) .
branch cuts are shown as a wavy line .
the circles indicate the two points @xmath98 and @xmath99 .
[ fig : shipstokes ] ] consider now the case of an angled step in a channel , as shown in fig .
[ formstep](a ) .
the problem is non - dimensionalized so that in the potential plane , the free surface is along @xmath20 , the channel bottom is along @xmath137 , and @xmath138 . in order to develop the analogous formulae to , we map the @xmath15 plane to the upper half - plane using @xmath139 , as shown in fig .
[ formstep](b ) where @xmath140 within the @xmath141-plane , the solid boundary lies along @xmath142 , the free surface along @xmath143 , and downstream infinity corresponds to @xmath144 .
the development of bernoulli s equation and the boundary integral equation follow parallel lines , and are given by with @xmath64 .
having non - dimensionalized with the channel height , we have the freedom to specify the locations of the step s stagnation point , @xmath145 , and corner point , @xmath146 , with @xmath147 .
the body geometry is described by the angle and speed , @xmath148 where the speed follows from .
this geometry corresponds to a step - up of angle @xmath149 to the horizontal , and reversing the flow or allowing a step - down is straightforward .
such step - geometries have been considered by _
e.g. _ @xcite and , like the semi - infinite ship , they provides a canonical model for free - surface flow in a channel .
-plane for the non - dimensional flow over the angled step .
the flow is contained within a strap in the complex potential @xmath44-plane .
the map from the strip to the upper - half plane is given by @xmath150 .
[ formstep],scaledwidth=100.0% ] in what follows , we must take care to transform the integration variables from @xmath44 to @xmath141 .
we will abuse notation , and instead of writing @xmath151 , we will write @xmath152 for @xmath153 . the analogous integral to and , written now for the @xmath141-plane is given by @xmath154 where @xmath155 using and .
the dummy variable is now @xmath156 , corresponding to integration in the @xmath141-plane .
we have also chosen to centre the singulant so that @xmath157 , and this only changes the numerical pre - factors embedded in @xmath158 . like the example of the ship , it is convenient to consider the initial point of integration @xmath159 at a finite non - degenerate point along the free - surface , @xmath160 , but we will eventually let @xmath161 , so as to impose the natural upstream boundary condition that requires @xmath162 as @xmath163 .
let us now consider the rectangular step , and set @xmath34 .
within integral for the singulant , we choose both branch cuts from @xmath145 and @xmath146 to lie in the direction of the positive real axis ; due to the combination of powers , this is equivalent to a single branch cut between the two points .
there is a further logarithmic branch cut that arises from the pole at @xmath164 , but crossing this cut does not cause any adjustment in the steepest descent paths . for the step geometry with @xmath165 , @xmath166 , and @xmath34 , shown in the @xmath167 plane .
the two thick contours correspond to steepest descent trajectories for an upstream and downstream point on the free surface .
the surface is generated from numerical integration of @xmath46 .
note that @xmath156 is the dummy variable for integration in the @xmath141-plane .
[ fig : surfstep],scaledwidth=100.0% ] using the algorithms described in
[ sec : riemann ] , we generate the riemann surface for @xmath46 , and present the projection into @xmath167-space in fig .
[ fig : surfstep ] . recall the index notation of . in this case ,
the riemann sheets are denoted by @xmath168 for the three branch points @xmath169 .
the main sheet , @xmath170 , is shown at the very top of the surface in fig .
[ fig : surfstep ] , and also shown are two adjacent sheets , @xmath171 and @xmath172 , achieved by passing through the logarithmic branch cut once or twice , respectively .
the reader may compare the three - dimensional surface representation with the planar projection in fig .
[ fig : stepstokes ] to see how the branch cuts and individual riemann sheets align .
examine now the lowest plane of fig .
[ fig : surfstep ] , where several representative steepest ascent / descent ( or constant - phase ) curves have been projected down from the surface .
locally setting @xmath173 , we find @xmath174 for @xmath175 . relating to fig .
[ fig : surfstep ] , we see dashed constant phase lines emerging from @xmath146 and forming two homoclinic orbits , separating the plane into three regions . within the interior of the right - most homoclinic ( formed by angles @xmath176 ) , the constant phase paths form closed orbits that become increasingly circular near @xmath164 .
the left - most homoclinic is formed by angles @xmath177 ) . outside
both homoclinic orbits , the constant phase paths also form close contours , approximately in the shape of a figure eight . note
that at the other critical point at @xmath145 , which corresponds to the stagnation point of the step , @xmath178 , and thus it is not a saddle point of the integrand .
we turn now to fig .
[ fig : stepstokes](a ) , and seek to deform the initial contour of integration in when the two endpoints @xmath159 and @xmath179 are chosen sufficiently far upstream from the step so as to lie outside the right homoclinic . in the figure , we use @xmath141 and @xmath156 interchangeably ( the latter is a dummy variable of integration ) . although the steepest descent path from each individual point forms a closed orbit in the @xmath156-plane , each full counterclockwise rotation about the figure eight trajectory is accompanied by a smaller value of @xmath180 .
this is seen in the three - dimensional representation of fig .
[ fig : surfstep ] .
thus , the final deformed contour begins from a , and follows the trajectory counterclockwise , achieving smaller and smaller values of @xmath180 . in order to connect to the figure - eight contour joined to b ,
the connection is made within the valley where @xmath118 .
once done , the contour unwinds in a clockwise direction , and finally joins with b. predictably , the result of this process is , and thus @xmath181 is approximated by the two endpoint contributions .
this yields the regular asymptotic expansion of the solution .
once we have understood this deformation process , it is simple to take the initial point @xmath182 , that is , to be upstream infinity .
we require the physical wave problem to impose a radiation condition such that the flow is uniform upstream and consequently , the only contribution to the integral at leading order is from an expansion about the point @xmath116 .
_ we conclude that when sufficiently far upstream from the inclined - face of the step , there are no free - surface waves in the @xmath7 limit . _ -plane .
arrows indicate directions of descent .
[ fig : stepstokes ] ] now consider the deformation process when the initial contour of integration connects the upstream point , @xmath159 , with a point , @xmath183 , chosen sufficiently far downstream so as to lie within the right - most homoclinic orbit .
this is shown in fig .
[ fig : stepstokes](b ) .
we split the integration range in an analogous manner to , with points @xmath184 and @xmath131 , chosen to lie just on the exterior and interior , respectively , of the homoclinic orbit .
the critical dividing line ( the stokes line ) , shown dash - dotted in the figure is the constant phase line from the step corner , @xmath185 .
the steepest descent process then proceeds similarly as for the case of the surface - piercing ship : endpoint contributions are obtained from @xmath186 and @xmath128 , and a saddle - point contribution is obtained from the corner of the step . unlike the steepest descent paths for the ship , which tended towards either @xmath187 or @xmath188 , the steepest descent contours for the rectangular step form closed orbits , with each rotation diminishing the value of the integrand .
this is shown most clearly in the three - dimensional surface of fig .
[ fig : surfstep ] . _
we conclude that when sufficiently far downstream from the rectangular step , the free - surface waves arise from switching - on the saddle - point contribution due to the corner of the step ( the stokes phenomenon)_. in both the rectangular ship and rectangular step , we have shown how the intersection of a stokes line from the corner of the body with the free - surface accounts for the production of water waves . in more exotic geometries , multiple points in the body may produce multiple relevant stokes lines . bodies with smooth geometries will also produce waves as well this does not invalidate our methods , and we return to discuss this in [ sec : discuss ] . for now , let us consider the accompanying deformation of the riemann surface in light of a change to the angle , @xmath103 , of the step in fig .
[ formstep ] . observe that stokes lines correspond to steepest descent curves emerging from one of the critical points of the integrand . with @xmath66 in
taken to be the point in question , then necessarily @xmath189 as the point is approached .
consider the stagnation point in the angled step , @xmath145 . from
, we would require @xmath190 in order for @xmath189 .
this criterion led chapman & vanden - broeck @xcite to conclude that a stokes line could only be produced by a point with an in - fluid angle greater than @xmath191 .
even at the proper angle , however , a stokes line may exist , but it may not intersect the free surface ( and thus makes no difference to the subsequent steepest descent paths ) .
let us take for instance a shallow angled step with @xmath192 .
for this example , both @xmath146 and @xmath145 produce a relevant stokes line . the riemann surface for the step geometry
is shown in fig .
[ fig : angledstep ] , and the picture reveals the remarkable complexity for even this moderately simple geometry .
we note that the shallow angle of the step , as compared to the rectangular case , forces certain steepest descent contours to now travel past the branch cut associated with @xmath145 , and move onto the sheet @xmath193 .
the steepest descent process will go through three possible configurations , depending on the extent of which @xmath194 in lies downstream from the step .
if @xmath194 has not passed the first stokes line from the corner , @xmath146 , then only the end - point contribution is derived from the steepest descent process and there are no waves .
once @xmath194 passes the first stokes line intersection from the corner , a single free - surface wave is switched on as in the rectangular case .
however , once @xmath194 passes the second stokes line , a second wave is switched on .
thus , for this most downstream point , we would modify to @xmath195 from the riemann surface , notice that saddle from @xmath196 lies lower than the one from @xmath185 .
therefore , the second free - surface wave will be exponentially subdominant to the first .
the work in this paper was inspired through three principal motivations .
the first motivation was to return to the work of tulin @xcite and tuck @xcite , both of whom had proposed reduced models for free - surface flows with moving bodies .
tulin s model relied on a strategic trigonometric substitution inspired by davies @xcite , while tuck s model was less rigorous and more pedagogic in nature .
both models were united by their dependence on simplifying or outright removing the hilbert transform , and also emphasized the importance of analytic continuation into the complex plane .
both models were also incomplete ; thus the search for a complete model led to the companion work @xcite , where a systematic reduction of the water - wave equations demonstrated why certain simplifications can be applied in the low - speed , @xmath7 , limit .
there , the analysis had depended on a general steepest descent argument , attributing surface waves to saddle - point contributions . despite the relative simplicity of such arguments
, the actual deformation process can be quite nontrivial , and in this paper , our objective was to demonstrate how this steepest descent analysis was performed for particular flow geometries .
our second motivation was to provide a powerful and visual methodology for the study of wave - structure interactions .
the method establishes a correspondence between the physical flow with the topology of the riemann surface that underlies paths of steepest descent .
for instance , through the visualizations of figs .
[ fig : surfship ] and [ fig : surfstep ] , we showed how to understand the effects of changing from flow past a surface - piercing ship , to flow over a step , simply by observing the associated changes in the riemann surfaces .
subtle geometrical changes can also be noted : for example in fig .
[ fig : angledstep ] , we changed the angle of the step , showing how the modification shifts paths of steepest descent onto previously untraveled areas of the riemann surface . finally , our last motivation was to explore the various connections between the work here , with other research where exponential asymptotics is used to study free - surface flows .
the successful application of exponential asymptotics to such problems has led to an exciting revival of classical problems in fluid mechanics @xcite , but the methodology has always been limited by its relative complexity .
there is a key simplification afforded by the study of _ gravity_-driven flow , and that is that bernoulli s equation is only first - order in its derivative .
our integral methodology exploits this fact , and allows the solution to be written in the quasi - closed form , and then studied using the method of steepest descents .
_ is it then possible to consider other problems where a similar visual methodology is developed ? _
we are equally motivated by a need to better understand free - surface waves with the combined effects of gravity and surface tension .
it is well known that in the case of gravity - capillary flows , the equivalent exponential asymptotics analysis is a great deal more complicated @xcite .
the difficulty with the inclusion of surface tension is that bernoulli s equation in becomes of second order ( due to the surface curvature terms ) , so the associated equations are not so easily studied using the method of steepest descents .
similarly , the study of time - dependent or three - dimensional problems using exponential asymptotics is still very much in its infancy @xcite .
_ what other geometries can be studied using this approach ? _ the reader may remark that , in reality , ships or channel topographies may not be piecewise - linear in shape as they are considered here . in our work ,
the generation of waves is attributed to critical points in the geometry ( corners or stagnation points ) , but surely a smooth object will still create a wave pattern ! in fact , our analysis does not stipulate that the body must contain singularities _ par se _ , but rather the analytic continuation of @xmath2 and @xmath27 .
consider , for example , the replacement of the bottom topography in fig .
[ formstep ] by any streamline of the flow above the step .
the free - surface waves must necessarily remain the same , and in fact , the steepest decents analysis previously applied will _ still _ apply , as the ( now - hidden ) complex singularities of the step remain in place .
there is no contradiction .
our work here also opens the possibility of studying flows past bodies with singularities weaker than the branch points found in the ship and step flows . in a forthcoming paper , we will study flows past bodies with singularities in the _ curvature _ ( rather than the angle ) , such as for the bulbous sterns considered in @xcite .
the general framework for the study of steady irrotational two - dimensional free - surface waves past a moving body is presented in the companion paper @xcite , and we also refer readers to the textbook by vanden - broeck @xcite for a more extensive review . here , we summarize the basic equations .
we begin with a set - up of the three relevant coordinate systems .
first , the physical plane with @xmath14 is chosen so that the body , which is moving at dimensional speed @xmath24 , is fixed in a traveling frame of reference .
second , in the complex potential plane with @xmath15 , the free surface is assumed to lie at @xmath20 and the fluid region in @xmath197 . in the case of the semi - infinite surface - piercing flow ,
the body lies on @xmath198 and the free - surface on @xmath19 , and a stagnation point attachment is assumed at @xmath199 . for channel flow , the fluid is confined within a finite strip , say @xmath200 , with @xmath138 .
the non - dimensional depth of the strip is chosen to be @xmath201 for convenience .
the third coordinate system is required only for the case of channel flow . here
, an additional mapping @xmath202 brings the fluid region to the upper half-@xmath141-plane .
the channel bottom now lies on @xmath203 , and the free surface on @xmath204 . for consistency of notation for the case of surface - piercing flows
, we also consider the lower half-@xmath141-plane using the identity map @xmath205 .
the relevant coordinate systems are shown in fig .
[ fig : outline ] ( for the ship ) and [ formstep ] ( for the step ) .
a boundary integral formulation of the potential flow equations yields the following system of equations for @xmath2 and @xmath27 along the free surface : [ govsys ] @xmath206(\xi ) \label{bdint}.\end{gathered}\ ] ] the first , , corresponds to bernoulli s equation , and imposes a constant pressure condition on the surface .
the second , , involves a cauchy principal value integral , and is equivalent to the imposition of laplace s equation in the fluid region ; it thus provides a second relationship between @xmath2 and @xmath27 that closes the system . in this latter equation , we have split the integral between the negative real axis ( the body ) and the positive real axis ( the free surface ) , and defined the _ body function _ @xmath207 and the hilbert transform operator @xmath208 according to [ qsh ] @xmath209(\xi ) = \frac{1}{\pi } \dashint_{0}^{\infty } \frac{\theta(\xi')}{\xi ' - \xi } \ { \operatorname{d\!}{}}{\xi'}. \tag{\theequation a,\,b}\ ] ] above
, we furthermore introduce the sign function @xmath210 which accounts for the fact that , in the surface - piercing geometry , the in - fluid - region lies to the right of the free - surface for increasing @xmath194 , but to the left for channel flows .
a given body geometry can be specified by providing the values of @xmath27 over @xmath203 and then , by evaluating the integral ( [ qsh]a ) , a body function @xmath207 is obtained , that encodes the appropriate geometry .
notice that when @xmath37 , @xmath211 from , and thus @xmath212 .
this is the leading - order solution of a perturbative expansion @xmath213 and thus , @xmath214 .
for completeness , we summarize the main reduction of the water wave equations to the integral form .
the full details of this reduction can be found in @xcite .
the first step is to analytically continue the free surface quantities @xmath215 and @xmath216 into the complex @xmath13 ( or @xmath194 plane ) .
we thus set @xmath217 and @xmath218 .
however , due to the correspondence between the ( physical ) complex velocity @xmath15 ( with @xmath219 ) and the new _ complexified _
@xmath13 , we can relabel the analytic continuations to use @xmath44 , with a similar relabeling for @xmath141 .
thus we write @xmath220 and @xmath221 .
this step of analytic continuation is explained in more detail in @xcite , and the reader is also encouraged to see the visualization in fig . 2 of @xcite .
under this analytic continuation , the boundary integral equation becomes @xmath222 \quad \text{where } \quad { \hat{{\mathscr{h}}}}[\theta](\zeta ) = \frac{1}{\pi } \int_{0}^{\infty } \frac{\theta(\xi')}{\xi ' - \zeta } \
{ \operatorname{d\!}{}}{\xi'},\ ] ] for analytic continuation into the upper half-@xmath44 or @xmath141-plane .
now the solutions , @xmath2 and @xmath27 are expressed in terms of a regular series expansion with an error term , _
i.e. _ @xmath223 , @xmath224 as in .
note that the leading - order @xmath65 is given by @xmath207 in ( [ qsh]a ) and @xmath38 , known as the rigid - body flow .
the combination of bernoulli s equation and the boundary integral equation gives the result of @xcite : [ simpsys ] @xmath226{\bar{q}}= r(w ; { \hat{{\mathscr{h}}}}[{\bar{\theta } } ] ) + { \mathcal{o}}({\bar{\theta}}^2 , { \bar{q}}^2).\ ] ] where @xmath227 \exp\left ( 3{\mathrm{i}}{\mathrm{j}}\int_{w^*}^w \frac{q_1(\varphi)}{q_0 ^ 4(\varphi ) } \ , { \operatorname{d\!}{}}{\varphi}\right ) \label{p1 } \\ r(w ; { \hat{{\mathscr{h}}}}[{\bar{\theta } } ] ) = -{\mathcal{e}_\mathrm{bern}}+ { \mathrm{i}}{\hat{{\mathscr{h}}}}[{\bar{\theta}}]\frac{\cos\theta_r}{q_r^2 } , \label{rfunc}\end{gathered}\ ] ] and the error term , @xmath228 , represents the error in bernoulli s equation , and is given by @xmath229 the initial point of integration , @xmath66 , in is typically chosen at an ( integrable ) singularity of @xmath65 , given by ( [ qsh]a ) .
the point of integration , @xmath230 , can be chosen anywhere the integral in is defined , and only changes the constant of integration , @xmath231 .
the solution of the first - order differential equation can then be written as the integral in .
the initial point of integration is taken so as to satisfy the boundary or radiation conditions .
for the case of the ship , we set @xmath100 ( for the stagnation point ) , while for the case of the step , we set @xmath232 ( for a wave - free upstream flow ) .
as discussed in @xcite , different choices for @xmath48 can be used in order to obtain different approximations to the waves .
a particularly convenient and often very accurate choice is the two - term approximation , @xmath233 ) \sim { \epsilon}^2 \left(-\frac{5{\mathrm{i}}{\mathrm{j}}q_1 ^ 2}{2 q_0 ^ 4 } + q_1 ' + 2{\mathrm{i}}\frac{{\hat{{\mathscr{h}}}}[\theta_1]q_1}{q_0 ^ 3}\right),\ ] ] which was used to illustrate the numerical example of fig .
[ fig : outline ] .
a comparison of different choices for @xmath48 is presented in table 2 of @xcite .
p. h. trinh .
exponential asymptotics and stokes line smoothing for generalized solitary waves . in h.
steinrck , editor , _ asymptotic methods in fluid mechanics : survey and recent advances _ , pages 121126 .
springerwiennewyork , 2010 .
vanden - broeck and e. o. tuck
. computation of near - bow or stern flows using series expansion in the froude number . in _
2nd international conference on numerical ship hydrodynamics _ , berkeley ,
california , 1977 .
university of california , berkeley .
|
the standard analytical approach for studying gravity free - surface waves generated by a moving body often relies upon a linearization of the physical geometry , where the body is considered asymptotically small in one or several of its dimensions . in this paper ,
a methodology that avoids any such geometrical simplification is presented for the case of flows at low speeds . the approach is made possible through a reduction of the water - wave equations to a complex - valued integral equation that can be studied using the method of steepest descents .
the main result is a theory that establishes a correspondence between a given physical flow geometry , with the topology of the riemann surface formed by the steepest descent paths .
then , when a geometrical feature of the body is modified , a corresponding change to the riemann surface is observed , and the resultant effects to the water waves can be derived
. this visual procedure is demonstrated for the case of two - dimensional free - surface flow past a surface - piercing ship and over an angled step in a channel .
|
strong correlations in impurity problems have been of tremendous theoretical interest.@xcite this interest has been spurred by the ability to engineer semiconducting quantum dots which are highly tunable.@xcite in such realizations of quantum dots , both the gate voltages of the quantum dot as well as the tunneling amplitudes between the dots and the leads can be tuned .
this gives one the ability to study these systems throughout their entire parameter space .
while the first reported dot systems involved single dots,@xcite more complicated dot structures can now be fabricated.@xcite with this ability to engineer multi - quantum dot systems comes the ability to realize more exotic forms of kondo physics . in double dot systems , for example , one can explore the competition between the kondo effect and the ruderman - kittel - kasuya - yosida ( rkky ) interaction.@xcite because of the non - perturbative nature of the kondo effect at low temperatures , one might expect this competition to be non - trivial .
we argue that it is in fact so for double dots arranged in parallel . the geometry that we wish to consider is sketched in fig . 1 . here
two closely spaced , single level dots do not directly interact with one another , either via tunneling or via a capacitive coupling .
they are however connected to one another via hopping processes involving the two leads .
we will want to consider a situation where a total of two electrons sit on the two dots and where the leads are symmetrically coupled to the two dots . in such a scenario only one continuum channel of electrons ( the even combination of the two leads ) couples to the two dots .
in such a system the kondo effect and the rkky interaction are antithetical to one another .
the kondo effect will promote many - body singlet formation while the rkky interaction will tend to encourage direct triplet formation between the two electrons on the two dots .
one can then imagine ( at least ) two possible scenarios for ground state formation as one lowers the temperature ( see fig .
2 ) . in the first scenario the rkky interaction dominates and at some intermediate temperature binds the two electrons on the two dots into a triplet . as the temperature is furthered lowered the kondo effect then comes into play partially screening the effective spin-1 object on the dots down to an effective spin-1/2 ( i.e. an underscreened kondo effect ) . here
the physics is non - fermi liquid with low temperature quantities such as conductance having a logarithmic dependence in temperature .
this scenario has been argued to hold in a number of studies including those based on a numerical renormalization group ( nrg).@xcite the particular implementation of the nrg there , however , has been criticized.@xcite a second scenario can be envisioned characterized by a lack of a clear separation of the rkky effect from kondo physics ( i.e. the persistence of higher temperature rkky - induced triplet formation down to temperatures below a putative kondo temperature ) . in this scenario
the low temperature state is an overall many - body singlet . at low temperatures
any effective binding by an rkky interaction into a triplet state comes undone .
rather the two electrons on the two dots participate in the singlet by screening one another in conjunction with the lead electrons .
the physics in this scenario is sharply differentiated from the first by being fermi liquid .
in particular the deviations in the conductance with temperature go as @xmath1 .
this scenario has been supported by both bethe ansatz computations@xcite as well as
slave boson mean field theory treatments.@xcite it is the purpose of this paper to provide supporting evidence for the second scenario . in it
we present a @xmath2 , @xmath0 treatment of the problem .
the method of @xmath0 expansion is well known@xcite and has been used in the context of quantum dots / impurities.@xcite this treatment is distinct from the use of slave boson mean field theory ( sbmft ) to argue the validity of the second scenario .
while sbmft is itself a @xmath0 technique , it represents some ( uncontrolled ) summation of @xmath0 diagrams . here
instead we consider a @xmath0 expansion in a systematic fashion .
in particular we argue that this @xmath0 expansion is consistent with the friedel sum rule a hallmark of fermi liquid physics and so indicative of the second scenario .
( in sbmft , the friedel sum rule is automatically satisfied by virtue of the quadratic form of the action . )
the paper is organized as follows . in section
ii we briefly describe the double - dot model we are interested in studying and the @xmath0 method used . in section
iii we present results for the dot green s functions , the system s partition function , and the dot occupancy . in section
iv we show that the friedel sum rule holds and so the system appears to be a fermi liquid , the central result of this paper . in section
v we compute the conductance using the friedel sum rule and compare it to slave boson mean field .
we then draw conclusions in section iv .
details related to computing diagrams in the @xmath0 expansion are relegated to appendices .
the anderson like hamiltonian that we study is given by @xmath3 the @xmath4 specify electrons with spin @xmath5 living on the two leads , @xmath6 .
the @xmath7 specify electrons found on the two dots @xmath8 .
electrons can hop from the leads to dots with tunneling strength @xmath9 .
we suppose that there is no interdot coulomb repulsion and that tunneling between the two dots is negligible .
we further consider this model in the limit that the strength of the coulomb repulsion on the two dots , @xmath10 , is taken to @xmath11 thus precluding any double occupancy .
the constraint of no double - occupancy is fulfilled by first introducing the slave boson formalism , one slave boson for each dot : @xmath12 here @xmath13 is the pseudofermion ( denoted later in feynman diagrams by dashed lines ) which annihilates one `` occupied state '' on dot @xmath14 and @xmath15 is a bosonic operator ( denoted later by wavy lines ) which creates an empty state on dot @xmath14 .
we subject these new degrees of freedom to the constraint , @xmath16 which we enforce by adding two lagrange multipliers @xmath17 and @xmath18 to the hamiltonian @xmath19 we thus understand the partition function of the system to be computed via @xmath20 in treating quantities in this model , we will perform a systematic diagrammatic expansion in @xmath0 where @xmath21 is the degeneracy of each dot .
we suppose the dot - lead couplings , @xmath9 , satisfy the following ratio condition , @xmath22 this condition ensures that only one effective channel of lead electrons couples to the two dots . to see this we reexpress the hamiltonian in terms of even and odd channels , @xmath23 where @xmath24 . in terms of these new degrees of freedom ,
the hamiltonian divides itself into an even and an odd sector , @xmath25 and we see that the odd sector of the hamiltonian is decoupled from the dots .
in this section we will compute the green s functions of the dot degrees of freedom , the dot partition function , and the dot occupancy to leading order and where necessary to prove the fsr the subleading order in @xmath0 . in terms of the slave bosons and fermions , the dot temperature green s functions
are defined as @xmath26 where the @xmath27 indicate which dot the electron is sitting on and @xmath28 is the part of the hamiltonian coupling the leads to the dots .
as computed in the appendices , the lowest order contributions in @xmath0 to these green functions are @xmath29\frac{k_{2}(e_{02})}{i\omega_{n}-t_{a_{2}}},\end{aligned}\ ] ] where @xmath30^{-1}\mid_{z = e_{01,02}}.\end{aligned}\ ] ] in eqn .
( [ eq : g11o1 ] ) one can make the replacement @xmath31 to get the corresponding expressions for @xmath32 and @xmath33 .
we see that the interdot correlation function , @xmath34 , first sees a contribution at @xmath35 . here
@xmath36 denotes the energy of the conduction electrons and @xmath37 is the fermi function .
finally , we note that @xmath38 is the most negative solution of @xmath39 and we correspondingly define @xmath40 as @xmath41 . for the purposes of proving the friedel sum rule , we will need not only these contributions to the dot green s functions but the imaginary part of the subleading ( @xmath35 ) correction to @xmath42 and @xmath43 as well .
these are as follows : @xmath44=-\frac{k_{1}^2(e_{01})\gamma_{1}}{\left(\omega - t_{a_{1}}\right)^{2}}-\pi\theta(-\omega - t_{a_{1}})r(\omega)-\pi\theta(\omega - t_{a_{1}})\left[s(\omega)+t(\omega)\right],\ ] ] where @xmath45 , and @xmath46 are defined by @xmath47^{2}\int_{t_{a_{1}}}^{-\omega}\frac{dk}{\left(\omega+\epsilon_{k}-t_{a_{1}}\right)^{2}}\frac{1}{\left(\epsilon_{k}-\left(\frac{n\gamma_{1}}{\pi}\right)\log(\frac{\epsilon_{k}}{t_{a_{1}}}-1)\right)^{2}+\left(n\gamma_{1}\right)^{2}};\cr\cr s(\omega ) & = & \frac{k_{1}(e_{01})}{n}\left[\frac{n\gamma_{1}}{\pi}\right]^{2}\frac{1}{\left(\omega - t_{a_{1}}\right)^{2}}\int_{t_{a_{1}}}^{\omega}\frac{dk}{\left(\epsilon_{k}-\left(\frac{n\gamma_{1}}{\pi}\right)\log(\frac{\epsilon_{k}}{t_{a_{1}}}-1)\right)^{2}+\left(n\gamma_{1}\right)^{2}};\cr\cr t(\omega ) & = & \frac{k_{1}(e_{01})}{n}\left[\frac{n\gamma_{1}}{\pi}\right]^{2}\left[\frac{1}{t_{a_{1}}}-\frac{1}{\omega}\right]\frac{1}{\left[\omega - t_{a_{1}}+\left(\frac{n\gamma_{1}}{\pi}\right)\log(\frac{\omega}{t_{a_{1}}})\right]^{2}}.\end{aligned}\ ] ] we obtain @xmath48 by swapping @xmath31 .
the above expression for @xmath49 undergoes a dramatic simplification when @xmath50 is set to 0 .
we find @xmath51=-\frac{k_1 ^ 2(e_{01})\gamma_1}{t_{a_1}^2}. \label{fsrlimit}\ ] ] this simplification will be useful for proving the friedel sum rule .
the partition function and the dot occupancy obey the simple relation @xmath52 the partition function at @xmath53 is given by @xmath54 and @xmath55 are defined as in section iii a. after some algebra we then find that @xmath56 where @xmath57 this @xmath58 computation of the dot occupancy is sufficient for proving the friedel sum rule at leading order in @xmath0 .
the friedel sum rule ( fsr ) states for a system with n - fold degeneracy that the scattering phase , @xmath59 , of an electron at the fermi surface @xmath60 is given in terms of the total occupancy , @xmath61 , of the dots by @xmath62 the friedel sum rule implies that the zero temperature linear response conductance is then given by @xmath63 where @xmath64 , @xmath8 . from ref .
, the linear response conductance can be expressed in terms of the dot green s functions as follows : @xmath65 \label{eq : meir}\ ] ] where @xmath66 is the full retarded green s function matrix in dot - space and @xmath67 is the bare hybridization matrix .
these are defined as @xmath68\label{eq : g_matrix},\ ] ] and @xmath69\label{eq : gamma_m}.\ ] ] thus to show that the fsr holds we need to show that @xmath70\label{eq : toprove}\ ] ] holds in terms of a @xmath0 expansion .
the left hand side of ( [ eq : toprove ] ) to @xmath71 is equal to @xmath72 where we only consider the dot occupancies to @xmath53 .
the right hand side of ( [ eq : toprove ] ) to the same order equals @xmath73&=&-\frac{2}{(1+\lambda^2)}\mbox{im}\biggl\ { v_{11}^{2}g_{11}^{\mbox{o(1/n)}}+v_{11}v_{12}g_{21}^{\mbox{o(1/n)}}\nonumber\\&+&v_{11}v_{12}g_{12}^{\mbox{o(1/n)}}+ v_{12}^{2}g_{22}^{\mbox{o(1/n)}}\biggr\ } \label{eq:2trc}\end{aligned}\ ] ] where all the green s functions were evaluated to the given order in the previous section .
note that @xmath74 and so does not appear in the above . using eqs .
[ fsrlimit ] and [ nd1nd2 ] one can then readily verify that the fsr ( [ eq : toprove ] ) holds for double dots to leading order in a @xmath35 expansion .
this then matches the proof of the fsr in single dots in a large - n expansion performed in ref . .
plot showing the linear response conductance for two different values of separation between the dots chemical potential . in the left panel we present results arrived at from the application of sbmft done in ref . and in the right panel the results coming from the large - n expansion .
] having proven the fsr , we will now use it to compute the linear conductance , @xmath75 , i.e. we will use the expression @xmath76}{n}\right),\label{eq : linear_cond_diag}\ ] ] where @xmath77 and @xmath78 are calculated from diagrams of the partition function @xmath79 ( see section iii b ) . in fig .
3 we present the conductance for two different spacings between the chemical potentials of the two dots ( one large and one small ) .
we compare the large - n computation done in this paper with the results of ref . .
we find remarkable agreement between the two approaches .
not only are the general trends in the conductances similar , but more detailed features such as inflection points are reproduced .
as the dot chemical potentials are lowered we see that the conductance vanishes . in fig .
4 we plot the corresponding total dot occupancies both from the @xmath0 expansion ( to lowest order in @xmath0 ) and sbmft ( from ref . ) .
we again see that there is a close correspondence between the two methods .
this is notable as in computing the dot occupancies from the @xmath0 expansion we only keep @xmath53 diagrams while the sbmft accounts for diagrams at all orders of @xmath0 .
plot showing the total occupation of the dots for two different values of separation between the dots chemical potential . as with fig .
2 we show the values of the dot occupancy both computed in sbmft and in the 1/n expansion . ]
in this paper we have studied a double - impurity model using a @xmath0 expansion where @xmath21 is the degeneracy of each dot level . using this @xmath0 expansion we demonstrated that the friedel sum rule holds , so providing complementary evidence that double dots in parallel in the absence of interdot and capacitive coupling have fermi liquid ground states . in previous works
we have argued that this ground state is fermi liquid using both the bethe ansatz@xcite and sbmft.@xcite this then provides a complementary piece of evidence for this fermi liquid nature .
such evidence is needed because there exists in the literature an opposite set of results coming from numerical renormalization group analyses.@xcite in ref .
we have suggested a possible reason for this discrepancy , namely in discarding modes that arise from the use of logarithmic basis , the nrg discards additional effective screening channels that would completely screen any putative spin-1 object that might form under a ferromagnetic rkky interaction .
but as also pointed out in ref .
, another possible reason for the discrepancy exists . in treating the double dots in parallel all of these methodologies ( bethe ansatz , sbmft , @xmath0 , nrg )
treat the dots as point scatterers in a continuum media .
then with two dots in close proximity to one another there exists the possibility that the physics might be influenced by the particular implementation of a uv cutoff that the methodology employs .
if the physics was sensitive to this implementation , one would expect to find non - universal results .
it is notable then that we were able to conclude in this study that the @xmath0 expansion produces results ( i.e. a fermi liquid ground state ) equivalent to that of sbmft and the bethe ansatz .
this finding then gives additional weight to the universal nature of the fermi liquid ground state of double dots in parallel .
we will end on a tangential note by stressing the remarkable similarity between results obtained for a double dot in a ( low order ) @xmath0 expansion and sbmft .
this would not be necessarily expected as the sbmft represents some uncontrolled sum over diagrams at all orders in @xmath0 and @xmath21 is merely 2 .
it is an interesting question why these higher order diagrams are suppressed strongly even when @xmath80 .
we note though that while the @xmath0 expansion agrees quantitatively with sbmft and while both of these methods agree on a qualitative level with the bethe ansatz , these methods do not produce fine details of the conductance of the double dots brought out by the bethe ansatz ( see ref . ) , presumably the most trustworthy method of the three as it represents an exact solution of the problem .
acknowledgements : m.k . was supported by the nsf under grant no .
dmr- 0906866 .
r.m.k . acknowledges support by the us doe under contract no .
de - ac02 - 98 ch 10886 .
we thank a. tsvelik and p. coleman for useful discussions .
following the definition of the green s function in eq .
[ eiiii ] , @xmath42 , can be written as @xmath82 a product of diagrams .
these diagrams are given in figs .
[ l1l2 ] and [ noleg ] and these are the only diagrams at @xmath58 that survive the projection @xmath83 involved in computing any correlation function .
the argument @xmath84 of @xmath85 and @xmath86 indicates the diagram involves dot electrons on dot @xmath87 .
@xmath88 involves a dressed boson propagator , @xmath89 , for dot 1 with self energy @xmath90 defined in eq .
[ eq : sigma12 ] .
this dressing occurs already at @xmath53 and is straightforwardly obtained via a dyson equation ( see fig . 4 in ref . as well as ref .
): @xmath91 where @xmath92 is the bare propagator of the boson line .
@xmath93 is then equal to @xmath94 in eqn .
[ eaii ] the summation over matsubara boson frequencies can be performed by the usual technique of converting the summation into a contour integration .
doing so we find @xmath95 here @xmath96 is the most negative solution of @xmath97 , @xmath98 is the bose function , and @xmath99 is given by @xmath100^{-1}\mid_{z = e_{01}}.\end{aligned}\ ] ] in order to perform the @xmath101 integration we note that the bose function can be expanded as @xmath102 . from this
it follows straightforwardly that @xmath103 to determine @xmath96 , we note that @xmath104 evaluates to @xmath105 eq .
[ eavii ] is obtained after converting the summation into an integration over k which brings in a factor of @xmath106 . as @xmath96 is the most negative solution of @xmath107
which satisfies @xmath108 we can write @xmath109 and see that @xmath110 is always positive .
@xmath110 is , in effect , the renormalized dot chemical potential of dot 1 .
given above are the diagrams involved in the evaluation of the numerator of the expression for @xmath42 at both @xmath53 ( @xmath93 ) and @xmath35 ( @xmath111 , and @xmath112 ) .
the argument @xmath87 of @xmath113 indicates the diagram involves only dot electrons of species j. ( and so @xmath114 does not have this argument as it involves both . )
the wavy thick lines indicate dressed boson propagators , the dashed lines , bare pseudofermion lines , and the solid ( internal ) lines , conduction electron propagators .
the labels and colors ( ( 1,black ) and ( 2,red ) ) indicate the lines belong to the bosons , @xmath115 , and fermions , @xmath116 , of different dots . ]
we will now compute @xmath117 a bubble diagram ( see fig .
[ l1l2 ] ) .
it differs from @xmath118 in that it also involves a conduction electron line ( represented as a solid line in fig .
[ l1l2 ] ) and involves no external frequency .
it is important to note that the boson propagator in this case is different from the boson propagator in the diagram represented by @xmath93 .
the boson propagator in @xmath119 involves summing over a class of diagrams , each of which comes with an additional factor of @xmath120.@xcite in order to make this sum correspond to the usual geometric series@xcite we introduce a _ coupling constant parameter _
@xmath121 by the transformation of the coupling constant @xmath122 . by suitably differentiating the series by @xmath121 one can turn this series into a regular geometric series.@xcite we thus arrive at the expression : @xmath123 after performing the matsubara summation over @xmath124 and the projection integral , and transforming the expression for the matsubara sum over @xmath125 to a contour integral , we obtain @xmath126 proceeding with the coupling constant integration in eqn .
[ eaviii ] , we find , @xmath127 if we now perform an integration by parts and then compute the contour integration about the pole at @xmath128 , we find , @xmath129 we finish up then with @xmath130 a similar expression for @xmath32 is found by interchanging all species indices @xmath131 .
given above are the diagrams involved in the evaluation of the denominator of the expression for @xmath42 at both @xmath53 ( @xmath119 ) and @xmath35 ( @xmath132 and @xmath133 ) . ]
@xmath136 is given by the following set of contributions , @xmath137 the first and third terms form the @xmath35 contribution to @xmath42 of a single dot while the second and fourth terms reflect the presence of a second dot .
we consider the latter first .
we will first consider the diagrams which contain both dot electron species , @xmath138 and @xmath133 ( see figs . 5 and 6 ) .
we will show that both of them go to zero in the low temperature limit .
let us look at @xmath138 first : @xmath139 performing the matsubara sums over @xmath140 and @xmath141 , we arrive at an expression given by ( dropping terms that will disappear once we apply the projection ) : @xmath142.\end{aligned}\ ] ] of the two terms in the square brackets in the above expression , the first dominates in the low temperature limit as it is proportional to @xmath143 and @xmath144 .
we can thus neglect the second term in this limit .
there are other such terms which are subdominant as @xmath145 goes to 0 and which are indicated by the ellipses .
if we perform the final matsubara sum together with the @xmath146 sum we obtain an expression involving terms of the form @xmath147 and @xmath148 \frac{1}{(\epsilon_{k}-t_{a_{2}})}\frac{1}{(\epsilon_{k}-t_{a_{1}})}.\ ] ] it is straightforward to show that these integrals vanish as @xmath149 in the large bandwidth limit .
now we will calculate the bubble diagram @xmath133 ( see fig .
we have @xmath150 performing the matsubara sums as well as the @xmath146 sum leaves us with an expression again involving terms of the form found in eqn .
( [ eqn ] ) and so like @xmath112 , @xmath133 goes to zero in the infinite bandwidth @xmath151 limit .
in the zero temperature , large bandwidth limit , @xmath136 reduces to @xmath152 but this is precisely the @xmath134 correction that is seen in single level dots .
we can thus use ref . to complete the evaluation of @xmath136 : @xmath153_{z = e_{01}};\cr b_{1}(i\omega_{n } ) & = & \frac{k_{1}(e_{01})}{(i\omega_{n}-t_{a_{1}})^{2}}\biggl[2\gamma_{1}\sum_{k } \frac{1-f(k)}{\left(e_{01}+i\omega_{n}-\epsilon_{k}-\sigma_{1}(e_{01}+i\omega_{n}-\epsilon_{k})\right)}\cr & & \hskip 1 in -k_{1}(e_{01})\sigma_{1}^{1/n}(e_{01})]\biggr];\cr b_{2}(i\omega_{n } ) & = & 2\gamma_{1}k_{1}(e_{01})\bigg ( -\sum_{k}\frac{f(k)}{(\epsilon_{k}-t_{a})^{2}\left(e_{01}-i\omega_{n}+\epsilon_{k}-\sigma_{1}(e_{01}-i\omega_{n}+\epsilon_{k})\right ) } \cr & & \hskip .9 in + 2n\gamma_{1}\sum_{k , k^{\prime}}f_{k}(1-f_{k^{\prime } } ) \frac{1}{(\epsilon_{k}-t_{a_{1}})^{2}}\frac{1}{(i\omega_{n}+\epsilon_{k}-\epsilon_{k^{\prime}}-t_{a_{1}})}\cr & & \hskip 1.2 in \times\frac{1}{(e_{01}+\epsilon_{k}-\epsilon_{k^{\prime}}-\sigma_{1}(e_{01}+\epsilon_{k}-\epsilon_{k^{\prime}}))^{2}}\bigg).\end{aligned}\ ] ] we have used the same notation as in ref . .
the lowest order contribution to @xmath154 occurs at @xmath134 and can be written as @xmath155 we already have @xmath156 from appendix a. the diagram representing @xmath157 is given in fig .
[ fig : m1 ] .
using calculations similar to appendix [ app : g11low ] one easily arrives at @xmath159 which in combination with the expression for @xmath156 gives , @xmath160 note that the term @xmath161 in eq .
[ ebiii ] results from the conduction line in fig .
[ fig : m1 ] .
|
a large - n diagrammatic approach is used to study coupled quantum dots in a parallel geometry .
we show that the friedel sum rule ( fsr ) holds at lowest order in a 1/n expansion for this system , thereby suggesting that the ground state is a fermi liquid . using the fsr together with the dot occupancy
, we compute the dot system s conductance .
our finding that the @xmath0 expansion indicates the system is a fermi liquid is in agreement with both prior results based on the bethe ansatz and slave boson mean field theory .
|
the south pole infrared explorer ( spirex ) telescope was a prototype system , developed to test the feasibility of building , operating , and maintaining an infrared ( ir ) telescope during an antarctic winter .
the initial driver for the spirex project was to exploit the conditions at the south pole that make it an excellent site for 35@xmath0 m observations that is the high altitude , the low temperatures , low precipitable water vapour content in the atmosphere , and the stable weather conditions ( burton et al .
1994 ; marks et al . 1996 ; hidas et al .
2000 ; marks 2002 ) . the spirex 60 cm telescope began operations at the amundsen scott south pole station in 1994 ( hereld et al .
spirex was initially used as part of a campaign to measure the south pole s thermal background , sky transparency , and the fraction of time useful for ir observations ( see e.g. ashley et al .
1996 , nguyen et al .
1996 , phillips et al .
1996 and chamberlain et al .
2000 for further details on the ir sky conditions ) .
due to the very low thermal background , spirex could perform longer integrations in the 35@xmath0 m regime , compared to temperate latitude facilities .
in addition to continuum emission which , at these wavelengths pinpoints young embedded objects , there are many astrophysically significant molecular lines accessible ; e.g. those from hydrogen and pah molecules
. observations at 35@xmath0 m are difficult at most temperate sites , making studies involving these lines limited , and thus the spirex dataset unique . from 19941997 spirex was equipped with the grim ( grism imager ) camera .
this detector contained 128@xmath4128 pixels and was sensitive from 12.5@xmath0 m . in 1998
the telescope was equipped and with the abu camera , which incorporated an engineering grade 1024@xmath41024 aladdin detector ( fowler et al .
this camera was sensitive from 2.45@xmath0 m , and provided a 10 field of view image with a 0.6 pixel scale .
six science filters were available with spirex / abu .
the three narrow - band filters were optimized to isolate emission from molecular hydrogen ( h@xmath2 at 2.42@xmath0 m ) , polycyclic aromatic hydrocarbons ( pahs at 3.29@xmath0 m ) , and hydrogen line emission ( br @xmath3 at 4.05@xmath0 m ) .
the three broad - band filters covered the l , l and m bands .
the filter parameters and achieved sensitivities are listed in table 1 ( see burton et al . ,
2000 ) .
[ cols="<,^,^,^,^,^ " , ] + @xmath5 + @xmath6 + @xmath7 + spirex / abu was well suited for studies of star forming complexes .
young stars containing circumstellar disks have a colour excess in the ir due to the absorption and re - emission of radiation from the central star by the surrounding material .
recent studies have found that the l band may be the optimal wavelength for detection of star and disk systems .
compared to the ( h k ) colour [ @xmath8 1.62.2@xmath0 m ] , the ( k l ) colour [ @xmath8 2.23.5@xmath0 m ] is more sensitive to the presence of a disk ( haisch et al .
2000 ; lada et al .
2000 ; kenyon & hartmann 1995 ) . observations of pah molecular line emission across star forming complexes , allows one to study their environments .
the fluorescent emission from pah molecules trace regions ( known as photodissociation regions or pdrs ) , where stellar uv radiation is heating the molecular gas ( hollenbach & tielens 1997 ) .
pah emission delineates externally heated molecular clouds and reveals the interactions between nearby massive stars and any remnant molecular material .
the first astronomical results from spirex were obtained when shoemaker - levy 9 collided with jupiter in 1994 . using the grim camera at a wavelength of 2.36@xmath0 m ,
images were obtained at 5 minute intervals and captured 16 of the fragments , showing evidence of impact with jupiter in 10 cases ( severson 2000 ) .
an extended halo around the edge - on spiral galaxy eso240g11 was also imaged using grim at 2.4@xmath0 m ( rauscher et al . , 1998 ) , reaching a sensitivity level of 25 mags / arcsec@xmath9 .
the following sections discuss a sample of data obtained from the two years of operation of spirex / abu ( 199899 ) at the south pole .
observations were conducted toward a number of different complexes .
the characteristics of these range from young to old , low- to high - mass , and near and far star forming complexes .
in particular , results are presented here for ngc 6334 , chamaeleon i , @xmath1 chamaeleontis , the carina nebula , 30 doradus , rcw 57 , rcw 38 , and the galactic centre .
in addition , spirex was used to search for an infrared counterpart to the gamma - ray burst grb990705 at 3.5@xmath0m(masetti et al . , 2000 ) , though no source was detected to a limit of 13.9 magnitudes after 2 hours of integration .
for each source position , a series of frames were obtained at the specified integration time and then averaged .
these parameters varied depending on the observing wavelength , the properties of the filter ( narrow- or broad - band ) and the weather conditions .
typical values for each filter are given in table 1 .
all observations were conducted by the winter - over scientists at the south pole station ; rodney marks ( 1998 ) and charlie kaminsky ( 1999 ) . the sequence for all observations consisted of a set of sky frames followed by two sets of object frames .
each set consisted of five averaged frames offset by from the previous .
this sequence was repeated allowing the easy removal of sky emission and artifacts from the array .
archived images were used for dark subtraction and flat - fielding .
observations of standard stars were obtained before and after all on - source observations for flux calibration .
the majority of the data presented here was reduced by joel kastner using the spirex / abu data pipeline ( the exceptions are ngc 6334 and the carina nebula ) .
the data pipeline was a joint project of the rochester institute for technology center for imaging science ( rit cis ) , the national optical astronomy observatories ( noao ) and the center for astrophysical research in antarctica ( cara ) . to cover the star forming complexes of ngc 6334 and the carina nebula ,
many adjacent positions were observed .
common stars in adjacent frames were used to align the images and create the final larger mosaic ( using iraf routines written by peter mcgregor ) . the final pah band mosaiced image for the carina nebula contained a total of 72 individual images , and for ngc 6334 the mosaic pah band image contained 304 images .
ngc 6334 is a young massive star forming region at a distance of 1.7kpc .
it contains seven distinct sites of ongoing star formation along a central molecular ridge .
1(a ) shows the 3.29-@xmath0 m data toward ngc 6334 .
the emission within this band comprises both pah and continuum emission .
a massive embedded star is seen at each of several sites of star formation along the ridge , each @xmath10pc apart .
they are surrounded by complex loops and filaments of pah emission .
these pdrs have formed on the edge of the remnant molecular cloud , as radiation from the young stars carves the material and ionizes their surroundings .
analysis of the pah emission and pdr features across the central ridge of ngc 6334 have been presented in jackson et al .
( 1999 ) and burton et al .
the data shown here expand on these studies , and further reveal the pdr structure adjacent to the central star forming ridge .
in addition to the pah
band images shown here , l band images were also obtained across ngc 6334 . when combined with near
ir data from the 2mass point source catalog they allow us to produce near ir colour colour diagrams ( e.g. , ( j h ) vs.(h k ) and ( j h ) vs.(k l ) diagrams ) .
results presented in rathborne et al .
( 2003 ) find 11 sources with a large ir excess using the ( k l ) colour , compared to just a single source using the ( h k ) colour excess , confirming that the ( k l ) colour is far more sensitive to the detection of circumstellar disks .
containing in excess of 100 pre - main sequence stars , the chamaeleon i dark cloud is one of the most active regions of nearby low - mass star formation .
the central 0.5deg@xmath11 of the complex was observed in the l band using spirex / abu .
these images reveal all of the known pre - main sequence stars ( to an l @xmath12 11 ) .
kenyon & & g ' omez ( 2001 ) combined jhk observations obtained at the ctio with the spirex / abu data , to construct near ir colour colour diagrams .
they find the fraction of sources with an ir excess to be 58 @xmath13 4% ( complete to an l @xmath14 11 ) .
in addition , they also confirm that sources with an ir excess are more easily identified when using the ( k l ) colour than ( h k ) .
the @xmath1 chamaeleontis cluster is one of the nearest to the sun , lying just 97pc away .
it is also of intermediate age for a pre - main sequence system , @xmath15myrs old , and somewhat older than a number of other clusters where the formation and evolution of circumstellar disks has been studied .
importantly , its proximity and compactness mean that there is a complete population census for its members 15 stars ranging from 0.2 to 3.4m@xmath16 in mass .
lyo et al.(2003 ) imaged the cluster using spirex / abu in the l band , finding 60% of the stars had ir excesses attributable to the presence of disks around them .
of those with disks , half showed a clear relation between the strength of the ir excess at 3.5@xmath0 m and the equivalent width of the h@xmath3 line emission , implying continuous accretion .
the lifetime of the disks in @xmath1 chamaeleontis ( @xmath17 9myrs ) is significantly longer than in a number of other systems that have been studied , with lifetimes of 36myrs ( e.g. , haisch et al . , 2001 ) .
the carina nebula is a massive star forming region at a distance of 2.2kpc .
because no protostellar objects had been found within this nebula , it was thought to be ` evolved ' and devoid of current star formation activity ( e.g. , cox & bronfman 1995 ) .
it does however , contain 53 o type stars and includes the clusters tr 14 and tr 16 . at its centre
lies the massive star @xmath1 car and the keyhole nebula .
optical and near ir images show a high amount of extinction toward this complex , with many dark lanes , globules , clumps , and filaments .
two large molecular clouds are located at the edges of the complex , in close proximity to the massive stars .
1(b ) shows the pah band emission across the central carina nebula ( these data are also discussed in brooks et al .
2000 and rathborne et al .
when combined with 2.12@xmath0 m h@xmath2 line emission , msx 8 and 21@xmath0 m images , sest molecular line maps , and most 843mhz radio continuum images , three different environments are revealed across the complex : ( i ) the keyhole nebula containing discreet , dense molecular clumps with pdrs on their surfaces ; ( ii ) pdrs at the edges of both the southern and northern molecular clouds ; and ( iii ) heated dust , intermixed with the pdrs surrounding tr 14 and the northern molecular cloud .
several 3.29 , 8 and 21@xmath0 m point sources were also located across the complex , with spectral energy distributions corresponding to compact hii regions .
interestingly , these were all found on the edges of pdrs which appear to have been carved out by the stellar winds and radiation from the nearby massive stars .
these results suggest that star formation is indeed ongoing within the carina nebula and may in fact be triggered by the interactions resulting from the nearby massive stars .
the 30 doradus region is the brightest hii region in the large magellanic cloud .
it contains a central cluster of massive stars surrounded by extended nebulosity in the near ir , with many protostellar candidates . to study the nature of the embedded stellar population in more detail ,
l band observations were obtained with spirex / abu .
the data presented in fig .
1(c ) are the most sensitive ever obtained of the 30 doradus complex . the faintest start seen in the image , which required 9.25 hours of on - source integration , has l = 14.5 magnitudes .
many sources with an ir excess are revealed when the l band data are combined with jhk observations .
several embedded massive stars are apparent and in addition , a population census of the young stellar objects can be conducted .
this still remains the deepest , wide - field l band image ever obtained , despite the small size of the telescope , with an extended source sensitivity ( @xmath18 ) of 18.2 magnitudes per square arcsecond .
rcw 57 is a bright southern hii region .
it contains a tight cluster of massive stars and shows extended infrared nebulosity .
band observations of the central 10 of this complex were obtained with spirex / abu ( fig .
this image reveals many point sources , in addition to both bright and diffuse extended emission . using jhk and the spirex / abu l band data
, studies are currently underway into the star formation history of this complex .
rcw 38 is also a bright hii region containing a tightly packed cluster of stars . near
ir images of this complex show extended nebulosity , with dust lanes and dark patches .
coincident with many of these features is extended 3.29@xmath0 m pah emission , as shown in fig .
these features trace the pdrs and delineate the molecular material .
in addition , they wrap around the ionized material seen at 2@xmath0 m .
observations of the galactic centre obtained with spirex / abu are far less affected by the extinction than at optical and near
ir wavelengths . the 4.05@xmath0 m image in fig .
1(f ) clearly demonstrates the advantages of this wavelength regime .
most of the 4.05@xmath0 m emission originates from heavily extincted sources , though there is also some br @xmath3 emission , pinpointing regions of ionized gas .
three sources , of roughly equal brightness , are prominent in the nucleus at 4@xmath0 m .
this is in contrast to the view at 2@xmath0 m , which is dominated by the source irs 7 .
these three sources are each separated by @xmath19 and centred roughly on the presumed nucleus , sgra@xmath20 . at the spatial resolution achieved with spirex
they are associated with the sources irs 1w , irs 7/irs 3 and irs
13e / irs 2l , respectively ( as imaged by clenet et al .
2001 , in the l band ) .
although spirex / abu was a prototype facility , it nevertheless obtained deep , high - resolution , wide - field images in the 35@xmath0 m regime toward many star forming complexes .
these data are not only unique , but when combined with complementary near , mid
ir , and radio continuum observations , reveal the inner environments of star forming complexes ( by tracing the pdrs ) , and identify the youngest objects with circumstellar disks ( through their high l band fluxes ) .
the success of the spirex / abu system lends strong support to current plans to build larger ir telescopes on the antarctic plateau , where they would provide the most sensitive facilities for 35@xmath0 m observations on the earth .
the spirex project was a collaboration between the usa and australia , involving the national optical astronomy observatories ( noao ) , the united states naval observatory ( usno ) , the center for astrophysical research in antarctica ( cara ) , boston university , goddard spaceflight center ( gsfc ) , ohio state university ( osu ) , rochester institute of technology ( rit ) , the university of chicago ( uc ) , the university of new south wales ( unsw ) , the australian national university ( anu ) and the universities space research association ( usra ) .
we are indebted to the dedicated efforts of our many colleagues here who together have demonstrated that infrared astronomy can indeed be conducted from the antarctic plateau .
ashley , m. c. b. et al . , 1996 , pasp , 108 , 721 brooks , k. j. et al .
, 2000 , mnras , 319 , 95 burton , m. g. et al . , 1994 , pasa , 11 , 127 burton , m. g. et al . , 2000 , apj , 542 , 359 cox , p. & bronfman , l. , 1995 , a&a , 299 , 583 chamberlain , m. a. et al . ,
2000 , apj , 535 , 501 clenet , y. et al . , 2001 , a&a , 376 , 124 fowler , a. et al .
, 1998 , proc .
, 3354 , 1170 haisch , k. e. , lada e. a. , lada c. j. , 2000 , aj , 120 , 1396 haisch , k. e. , lada , e. a. , lada , c. j. , 2001 , apjl , 553 , l153 hereld , 1990 , proc .
spie , 1235 , 43 hidas , m. g. et al . , 2000 , pasa , 17 , 260 hollenbach d. , tielens a. g. g. m. , 1997 , ara&a , 35 , 179 jackson , j. m. et al . , 1999 ,
baas , 194 , 7117 kenyon , s. j. , g ' omez , m. , 2001 , aj , 121 , 2673 kenyon , s. j. , hartmann l. w. , 1995 , apjs , 101 , 117 lada , c. j. et al .
, 2000 , apj , 120 , 3162 lyo , a.r .
et al . , 2003 ,
mnras , 338 , 616 marks , r. d. , 2002 , a&a , 385 , 328 marks , r. d. et al . , 1996 , a&as , 118 , 385 masetti , n. et al . , 2000 , a&a , 354 , 473 nguyen , h. et al . , 1996 , pasp , 108 , 718 phillips , a. et al . , 1999 ,
apj , 527 , 1009 rathborne , j. m. et al . , 2003 , in preparation rathborne , j. m. et al . , 2002 ,
mnras , 331 , 85 rauscher , b. j. et al . , 1998 , apj , 506 , 116 severson , s. 1998 , phd thesis , university of chicago
|
the spirex telescope , located at the amundsen scott south pole station , was a prototype system developed to exploit the excellent conditions for ir observing at the south pole .
observations over two winter seasons achieved remarkably deep , high - resolution , wide - field images in the 35@xmath0 m wavelength regime .
several star forming complexes were observed , including ngc 6334 , chamaeleon i , @xmath1 chamaeleontis , the carina nebula , 30 doradus , rcw 57 , rcw 38 , as well as the galactic centre .
images were obtained of lines at 2.42@xmath0mh@xmath2 , 3.29@xmath0 m pah and 4.05@xmath0 m br @xmath3 , as well as 3.5@xmath0 m l band and 4.7@xmath0 m m band continuum emission .
these data , combined with near
ir , mid
ir , and radio continuum maps , reveal the environments of these star forming sites , as well as any protostars lying within them .
the spirex project , its observing and reduction methods , and some sample data are summarized here .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in
|
in a series of recent publications we studied the extraction of the ground - state parameters from svz sum rules @xcite ( see also , e.g. , @xcite ) .
we made use of a quantum - mechanical potential model since this is essentially the only case where the standard procedures adopted in the method of sum rules may be tested : the estimates for the ground - state parameters obtained by these procedures may be compared with the actual values of the ground - state parameters calculated from the schrdinger equation , thus providing an unambiguous check of the reliability of the method .
the main results of our papers may be summarized as follows : ( i ) the standard approximation of a constant effective continuum threshold does not allow one to probe the accuracy of the extracted hadron parameter @xcite .
( ii ) allowing for a borel - parameter - dependent effective continuum threshold ( we denote the borel parameter @xmath0 in qcd and @xmath1 in the potential model ) and fixing this quantity by using the information on the ground - state mass leads to a considerable improvement of the accuracy of the method @xcite .
the goal of this letter is to demonstrate that the results obtained at each step of the extraction procedure both in qcd and in potential models follow the same pattern .
this similarity gives a strong argument that all our findings concerning the extraction of bound - state parameters from correlators obtained in potential model apply also to qcd .
in particular , it points out a way of improving the results for bound - state parameters obtained from various correlators in qcd .
the paper is organized as follows : in section [ sect : qcd ] we recall the qcd results from ref .
@xcite for the vacuum correlator of two pseudoscalar currents the basic object for the extraction of @xmath2 within the framework of qcd sum rules .
section [ sect : qm ] provides the analogous results for a quantum - mechanical model for the case of a potential containing confining and coulomb interactions .
section [ sect : sr ] compares the procedures of extracting the decay constant .
section [ sect : conclusions ] summarizes our conclusions .
let us consider the correlator @xmath3 of two pseudoscalar currents @xmath4 .
the borel - transformed operator product expansion ( ope ) series for this correlator has the form @xmath5 where the perturbative spectral density reads @xmath6 @xmath7 being the renormalization scale .
we make use of the results for @xmath8 reported in @xcite and do not reproduce the explicit expression for this quantity here . following the argument of @xcite we work in terms of the running masses in the @xmath9 scheme .
therefore , in all expressions in this section the quark masses @xmath10 and @xmath11 , and @xmath12 are the @xmath9 running quantities at the scale @xmath7 .
recall that the full borel - transformed correlator ( [ ope_qcd ] ) does not depend on the renormalization scale @xmath7 ; however , both the perturbative expansion truncated to a fixed order in @xmath12 and the truncated power corrections depend on @xmath7 .
we provide numerical estimates for @xmath13 ; for this choice of the scale the known terms of the perturbative expansion exhibit a good hierarchy .
we set @xmath14 gev and for other qcd parameters make use of the central values reported in table i of @xcite . and on the specific choice of the renormalization scale @xmath7 .
however , a discussion of this dependence is far beyond the scope of this paper .
a detailed analysis of @xmath2 in qcd is deferred to a separate publication . ]
the power corrections have been also considered in @xcite : @xmath15 + \frac{1}{12}\langle{\frac{\alpha_s}{\pi } ff}\rangle \right\}.\end{aligned}\ ] ] the parameter @xmath16 describes the contribution of the four - quark condensate .
notice that radiative corrections to the condensates increase rather fast with @xmath0 .
the correlator ( [ pi_qcd ] ) may be calculated in terms of hadron intermediate states : @xmath17 where @xmath2 is the decay constant of the @xmath18-meson , defined by @xmath19 for large values of @xmath0 the contributions of the excited states decrease faster than the ground - state contribution and @xmath20 is dominated by the ground state .
unfortunately , the truncated ope does not allow to evaluate the correlator at sufficiently large @xmath0 , so the excited states give a sizeable contribution to @xmath20 for the considered values of @xmath0 . according to the duality assumption ,
the contribution of the excited states is described by the perturbative contribution above some effective continuum threshold @xmath21 .
then one obtains the following relation : @xmath22 with @xmath23 in the region near the physical continuum threshold at @xmath24 , the perturbative spectral density and the hadron spectral density are very different .
consequently , the effective continuum threshold as defined in ( [ sr_qcd ] ) turns out to be necessarily a function of the borel parameter @xmath0 .
the necessity of the @xmath0-dependence of @xmath21 may be understood by comparing the left - hand side ( l.h.s . ) and right - hand side ( r.h.s . ) of ( [ sr_qcd ] ) : the only way to obtain a single exponential on the l.h.s . for a given spectral density of the integral representation on the r.h.s .
is to have a @xmath0-dependent @xmath21 .
the @xmath0-dependence of @xmath21 may be also demonstrated explicitly : for any value of the ground - state parameters on the l.h.s . of ( [ sr_qcd ] )
one can obtain numerically @xmath21 and see that it does depend on @xmath0 .
one should be aware of the fact that the @xmath0-dependence of @xmath21 can not and does not contradict to any principles of field theory : the dual correlator is a _ hand - made _ object ; such an object does not emerge in field theory .
therefore , the properties of the dual correlator ( e.g. , its analytic properties ) are very different from the properties of the field - theoretic correlators .
clearly , the standard assumption of a @xmath0-independent @xmath21 is a possible assumption .
we shall demonstrate , however , that relaxing this assumption leads to a visible improvement of the obtained results .
we define the dual decay constant and the dual invariant mass by the relations @xmath25 notice that the deviation of the dual mass from the actual mass of the ground state gives an indication of the excited - state contributions picked up by the dual correlator .
in parallel to qcd , let us consider a quantum - mechanical model with a potential containing a confining part , for which we take the ho form , and an attractive coulomb interaction : @xmath26 a quantum - mechanical analogue of the borelized two - point function has the form @xcite @xmath27 we construct the analogue of the ope series for this correlator by retaining , similar to the qcd case , the perturbative contributions up to @xmath28 ( three loops of the non - relativistic field theory ) and two power corrections , including @xmath29 corrections to them .
the resulting expression reads ( the correlator for a pure coulomb potential can be found in @xcite ) : the system behaves like a free system , since the contribution of the confining potential as well as that of the radiative corrections vanish for small @xmath1 . according to @xcite this is a signature of asymptotic freedom .
so , a non - relativistic potential model ( [ h ] ) even with a constant @xmath30 behaves like an asymptotically free theory . ]
@xmath31 , \nonumber\\ \pi_{\rm power}(t)&= & \left(\frac{m}{2\pi t}\right)^{3/2 } \left[-\frac{1}{4}\omega^2 t^2\left(1+\frac{11}{12}\sqrt{2\pi m t}\alpha\right ) + \frac{19}{480}\omega^4 t^4\left(1+\frac{1541}{1824}\sqrt{2\pi m t}\alpha\right)\right].\end{aligned}\ ] ] now , according to the standard procedures of the method of sum rules , the dual correlator is obtained as follows : we represent the perturbative contribution as a single spectral representation in the relative kinetic energy @xmath32 of the interacting quarks and cut this representation at @xmath33 : @xmath34 + \pi_{\rm power}(t).\end{aligned}\ ] ] by construction , the dual correlator is related to the ground - state contribution by @xmath35 as we have shown in our previous studies of potential models , the effective continuum threshold defined according to ( [ pidual ] ) is a function of the borel time parameter @xmath1 . for our numerical analysis
, we adopt the following parameter values : a reduced quark mass of @xmath36 gev , which corresponds to a constituent quark mass of 0.350 gev relevant for nonrelativistic computations ; @xmath37 gev , which leads to a realistic radius of the @xmath38 system , and @xmath39 .
the energy and wave function of the ground state are found by solving numerically the schrdinger equation with the help of the mathematica code provided in @xcite : the ground - state energy is @xmath40 gev [ for comparison , a pure ho model yields @xmath41 gev ] ; the ground - state wave function at the origin has the value @xmath42
gev@xmath43 [ @xmath44 gev@xmath43 ] .
obviously , the effect of the coulomb interaction is not small .
whether a @xmath0-independent or some @xmath0-dependent effective continuum threshold is considered , the crucial problem is the choice of the criterion for fixing this quantity .
we shall proceed as follows : first , we must fix the working @xmath0 window where , on the one hand , the ope gives an accurate description of the exact correlator ( i.e. , the higher - order radiative and power corrections are small ) and , on the other hand , the ground state gives a sizeable contribution to the correlator . in qcd , we set the window as follows : @xmath46 in the region @xmath47 gev@xmath48 the @xmath12 and @xmath49 terms to @xmath20 contribute less than 10% and 3% of the leading term , respectively .
power corrections give about 20% of the leading term .
we point out that the radiative corrections to the condensates increase rather fast with @xmath0 , so it is preferable to stay at relatively low values of @xmath0
. therefore our window is located at the lower values of @xmath0 compared to the window adopted in @xcite ; in the region ( [ our_window ] ) the accuracy of the truncated ope is higher .
it is known that the experimental value of the @xmath18 meson decay constant is @xmath50 mev . adopting this value
, we may calculate the relative contribution of the ground state to the correlator .
in our window it does not exceed 50% .
in the potential model , we choose the window as @xmath51 for these values of @xmath1 the omitted unknown higher - order power corrections are negligible , so the correlator is known with good accuracy .
the relative contribution of the ground state to the correlator amounts to 10% at @xmath52 and to 50% at @xmath53 .
we shall see that for a relative ground - state contribution of this size our procedure allows one to extract the decay constant with a reasonable accuracy .
widely used is the so - called stability criterion : one looks for that constant value of @xmath21 for which the extracted decay constant is most stable in the window .
this borel stability is an implementation of a self - evident statement that the physical observable can not depend on @xmath0 , an auxiliary parameter of the method .
the problem is , however , that the independence of a hadron decay constant of @xmath0 , being a necessary condition , is not sufficient to guarantee the extraction of the right value .
we have given several examples for potential models @xcite which nicely demonstrate that assuming a @xmath0-independent effective continuum threshold and fixing its value by requiring maximal stability in the borel window may lead to the extraction of a very inaccurate value .
in this paper , we consider a different algorithm for the extraction of @xmath54 which makes use of the knowledge of the ground - state mass @xcite . in parallel
to qcd we present also the results for a quantum - mechanical model ( [ h ] ) .
this is done in order to demonstrate the way our algorithm works in the case where the exact value of the decay constant is known .
a comparison makes clear that , with respect to the extraction procedure , there are no essential differences between qcd and quantum mechanics .
the algorithm developed in our previous works and established to work well for different correlators in the potential model is very simple : we consider a set of @xmath0-dependent anstze for the effective continuum threshold ( for the case of the potential model one just replaces @xmath55 and @xmath56 ) : @xmath57 obviously , the standard @xmath0-independent effective continuum threshold is also taken account by ( [ zeff ] ) . now
, we fix the parameters on the r.h.s . of ( [ zeff ] )
as follows : we calculate the dual mass squared according to ( [ mdual ] ) for the @xmath0-dependent @xmath21 of eq .
( [ zeff ] ) .
we then evaluate @xmath58 at several values of @xmath59 ( @xmath60 , where @xmath61 can be taken arbitrarily large ) chosen uniformly in the window .
finally , we minimize the squared difference between @xmath62 and the known value @xmath63 : @xmath64 ^ 2.\end{aligned}\ ] ] this gives us the parameters of the effective continuum thresholds .
as soon as the latter are fixed , it is straightforward to obtain the decay constant .
figure [ plot:1 ] shows the results for qcd and for our potential model ; in the latter the actual value of the decay constant has been found from the schrdinger equation , so that we may control each step of the extraction procedure . [
cols="^,^ " , ] first , let us notice that the borel - parameter - dependent effective thresholds corresponding to @xmath65 and @xmath66 lead to a visible improvement of the stability of the dual mass [ figs .
[ plot:1](c ) and ( d ) ] compared to the constant threshold .
this means that the dual correlator for @xmath67 is less contaminated by the excited states ; according to the philosophy of qcd sum rules the better stability of @xmath68 with @xmath67 is an important achievement for the trustability of the results .
according to fig .
[ plot:1 ] , in the potential model the true value of the decay constant lies in the band provided by the linear ( @xmath65 ) and the quadratic ( @xmath66 ) fits . we have checked that this result holds in a broad range of the parameters of the potential model .
the similarities of each step of the extraction procedure in qcd and in the potential model are evident .
therefore , it is tempting to expect that also in qcd the decay constant lies in the range provided by the linear and the quadratic fits .
anyway , the difference of the results obtained for @xmath65 and @xmath66 constitutes a realistic estimate of the intrinsic uncertainty of the extracted decay constant .
if one considers only the standard constant ansatz ( @xmath69 ) for the effective continuum threshold , the accuracy of the extracted decay constant can not be probed .
we have presented a detailed analysis of the extraction of the decay constant from the two - point function in qcd and in a potential model .
our results may be summarized as follows : * the comparison presented in this work makes obvious that , with respect to the extraction of the ground - state parameters , there are no essential differences between qcd and quantum mechanics : as soon as the parameters of the lagrangian are fixed , and the truncated ope is calculated with a reasonable accuracy ( taking into account also the relevant choice of the renormalization scale in qcd ) , the extraction procedures are very similar .
+ at first glance , this similarity might look surprising since we know that the structure of bound states in potential models and in qcd are rather different .
however , the method of dispersive sum rules does not make use of ( and does not provide ) information about the details of the ground - state structure .
what really matters for extracting the ground - state parameters in this method is the structure of the ope for a given correlator . since the structure of the ope in qcd and of its analogue in potential models is rather similar , it should not be surprising at all that the extraction procedures in potential models and in qcd are similar , too .
+ in view of the above similarity , our previous results for the extraction of the ground - state parameters ( including also the form factors @xcite ) obtained in potential models have direct implications for the corresponding analyses in qcd and should be taken quite seriously . *
allowing for @xmath0-dependent anstze for the effective continuum threshold leads to two essential improvements : + \(a ) the stability of both the dual mass and the dual decay constant in the window is considerably improved if one proceeds from the standard @xmath0-independent to the @xmath0-dependent ansatz for the effective continuum threshold .
+ \(b ) in the potential model , where the exact decay constant has been calculated from the schrdinger equation , allowing for a @xmath0-dependent effective continuum threshold and fixing its parameters according to ( [ chisq ] ) i.e. , by minimizing the deviation of the dual mass from the known ground - state mass in the window leads to the extraction of a more accurate value .
as follows from our analysis , a realistic band of values of the decay constant is provided by the numerical results obtained with the linear and quadratic anstze for the effective continuum threshold . the intrinsic uncertainty ( i.e.
, the one related to the extraction procedure ) of the decay constant found in this way is expected to be at the level of a few percent .
+ although not rigorous in the mathematical sense , this estimate for the systematic uncertainty may be considered as a realistic educated guess supported by findings in models where the true value of the decay constant is known . moreover , we have doubts that a more rigorous estimate of the intrinsic error of the method of sum rules may be obtained in principle
. _ acknowledgements _ : we are grateful to matthias jamin for providing us with his mathematica code for the calculation of @xmath2 .
d. m. gratefully acknowledges financial support from the austrian science fund ( fwf ) under project p20573 and from the alexander von humboldt - stiftung .
d. m. expresses his gratitude to the institute of theoretical physics of the heidelberg university for hospitality during his visit , where this work was started .
the work was supported in part by federal agency for science and innovation of russian federation under state contract 02.740.11.0244 and by eu contract no .
mrtn - ct-2006 - 035482 , `` flavianet '' .
30 m. shifman , a. vainshtein , and v. zakharov , nucl .
* b147 * , 385 ( 1979 ) .
a. i. vainshtein , v. i. zakharov , v. a. novikov , and m. a. shifman , sov .
. phys . * 32 * , 840 ( 1980 ) .
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 ] .
colangelo and a. khodjamirian , _ qcd sum rules : a modern perspective _ , hep - ph/0010175 .
w. lucha , d. melikhov , and s. simula , phys .
* d76 * , 036002 ( 2007 ) ; phys .
* b657 * , 148 ( 2007 ) ; + phys .
* 71 * , 1461 ( 2008 ) .
w. lucha , d. melikhov , and s. simula , phys .
* b671 * , 445 ( 2009 ) .
d. melikhov , phys .
* b671 * , 450 ( 2009 ) . w. lucha , d. melikhov , and s. simula , phys .
rev . * d75 * , 096002 ( 2007 ) ; phys
. atom .
nucl . * 71 * , 545 ( 2008 ) .
w. lucha , d. melikhov , and s. simula , phys .
rev . * d79 * , 096011 ( 2009 ) ; arxiv:0905.0963 [ hep - ph ] ; w. lucha , d. melikhov , h. sazdjian , and s. simula , arxiv:0910.3164 [ hep - ph ] . m. jamin and b. o. lange , phys . rev . * d65 * , 056005 ( 2002 ) .
m. b. voloshin , int . j. mod
. phys . *
a10 * , 2865 ( 1995 ) . w. lucha and f. schberl , int .
j. mod .
c10 * , 607 ( 1999 ) .
|
we compare the extraction of the ground - state decay constant from the two - point correlator in qcd and in potential models and show that the results obtained at each step of the extraction procedure follow a very similar pattern .
we prove that allowing for a borel - parameter - dependent effective continuum threshold yields two essential improvements compared to employing a borel - parameter - independent quantity : ( i ) it reduces considerably the ( unphysical ) dependence of the extracted bound - state mass and the decay constant on the borel parameter .
( ii ) in a potential model , where the actual value of the decay constant is known from the schrdinger equation , a borel - parameter - dependent threshold leads to an improvement of the accuracy of the extraction procedure .
our findings suggest that in qcd a borel - parameter dependent threshold leads to a more reliable and accurate determination of bound - state characteristics by the method of sum rules .
|
an important task within modern particle phenomenology consists in the determination of the quark masses , being fundamental parameters of the standard model . in the past , qcd moment sum rule analyses have been successfully applied for extracting the charm and bottom quark masses from experimental data on the charmonium and bottomium systems respectively @xcite .
the basic quantity in these investigations is the vacuum polarisation function @xmath5 : @xmath6 where the relevant vector current is represented either by the charm @xmath7 or the bottom current @xmath8 . via the optical theorem ,
the experimental cross section @xmath9 is related to the imaginary part of @xmath10 : @xmath11 usually , moments of the vacuum polarisation are defined by taking derivatives of the correlator at @xmath12 . however , in this work we allow for an arbitrary evaluation point @xmath13 to define the dimensionless moments @xcite : @xmath14 where @xmath15 is the velocity of the heavy quark . the parameter @xmath16 encodes much information about the system . by taking @xmath16 larger
the evaluation point moves further away from the threshold region .
consequently , the theoretical expansions show a better convergence , but at the same time the sensitivity on the mass is reduced .
the moments can either be calculated theoretically , including coulomb resummation , perturbation theory and nonperturbative contributions , or be obtained from experiment . in this way
one can relate the heavy quark masses to the hadronic properties of the quark - antiquark systems .
a characteristic feature of these heavy - heavy bound state systems is the coulomb - like form of the potential . developing the quantum mechanical sum rules for coulomb systems , in @xcite it has been shown that the application of fixed - order perturbation theory in such systems results in unstable predictions for the masses . to obtain a stable sum rule
it is necessary to incorporate the threshold behaviour which can be calculated in the framework of non - relativistic qcd ( nrqcd ) . a natural choice for the mass in eq .
( [ eq:1.c ] ) is the pole mass @xmath17 .
first we will employ the pole mass scheme to extract the pole masses .
however , as the pole masses suffer from renormalon ambiguities @xcite , we shall then use the potential - subtracted ( ps- ) mass @xmath18 @xcite to extract the @xmath19-masses .
the theory of nrqcd provides a consistent framework to treat the problem of heavy quark - antiquark production close to threshold .
the contributions can be described by a nonrelativistic schrdinger equation and systematically calculated in time - independent perturbation theory ( tipt ) .
the correlator is expressed in terms of a green s function @xmath20 @xcite : @xmath21 where @xmath22 and @xmath17 represents the pole mass .
the constant @xmath23 is a perturbative coefficient needed for the matching between the full and the nonrelativistic theory and naturally depends on the hard scale . to calculate the moments from the green s function
we will directly perform the derivatives at @xmath24 according to eq .
( [ eq:1.c ] ) . since the green s function is known analytically @xcite as a function of @xmath25 , this can be done numerically . in this way
we take advantage of the fact that the perturbative expansion parameter depends on the evaluation point . in addition
, we can extract the spectral density above threshold by taking the imaginary part of the green s function @xcite .
the moments depend on the three scales @xmath26 , @xmath27 and @xmath28 , the soft , factorisation and hard scale respectively .
the residual dependence on these scales will turn out to give the dominant error in the determination of the masses .
the large corrections are partly due to the definition of the pole mass
. these contributions can be reduced by using an intermediate mass definition . in this analysis
we will use the potential - subtracted ( ps ) mass @xcite where the potential below a separation scale @xmath29 is subtracted .
this mass definition leads to an improved scale dependence and a more precise determination of the @xmath19-masses .
the perturbative spectral function @xmath30 can be expanded in powers of the strong coupling constant , @xmath31 from this expression the corresponding moments @xmath32 can be calculated via the integral of eq .
( [ eq:1.c ] ) .
the first two terms are known analytically and can for example be found in ref .
@xmath33 is still not fully known analytically .
we employ a method based on pad - approximants to construct the spectral density in the full energy range @xcite .
it uses available information around @xmath34 , at threshold and in the high energy region .
it has the advantage that it gives a good description until relatively close to threshold , the region on which the quark masses are most sensitive .
the non - perturbative effects on the vacuum correlator are parametrised by the condensates .
the leading correction is the gluon condensate contribution which is known up to next - to - leading order @xcite .
furthermore , in @xcite the dimension 6 and 8 condensates have been calculated . from the numerical analysis it turns out that the absolute contribution of the condensates to the moments is negligible for the bottomium and of little influence for the charmonium .
the relative suppression of the gluon condensate to former charmonium analyses is due to three reasons : first , the absolute value of the theoretical moments increases from the coulomb resummation .
then we evaluate the moments at larger @xmath16 and smaller @xmath35 where the nonperturbative contributions are relatively small .
finally , since we obtain a larger pole mass than former analyses , the condensates , starting with a power of @xmath36 , are suppressed further .
experimentally , the six lowest lying @xmath37 and @xmath38resonances have been observed .
furthermore , recent measurements of bes @xcite in the charmonium region have improved the cross section between 3.7 and 4.8 gev .
since the widths of the poles are very small compared to the masses , the narrow - width approximation provides an excellent description of these states . to model the contributions above the 6th resonance in the bottomium system we use the assumption of quark - hadron - duality and integrate the theoretical spectral density above @xmath39 gev . in the charmonium system
we include the two lowest resonances , the bes data and the theoretical spectral density above 4.8 gev .
the theoretical part of the correlator contains the poles of the green s function , the spectral density above threshold and the condensates . for high velocities the spectral density
is well described by the perturbative expansion whereas the resummed spectral density gives a good description for low values of @xmath40 .
therefore we construct a theoretical spectral density in the full energy range which includes the essential information in both regions of @xmath40 . for a more detailed discussion
the reader is referred to @xcite .
now we discuss the most important points in the numerical analysis of the bottomium and charmonium systems respectively .
first we discuss the values of @xmath16 and @xmath35 .
since the bottom quark is relatively heavy , even for @xmath41 the nonrelativistic and perturbative expansions converge reasonably well .
nevertheless , the contributions from the poles of the green s function still dominate the theoretical part . to reduce their influence and to spread the theoretical contributions more equally among the poles , the resummed spectral density and the perturbative spectral density we must choose a higher @xmath16 .
however , for @xmath42 the moments loose sensitivity on the mass and the error from the input parameters increases
. therefore we use a central value of @xmath43 and vary @xmath16 between @xmath44 . since the relevant scale for the evaluation point is the lowest bound state energy , values of @xmath45 or 1 already correspond to well separated evaluation points . for @xmath35
we use a range of @xmath46 where the theoretical expansion and the phenomenological uncertainty are under control . as central values for our scales
we have selected @xmath47 gev , @xmath48 gev and @xmath49 gev .
for the error estimate we vary these values between @xmath50 , @xmath51 and @xmath52 . for the separation scale @xmath29 which appears as additional parameter in the definition of the ps - mass
we employ a value of @xmath53 gev .
the analysis is performed independently in the pole- and ps - scheme . in figure [ fig:8.b ]
we have plotted the ps - mass as a function of @xmath35 and the influence of the scale variations . the largest contribution to
the error comes from the soft scale . adding the errors from all input parameters quadratically , our final results for the masses are @xmath54 as in the bottom case we use @xmath43 . at this value
the pole contributions still represent the dominant part . in principle
one would like to choose a higher value where the theoretical expansions converge better .
however , the contribution from the theoretical poles varies significantly with the scales ; for @xmath55 the mass depends too strongly on these variations .
thus we again use a range of @xmath44 .
since the perturbative expansions converge more slowly than for the upsilon we restrict the analysis to smaller values of @xmath56 . as central values for our scales
we have selected @xmath57 gev , @xmath58 gev and values of @xmath59 gev and @xmath60 gev in the pole- and ps - schemes respectively .
for the error estimate we have varied the scales between @xmath61 , @xmath62 and @xmath63 . for the separation scale
we choose @xmath64 . in figure [ fig:9.b ]
we have plotted the ps - mass and the corresponding error from the scales .
finally we obtain the masses : @xmath65
the method of qcd sum rules is a very powerful tool to extract the masses since - by the choice of @xmath35 and @xmath16 - it can react very sensitive to threshold .
thus , large theoretical uncertainties only lead to a relatively small shift in the masses .
we have tried to develop a consistent framework to describe the physics of the relevant energy region .
the masses show a stable behaviour over a large range of @xmath35 and the dominant uncertainties originate from the threshold expansion of nrqcd .
i would like to thank s. narison for the invitation to this pleasant and interesting conference .
this work has been supported in part by tmr , ec contract no .
erb fmrx - ct98 - 0169 , by mcyt ( spain ) under grant fpa2001 - 3031 , and by erdf funds from the european commission .
i thank the deutsche forschungsgemeinschaft for financial support .
* 317 * ( 1999 ) 1 .
* b 434 * ( 1998 ) 115 .
_ * d 43 * ( 1991 ) 1500 . ,
_ phys . rev .
_ * d 58 * ( 1998 ) 114023 .
* b 549 * ( 1999 ) 217 .
* b 498 * ( 2001 ) 203 .
* b 507 * ( 1997 ) 334 .
* b 482 * ( 1996 ) 213 .
* b 505 * ( 1997 ) 40 .
* b 329 * ( 1994 ) 103 .
* b 213 * ( 1983 ) 285 .
* b 124 * ( 1983 ) 243 .
* 88 * ( 2002 ) 101802 .
|
in this work the charm and bottom quark masses are determined from qcd moment sum rules for the charmonium and upsilon systems . in our analysis
we include both the results from non - relativistic qcd and perturbation theory at next - next - to - leading order . for the pole masses
we obtain @xmath0 gev and @xmath1 gev . using the potential - subtracted mass in intermediate steps of the calculation the @xmath2-masses are determined to @xmath3 gev and @xmath4 gev .
|
neutron stars ( ns ) are gravitationally bound , therefore the precise measurement of mass and radius of a ns should provide a very fine probe for the equation of state ( eos ) of dense matter .
the first reasonable ideas about the composition of compact stars argued that matter are under extreme densities and is mainly composed of neutrons with small fractions of protons and electrons .
further theoretical developments and modern experimental results opened the window to other possibilities .
the densities in the interior of neutron stars is about @xmath1 times that of the nuclear saturation density ( @xmath2@xmath3 ) . at such high densities in their interiors ,
the matter there is likely to be in a deconfined and chirally restored quark phase @xcite .
the strange matter hypothesis was first proposed by itoh and bodemer @xcite and was then improved by witten @xcite .
it states that , matter at extreme density and/or temperature are composed of almost equal number of up , down and strange quarks , called strange quark matter ( sqm ) .
it is also the ground state of strongly interacting matter at such extreme conditions .
if this is true , then matter at such extreme conditions is likely to eventually convert to sqm .
such a high density scenario is present in the interiors of a ns and therefore normal nuclear matter is likely to undergo a phase transition and converts to sqm .
the strange matter hypothesis was first extensively studied in the simple mit bag model by farhi & jaffe @xcite .
the conversion process and the phase transition was further analyzed by alcock et al . @xcite .
the phase transition in a ns may continue up to the surface of the star or may stop inside the star . depending up on where this , a quark star ( qs ) may be of two types , a strange star ( ss ) or a hybrid star ( hs ) .
ss are stars composed only of sqm , while hs has a quark core and a hadronic exteriors . in the region between the quark core and hadronic outer matter
, there may exist a mixed phase region where both quarks and hadrons are present .
thus the observed pulsars are still very much model dependent .
recently , demorest et al .
@xcite found a new maximum mass limit for compact stars by measuring very precisely the mass of the millisecond pulsar psr j1614 - 2230 to be @xmath4m@xmath5 .
this value is much higher than any previously measured pulsar mass .
this measurement , has imposed a very severe constraints on the eos of matter describing the compact objects .
the model of ns , without hyperons , can easily satisfy the new mass constraint .
however the presence of strangeness , either in the form of hyperons in nuclear matter or in the form od strange quarks in quark matter , can not easily satisfy the mass limit .
so , new studies had been carried out to make the hyperonic eos and quark eos to satisfactorily explain the new mass constraint . basically to satisfy the new mass limit
, one has to make the eos stiffer , which usually is softened by the presence of strangeness . in the hyperonic nuclear matter sector
, recent studies have suggested that the stiffening of hyperonic eos is possible at par with the new experimental results @xcite .
authors also had revisited the role of vector meson - hyperon coupling @xcite and hyperon potentials @xcite , to calculate the maximum mass .
studies prior to the discovery pulsar psr j1614 - 2230 have suggested the stiffening quark matter eos from the effect of strong interactions , such as one - gluon exchange or color - superconductivity @xcite , which can satisfy the new constraint .
ozel @xcite and lattimer @xcite gave first studies on the implications of the new mass limits from psr j1614 - 2230 for quark and hybrid stars in the quark bag model .
recently , bonanno & sedrakian @xcite has succeeded in obtaining massive hs .
they employed color - superconducting quark core and very stiff hadronic eos ( like the nl3 hyperonic model or the gm3 nuclear model ) . in this work
i perform an extensive study of hybrid star mass using the relativistic mean - field hadronic eos together with a simple three - flavor mit bag model quark eos .
the model of the hs has a mixed phase intermediate region .
i would also discuss as how the understanding of more precise astrophysical measurements of the mass and radius of neutron stars can help revealing the viability of exotic quark star models .
the paper is organized as follows : in section ii , i describe the hadronic phase and in section iii i describe the mit bag model .
the mixed phase eos is constructed in section iv , using the glendenning construction .
i present my plots and extensively describe my results for the eos and the mass - radius curve in section v. the maximum mass for the hybrid star is also calculated in this section . finally in section vi
, i summarize my results and draw important conclusion from them .
at the outermost region of the star , at comparatively low densities the matter is mainly composed of hadrons .
i use the non linear relativistic mean field ( rmf ) model with hyperons ( tm1 parametrization ) to describe the hadronic phase eos . in this model
the baryons interact with mean meson fields @xcite .
the model lagrangian density includes nucleons , baryon octet ( @xmath6 ) and leptons @xmath7\psi_{b } + \frac{1}{2}({\partial_\mu \sigma \partial^\mu \sigma - m_{\sigma}^2 \sigma^2 } ) \nonumber \\ & - & \frac{1}{4 } \omega_{\mu \nu}\omega^{\mu \nu}+ \frac{1}{2 } m_{\omega}^2 \omega_\mu \omega^\mu - \frac{1}{4 } \vec \rho_{\mu \nu}.\vec \rho^{\mu \nu } \nonumber \\ & + & \frac{1}{2 } m_\rho^2 \vec \rho_{\mu}. \vec \rho^{\mu } -\frac{1}{3}bm_{n}(g_{\sigma}\sigma)^{3}- \frac{1}{4}c(g_{\sigma}\sigma)^{4 } + \frac{1}{4}d(\omega_{\mu}\omega^{\mu})^2 \nonumber \\ &
+ & \sum_{l } \bar{\psi}_{l } [ i \gamma_{\mu } \partial^{\mu } - m_{l } ] \psi_{l}.\end{aligned}\ ] ] leptons @xmath8 are non - interacting but the baryons are coupled with the scalar @xmath9 mesons , the isoscalar - vector @xmath10 mesons and the isovector - vector @xmath11 mesons .
the model constants are fitted according to the experimental results of bulk properties of nuclear matter @xcite .
the tm1 model explains the nuclear saturation of but can not sufficiently models the hyperonic matter , as it fails to reproduce the strong observed @xmath12 attraction .
this defect can be remedied by mishustin & schaffner @xcite by the addision of iso - scalar scalar @xmath13 mesons and the iso - vector vector @xmath14 mesons , coupling only with the hyperons .
the detailed eos calculation can be found in the above mentioned references @xcite , and i do not repeat them here .
the total energy density takes the form @xmath15 and the pressure can be represented as @xmath16 where @xmath17 and @xmath18 is the chemical potential and number density of particle species @xmath19 .
the quark phase is modeled according to the simple mit bag model @xcite .
the current masses of up and down quarks are extremely small , e.g. , @xmath20 and @xmath21 mev respectively , whereas , for strange quark the current quark mass is not well established , and i vary it in my calculation . for the bag model
the energy density and pressure can be written as @xmath22 where @xmath23 and @xmath24 is the fermi momentum and degeneracy factor of quarks of species @xmath19 .
@xmath25 is the energy density difference between the perturbative vacuum and the true vacuum , i.e. , the bag constant . in this sense @xmath25
can be considered as a free parameter .
both the hadronic and quark matter , maintains baryon number conservation , and are beta - equilibrated and charge neutral .
with the previously described hadronic and quark eos , glendenning construction @xcite gives the mixed phase regime .
the mixed phase is the baryong density range where both quarks and hadrons are present . in the mixed phase the hadron and the quark phases
are separately charged but the mixed phase is charge neutral as a whole .
thus the matter can be parametrized by the pair of electron and baryon chemical potentials @xmath26 and @xmath27 .
pressure of the two phases are made equal to maintaining mechanical equilibrium . to satisfy the chemical and beta equilibrium conditions the chemical potential of different particles are related to each other .
the gibbs criterion gives the mechanical and chemical equilibrium between two phases , and is written as @xmath28 the solution of above equation gives the equilibrium chemical potentials of the mixed phase .
as the two phases intersects one can calculate the corresponding charge densities of the hadronic components @xmath29 and quark components @xmath30 separately in the mixed phase .
the volume fraction occupied by quark matter in the mixed phase @xmath31 is given by @xmath32 the mixed phase energy density @xmath33 and the number density @xmath34 can be written as @xmath35 therefore the eos is now a system having a charge neutral hadronic phase at lower densities , a charge neutral mixed phase in the intermediate region and a charge neutral quark phase at higher densities .
0.2 in and @xmath36mev.,title="fig:",width=288 ] 0.2 in and @xmath36mev.,title="fig:",width=288 ] the eos are constructed to describe the properties of matter inside a ns , therefore the eos properties would also resemble the properties of a ns .
the central region of the star has maximum density ( few times @xmath37 ) , therefore the matter at the core is most likely to have a phase transition .
therefore the central region would have stable strange matter ( or a colour superconducting matter ) . as the density decreases radially outwards some nuclear matter ( nucleons )
starts appearing and so in the intermediate region there is likely to have a mixed phase .
much further outwards i have only matter consisting of only nucleons .
the crust consisting mainly of free electrons and nuclei , which completes the star structure .
the hadronic eos , i assume a fixed tm1 parameter set , which satisfactorily explains the properties of hadronic matter at extreme condition .
i can control the quark eos by changing the masses of the quarks and the bag constant .
the masses of the light quarks are quite bounded and take them to be @xmath20mev ( u ) and @xmath21mev ( d ) .
the mass of s - quark is still not well established , but expected to lie between @xmath38mev .
i would vary the mass of the s - quark within this bounded mass range .
i would also vary the bag constant ( @xmath25 ) to regulate the mixed phase region .
this parametrization of the eos of the hadron and quark matter is responsible for characterization of the matter in the mixed phase region . using the glendenning construction to construct the mixed phase , and plot curves of pressure against energy density as seen in fig [ fig1 ] . in fig [ fig1 ]
i have plotted the mixed phase eos with bag pressures @xmath39mev and @xmath36mev .
actually the relation runs as @xmath40mev , but for simplicity i will denote @xmath41 . for this case the mass of the s - quark ( @xmath42 ) is taken to be @xmath43mev . with a constant bag pressure , lower bag pressure can not generate a mixed phase region .
i do not go above @xmath44mev , as for that case the eos becomes very flat , and the maximum mass of the star becomes less . in the curves , the lower portion is nuclear phase ( dotted / dash line ) , the intermediate region is the mixed phase ( bold line ) and the higher region is the quark phase ( dotted / dash line ) .
fig [ fig2 ] , shows the pressure against baryon density for bag constant @xmath39mev and @xmath36mev .
the mixed phase starts at @xmath45 and ends at @xmath46 for bag pressure @xmath39mev . for bag pressure @xmath36mev
the mixed phase region is in between @xmath47 and @xmath48 .
the curve with bag constant @xmath39mev is much stiffer than the curve with bag pressure @xmath36mev , because the bag pressure adds negatively to the matter pressure , making the effective pressure low .
the above curves also shows that as the bag pressure increases the range of mixed phase region also increases . as the variation of pressure with both energy density and
baryon density is quite similar , from now on i would only plot curve showing pressure as function of energy density . 0.2 in mev.,title="fig:",width=288 ] with such high bag pressure it is impossible to attain the mass limit set by psr j1614 - 2230
. therefore i have to devise some other mechanism which would give stiffer eos , thereby increasing the maximum mass of the hs .
for that , i assume a density dependent bag constant . in the literature
there are several attempts to understand the density dependence of @xmath49 @xcite ; however , currently the results are highly model dependent and still there is no definite picture .
i parametrized the bag constant in such a way that it attains a value @xmath50 , asymptotically at very high densities .
the range of value of @xmath51 obtained from experiments can be found in burgio et al .
@xcite , and i assume it to be @xmath52mev , the lowest value mentioned there . with such assumptions
i then construct a gaussian parametrization given as @xcite @xmath53 \ : .
\label{bag}\end{aligned}\ ] ] the lowest value of @xmath54 , which is its value at the asymptotic high density in quark matter , is fixed at @xmath52mev .
the bag pressure quoted would be the value of the bag constant at the starting of the mixed phase region on the low density regime ( @xmath49 in the equation ) .
as the density increases the bag pressure decreases and reaches @xmath52mev asymptotically , the decrease rate is controlled by @xmath55 . 0.2 in and @xmath43mev.,title="fig:",width=288 ] 0.2 in mev.,title="fig:",width=288 ] in fig [ fig4 ] i have plotted curves showing the difference in the slope of the curves with and without the variation of bag pressure ( for @xmath56mev ) .
for the varying bag pressure the mixed phase region shrinks , becomes flatter but the quark phase region becomes stiffer .
the mixed phase region now only extends up to baryon density @xmath57 .
the change in the mixed phase region is about @xmath58 .
this is because , going to higher densities ( or higher energy density towards the core ) the effective matter pressure increases with the decrease in bag pressure ( bag pressure adds negatively to the matter pressure ) .
with such a density dependent bag constant i can have a significant mixed phase region with lower values of bag pressure .
as shown in fig [ fig5 ] i can have mixed phase region with bag pressure @xmath49 , for @xmath59mev and @xmath43mev . for the @xmath59mev eos the s - quark mass @xmath60mev and for the @xmath43mev curve
the s - quark mass is @xmath61mev . with bag pressure , @xmath49 , @xmath59 and
@xmath43mev the mixed phase region is of considerable small . for bag constant @xmath59mev
the mixed phase region starts at density @xmath62 and ends at @xmath63 . with
bag constant @xmath43mev the mixed phase region starts at density @xmath64 and ends at @xmath65 . in fig [ fig6 ]
i have separately plotted the eos for @xmath66mev showing the mixed phase region clearly .
as it would be shown later that with only such choice of quark matter parameters i can attain the mass limit set by psr j1614 - 2230 . assuming the star to be stationary and spherical , the tolman - oppenheimer - volkoff ( tov ) equations @xcite gives the solution for the pressure @xmath67 and the enclosed mass @xmath68 , @xmath69 \left [ 1 + { 4\pi r^3 p(r ) / m(r ) } \right ] \over 1 - { 2 g m(r)/ r } } \ : , \\ { dm(r ) \over dr } & = & 4 \pi r^2 \epsilon(r ) \:,\end{aligned}\ ] ] @xmath70 being the gravitational constant . starting with a fixed central energy density @xmath71
, i integrate radially outwards until the pressure on the surface equals the one corresponding to the density of iron .
this gives the star s
radius @xmath72 having gravitational mass @xmath73 for the ns crust , in the medium density range we add the hadronic eos by negele and vautherin @xcite , and for the outer crust we add the eos by feynman - metropolis - teller @xcite and baym - pethick - sutherland @xcite . 0.2 in mev.,title="fig:",width=288 ] 0.2 in mev and @xmath43 mev.,title="fig:",width=288 ]
fig [ fig7 ] shows the gravitational mass @xmath74 ( in units of solar mass @xmath75 ) as a function of radius @xmath72 , for constant and varying bag pressure @xmath56 mev .
as the bag pressure varies and decreases towards the center of the star ( at higher densities ) the curve becomes stiffer as the effective matter pressure increases ( bag pressure being negative ) .
i find that a flatter eos corresponds to a flatter mass - radius curve , and therefore the maximum mass of the star with varying bag pressure is higher than the non varying one .
with such varying bag constant i plot the mass - radius curve with @xmath76 mev and @xmath43 mev ( fig [ fig8 ] ) . with the same qualitative aspect
i find that the maximum mass of a mixed hybrid star obtained with @xmath76mev is @xmath77 .
the maximum mass with @xmath66mev and @xmath61mev , is @xmath78 solar mass .
the discovery of high - mass pulsar psr j1614 - 2230 @xcite with mass of about @xmath79 , has set a stringent condition on the eoss describing the interior of a compact star .
they @xcite quote the typical values of the central density of j1614 - 2230 , for the allowed eoss in the range 2@xmath37 - 5@xmath37 , whereas consideration of the eos independent analysis of @xcite sets the upper central density limit at @xmath80 .
the maximum mass of a mixed phase eos star with @xmath60mev is calculated to be @xmath81 solar mass .
the maximum mass for the mixed hybrid star can be increased to @xmath78 solar mass , with @xmath61mev having a varying bag pressure of @xmath66mev .
only such choice of the quark matter parametrization can give rise to star which would satisfy the mass set by psr j1614 - 2230 .
but with such choice of parameters the mixed phase region is small .
this maximum mass limit is for this hadronic and quark matter eoss .
stiffer eos sets ( like hadronic nl3 and quark quark njl model ) for the mixed hybrid star can produce much higher maximum mass @xcite . from the figure it is also clear that the maximum mass of the star corresponds to a radius of about @xmath21 km .
previous calculations have shown the maximum mass of a ns have radius greater than @xmath82 km , whereas the maximum mass of a ss corresponds to a radius of less than @xmath83 km .
therefore it is clear from my calculation that the mixed hybrid star has radius corresponding to the maximum mass , quite different from the neutron and strange star .
they are not as compact as strange stars and their radius lies between the nuclear and strange star .
in this work i have studied the maximum mass of a hybrid star having a mixed phase region . with the hadronic matter eos having hyperons , and remaining in
the simple mit bag model i wanted to study what parameters value could give such high masses for a hs having a mixed phase region .
the star has a dense quark core , a mixed phase intermediate region and hadronic outer region .
the hadronic and quark matter eos is simultaneously constructed according to relativistic mean field approach and mit bag model .
the mixed phase is determined in accordance with the glendenning construction .
all the phases are at chemical and mechanical equilibrium , and also they are charge neutral as a whole . with constant bag pressure @xmath49 of @xmath39 and @xmath36mev ( and @xmath60mev )
i get eos with considerable mixed phase region but with such parametrization the maximum mass of the star is about @xmath84 solar mass .
i therefore consider a density dependent bag pressure @xmath49 , parametrized according to the gaussian parametrization .
the asymptotic value of the bag constant at high density is fixed at @xmath52mev , which is its lowest value known from the experiments @xcite . with such varying bag pressure
i can have a mixed phase region with @xmath76mev , but still the mass of the star is below @xmath85 solar mass . to reach the mass limit set by psr j1614 - 2230 , for a mixed phase hs
, i build the eos with bag pressure of @xmath66mev , having s - quark mass @xmath61mev . for such choice of parameters values
, the mixed phase region is small .
further lowering of bag pressure is not possible , as then the mixed phase disappears .
the maximum mass for a mixed hybrid star with the given set of parameters is @xmath0 .
another important results of my calculation is that the hs , with mixed phase , has radius ( for the maximum mass ) quite different from the neutron or strange star , their radius lying in between the neutron and strange star .
after the discovery of psr j1614 - 2230 , setting the mass limit to @xmath86 solar mass , new eoss model has been proposed .
weissenborn et al .
@xcite showed that absolutely strange star can have mass above @xmath86 solar mass is the effect of strong coupling constant and color superconductivity is taken into account .
bednarek et al .
@xcite argued that eos with hyperons having quartic terms involving hidden strangeness vector meson can reach such limit .
matsuda et al .
@xcite extended their calculation to hybrid stars , having a smooth crossover from hadronic to quark matter .
for the mass to reach the maximum mass limit they showed that the crossover has to take place at low density and the quark matter has to be strongly interacting . using very stiff eos sets ( hadronic nl3 and quark quark njl model ) the maximum mass limit for the hybrid star can be raised much higher as shown by lenzi & lugones @xcite . in my work
, i also have shown that the maximum mass limit can be reached by a hs with mixed phase even with simple hyperonic nuclear matter eos and mit bag model quark matter
eos if i assume a relatively low density dependent bag pressure .
observationally the ns is characterised only by signals coming to us from its surface .
developments has been made on them to measure accurately the mass of compact stars but same can not be done for their radius .
reasonable measurement of the radius of a compact stars could differentiate ns , ss and hs , as we have seen here that different eos of matter gives different mass - radius relationship . as it is clear from my calculation and also from previous calculations that by suitable tuning of the parameters or by invoking new terms in the eoss calculations the mass limit set by psr j1614 - 2230 can be reached . therefore to have a full understanding of the matter at extreme densities we need results not only from astrophysical observations but also from earth based experiments .
weber f. _ pulsars as astrophysical laboratories for nuclear and particle physics _ , ( iop publishing , bristol , 1999 ) itoh n. 1970 , prog .
. phys . * 44 * , 291 bodmer a. r. 1971 , phys .
d * 4 * , 1601 witten e. 1984 , phys .
d * 30 * , 272 farhi e. & jaffe r. l. 1984 , phys .
d * 30 * , 2379 alcock c. , farhi e. and olinto a. 1986 , astrophys .
j * 310 * , 261 demorest , p. , pennucci , t. , ransom , s. , roberts , m. , & hessels , j. 2010 , nature * 467 * , 1081 bednarek i. , haensel p. , zdunik j. l. , bejger m. & manka r. 2011 , arxiv:1111.6942 [ astro-ph.sr ] weissenborn s. , chatterjee d. , & schaffner - bielich 2011 , arxiv:1112.0234 [ astro-ph.he ] weissenborn s. , chatterjee d. & schaffner - bielich 2012 , nucl .
a * 881 * , 62 lugones g. & horvath j. e. 2003 , astron . & astrophys .
* 403 * , 173 horvath , j. e. & lugones , g. 2004 , astron .
& astrophys .
* 442 * , l1 rster , s. b. & rischke , d. h. 2004 , phys .
d * 69 * , 045011 alford , m. , blaschke , d. , drago , a. , et al .
2007 , nature , * 445 * , 7 fischer , t. , sagert , i. , pagliara , g. , et al .
2010 , astrophys .
* 194 * , 39 kurkela , a. , romatschke , p. , & vuorinen , a. 2010 , phys .
d * 81 * , 105021 kurkela , a. , romatschke , p. , vuorinen , a. , & wu , b. 2010 , arxiv:1006.4062 [ astro-ph.he ] zel f. , psaltis d. , ransom s. , demorest p. & alford m. 2010 ,
j. lett . * 724 * , l199 lattimer j. m. & prakash m. 2010 , arxiv:1012.3208 [ astro-ph.sr ] bonanno l. & sedrakian a. 2012 , astron . & astrophys .
* 539 * , a16 boguta j. & bodmer r. a. 1977 , nucl .
a * 292 * , 413 glendenning n. k. & moszkowski s. a. 1991 , phys .
lett . * 67 * , 2414 ghosh s. k. , phatak s. c. & sahu p. k. 1995 , z. phys .
a * 352 * , 457 sugahara y. & toki h. 1994 , nucl .
a * 579 * , 557 schaffner j. & mishustin i. n. 1996 , phys .
c * 53 * , 1416 chodos a. , jaffe r. l. , johnson k. , thorn c. b. and weisskopf v. f. 1974 , phys . rev .
d * 9 * , 3471 glendenning n. k. 1992 , phys .
d * 46 * , 1274 adami c. & brown g. e. 1993 , phys .
rep . * 234 * , 1 ( 1993 ) ; xue - min jin & jennings b. k. 1997 , phys .
c * 55 * , 1567 blaschke d. , grigorian h. , poghosyan g. , roberts c. d. & schmidt s. 1999 , phys . lett .
b * 450 * , 207 burgio g. f. , baldo m. , sahu p. k. & schulze h. j. 2002 , phys .
c * 66 * 025802 mallick r. & sahu p. k. 2012 , arxiv:1207:4870 [ astro - ph , he ] shapiro s. l. & teukolsky s. a. _ black holes , white dwarfs , and neutron stars _ , ( john wiley & sons , new york , 1983 ) negele j. w. & vautherin d 1973 , nucl .
a * 207 * , 298 feynman r. , metropolis f. & teller e. 1949 , phys . rev . * 75 * , 1561 baym g. , pethick c. & sutherland d. 1971 , astrophys .
j * 170 * , 299 lattimer j. m. & prakash m. 2005 , phys .
* 94 * , 111101 lenzi c. h. & lugones g. 2012 , arxiv:1206.4180 [ astro-ph.sr ] weissenborn s. , sagert i. g. , hempel m. & schaffner - bielich j. 2011a , astrophys .
j * 740 * , l14 masuda k. , hatsuda t. & takatsuka t. 2012 , arxiv:1205.3621 [ nucl - th ]
|
recent observation of pulsar psr j1614 - 2230 with mass about 2 solar masses poses a severe constraint on the equations of state ( eos ) of matter describing stars under extreme conditions .
neutron stars ( ns ) can reach the mass limits set by psr j1614 - 2230 .
but stars having hyperons or quark stars ( qs ) having boson condensates , with softer eos can barely reach such limits and are ruled out . qs with pure strange matter also can not have such high mass unless the effect of strong coupling constant or color superconductivity are considered . in this work
i try to calculate the upper mass limit for a hybrid stars ( hs ) having a quark - hadron mixed phase .
the hadronic matter ( having hyperons ) eos is described by relativistic mean field theory and for the quark phase i use the simple mit bag model .
i construct the intermediate mixed phase using glendenning construction .
hs with a mixed phase can not reach the mass limit set by psr j1614 - 2230 unless i assume a density dependent bag constant . for such case
the mixed phase region is small .
the maximum mass of a mixed hybrid star obtained with such mixed phase region is @xmath0 .
|
ad - hoc wireless networks are composed of a set of nodes that exchange data with each other through wireless links without relying on any pre - existing infrastructure @xcite .
typically , the network topology follows a nearest neighbor criterion to allow for low - power communications .
the diffusion of information through a network presumes connectivity of the network .
furthermore , many distributed algorithms running over a graph , such as consensus , diffusion , or swarming algorithms , have a convergence time strictly dependent on the graph connectivity @xcite-@xcite . for example
, highly connected networks generally have significantly faster convergence thanks to a more efficient in - network information diffusion . in many practical examples
, connectivity can only be assumed to hold in probability because the links among the nodes may be on or off depending on channel conditions . in most applications ,
channel variability may depend on several factors , such as mobility of the nodes , as in vehicular networks , channel fading due to propagation over multipath channels , or packet collisions due to random mac strategies working on a collision avoidance regime .
it is then of interest to look at distributed mechanisms to estimate and control the network connectivity in the presence of realistic channel models .
spectral graph theory @xcite has been demonstrated to be a very powerful tool for topology inference .
the eigenvalues and/or eigenvectors of the laplacian matrix of the graph have been exploited , e.g. , to estimate the connectivity of the network @xcite-@xcite , to find densely connected clusters of nodes @xcite-@xcite , and to search for potential links that would greatly improve the connectivity if they were established @xcite . a recent tutorial work that gives an excellent view of this topic is @xcite . in all these works
, it was shown that the connectivity properties of a graph can be assessed by looking at the second smallest eigenvalue of the laplacian matrix , also known as the _ algebraic connectivity _ , whereas a significant role in graph partitioning is plaid by the eigenvector associated to the algebraic connectivity , also known as _
fiedler vector _
it is then important to find efficient algorithms to compute these connectivity parameters .
the problem of distributed estimation of the eigenvalues of the laplacian matrix has been considered in several previous works , e.g. , @xcite-@xcite . in @xcite
a distributed algorithm is proposed to find the @xmath0 eigenvectors corresponding to the @xmath0 largest eigenvalues of the laplacian matrix or the ( weighted ) adjacency matrix , based on power iteration and random walk techniques .
the work in @xcite evaluates the eigenstructure of the laplacian matrix by letting the nodes oscillate at the eigenfrequencies corresponding to the network topology . a distributed algorithm that computes
the fiedler vector and the algebraic connectivity , with application to topology inference in ad hoc networks , has been also proposed in @xcite-@xcite .
the algorithms in @xcite-@xcite allow to estimate the algebraic connectivity and the fiedler vector without the need for nested consensus iterations , thus significantly reducing the communication burden .
distributed methods aimed at controlling the algebraic connectivity for flocking maintenance have also been proposed in @xcite-@xcite . in @xcite ,
the authors propose a distributed algorithm that allows each node to estimate and track the algebraic connectivity of the graph in mobile wireless sensor networks .
then , based on this estimator , a decentralized gradient controller for each agent helps maintain global connectivity during motion .
the work in @xcite enforces network connectivity by means of distributed topology control that decides on both deletion and creation of communication links between agents . with this protocol ensuring network connectivity
, a decentralized motion controller aligns agent velocity vectors and regulates inter - agent distances to maintain existing network links . in @xcite , a theoretical framework for controlling graph connectivity in mobile robot networks
is given , thus providing centralized and distributed algorithms to maintain , increase , and control connectivity .
finally , references @xcite-@xcite propose distributed methods for estimating the algebraic connectivity with the aim of optimizing the performance of average consensus algorithms . in @xcite a consensus method using chebyshev polynomials
is considered and the authors proposed a distributed algorithm to compute the parameters that enable the method to maximize the convergence rate .
then , the work in @xcite proposes a distributed algorithm to estimate the algebraic connectivity of a graph , thus applying this method to an event - triggered consensus scenario , where the most recent estimate of the algebraic connectivity is used for adapting the behavior of the average consensus algorithm . finally , in @xcite the authors proposed a topology - aware distributed algorithm for on - line adaptation of the laplacian weighting rule , when applied in an in - network averaging procedure .
all these previous works assumed ideal communications among the network nodes .
however , in a realistic scenario , the wireless channels are affected by random fading and additive noise , which induce errors in the received packets .
furthermore , realistic random mac protocols may determine packet collisions during the exchange of data among the nodes .
in such a case , the receiving node could request the retransmission of the erroneous packets , but this would imply random delays in the communication among the nodes and it would be complicated to implement over a totally decentralized system .
it is then of interest to analyze networks where the erroneous packets are simply dropped , without requiring a retransmission .
random packet dropping can be modeled as having random switching graph topologies .
the effect of random graphs on distributed algorithms has been thoroughly studied in a series of works , mainly focused on the convergence of consensus algorithms , e.g. , @xcite-@xcite , and of swarming algorithms for radio resource allocation @xcite . in this work ,
we propose a distributed algorithm , based on a stochastic power iteration method , whose aim is to estimate the algebraic connectivity and the related fiedler vector of the expected laplacian matrix of a random graph , incorporating random impairments in the exchange of data among the nodes .
the basic contributions of this paper are the following : 1 ) a novel algorithm to estimate in a distributed fashion the spectral connectivity parameters of the expected laplacian matrix of a random graph ; 2 ) the derivation of the convergence properties of the proposed algorithm in the presence of random link failures in the communications among nodes ; 3 ) the control of the expected connectivity through the adaptation of the power transmitted by each node , in order to maximize the network connectivity in the presence of realistic mac protocols or simply to drive it toward a desired target value .
the paper is organized as follows . in section
ii we first recall some basic concepts from algebraic graph theory that will be used throughout the paper .
then , we describe the proposed stochastic power iteration method for estimating the connectivity of a random graph , thus illustrating the distributed implementation based on consensus algorithms to decentralize the computation .
the convergence properties of the proposed algorithm in the presence of random link failures are also investigated . in section iii , exploiting the proposed strategy for connectivity estimation , we propose a simple power control method aimed at controlling the expected connectivity of a network .
section iv then shows the effect of collisions , which are induced by a realistic random medium access control protocol , on the network connectivity .
in particular , it is shown how , by choosing a too large transmission power , the network connectivity may be heavily degraded due to an increase in the collision probability .
then , we propose a distributed algorithm to evaluate the optimal transmission power that maximizes the connectivity in the presence of realistic mac protocols . finally , section v draws some conclusions .
we consider a network composed of @xmath1 nodes interacting according to a communication topology . the interaction among the nodes
is modeled as an undirected graph @xmath2 , where @xmath3 denotes the set of nodes and @xmath4 is the edge set .
the structure of the graph is described by a symmetric @xmath5 adjacency matrix @xmath6 , whose entries @xmath7 are either positive or zero , depending on wether there is a link between nodes @xmath8 and @xmath9 or not , i.e. , if the distance between nodes @xmath8 and @xmath9 is less than a coverage radius , which is dictated by nodes transmit power and the channel between them .
the set of neighbors of a node @xmath8 is @xmath10 , defined as @xmath11 .
node @xmath8 communicates with node @xmath9 if @xmath9 is a neighbor of @xmath8 ( or @xmath12 ) .
the graph has no self loops , i.e. , @xmath13 for all @xmath8 .
denoting by @xmath14 the degree of node @xmath8 , the degree matrix @xmath15 is a diagonal matrix with entries @xmath16 that are the row sums of the adjacency matrix @xmath17 .
the graph laplacian @xmath18 is an @xmath5 matrix defined as @xmath19 the spectral properties of @xmath18 have been shown to be critical in many multiagent applications , such as formation control @xcite , consensus seeking @xcite and direction alignment @xcite .
we denote by @xmath20 , @xmath21 , the eigenvalues of @xmath18 , ordered in increasing sense .
the matrix @xmath18 always has , by construction , a null eigenvalue @xmath22 , with associated eigenvector @xmath23 composed of all ones . for a connected graph
, the nullspace of @xmath18 has dimension @xmath24 and it is spanned by the vector @xmath23 .
the second smallest eigenvalue @xmath25 is known as the _ algebraic connectivity _ of the graph .
this eigenvalue is greater than 0 if and only if @xmath26 is a connected graph .
the magnitude of this value reflects how well connected the overall graph is .
for this reason , it has been used for example in analysing the synchronizability of networks @xcite-@xcite , @xcite-@xcite , in maintaining stable flocking @xcite , and for routing optimization in cognitive radio ad - hoc networks @xcite . _ random link failures : _ in a realistic communication scenario , the packets exchanged among the nodes may be received with errors , because of collisions , channel fading or noise .
the retransmission of erroneous packets can be incorporated into the system , but packet retransmission introduces a nontrivial additional complexity in decentralized implementations and , more importantly , it also introduces an unknown delay and delay jitter .
it is then of interest to examine protocols where erroneous packets are simply dropped .
we take into account random packet dropping by modeling the coefficient @xmath7 describing the network topology as statistically independent random variables . then , the laplacian of the graph varies with time as a sequence of i.i.d .
matrices @xmath27\}$ ] , which can be written , without any loss of generality , as @xmath28 = \bar{\bl } + \tilde{\bl}[k]\ ] ] where @xmath29 denotes the expected matrix and @xmath30=\{\tilde{l}_{ij}[k]\}$ ] are i.i.d .
perturbations around the mean . the i.i.d .
fluctuations @xmath31 $ ] affect only the active links , i.e. the links for which @xmath32 ; for all other inactive links , the perturbations are equal to zero .
we do not make any assumptions of symmetry of the failures , i.e. @xmath31 $ ] may be not equal to @xmath33 $ ] , or about the link failure model . although the link failures are independent over time , during the same iteration , the link failures can still be spatially correlated .
it is important to remark that , in the ensuing analysis and derivations , we do not require the random instantiations @xmath34 $ ] of the graph be connected for all @xmath35 .
we only require the graph to be connected on average .
this condition is captured by requiring @xmath36 . in this section
, we propose a novel algorithm aimed at assessing the connectivity of a random graph by estimating the second smallest eigenvalue of the expected laplacian matrix @xmath37 .
since in our setting the network graph is random due to the presence of link failures , we introduce a stochastic power iteration method to handle the randomness introduced by the graph fluctuation .
let us introduce the matrix @xmath38 $ ] given at time @xmath35 by : @xmath39=\bi-\bar{\varepsilon}\bl[k]=\bar{\bw}+\tilde{\bw}[k]\ ] ] where @xmath40 is a positive parameter , @xmath41 is the mean matrix , and @xmath42=-\bar{\varepsilon}\tilde{\bl}[k]$ ] are i.i.d .
fluctuations around the mean .
the matrix @xmath38 $ ] in ( [ consensus_matrix ] ) was used , for example , as the iteration matrix of consensus algorithms over random graphs , see e.g. @xcite-@xcite , @xcite-@xcite . from ( [ consensus_matrix ] ) , the eigenvalues of the expected laplacian matrix @xmath37
are directly related to those of the expected matrix @xmath43 in ( [ consensus_matrix ] ) through the relation @xmath44 in particular , the algebraic connectivity is given by @xmath45 .
furthermore , the eigenvector @xmath46 associated to the second largest eigenvalue of the expected matrix @xmath43 coincides with @xmath47 , which is the one associated to the second smallest eigenvalue of the expected laplacian matrix @xmath37 , also known as the fiedler vector .
the coefficient @xmath40 in ( [ consensus_matrix ] ) satisfies @xmath48 which , combined with the condition @xmath36 , ensures that the mean matrix @xmath43 is a perron matrix having a single unitary eigenvalue @xcite .
since we want to track the second largest eigenvalue of the mean matrix @xmath43 , we deflate the original matrix @xmath38 $ ] by removing its largest eigenvalue , thus obtaining the matrix @xmath49 $ ] given by : @xmath50 = \bw[k]-\frac{1}{n}\mathbf{1}\mathbf{1}^t:=\bar{\bb}+\tilde{\bb}[k]\end{aligned}\ ] ] where @xmath51 and @xmath52=\tilde{\bw}[k]=-\bar{\varepsilon}\tilde{\bl}[k]$ ] . in this way , the maximum eigenvalue of the deflated expected matrix @xmath53 coincides with the second largest eigenvalue of @xmath43 . to handle the randomness of the graph , we introduce also the deflated matrix @xmath54=\bi-\varepsilon[k]\bl[k]-\frac{1}{n}\mathbf{1}\mathbf{1}^t=\bw_2[k]-\frac{1}{n}\mathbf{1}\mathbf{1}^t,\end{aligned}\ ] ] where @xmath55=\bi-\varepsilon[k]\bl[k]$ ] , with @xmath56 $ ] denoting a positive diminishing sequence that we will choose in the sequel .
the matrices @xmath57 $ ] and @xmath49 $ ] have exactly the same eigenvectors , but different eigenvalues , at each time @xmath35 , due to the time - varying sequence @xmath56 $ ] .
we consider first a centralized implementation of the stochastic power iteration algorithm , whose main steps are listed in table 1 .
a distributed implementation of the algorithm will be illustrated later on .
initialize @xmath58 $ ] , @xmath59 $ ] , and @xmath60 $ ] randomly .
then , set @xmath61 and perform the following steps : 1 .
build the deflated matrices @xmath62&=\bi-\bar{\varepsilon}\bl[k]-\frac{1}{n}\mathbf{1}\mathbf{1}^t\\ \bb_2[k]&=\bi-\varepsilon[k]\bl[k]-\frac{1}{n}\mathbf{1}\mathbf{1}^t \end{aligned}\ ] ] where @xmath40 and @xmath56 $ ] satisfy ( [ condition ] ) and ( [ epsilonk ] ) , respectively ; 2 .
evaluate the estimate @xmath63 $ ] of @xmath64 as : @xmath65&=\frac{\bx^t[k]\bb[k]\bx[k]}{\bx^t[k]\bx[k]}\label{lambdan-1_2}\\ y[k+1]&=y[k]+\alpha[k]\left(y_0[k]-y[k]\right)\label{lambdan-1 } \end{aligned}\ ] ] where @xmath66 $ ] is a time varying step - size satisfying ( [ step_size ] ) ; 3 .
compute the estimate @xmath67 $ ] of @xmath68 as : @xmath69=\frac{1-y[k+1]}{\bar{\varepsilon}};\ ] ] 4 .
perform the power iteration step @xmath70=\frac{\bb_2[k]\bx[k]}{\|\bb_2[k]\bx[k]\|};\ ] ] 5 . if convergence is achieved stop , otherwise set @xmath71 and go to step 1 .
the aim of the stochastic power iteration steps in ( [ lambdan-1_2])-([pow_iter ] ) is to estimate the largest eigenvalue of the expected matrix @xmath53 ( i.e. @xmath64 ) , which is directly related to the second eigenvalue @xmath68 of the expected laplacian through ( [ lambda2_est ] ) . to achieve this goal
, the power iteration in ( [ pow_iter ] ) runs over the matrix @xmath57 $ ] , thus providing an estimate @xmath72 $ ] of the eigenvector @xmath73 associated to the largest eigenvalue of the expected matrix @xmath53 ( remember that @xmath57 $ ] and @xmath49 $ ] have the same eigenvectors ) .
the eigenvector estimate @xmath72 $ ] is then used in the rayleigh ratio ( [ lambdan-1_2 ] ) to provide an estimate @xmath74 $ ] for @xmath64 . in both adaptations ,
it is fundamental to choose the step sizes @xmath56 $ ] in ( [ defl_matrix3 ] ) and @xmath66 $ ] in ( [ lambdan-1 ] ) . in particular , we make the following assumptions , which are standard in stochastic approximation and adaptive signal processing @xcite-@xcite : _ assumption a.1 _ : ( _ persistence _ ) the sequence @xmath56 $ ] in ( [ defl_matrix3 ] ) and the step - size sequence @xmath66 $ ] in ( [ lambdan-1 ] ) satisfy the conditions @xmath75>0 , \hspace{.4 cm } \sum_{k=0}^{\infty}\varepsilon[k]=\infty , \hspace{.4 cm } \sum_{k=0}^{\infty}\varepsilon^2[k]<\infty,\label{epsilonk}\\ & \alpha[k]>0 , \hspace{.4 cm } \sum_{k=0}^{\infty}\alpha[k]=\infty , \hspace{.4 cm } \sum_{k=0}^{\infty}\alpha^2[k]<\infty.\label{step_size}\end{aligned}\ ] ] conditions ( [ epsilonk])-([step_size ] ) ensure that the step size sequences decay to zero , but not too fast .
an example of sequences satisfying ( [ epsilonk])-([step_size ] ) is @xmath76=\frac{\varepsilon_0}{(k+1)^{\gamma } } , \quad\quad \alpha[k]=\frac{\alpha_0}{(k+1)^{\beta}},\end{aligned}\ ] ] @xmath77 we are now able to state the main theorem on the convergence of the proposed stochastic power iteration method .
_ theorem 1 _ : let @xmath78 $ ] and @xmath72 $ ] be the sequences generated in ( [ lambda2_est ] ) and ( [ pow_iter ] ) by the stochastic power iteration .
if @xmath79 , and condition ( [ condition ] ) and assumption a.1 hold , we have @xmath80=\lambda_{2}(\bar{\bl } ) , \quad\hbox{and}\quad \lim_{k\rightarrow\infty } \bx[k]=\hat{\bu}_{2}(\bar{\bl}),\end{aligned}\ ] ] almost surely ( w.p.1 ) , where @xmath81 denotes the normalized fiedler vector of the expected laplacian matrix @xmath37 .
see appendix b. theorem 1 establishes the almost sure convergence of the stochastic power iteration method to the fiedler vector @xmath81 and to the algebraic connectivity @xmath68 .
as shown in appendix c , an upper bound on the slowest ( undesired ) decaying mode @xmath82 $ ] of the algorithm is given by @xmath83&\leq \exp \left(-\big(\lambda_3(\bar{\bl})-\lambda_2(\bar{\bl})\big)\sum_{l=0}^k\varepsilon[l]\right).\end{aligned}\ ] ] from ( [ conv_mean_part2 ] ) , we see how the convergence rate depends on the difference between the third and the second eigenvalues of the expected laplacian @xmath37 , and on the sequence @xmath56 $ ] .
in particular , ( [ conv_mean_part2 ] ) makes clear that , because of ( [ epsilonk ] ) , the slowest decaying mode goes to zero as @xmath84 .
_ remark : _ as mentioned in the introduction , the eigenvectors of the laplacian matrix give useful information about how the network can be partitioned , i.e. , how to find clusters of nodes in the network .
it has been shown in several works , e.g. , @xcite-@xcite , that spectral clustering can infer more topological properties of the graph if more eigenvectors of the laplacian matrix , besides the one associated to the algebraic connectivity , are known . up to now
, it was shown how the proposed stochastic power iteration can compute only the second smallest eigenvalue and the corresponding eigenvector of the expected laplacian matrix of the graph .
however , the trick of deflating the matrix @xmath38 $ ] in ( [ defl_matrix2 ] ) can be sequentially iterated in order to estimate the third order eigenvalue and the associated eigenvector . as an example , once the algebraic connectivity @xmath85 and the fiedler vector @xmath47 have been estimated through a first stage power iteration , the third order eigenparameters can be estimated by applying again power iteration using the pair of deflated matrices @xmath86&=\bb[k]-(1-\bar{\varepsilon } \lambda_{2}(\bar{\bl}))\bu_{2}(\bar{\bl})\bu^t_{2}(\bar{\bl}),\nonumber\\ \bc_2[k]&=\bb_2[k]-(1-\varepsilon[k ] \lambda_{2}(\bar{\bl}))\bu_{2}(\bar{\bl})\bu^t_{2}(\bar{\bl } ) , \nonumber\end{aligned}\ ] ] which take the role of @xmath49 $ ] and @xmath57 $ ] in table 1 , respectively . in a similar way
, we can estimate also the higher order eigen - parameters by sequential deflation and power iteration .
_ distributed implementation : _ the stochastic power iteration method described before requires a centralized implementation . in the following , we propose a decentralized implementation based on average consensus @xcite,@xcite .
the two operations to be distributed are the rayleigh ratio in ( [ lambdan-1_2 ] ) and the power iteration in ( [ pow_iter ] ) , whereas all other computations can be performed locally . setting @xmath87=\bb[k]\bx[k]$ ] and @xmath88=\bb_2[k]\bx[k]$ ] ,
the @xmath8-th components @xmath89 $ ] and @xmath90 $ ] of the vectors @xmath87 $ ] and @xmath88 $ ] can be evaluated locally .
in fact , exploiting the structure of the matrices in ( [ defl_matrix2 ] ) and ( [ defl_matrix3 ] ) , we have @xmath91=x_i[k]+\bar{\varepsilon}\sum_{j=1}^na_{ij}[k](x_j[k]-x_i[k])-m[k]\label{b_i}\\ b_{2,i}[k]=x_i[k]+\varepsilon[k]\sum_{j=1}^na_{ij}[k](x_j[k]-x_i[k])-m[k]\label{b_i2}\end{aligned}\ ] ] where @xmath92=\frac{1}{n}\mathbf{1}^t\bx[k]$ ] is a global parameter .
the value @xmath93 $ ] is given by the average of the values @xmath94 $ ] stored locally at each node , and can be computed in a decentralized fashion using a round of average consensus protocol .
the next step is to evaluate the rayleigh ratio in ( [ lambdan-1_2 ] ) in a distributed fashion . to this end
, we notice that expression ( [ lambdan-1_2 ] ) can be recast as @xmath95\bb[k]\bx[k]}{\bx^t[k]\bx[k ] } = \frac{\sum_{i=1}^n x_i[k]b_i[k]}{\sum_{i=1}^n x^2_i[k]},\end{aligned}\ ] ] where both numerator and denominator are written as inner products .
this notation is convenient because it enables us to compute this expression through a step of weighted average consensus @xcite,@xcite , which evaluates in a distributed manner the ratio in ( [ lambdan-1_3 ] ) .
thus , at this stage , each node is able to compute ( [ lambdan-1_2 ] ) and ( [ lambdan-1 ] ) locally . to complete the series of operations of the stochastic power iteration algorithm ,
node @xmath8 still needs to evaluate ( [ pow_iter ] ) in a distributed fashion .
then , each node @xmath8 computes the @xmath8th component of vector @xmath96 $ ] in ( [ pow_iter ] ) as : @xmath97=\frac{b_{2,i}[k]}{\|\bb_2[k]\|}.\ ] ] since the numerator has been already computed through ( [ b_i2 ] ) , we only need to compute the denominator of ( [ pow_iter3 ] ) .
in particular , we consider the evaluation of @xmath98\|=\sqrt{\frac{1}{n}\sum_{i=1}^n b^2_{2,i}[k]},\ ] ] which is a scaled version of @xmath99\|$ ] , and can be computed in a distributed fashion by taking the square root of the output of an average consensus step .
each node then computes @xmath100=\sqrt{n}\frac{b_{2,i}[k]}{\|\bb_2[k]\|}$ ] , which is a scaled version of the true value @xmath94 $ ] that the algorithm should compute in ( [ pow_iter3 ] ) .
however , even in the presence of such update , the method still works correctly because , at time @xmath101 , the step in ( [ lambdan-1_2 ] ) is a rayleigh ratio , whose result is not affected by the scaling @xmath102 , thus leading to the correct update of the algorithm .
the main steps of the decentralized implementation are summarized in table 2 .
we also define @xmath103-convergence of a sequence @xmath104 $ ] the event @xmath105-c[k]|}{|c[k]|}\leq\delta$ ] .
each node initializes @xmath106 $ ] , @xmath59 $ ] , and @xmath60 $ ] randomly
. then , set @xmath61 and performs the following steps : 1 .
run a consensus round to get @xmath92=\mathbf{1}^t\bx[k]/n$ ] until @xmath107-convergence ; 2 . evaluate @xmath89 $ ] and @xmath90 $ ] , @xmath108 , using ( [ b_i])-([b_i2 ] ) ; 3 . run a consensus round to compute the rayleigh ratio in ( [ lambdan-1_3 ] ) ( i.e. @xmath74 $ ] in ( [ lambdan-1_2 ] ) ) and the scaled norm in ( [ scaled_norm ] ) until @xmath109-convergence ; 4 . compute the estimate @xmath63 $ ] of @xmath64 using ( [ lambdan-1 ] ) ; 5 .
compute the estimate @xmath67 $ ] of @xmath85 using ( [ lambda2_est ] ) ; 6 .
perform the power iteration in ( [ pow_iter3 ] ) , @xmath108 .
if convergence is achieved stop , otherwise set @xmath71 and go to step 1 .
_ remark 2 : _ the distributed implementation of the stochastic power iteration proposed in table 2 is based on the use of two rounds of average consensus algorithm @xcite,@xcite , which allow the computation of the global quantities in the steps 1 and 3 in table 2 .
of course , also the consensus algorithm will be affected by the presence of random link failures .
however , if the expected graph is connected ( i.e. , @xmath36 ) and the matrices @xmath38 $ ] in ( [ consensus_matrix ] ) are doubly stochastic for every @xmath35 , it is well known that consensus algorithm is robust to the presence of link failures @xcite-@xcite , thus guaranteing convergence to the desired average value .
thus , differently from the centralized case in table 1 , where no symmetry assumptions are required on the link failures , in the distributed implementation the network graph must be balanced at every iteration . regarding the communication demands of the proposed distributed implementation , in the first consensus round
each node must broadcast a scalar value to its neighbors , whereas in the second round it is necessary to transmit two scalar values .
thus , letting @xmath110 be the cost associated to the transmission of a scalar value , the communication cost per iteration @xmath35 of the power iteration algorithm is equal to @xmath111 , where @xmath112 and @xmath113 are the number of iterations needed by the first and second consensus rounds to converge within prescribed accuracies @xmath107 and @xmath109 , respectively .
the communication demand is then determined by the convergence rate of the consensus algorithm , which depends on the value of @xmath68 , as shown in several previous works , see , e.g. , @xcite , @xcite .
in particular , the more connected is the expected graph ( i.e. , larger values of @xmath68 ) the faster is convergence of consensus , as illustrated in the following numerical examples .
_ numerical example - convergence : _ the aim of this example is to corroborate the theoretical results in theorem 1 , which establishes the almost sure convergence of the stochastic power iteration method in table 1 .
we consider a connected network composed of 20 nodes , whose ideal topology ( in the absence of link failures ) is shown in fig .
the communication among nodes is impaired by random link failures so that each link in fig .
[ net ] is on with a certain probability @xmath114 , which is here assumed to be constant over all links . in fig .
[ fiedler ] , we report the behavior of the estimate of five components of the fiedler vector @xmath73 versus the iteration index , obtained setting @xmath115 . the theoretical value of each component is also reported as a horizontal dashed line .
the sequences @xmath66 $ ] and @xmath56 $ ] are chosen as in ( [ step_size2 ] ) , with @xmath116 , @xmath117 , @xmath118 , @xmath119 , in order to satisfy ( [ step_size ] ) . as we can notice from fig .
[ fiedler ] , the algorithm asymptotically converges to the theoretical value of fiedler vector , thus confirming the theoretical results obtained in theorem 1 .
moreover , in fig . [ lambda2 ] , we report the behavior of the estimate of @xmath68 versus the iteration index , considering different probabilities to establish a communication link .
the theoretical values of @xmath68 are also reported as horizontal dashed lines .
the parameters are the same as in the previous simulation . as we can notice from fig .
[ lambda2 ] , the algorithm converges to the theoretical value of the algebraic connectivity . to validate the almost sure convergence claimed in theorem 1 numerically , in fig .
[ mse_lambda2 ] , we report the behavior of the mean square error ( mse ) on the estimate of @xmath68 , i.e. @xmath120-\lambda_2(\bar{\bl}))^2\}$ ] , considering different probabilities to establish a communication link . the results are averaged over 100 independent realizations .
the parameters are chosen as @xmath116 , @xmath121 , @xmath122 , @xmath123 .
the ideal case , corresponding to @xmath124 , is also reported as a benchmark . from fig .
[ mse_lambda2 ] , we notice how the mse goes to zero as @xmath84 , for any value of the probability @xmath114 to establish a link .
as expected , we can also see how , reducing the probability to establish a communication link , the convergence rate of the algorithm decreases . versus iteration index , for different probabilities to establish a communication link.,width=321 ] versus iteration index.,width=306 ] _ numerical example - adaptation to time - varying scenarios : _ in many practical applications , the connectivity may change over time due to many factors like , e.g. , nodes mobility , nodes failures , variation of channel conditions , etc .
it is then of interest to devise adaptive techniques that are able to track the temporal variation of the expected graph s connectivity .
the stochastic power iteration in ( [ lambdan-1_2])-([pow_iter ] ) converges to the desired eigenvalue almost surely thanks to the effect of the diminishing sequences in ( [ step_size])-([step_size2 ] ) , which asymptotically drives to zero the noise variance . however , this method is not adaptive since the use of a diminishing step - size would make impossible a short - term adaptation to temporal variations of the expected graph connectivity .
it is then of interest to check the tracking capabilities of the algorithm , assuming constant step sizes , i.e. @xmath66=\alpha_0 $ ] in ( [ lambdan-1 ] ) and @xmath56=\bar{\varepsilon}$ ] in ( [ defl_matrix3 ] ) . to assess the adaptation capability of the proposed method to temporal changes in the algebraic connectivity of the network expected graph
, we consider a scenario where the ideal connectivity of the graph varies with time between three different values .
the probability to establish a communication link is kept fixed at @xmath125 , and @xmath126 . in fig .
[ tracking ] we illustrate the behavior of the estimate of @xmath68 versus the iteration index , considering the adaptive stochastic power iteration method with two different constant step - size values @xmath127 .
the theoretical value of @xmath68 is also reported to illustrate the convergence properties . as we can see from fig .
[ tracking ] , the adaptive implementation allows online tracking of the algebraic connectivity of the expected graph . in particular , we can notice from fig .
[ tracking ] how a larger step - size leads to a better adaptation capability , at the cost of larger variance of the estimation error .
.,width=313 ] _ numerical example - performance of the distributed implementation : _ the distributed implementation of the stochastic power iteration method in table 2 relies on two nested consensus loops per iteration .
each of them runs until it reaches a certain precision determined by the parameters @xmath107 and @xmath109 .
since this premature stop inevitably introduces an approximation error in the evolution of the stochastic power iteration , it is of interest to evaluate the effect of such an error on the performance of the algorithm .
then , in fig
. [ mse_delta ] , we illustrate the behavior of the mse on the estimate of @xmath68 versus the iteration index , achieved by the distributed stochastic power iteration algorithm in table 2 , considering different values of the approximation parameter @xmath109 . the results are averaged over 100 independent simulations .
the network topology is achieved from the one in fig .
[ net ] by increasing the algebraic connectivity of the ideal graph ( with no failures ) to @xmath128 .
the link failure probability is set to @xmath129 , and the approximation parameter @xmath130 .
the sequences @xmath66 $ ] and @xmath56 $ ] are chosen as in ( [ step_size2 ] ) , where @xmath131 , @xmath132 , @xmath133 , @xmath119 . from fig .
[ mse_delta ] , we notice how the mse converges to a finite value due to the presence of a bias introduced by the approximation errors in the estimate of @xmath68 .
as expected , we can see how , using a smaller value of @xmath109 ( higher precision in the final consensus value ) , the bias is reduced and the algorithm shows better performance in terms of mse .
however , this benefit does not come without a price as , reducing the value of the approximation parameter @xmath109 , the inner consensus loops will require more iterations to converge with the desired accuracy . to give an example of the overall communication burden , in fig .
[ iter_delta ] we report the number of communication rounds , needed by the proposed distributed algorithm to converge , versus the approximation parameter @xmath109 .
we consider different values of the algebraic connectivity of the expected graph , which are obtained by varying the connectivity of the underlying ideal graph while keeping fixed the link failure probability to @xmath129 .
the behaviors are averaged over 200 independent simulations . to obtain these results ,
we have defined a convergence criterion also for the sequence @xmath78 $ ] in ( [ lambda2_est ] ) . in particular ,
we consider a @xmath134-convergence criterion , with @xmath135 . as we can notice from fig .
[ iter_delta ] , reducing the value of @xmath109 , the algorithm requires a larger communication burden to converge within the desired accuracy .
furthermore , increasing the algebraic connectivity of the expected graph , the overall number of communication decreases due to the improved convergence rate of the power iteration and the inner consensus loops . in summary , the choice of the precision parameters @xmath107 , @xmath109 , and @xmath134 introduces a tradeoff between achievable performance at convergence and communication burden of the algorithm . , for different values of @xmath68.,width=302 ]
the proposed algorithm can be applied to estimate the connectivity of all kinds of random networks . assuming the nodes are deployed according to a random geometric graph ( rgg ) @xcite , the stochastic power iteration method can be used to adapt the power transmitted by each node in order to drive the network connectivity toward a desired value .
this can be obtained through a power control step , where each node updates its transmission power as : @xmath136=p_t[k]+\mu\left(\lambda^*-z[k+1]\right)\ ] ] for @xmath137 , where @xmath138 is a positive constant used to enforce a desired connectivity value , @xmath139 is a positive step - size , and @xmath67 $ ] is the estimate of @xmath68 , at time @xmath101 , carried out by the stochastic power iteration method in ( [ lambda2_est ] ) .
intuitively , the proposed controller increases the power transmitted by each node if the current estimated value of @xmath68 is lower than the desired value , whereas it reduces the transmitted power in the opposite case . the power update in ( [ power_update ] ) can be inserted as fifth step of the stochastic power iteration in ( [ lambdan-1_2])-([pow_iter ] ) , thus leading to a dynamic change of the network topology toward a desired connectivity value .
_ numerical example : _ in this example we combine the stochastic power iteration step in ( [ lambdan-1_2])-([pow_iter ] ) with the power control step in ( [ power_update ] ) , thus illustrating the capability of the resulting strategy to control the connectivity of the network random graph . as a starting point
, we consider a network composed of 25 nodes deployed over a geographic area of 1600 @xmath140 according to a certain initial topology having an initial value of algebraic connectivity @xmath141 .
the initial power transmitted by each node is @xmath142=1 $ ] mw , whereas the minimum received power needed to establish a communication link among two nodes is @xmath143 mw .
we assume a free - space path loss as a propagation environment , i.e. @xmath144 . our goal is to use the proposed algorithm to drive the connectivity of the expected graph toward a desired value @xmath145 , considering two different values of probability to establish a link , e.g. , @xmath124 , and @xmath125 . in fig .
[ lambda2pc ] , we report the behavior of the estimate of @xmath68 in ( [ lambda2_est ] ) versus the iteration index . the continuous curve illustrates the case @xmath124 , whereas the dashed curve the case @xmath125 .
the step - sizes are chosen as in ( [ step_size2 ] ) , where @xmath131 , @xmath146 , @xmath147 , @xmath148 , and @xmath149 . as we can notice from fig .
[ lambda2pc ] , the value of the algebraic connectivity of the expected graph converges very close to the desired value @xmath138 for both values of @xmath114 . versus iteration index.,width=302 ] in fig .
[ sum_power ] , we illustrate the temporal behavior of the power transmitted by each node . form fig .
[ sum_power ] , we can see how the power converges to a fixed value , which depends on the probability to establish a communication link .
in particular , reducing the probability to establish a communication link with respect to the ideal case , the algorithm will increase the number of links of the resulting network in order to counteract the effect of failures and reach the target value of connectivity .
then , each node will transmit more power to enlarge its own subset of neighbors .
in a realistic communication scenario , nodes communicate with each other by accessing to a shared channel according to a specified mac protocol .
let us assume that , in the considered wireless ad - hoc scenario , each node has @xmath150 wireless channels that are dedicated to the exchange of data with its own neighbors . to establish a communication , a node randomly selects one of these channels independently of the choices of its neighbors .
let us further assume that the nodes are deployed according to an rgg .
it is well known that asymptotically , as the number of nodes goes to infinity , rgg networks tend to satisfy a regularity condition , i.e. , each node has the same number @xmath151 of neighbors on average @xcite .
the average number @xmath151 of neighbors depends on the covering radius of each node , which is dictated by the transmitted power and the channel conditions .
let us assume a simple free - space propagation model so that the power received by a node is related to the transmitted power as @xmath152 , where @xmath153 is the covered distance .
now , setting a minimum threshold value @xmath154 for the power at the receiver node , the covering radius is just obtained by inverting the previous expression as @xmath155 .
the average number @xmath151 of neighbors is then related to the covering radius and , consequently , to the transmitted power @xmath156 , as @xmath157 where @xmath158 is the spatial density of nodes inside a circle of area @xmath159 . in this
setting , it is clear that the number @xmath150 of channels used to establish a communication must be designed with respect to the average number @xmath151 of neighbors , in order to keep the probability to have a collision among the communications of two nodes sufficiently small . assuming independence among the channel selections of different nodes and exploiting ( [ node_degree ] ) , the probability that a packet is correctly exchanged over the selected channel is given by @xmath160 where @xmath161 . as expected , the probability to establish correctly a communication link in ( [ p_c ] ) gets worse by increasing the transmitted power @xmath162 , because it translates in having more neighbors to communicate with , whereas , for a fixed transmitted power , it of course improves by taking a larger number of channels @xmath150 . versus the power transmitted by each node , for different number @xmath150 of available channels.,width=302 ] in an ideal communication case where no collisions occur
, the increment of the power transmitted by each node leads to a monotonic increment of the network connectivity .
thus , in an ideal case , it is always convenient to increase the power in order to increase the connectivity of the network , until full connectivity is reached .
nevertheless , in a real communication case , the presence of collisions due to the adoption of a random medium access protocol , e.g. the one introduced before , makes the graph describing the network a random graph , where each link is on with a probability given by ( [ p_c ] ) .
it is then of interest to check the effect of collisions on the connectivity of the expected graph , which is actually the effective connectivity of the network .
an example is given in fig .
[ mean_lambda_power ] , where we show the behavior of the algebraic connectivity of the expected graph @xmath68 versus the power transmitted by each node , for different number @xmath150 of available channels .
the simulation considers a network composed of @xmath163 nodes randomly deployed over a geographic area of @xmath164 m@xmath165 . the threshold power value at the receiver node is given by @xmath143 mw .
as we can notice from fig .
[ mean_lambda_power ] , @xmath68 shows approximatively a quasi - concave behavior with respect to the transmitted power @xmath162 .
in fact , at low power values , the algebraic connectivity of the expected graph increases due to an increment of the links among neighbor nodes , whereas , at high power values , the number of neighbors becomes too large and the probability of having a collision increases , thus leading to a reduction of the overall connectivity of the network . from fig .
[ mean_lambda_power ] , as expected , we also notice how , increasing the number of available channels @xmath150 for a fixed transmitted power , the connectivity of the expected graph improves .
the behavior of @xmath68 shows that there is an optimal transmitted power that nodes should use to maximize the connectivity of the expected graph .
an increment of the power with respect to this threshold value would lead to a waste of energy due to the effect of collisions , which becomes the dominant effect that drives to zero the connectivity . in summary , while in an ideal communication scenario nodes would always improve the network connectivity by increasing their transmitted power , considering a realistic random mac , a too large transmission power may degrade the connectivity due to an increase in the collision probability .
_ distributed connectivity maximization : _ in the previous section , we have shown that , for a sufficiently large number of nodes composing the network , the behavior of the algebraic connectivity of the expected graph @xmath68 versus the power @xmath162 transmitted by each node is unimodal , thus leading to the presence of a unique maximum point ( see fig .
[ mean_lambda_power ] ) .
the goal of this section is to find the optimal power value @xmath166 that maximizes the connectivity of the expected graph , without assuming knowledge of the analytical relation between @xmath68 and @xmath162 .
the method is based on a stochastic algorithm that approximates the derivative of the function on the basis of noisy measurements of @xmath68 . for a given power value @xmath162 , using the stochastic power iteration method in ( [ lambdan-1_2])-([lambda2_est ] ) ,
it is possible to get an estimate @xmath167 of the second smallest eigenvalue of the expected laplacian matrix in a totally distributed fashion . in practice
, stopping the iterative method in ( [ lambdan-1_2])-([pow_iter ] ) at a finite number of iterations , induces an inevitable estimation error , so that we can write @xmath168 where @xmath169 is a realization of a zero - mean random variable with bounded variance @xmath170 .
now , exploiting the noisy measurements in ( [ noisy_measurements ] ) , we can use a kiefer - wolfowitz ( kw ) stochastic approximation method @xcite to find the maximum of the function @xmath171 .
the algorithm runs in parallel over each node , which updates its own transmitted power according to the recursive rule : @xmath172=p_t[t]+q[t]\frac{\hat{z}(p_t[t]+c[t])-\hat{z}(p_t[t]-c[t])}{2c[t]}\end{aligned}\ ] ] @xmath173 , where @xmath174 $ ] is chosen at random , and @xmath175 $ ] and @xmath176 $ ] are two positive sequences that satisfy ( [ step_size ] ) and the further conditions @xmath176\rightarrow 0 $ ] , and @xmath177}{c^2[t]}<\infty$ ] . for any @xmath178 , the stochastic power iteration method in ( [ lambdan-1_2])-([lambda2_est ] )
must be run twice in order to get @xmath179+c[t])$ ] and @xmath179-c[t])$ ] .
the procedure in ( [ kiefer_wolfowitz ] ) is then repeated until convergence . versus iteration index , for different number @xmath1 of nodes.,width=302 ] a numerical example
is shown in fig .
[ max_conv ] , where we illustrate the behavior of the algebraic connectivity of the expected graph @xmath68 versus iteration index , for different values of the number of network nodes @xmath1 , obtained by using the kw method in ( [ kiefer_wolfowitz ] ) .
the maximum values of the algebraic connectivity are also reported for comparison purposes .
the simulation considers a network composed of @xmath1 nodes randomly deployed over a geographic area of @xmath164 m@xmath165 . the threshold power value at the receiver node is given by @xmath143 mw .
the number of channels is set to @xmath180 .
the step - size sequences are chosen as @xmath175=1/t$ ] , and @xmath176=1/\sqrt[3]{t}$ ] . as we can notice from fig .
[ max_conv ] , when the number of nodes is sufficiently large , the kw method in ( [ kiefer_wolfowitz ] ) is able to find the maximum of @xmath171 in a few iterations . at the same time , reducing the number of nodes in the network , the behavior of @xmath171 looses its unimodality .
this implies that the algorithm in ( [ kiefer_wolfowitz ] ) can get stuck in some local maximum , thus explaining the gap between the maximum connectivity value and the kw method in fig .
[ max_conv ] , at low number of nodes @xmath1 .
in this paper we have proposed a stochastic power iteration algorithm aimed at estimating the algebraic connectivity of an ad - hoc network in the case the communication among nodes is affected by random link failures .
we have proved that the algorithm converges almost surely to the second smallest eigenvalue of the expected laplacian of the network graph and to its related eigenvector .
several numerical results confirm our theoretical findings .
a distributed implementation of the proposed algorithm based on nested consensus loops is also proposed , and its performance is analyzed through several numerical simulations . the proposed estimation method is then coupled with a power adaptation mechanism to implement a network connectivity control used to drive connectivity to a desired value . finally , the behavior of the connectivity of the expected communication graph over a realistic mac has been investigated .
an interesting result is that , contrarily to what happens in the absence of packet collisions , in a realistic scenario the nodes should not increase too much their transmission power , because this might heavily degrade the network connectivity due to an increase of collisions .
building on such a result , we have proposed a distributed kw stochastic approximation algorithm to find the transmit power that maximizes the connectivity in the presence of collisions .
in this section we report two theorems from stochastic approximation theory @xcite,@xcite , concerning the convergence properties of stochastic recursive procedures .
we start introducing the notation from @xcite , which is instrumental for the definition of the theorems .
let @xmath181\}$ ] be a markov process on @xmath182 . given a nonnegative function @xmath183 , the generating operator @xmath184 is defined as @xmath185 ) |
\bw[k]=\bw \right]-v(k,\bw)\end{aligned}\ ] ] @xmath186 , @xmath187 , provided the conditional expectation exists .
we say that @xmath188 in a domain @xmath189 , if @xmath184 is finite for all @xmath190 . denoting the euclidean distance between two points @xmath191 and @xmath192 by @xmath193 , the @xmath194-neighborhood of a set @xmath195 and its complement
are defined as @xmath196 we recall next two theorems from @xcite that will be useful in the sequel . _ theorem 2 _ : let @xmath197 be a markov process with generating operator @xmath198 .
let there exist a nonnegative function @xmath188 in the domain @xmath186 , @xmath187 , with the following properties : @xmath199(1+v(k,\bw))-\beta[k]\phi(k,\bw ) \label{lv}\end{aligned}\ ] ] where @xmath200>0 , \quad \sum_{k\geq 0 } \beta[k]=\infty \\ & \hspace{-1 cm } g[k]>0 , \quad \sum_{k\geq 0 } g[k]<\infty\end{aligned}\ ] ] then , the markov process @xmath181\}_{k\geq 0}$ ] with arbitrary initial distribution converges a.s .
to @xmath195 as @xmath84 , i.e. @xmath201,s)=0\right)=1,\end{aligned}\ ] ] with @xmath202 denoting the probability of the event @xmath203 .
the proof can be found in @xcite,@xcite . _ theorem 3 _ : let @xmath204\}_{k\geq0}$ ] be a random vector generated by a markov process defined by the difference equation @xmath205=\bz[k]+\alpha[k]\big[\bh(\bz[k])+\boldsymbol{\varphi}(k,\bz[k],\omega)\big]\end{aligned}\ ] ] with initial condition @xmath206=\bz_0 $ ] , where @xmath207 is borel - measurable , @xmath208,\omega)$ ] is a family of zero - mean random vectors in @xmath182 , defined on some probability space @xmath209 , and @xmath210 is a canonical element of @xmath211 . consider the following set of conditions : _ condition c.1 _ : there exists a nonnegative function @xmath212 with bounded second - order partial derivatives and a point @xmath213 satisfying the conditions @xmath214 where @xmath215 denotes the inner product operator . + _ condition c.2 _ : there exist two constants @xmath216 such that @xmath217 _
condition c.3 _ : the step - size sequence @xmath218\}_{k\geq0}$ ] satisfies the persistence conditions in ( [ step_size ] ) .
let the conditions _ c.1_-_c.3 _
hold for the process @xmath204\}_{k\geq0}$ ] given by ( [ stoch_recursion ] ) .
then , @xmath204\}_{k\geq0}$ ] is a markov process and , starting from an arbitrary initial condition @xmath219 , it converges almost surely to @xmath220 as @xmath221
. the proof can be found in @xcite .
the first part of the proof shows that the sequence @xmath72 $ ] in ( [ pow_iter ] ) converges a.s . to the normalized fiedler vector @xmath222 .
the power iteration in ( [ pow_iter ] ) can be cast as : @xmath223=\frac{\prod_{l=0}^k\bb_2[l]\bx_0}{\left\|\prod_{l=0}^k\bb_2[l]\bx_0\right\|}\ ] ] where @xmath224 is the initialization and @xmath57 $ ] is given by ( [ defl_matrix3 ] ) .
let us focus on the numerator of ( [ pow_iter2 ] ) , which can be rewritten as : @xmath225=\prod_{l=0}^k\left(\bw_2[l]-\frac{1}{n}\mathbf{11}^t\right)\bx_0.\end{aligned}\ ] ] we first prove the boundness of the sequence @xmath226 $ ] in ( [ numer ] ) .
exploiting the fact that @xmath55 $ ] is a sequence of right stochastic matrices , we have @xmath227=\left(\bw_2[0]-\frac{1}{n}\mathbf{11}^t\right)\left(\prod_{l=1}^k\bw_2[l]\right)\bx_0.\end{aligned}\ ] ] since @xmath55 $ ] is a sequence of finite , primitive , ( right ) stochastic matrices , the sequence @xmath228 $ ] in ( [ numer2 ] ) has a finite limit as @xmath84 , @xcite,@xcite .
thus , assuming a finite initial value @xmath224 , from ( [ numer2 ] ) we have @xmath229\|\leq c , \quad\quad \forall k,\end{aligned}\ ] ] where @xmath230 is a positive , finite constant .
let us now introduce the notation @xmath231 .
the proof follows by showing that the sequence @xmath226 $ ] in ( [ numer ] ) converges a.s . to @xmath232 , as @xmath84 , thus guaranteeing the a.s .
convergence of the sequence @xmath72 $ ] in ( [ pow_iter2 ] ) to the normalized fiedler vector @xmath233 , thanks to the presence of the normalization in ( [ pow_iter2 ] ) .
the sequence @xmath234 $ ] in ( [ numer ] ) can be equivalently recast as the following recursive rule : @xmath235=\left(\bi-\varepsilon[k]\bar{\bl}-\frac{1}{n}\mathbf{11}^t-\varepsilon[k]\tilde{\bl}[k]\right)\bw[k],\end{aligned}\ ] ] with @xmath236=\bx_0 $ ] .
we will now use theorem 2 to prove that the sequence in ( [ sequence2 ] ) converges to @xmath232 . to avoid overcrowding of formulas , in what follows we use the notation @xmath237 and @xmath238 , @xmath239 , for the eigenparameters of the expected laplacian , whereas the dependency is explicitly written in the case of other matrices . under the assumption of temporal independence of the sequence
@xmath30 $ ] , the sequence @xmath226 $ ] in ( [ sequence2 ] ) is a markov process . let us define the potential function ( independent of @xmath35 ) : @xmath240 where @xmath241 is a positive semi - definite matrix such that its nullspace satisfies @xmath242 .
let us then introduce the set @xmath243 .
the potential function @xmath244 is non - negative . since @xmath245
, we have @xmath246 the second condition in ( [ cond1 ] ) comes from the continuity of the potential @xmath247 .
let us consider the orthogonal decomposition @xmath248 .
then , @xmath249 . by the definitions in ( [ neigh1 ] )
, we have that @xmath250 implies @xmath251 .
hence , we obtain @xmath252 with @xmath253 .
now , consider @xmath254 in ( [ gen_operator ] ) . exploiting ( [ sequence2 ] ) and ( [ potential_v2 ] ) in ( [ gen_operator ] ) , and since @xmath255\}=\mathbf{0}$ ] , we obtain @xmath256 ) | \bw[k]=\bw \right]-v(\bw ) \nonumber\\ & = \bw^t\left(\bi-\varepsilon[k]\bar{\bl}-\frac{1}{n}\mathbf{11}^t\right)\bf\left(\bi-\varepsilon[k]\bar{\bl}-\frac{1}{n}\mathbf{11}^t\right)\bw \nonumber\\ & \quad+ \varepsilon^2[k]\bw^t\bp\bw -\bw^t\bf\bw\end{aligned}\ ] ] with @xmath257\bf\tilde{\bl}[k ] \right\}$ ] .
let us now consider the following relations @xmath258\frac{1}{n}\mathbf{11}^t .
\label{ineq_5}\end{aligned}\ ] ] the relation in ( [ ineq_4 ] ) holds true by the gershgorin theorem @xcite , because the elements of the matrix @xmath30 $ ] ( thus , @xmath259 ) are taken from a finite set . under assumption a.1 , the sequence @xmath56 $ ] is positive and diminishing , and the inequality in ( [ ineq_5 ] ) holds true , for all @xmath35 , if @xmath260\leq1 $ ] .
otherwise , even if @xmath260>1 $ ] , it exists a finite instant @xmath261 such that , for @xmath262 , the inequality in ( [ ineq_5 ] ) is satisfied .
then , exploiting ( [ boundw ] ) and ( [ eq_1])-([ineq_5 ] ) in ( [ gen_operator2 ] ) , we get @xmath263(\bw^t\bf_3\bw+\eta\cdot c)-2\varepsilon[k ] \bw^t\bf_2\bw,\end{aligned}\ ] ] where @xmath264 are positive semidefinite matrices . considering the orthogonal decomposition @xmath248 and since @xmath265
, we have @xmath266 thus , exploiting ( [ ineqf3f ] ) , the inequality in ( [ gen_operator3 ] ) can be recast in the form of ( [ lv ] ) , where @xmath267=\varepsilon[k]$ ] , and @xmath268&=\varepsilon^2[k]\max\left(\frac{\lambda^3_n}{\lambda_2(\bf)},\eta\cdot c\right).\end{aligned}\ ] ] it is easy to see how the conditions of theorem 2 on the sequences @xmath267 $ ] and @xmath269 $ ] are guaranteed by the choice of the sequence @xmath56 $ ] in ( [ epsilonk ] ) made in assumption a.1 . finally ,
since @xmath250 implies @xmath270 , we obtain @xmath271 with @xmath272 .
all the conditions of theorem 2 are then satisfied , thus guaranteing that the sequence @xmath226 $ ] in ( [ sequence2 ] ) converges almost surely to @xmath243 as @xmath84 .
this result , combined with ( [ pow_iter2 ] ) , ensures that @xmath273=\hat{\bu}_{2},\end{aligned}\ ] ] almost surely ( w.p.1 ) , where @xmath274 denotes the normalized fiedler vector of the expected laplacian matrix @xmath37 .
this completes the first part of the proof .
the second part of the proof aims to show the convergence of the sequence of estimates @xmath78 $ ] in ( [ lambda2_est ] ) to the algebraic connectivity @xmath275 . to prove it
, we notice that , since @xmath276=\hat{\bu}_2 $ ] , and further we have @xmath277 , the behavior of ( [ lambdan-1_2 ] ) at time @xmath35 can be written w.l.o.g . as @xmath278=\frac{\bx^t[k]\bb[k]\bx[k]}{\bx^t[k]\bx[k]}= \hat{\bu}_2\bb[k]\hat{\bu}_2+e[k]\ ] ] where @xmath279 $ ] , such that latexmath:[\[\label{eps_properties } is a random error due to the fact that , at time @xmath35 , the power iteration in ( [ pow_iter ] ) has not converged yet . now , since @xmath281=\bar{\bw}-\frac{1}{n}\mathbf{1}\mathbf{1}^t-\bar{\varepsilon}\tilde{\bl}[k]$ ] and @xmath282 , from ( [ conv_y ] ) we have @xmath283=\ ; & \hat{\bu}_2^t\bar{\bw}\hat{\bu}_2-\bar{\varepsilon}\cdot\hat{\bu}_2^t\tilde{\bl}[k]\hat{\bu}_2+e[k ] \nonumber\\ = \ ; & \lambda_{n-1}(\bar{\bw})+e[k]-\bar{\varepsilon}\cdot\hat{\bu}_2^t\tilde{\bl}[k]\hat{\bu}_2\end{aligned}\ ] ] then , letting @xmath284=-\hat{\bu}_2^t\tilde{\bl}[k]\hat{\bu}_2 $ ] , the a.s .
behavior of the recursion in ( [ lambdan-1 ] ) is given by @xmath285=y[k]+\alpha[k]\big(\lambda_{n-1}(\bar{\bw})-y[k]+e[k]-\bar{\varepsilon}\tilde{n}[k]\big)\end{aligned}\ ] ] from ( [ asymptotic_recursion ] ) , exploiting expression ( [ lambda2_est ] ) , it is possible to derive the recursion for the sequence @xmath78 $ ] as : @xmath286= \hspace{.1cm}z[k]+&\alpha[k]\left(\lambda_{2}-z[k]+\tilde{n}[k]+e_{\varepsilon}[k]\right)\end{aligned}\ ] ] where @xmath287=e[k]/\bar{\varepsilon}$ ] .
since @xmath287 $ ] is a random quantity , it can be written w.l.o.g .
as @xmath287=\bar{e}_{\varepsilon}[k]+\tilde{e}_{\varepsilon}[k]$ ] , where @xmath288 $ ] is the mean part and @xmath289 $ ] is a zero - mean random fluctuation .
thus , ( [ asymptotic_recursion2 ] ) can be recast in the notation of theorem 3 , where : @xmath290)&=\lambda_{2}-z[k]+\bar{e}_{\varepsilon}[k ] \\
\varphi(k , z[k],\omega)&=\tilde{n}[k]+\tilde{e}_{\varepsilon}[k ] \label{gamma_function}\end{aligned}\ ] ] the proof follows by showing that the process @xmath291\}_{k\geq0}$ ] , generated by the recursion in ( [ asymptotic_recursion ] ) , satisfies the conditions _
c.1_-_c.3 _ of theorem 3 .
consider the filtration of the @xmath292-algebra generated by the initial point @xmath60 $ ] and the stochastic error sequence @xmath293\}$ ] , for @xmath294 , i.e. , @xmath295,\{\tilde{\bl}[l]\}_{0\leq l < k}\right)$ ] . the random family @xmath296 in ( [ gamma_function ] ) is @xmath297 measurable , zero mean and independent of @xmath298 , thus making the random process @xmath291,\mathcal{f}_k\}_{k\geq0}$ ] a markov process .
we will show now the existence of a potential function @xmath299)$ ] such that the recursion in ( [ asymptotic_recursion ] ) satisfies the conditions _
c.1_-_c.3_. to this end , we define @xmath300)=\left(\lambda_{2}-z[k]\right)^2.\end{aligned}\ ] ] it is easy to see how @xmath299)\in c_2 $ ] and is non - negative .
furthermore , setting @xmath301 , we have @xmath302 , @xmath299)>0 $ ] for @xmath78\ne z^*$ ] , @xmath303|\rightarrow \infty}v(z[k])=\infty$ ] , and @xmath304-z^*|<1/\mu}\left(h(z[k]),\frac{dv(z[k])}{dz}\right ) \label{lyap_cond}\\ & = \sup_{\mu<|z[k]-z^*|<1/\mu}-2(z^*-z[k])^2 - 2\bar{e}_{\varepsilon}[k](z^*-z[k ] ) \nonumber\\ & < -2\mu^2+\frac{2}{\mu}\bar{e}_{\varepsilon}[k ] \nonumber\end{aligned}\ ] ] since @xmath305=\lim_{k\rightarrow\infty}\bar{e}[k]/\bar{\varepsilon}=0 $ ] due to ( [ eps_properties ] ) , it exists a finite time instant @xmath306 such that , for @xmath307 , the lyapunov condition in ( [ lyap_cond ] ) is always satisfied , for any choice of the positive parameter @xmath139 . to check condition _
c.2 _ , we note that @xmath308\right|^2 + 2\mathbb{e}\left|\tilde{e}_{\varepsilon}[k]\right|^2\end{aligned}\ ] ] from ( [ eps_properties ] ) , the error sequence @xmath309|^2=|\tilde{e}[k]|^2/\bar{\varepsilon}^2 $ ] is upper bounded by a finite constant .
furthermore , the matrix @xmath30 $ ] takes values from a finite set , thus implying that its eigenvalues are finite @xcite .
thus , the variance in ( [ gamma ] ) can be upper bounded by a positive constant @xmath310 , i.e. , @xmath311 . then , we have @xmath312(z^*-z[k])+c_2 \nonumber\\ & \hspace{-1cm}=c_2+\bar{e}^2_{\varepsilon}[k ] + \frac{1}{2}(z^*-z[k])^2+\frac{1}{2}(z^*-z[k])^2 \nonumber\\ & \hspace{1cm}+2\bar{e}_{\varepsilon}[k](z^*-z[k ] ) \nonumber\\ & \hspace{-1 cm } \leq k_1(1+v(z[k]))-k_2\left(h(z[k]),\frac{dv(z[k])}{dz}\right)\end{aligned}\ ] ] where @xmath313 and @xmath314 .
this verifies also condition _
c.2 _ of theorem 3 and condition _
c.3 _ is satisfied by the choice of @xmath218\}_{k\geq0}$ ] in ( [ step_size ] ) made in assumption a.1 .
the previous analysis shows that there exists a finite time instant @xmath306 such that , for @xmath307 , all the conditions of theorem 3 are satisfied , thus ensuring the convergence result @xmath80= z^*=\lambda_{2 } \quad \hbox{a.s.}\end{aligned}\ ] ] this concludes the proof of theorem 1 .
let us consider the behavior of @xmath315 $ ] in ( [ numer ] ) : @xmath316=\prod_{l=0}^k\bar{\bb}_2[l]\bx_0=\prod_{l=0}^k\left(\bar{\bw}_2[l]-\frac{1}{n}\mathbf{11}^t\right)\bx_0\ ] ] now , since we have @xmath317-\frac{1}{n}\mathbf{11}^t=\sigma_{n-1}[l]\bu_{n-1}\bu_{n-1}+\sum_{i=1}^{n-2}\sigma_i[l]\bu_i\bu_i^t \nonumber\ ] ] where @xmath318=\lambda_{i}(\bar{\bw}_2[l])=1-\varepsilon[l]\lambda_{n - i+1}$ ] and @xmath319)=\bu_{n - i+1}$ ] is the corresponding eigenvector , the expression in ( [ exp_numer ] ) can be recast as @xmath320=\left(\prod_{l=0}^k\sigma_{n-1}[l]\right)\cdot\bigg(\bu_{2}\bu_{2}^t \nonumber\\ & \quad\quad\quad+\sum_{i=1}^{n-2}\prod_{l=0}^k\left(\frac{\sigma_i[l]}{\sigma_{n-1}[l]}\right)\bu_{n - i+1}\bu_{n - i+1}^t\bigg)\bx_0\end{aligned}\ ] ] from ( [ exp_numer2 ] ) , neglecting the shrinking factor @xmath321\right)$ ] that is canceled by the normalization of the power iteration , the slowest mode in ( [ exp_numer2 ] ) is given by @xmath322=\prod_{l=0}^k\left(\frac{\sigma_{n-2}[l]}{\sigma_{n-1}[l]}\right)=\prod_{l=0}^k\left(\frac{1-\varepsilon[l]\lambda_3}{1-\varepsilon[l]\lambda_2}\right).\end{aligned}\ ] ] since @xmath323 , for @xmath324 and @xmath325 , the slowest mode @xmath82 $ ] is bounded as : @xmath326&\leq\prod_{l=0}^k\exp\big(-(\lambda_3-\lambda_2)\varepsilon[l]\big ) \nonumber\\ & = \exp \left(-(\lambda_3-\lambda_2)\sum_{l=0}^k\varepsilon[l]\right)\end{aligned}\ ] ]
the authors thank the anonymous reviewers for the detailed suggestions and corrections that improved the manuscript .
s. barbarossa , s. sardellitti , and p. di lorenzo , `` distributed detection and estimation in wireless sensor networks , '' e - reference signal processing , r. chellapa and s. theodoridis , eds . ,
elsevier , 2013 .
g. scutari , s. barbarossa , and l. pescosolido , `` distributed decision through self - synchronizing sensor networks in the presence of propagation delays and asymmetric channels , '' _ ieee transactions on signal processing _ , vol .
4 , pp . 16671684 , 2008 .
g. mateos , i. d. schizas , and g. b. giannakis , `` performance analysis of the consensus - based distributed lms algorithm , '' _ eurasip journ . on adv . in signal processing _ , vol .
57 , no . 6 , pp . 23652382 , june 2009 .
p. di lorenzo , s. barbarossa , and ali h. sayed , `` bio - inspired decentralized radio access based on swarming mechanisms over adaptive networks , '' _ ieee transactions on signal processing _
31833197 , 15 june 2013 .
m. franceschelli , a. gasparri , a. giua , and c. seatzu , `` decentralized laplacian eigenvalues estimation for networked multi - agent systems , '' in _ proc .
ieee conf . on decision and control
2009 , pp .
27172722 .
p. yang , r. a. freeman , g. j. gordon , k. m. lynch , s. s. srinivasa , r. sukthankar , `` decentralized estimation and control of graph connectivity in mobile sensor networks , '' _ proc .
american control conference _
, pp . 2678 - 2683 , 1113 june 2008 , evanston , il .
a. simonetto , t. keviczky , r. babuska , `` on distributed maximization of algebraic connectivity in robotic networks , '' _ proc .
of american control conference _ , pp . 21802185 , june - july 2011 , san francisco , california .
e. montijano , j.i .
montijano , c. sagues , `` adaptive consensus and algebraic connectivity estimation in sensor networks with chebyshev polynomials , '' _ ieee conference on decision and control and european control conference _ , pp . 42964301 , orlando , florida , dec .
r. aragues , g. shi , d. v. dimarogonas , c. sagues , and k. h. johansson , `` distributed algebraic connectivity estimation for adaptive event - triggered consensus , '' in _ proc . of the american control conference _ , pp .
3237 , montreal , canada , june 2012 .
a. bertrand and m. moonen , `` topology - aware distributed adaptation of laplacian weights for in - network averaging , '' _ proc . of the european signal processing conference ( eusipco )
_ , marrakech , morocco , sept . 2013 .
l. hagen and a. kahng , `` new spectral methods for ratio cut partitioning and clustering , '' _ ieee trans . on computer - aided design of integrated circuits and systems _ , vol .
9 , pp . 10741085 , sep 1992 .
a. jadbabaie , j. lin , and a. s. morse , `` coordination of groups of mobile autonomous agents using nearest neighbor rules , '' _ ieee transactions on automatic control _ ,
48 , no . 6 , pp .
9881001 , june 2003 .
y. hatano , a. k. das , and m. mesbahi , `` agreement in presence of noise : pseudogradients on random geometric networks , '' in _ proc .
ieee conf . on decision and control
_ , seville , pp . 63826387 , december 2005 . e. oja and j. karhunen , `` on stochastic approximation of the eigenvecotrs and eigenvalues of the expectation of a random matrix , '' _ journal of mathematical analysis and applications _ , vol .
1 , feb . 1985 .
s. kar and j.m.f .
moura , `` distributed consensus algorithms in sensor networks with imperfect communication : link failures and channel noise , '' _ ieee trans . on signal processing _ vol .
5 , pp . 355369 , jan .
s. kar and j.m.f .
moura , `` distributed consensus algorithms in sensor networks : quantized data and random link failures , '' _ ieee transactions on signal processing _ , vol .
13831400 , march 2010 .
p. di lorenzo , s. barbarossa , and ali h. sayed , `` decentralized resource assignment in cognitive networks based on swarming mechanisms over random graphs , '' _ ieee transactions on signal processing _
37553769 , july 2012 .
p. di lorenzo and s. barbarossa , `` decentralized estimation and control of algebraic connectivity of random ad - hoc networks , '' _ proc . of the ieee international conference on acoustics , speech , and signal processing ( icassp ) _ , pp .
44744478 , vancouver , may 2013 .
|
in this paper we propose a distributed algorithm for the estimation and control of the connectivity of ad - hoc networks in the presence of a random topology .
first , given a generic random graph , we introduce a novel stochastic power iteration method that allows each node to estimate and track the algebraic connectivity of the underlying expected graph . using results from stochastic approximation theory ,
we prove that the proposed method converges almost surely ( a.s . ) to the desired value of connectivity even in the presence of imperfect communication scenarios .
the estimation strategy is then used as a basic tool to adapt the power transmitted by each node of a wireless network , in order to maximize the network connectivity in the presence of realistic medium access control ( mac ) protocols or simply to drive the connectivity toward a desired target value .
numerical results corroborate our theoretical findings , thus illustrating the main features of the algorithm and its robustness to fluctuations of the network graph due to the presence of random link failures .
spectral graph theory , random graph , stochastic power iteration , algebraic connectivity , fiedler vector , topology control , stochastic approximation , distributed computation .
|
Subsets and Splits
Data Mining Abstracts
This query searches for and retrieves 2000 documents containing the phrase 'data mining' in the abstract, providing limited but specific document samples.